Subcutaneous Transmitter Development

© 2013-2018, Kevan Hashemi, Open Source Instruments Inc.

Contents

Introduction
Accelerated Aging
Circuit Versions
2013
2014
2015
2016
JAN-16FEB-16MAR-16APR-16
MAY-16JUN-16JUL-16AUG-16
SEP-16OCT-16NOV-16DEC-16
2017
JAN-17FEB-17MAR-17APR-17
MAY-17JUN-17JUL-17AUG-17
SEP-17OCT-17NOV-17DEC-17
2018
JAN-18FEB-18

Introduction

Here we record our work on the development, improvement, and quality assurance of our Subcutaneous Transmitters. We begin with a summary table showing the results of our accelerated aging reliability testes. We present the various circuit board versions we have made in our pursuit of ease and reliability of encapsulation. Our day-to-day record of work starts with the introduction of the A3028 transmitter in September 2013. For earlier records of development, see A3019 (2010-2014), and A3013 (2007-2011) manual pages. For early work on encapsulation see SCT Encapsulation. For radio frequency isolation see Faraday Enclosures. For flexible leads see Flexible Wires. For comparison of electrodes see Electrodes. For work on radio frequency reception see Data Transmission and Reception.

Accelerated Aging

[29-MAR-17] We started accelerated aging tests after discovering that a lot of one hundred A3028AV2 circuits was plagued by cracked capacitors as a result of excessive placement force during automated assembly. In the warmth and humidity of an animal's body, these cracks corroded and caused the capacitors to fail within a few days of implantation. Such corrosion takes place more quickly at higher temperatures, as presented in Hallberg and Peck. We began accelerated aging test in March, 2015. Since then, we have poached roughly one hundred transmitters in hot water until they failed by one means or another.

According to statistical mechanics, the rate at which a chemical reaction proceeds is proportional to eE/kT, where T is absolute temperature, E is the activation energy of the rate-determining step in the chemical reaction, and k = 8.6×10−5 eV is the Boltzmann constant. We expect the mean time to failure of our subcutaneous transmitters to be proportional to eE/kT. In Hallberg and Peck, the authors show that this relationship applies well to measurements of the mean time to failure for temperatures 20-150 °C and relative humidity 20%-100% when they use an activation energy of 0.9 eV. Their measurements came from studies of circuits assembled with tin-lead solder. For modern circuits, which are usually made with silver-tin solder, a better value for the activation energy may be 1.1 eV. The higher the value of E, the greater the acceleration of aging at a higher temperature. We assume an activation energy of 0.9 eV for corrosion in our circuits, so as to give us a conservative estimate of the acceleration caused by poaching at higher temperatures. Our devices operate at the rodent body temperature of 37°C = 310 K. When we perform an accelerated aging test at 60°C = 333 K, we expect an acceleration of at least ×10. At 80°C = 353 K we expect acceleration of at least ×60.

The following table summarizes our accelerated aging tests. All tests in water in a sealed jar. Failures recorded on the day they are first detected. An "artifact" is a severe corruption of the EEG signal. Examples of severe corruption are: gain versus frequency for 100-kΩ source is wrong by more than 3 dB, steps changes of ≥100 μV once every ten seconds, dynamic range compressed to less than 10 mV. A "failure" is a failure to transmit an EEG signal, however corrupted that signal might be by artifacts. Search the notes below with the transmitter serial number to get the details of each transmitter we poached.

Circuit Battery Qty Start First
Artifact
(days)
First
Failure
(days)
Comments
RV1 BR2330 8 09MAR15 21 35 60°C Mostly CC failures, no FL. Thermal shock possible during checks.
AV3 BR1225 9 12MAY15 9 9 60°C Some CC, some UD, some FL21d. Thermal shock possible during checks.
RV1 BR1225 8 23JUN15 11 18 60°C Some CC, some UD, some FL21d. Thermal shock always.
AV4 BR1225 1 15SEP15 19 19 60°C TX1.1: UD19d. Thermal shock always.
AV4 BR2330 1 30SEP15 17 22 60°C E88.5: CC22d. Thermal shock always.
AV4 BR2330 2 09OCT15 14 14 60°C E89.1: CC14d. E89.2: RS14d. Thermal shock always.
AV3 BR2330 1 20OCT15 21 34 60°C E90.14: RS34d, fine after drying.
AV4 BR2330 2 17NOV15 7 17 60°C E93.2: RS7d. E93.9: RS14d, CS17d.
AV3 BR1225 8 25NOV15 12 5 60°C Rotated and three coats of MED10-6607. B91.1: GE13d, UD16d. B91.2: CC16d. B91.3: UD5d. B91.10: RS9d. B91.6: CS9d. B91.9: UD12d. B91.8: GE12d, CC16d. B91.12 FL23d.
AV5 BR1225 2 08DEC15 25 25 60°C C94.3: UD25d. C94.4: UD25d.
AV5 BR2330 2 11DEC15 11 11 60°C E94.12: RS11d. E94.3: RS12d
AV5 BR1225 2 22DEC15 17 17 60°C C96.4: CD17d. C96.5: UD29d.
AV5 BR2330 2 28DEC15 28 31 60°C E97.2: UD37d. E97.3: UD31d.
AV5LF BR2330 1 12JAN16 14 14 60°C E98.13: CC14d.
AV5LF BR2330 2 02FEB16 45 45 60°C E100.13: UD45d, E100.14: UD66d.
AV5LF BR1225 3 09FEB16 9 13 60°C C101.11: FL39d, C101.13: UD22d, B102.12: TM13d.
RV2 BR2330 2 10MAR16 27 32 60°C R105.12: CC32d, R106.10: CC69d.
AV5LF BR2330 2 30MAR16 36 36 60°C E107.9: UD48d, E108.8: CC36d.
AV5LF BR1225 2 06APR16 19 19 60°C B108.14: UD19d, B109.3: FL26d.
RV3 BR1225 2 29APR16 14 19 60°C C110.13: UD25d, C110.14: UD19d.
RV3 BR2330 1 17MAY16 7 14 60°C R112.11: RS14d. Hand-made, no epoxy top-coat, 5 coats MED10-6607.
RV3 BR2330 2 24MAY16 40 125 60°C E113.9 (24th): CC125d, E113.10 (27th) FL125d. Rotator, 3 coats MED10-6607.
RV3 BR1225 2 31MAY16 27 27 60°C B113.13: FL27d, B114.4: FL27d. Rotator, 4 coats MED10-6607.
RV3 BR2330 2 09JUN16 31 61 60°C R114.10: CC78d, R114.11: CC61d. Rotator, 5 coats MED10-6607.
RV3 BR2330 2 14JUN16 55 104 60°C E115.12: UD107d, E116.4: RS104d. Rotator, 3 coats MED10-6607.
RV3 BR1225 2 09AUG16 27 27 60°C B121.13: FL27d, B122.11: FL27d. Rotator, 5 coats MED10-6607.
RV3 BR2330 4 05SEP16 59 60 60°C until 27OCT16, then 80°C. R124.9 CC63d, E126.8 TS60d, E127.1 CC65d, E127.11 UD71d. Rotator, 5 coats silicone.
RV3 BR2330 1 16SEP16 56 56 60°C until 27OCT, then 80°C. R129.2 TS56d. Rotator, 3 coats SS-5001 silicone.
RV3 BR2330 2 30SEP16 12 12 60°C until 27OCT, then 80°C. R129.5 UD53d, R129.7 FE12d. Rotator, 5 coats MED10-6607, wrinkles and noise.
RV3 BR1225 2 30SEP16 28 28 60°C B129.14 FL28d, B130.9 FL28d. 4.5 coats MED10-6607, wrinkles.
RV3 BR1225 7 27OCT16 11 11 80°C A135.1 TS13d, A135.3 CC13d, A133.5 FL15d, A135.7 UD13d, A135.9 RS11d, A134.11 CC13d, A134.13 UD13d, 1 coat SS-5001.
RV3 BR1225 2 16DEC16 22 91 60°C/80°C B146.8 CC91d, B146.9 CC91d. Test of shelf life: turned off. 1 coat SS-5001.
RV3 CR2450 2 17FEB17 11 31 80°C Q154.56 CS31d, Q154.73 UD31d. 1 coat SS-5001.
RV3 BR2330 2 17FEB17 11 31 80°C E144.10 CC31d, E150.6 CS31d. 1 coat SS-5001.
RV3 BR2330 2 17MAR17 15 27 80°C E153.7 UD35d, 1 coat SS-5001. E154.1 UD27d, 2 coats SS-6001.
RV3 BR2330 2 13APR17 19 36 80°C R129.6 CC39d, 5 coats MED10-6607. E153.14 CC36d, 2 coats SS-6001.
RV3 BR2330 4 19MAY17 14 14 80°C E146.2 CC33d, E154.3 UD26d, E155.21 CC26d, E155.22 UD14d.
1 coat SS-5001 + 1 coat MED10-6607 after 2 months.
RV3 BR2330 2 23JUN17 18 39 80°C E201.57 CC39d, E201.61. UD39d. 1 coat SS-5001 + 1 coat MED10-6607 after 2 hrs.
RV3 BR1225 1 04AUG17 14 14 80°C C201.102 UD14d. 1 coat SS-5001 + 1 coat MED-6607 after 2 hrs.
RV3 BR2330 1 04AUG17 21 35 80°C E201.114 CC35d. 1 coat SS-5001 + 1 coat MED-6607 after 2 hrs.
RV3 LiPo 2 08SEP17 10 10 80°C No1 FL10d, No3 FL10d. 19 mA-hr Battery. 1 coat SS-5001 + 1 coat MED-6607 after 2 hrs.
RV3 LiPo 1 08SEP17 4 4 80°C No5 OG4d. 190 mA-hr Battery. 1 coat SS-5001 + 1 coat MED-6607 after 2 hrs.
RV3 LiPo 1 03OCT17 66 70 60°C ER.8 UD70d. 190 mA-hr Battery. 3 coats MED-6607.
RV3 BR1225 2 03NOV17 14 14 60°C U201.161 FL14d, U201.163 FL14d. 3 MED-6607.
RV3 BR1225 2 14NOV17 15 15 60°C J204.45 FL15d, J204.49 FL15d. 3 MED-6607.
RV3 BR2330 2 05DEC17 31 41 60°C G201.183 FL43d, G201.185 FL41d. 3 MED-6607.
GV1 BR1225 10 16JAN18 3 6 80°C, J3 CC6d, J13 UD6d, J5/J21 UD13d, J1/J7/J9/J11/J17/J19 FL14d. 1 SS-5001.
Table: Summary of Accelerated Failure Tests. The circuit problems are identified by two-letter codes. RS is "resistive switch", CC is "corroded capacitor", CS is "corrosion short", CD is "cavity drain", UD is "unidentified drain", GE is "gain error", TM is "transmit malfunction", TS is "Temporary Shutdown", "Out Gassing" and FE is "Faulty Encapsulation". We also say FL for "full life". Time until we discovered the failure is after the failure code. The actual failure time will be from one to five days before the discovery, depending upon how frequently we were checking the test.

In the first tests, we did not take note of how we cooled the devices when we removed them from the oven to check them. In some tests we deliberately put them straight into cold tapwater to cool them rapidly. When three such transmitters failed within minutes of cooling, we suspected that thermal shock was a factor in the failures. For several months, we removed the devices from their poaching water and let them cool in air before testing. The resistive switch problem persisted, however, which suggests it is caused by condensation, not contraction. The resistive switch problem turned out to be due to dendrites growing between the pins of U1. We no longer take any particular precautions heating and cooling transmitters to and from 60°C, but we monitor their behavior during and after such changes, so as to catch failures that occur during expansion or condensation.

[15-NOV-16] Seven dual-channel A3028A mouse transmitters provided full performance in 80°C water for 11 days, during which corrosion equivalent to at least 660 days at 37°C took place. Four single-channel A3028E rat transmitters survived corrosion equivalent to at least 940 days at 37°C over the course of ten weeks at 60°C and 80°C. We conclude that that our transmitters will survive corrosion within an animal body for two years. This corrosion lifetime is separate from the operating life of the battery, which is set by the battery capacity and the transmitter's current consumption.

Circuit Versions

Aside from the assembly versions, we have versions of the circuit board that we use to build the A3028. The same circuit version can be used to make multiple assembly versions. The same assembly version can be constructed using multiple circuit versions. Here we list the more recent versions, the name of the printed circuit board they use, the type of capacitor, and what we know about the component assembly process.

Version PCB Large Capacitor Solder Wash Details
RV1 A302801D 10μF 10V X5R Samsung lead-free water X and Y 160 Hz, protected inputs, square pads, U3-1 rounded, lead-free BGA, solder paste stencil.
AV3 A302801E 10μF 10V X5R Samsung lead-free water X and Y 160 Hz, round pads, U3-1 square, lead-free BGA, fits BR1225, solder paste stencil
AV4 A302801E 2.2μF 10V X5R Taiyo-Yuden lead water X and Y 160 Hz, round pads, U3-1 square, lead-free BGA, fits BR1225, solder paste stencil
AV5 A302801E 10μF 6.3V X5R TDK lead chemical X 160 Hz, Y 80 Hz, round pads, U3-1 square, lead-free BGA, fits BR1225, solder paste printer
AV5LF A302801E 10μF 6.3V X5R TDK lead-free chemical X 160 Hz, Y 80 Hz, round pads, U3-1 square, lead-free BGA, fits BR1225, paste printer
RV2 A302801F 10μF 6.3V X5R TDK lead-free chemical X and Y 160 Hz, input protection, octagonal pads, U3-1 rounded, lead-free BGA, fits BR1225 or BR2330, solder paste printer, mounting wire pad
RV3 A302801F 10μF 6.3V X5R TDK lead-free chemical X 160 Hz, Y 80 Hz, input protection, octagonal pads, U3-1 rounded, lead-free BGA, fits BR1225 or BR2330, solder paste printer, mounting wire pad
RV4 A302801F 10μF 6.3V X5R TDK lead-free chemical X 80 Hz, Y 80 Hz, otherwise as RV3
RV5 A302801F 10μF 6.3V X5R TDK lead-free chemical X 160 Hz, Y 160 Hz, otherwise as RV3
RV6 A302801F 10μF 6.3V X5R TDK lead-free chemical X 320 Hz, Y 320 Hz, otherwise as RV3
RV7 A302801F 10μF 6.3V X5R TDK lead-free chemical X 640 Hz, Y 640 Hz, otherwise as RV3
GV1 A302801G 4×10μF 6.3V X5R TDK lead-free chemical X 160 Hz, Y 80 Hz, eliminate mosfet power switch, two anchor holes for mounting wire pad, dramatically reduced switching noise as a result of increased decoupling capacitors.
Table: Summary of Circuit Board Versions. Links are to bills of materials.

The GV1 comes ready-made from the assembly house for single-channel 160-Hz and single-channel 80-Hz transmitters, both 1.4-ml and 2.8-ml sizes. The GV1 eliminates the mosfet power switch present on earlier versions, which turned out to be redundant, and which was the source of the "resistive switch" failure in accelerated aging. The GV1 provides a total of 40 μF of decoupling on the battery voltage, which reduces switching noise down to below 1 μV rms in encapsulated transmitters. The GV1 offers input protection, which has no significant effect upon the recording of EEG, but which guarantees that the circuit will not be endangered by stimulation voltages applied to the brain near the EEG electrodes.

2013

See Development Archive.

2014

See Development Archive.

2015

See Development Archive.

2016

JAN-16

[02-JAN-16] E94.3 no longer transmitting, even after cool-down. Dissect. Battery voltage 0.6 V. With external 2.6-V supply, inactive current 1.6 μA, active 11 mA, transmitter turns on, perfect reception, VA = 2.5 V. Replace C4, no change. Reduce external supply to 2.2 V, active current 9 mA. Replace C3. Current 9 mA. No other capacitor failure can give these symptoms. Classify as Unidentified Drain.

E94.4 transmits briefly then stops. Dissect. Battery voltage 1.7 V. Apply external 2.6 V. Inactive current 1.7 μA, active 50.0 μA. Reception perfect. VA = 2.5 V. At 50 μA, this transmitter should have run for 40 days. But here we see it expiring after 25 days. Classify as Unidentified Drain. This is our longest-lived transmitter equipped with a BR1225 battery so far.

E97.2, E97.3, C96.4, C96.5 running well. Reception robust. Battery voltage and noise are E97.2 2.80 V and 6.5 μV, E97.3 2.83 V and 6.9 μV, C96.4 2.75 V and 5.9 μV, C96.5 2.77 V and 5.9 μV. Put them back in the oven.

[05-JAN-16] E97.2, E97.3, C96.4, C96.5 running well. Reception robust. Battery voltage and noise are E97.2 2.82 V and 6.3 μV, E97.3 2.84 V and 6.5 μV, C96.4 2.81 V and 4.8 μV, C96.5 2.77 V and 4.9 μV. Put them back in the oven.

[06-JAN-16] All four poaching devices are transmitting, but we perform no other tests.

[08-JAN-16] C96.4 no longer transmitting. E97.2, E97.3, and C96.5 perfect reception. E97.2 2.78 V, 8.6 μV. E97.3 2.81 V, 8.9 μV. C96.5 2.73 V, 9.1 μV. We dissect C96.4. After removing silicone, we break through a thin layer of epoxy and enamel, to find this cavity with metal visible at the bottom. The cavity is above C10, C11, U5-4.


Figure: Cavity in Epoxy of C96.4.

Battery voltage 0.1 V. Apply external 2.6 V. Inactive current 1.7 μA, active 2.4 mA. We record full-range mains hum. Remove R4, active 52 μA. The X signal is stuck at 0. Because R4 supplies VA and is 1 kΩ, 2.4 mA would be consistent with VA = 0.2 V, which precludes detection of mains hum. When we remove R4, current returns to normal, and U7 is disabled. Measure R4 as 1.000 kΩ. Replace C2, C5, and R4. Active current 2.5 mA. VA = 0.2 V. Clean and dry. VA = 0.2 V, 1V8 = 1.8 V. Clear epoxy from around these parts. Transmitter now turns on, records mains hum, current 55 μA, VA = 2.6 V. It looks like a short developed beneath the cavity. We removed the short when we cleared the epoxy. The one confounding observation is mains hum when we were apparently consuming 2.4 mA, but our 2.4 mA and mains hum observations were consecutive, not simultaneous. We invent a new classification for this problem: Cavity Drain.

We take B84.9 and B86.1 off the shelf and measure their frequency response. We plan to ship these to Newcastle as additional samples.

[11-JAN-16] Transmitter E97.2 perfect reception, VA = 2.83 V, noise 6.6 μV. E97.3 robust reception, VA = 2.85 V, noise 6.9 μV. C96.5 perfect reception, VA = 2.75 V, noise 5.3 μV.

[12-JAN-16] Transmitter E97.2 robust reception, VA = 2.81 V, noise 7.3 μV. E97.3 perfect reception, VA = 2.84 V, noise 6.9 μV. C96.5 robust reception, VA = 2.72 V, noise 8.6 μV.

We have batch E98.3-E99.4, a batch of sixteen A3028E. All turn on and off except E98.14. Dissect. Battery 1.4 V. Apply 2.6 V. Inactive current 1.6 μA, active 84 μA. Reception perfect. Picks up mains hum. The remaining 15 have perfect reception, switching noise less than 4 μV. All have gain versus frequency within range ±0.4 dB, E98_G_vs_F. We ship 14 transmitters and keep E98.13, which we put in the oven to poach.

In making batch E99.5-100.2 we notice after vacuum-cycling the top coat that there are bubbles 0.5 mm or so in diameter pressing up against the surface tension of the epoxy. We burst these with a metal point, and find in many cases that the bubble underneath is of diameter far larger than the visible circle on the epoxy, so that when we pop the bubble we see 1 mm down into the epoxy, sometimes glimpsing the components. In the past three months, we have been using a new vacuum chamber that provides a better seal and lower vacuum. After bursting the bubbles, more appear, and we burst these too. After the fourth run through the batch, there are no more large bubbles left.

[14-JAN-16] For E97.2, E97.3, C96.5, and E98.13 we get the following from a one-second interval. No2: 512 samples, VA = 2.81 V, noise = 6.7 μV; No3: 512 samples, VA = 2.84 V, noise = 7.0 μV; No5: 256 samples, VA = 2.71 V, noise = 8.8 μV; No13: 510 samples, VA = 2.80 V, noise = 7.7 μV.

We have batch C99.5-100.2, 12 of A3028C-AA. These are the first transmitters we have made with the BR1225 battery with corners cut off, as shown below.


Figure: BR1225 Battery, As-Purchased (left) and Trimmed (right).

We stuff them all into a 50-ml beaker and cover them with water, antennas and all. The water comes up to the 40-ml mark. We remove the transmitters. The water drops to 24 ml. The volume of each device is 1.3 ml. All of them turn on and off and give robust reception. We put them in the oven to burn in.

[15-JAN-16] Batch C99.5-100.2 have survived burn-in. Battery voltages 2.67-2.75 V. Noise less than 11 μV. Reception robust. We turn them all off and put them in room temperature water. For E97.2, E97.3, C96.5, and E98.13 we get the following from a one-second interval: No2: 512 samples, VA = 2.82 V, noise = 6.6 μV; No3: 512 samples, VA = 2.85 V, noise = 7.2 μV; No5: 256 samples, VA = 2.70 V, noise = 15.4 μV; No13: 512 samples, VA = 2.81 V, noise = 7.8 μV;.

[16-JAN-16] Batch C99.5-100.2 have soaked for 24 hours. We measure gain versus frequency, C99_G_vs_F. C99.6 gain is higher than usual and its battery voltage is lower, which are correlated symptoms. We apply 5-mV sweep and scale the response to account for lower battery life and amplitude, and we get good agreement with the other transmitters. Simultaneous reception from all twelve gives us: No1: 247 samples, VA = 2.59 V, noise = 9.5 μV; No2: 256 samples, VA = 2.59 V, noise = 7.4 μV; No5: 251 samples, VA = 2.58 V, noise = 6.2 μV; No6: 255 samples, VA = 2.48 V, noise = 6.5 μV; No7: 256 samples, VA = 2.54 V, noise = 7.5 μV; No8: 256 samples, VA = 2.55 V, noise = 6.4 μV; No9: 256 samples, VA = 2.55 V, noise = 7.1 μV; No10: 256 samples, VA = 2.51 V, noise = 8.4 μV; No11: 255 samples, VA = 2.51 V, noise = 9.4 μV; No12: 253 samples, VA = 2.51 V, noise = 8.9 μV; No13: 257 samples, VA = 2.55 V, noise = 7.4 μV; No14: 249 samples, VA = 2.56 V, noise = 8.5 μV.

For E97.2, E97.3, C96.5, and E98.13 we get the following in one second. No2: 509 samples, VA = 2.83 V, noise = 6.4 μV; No3: 135 samples, VA = 2.86 V, noise = 6.9 μV; No5: 207 samples, VA = 2.30 V, noise = 11.7 μV; No13: 511 samples, VA = 2.84 V, noise = 7.6 μV. C96.5's battery appears to be failing. Given low battery voltage, gain versus frequency still looks perfect, and reception is perfect.

[20-JAN-16] E97.2, E97.3, and C98.13 still running. C96.5 stopped.

[21-JAN-16] E97.2, E97.3, and E98.13 still running.

[26-JAN-16] E97.2, E97.3 still running. No2: 508 samples, VA = 2.53 V, noise = 7.1 μV; No3: 501 samples, VA = 2.85 V, noise = 7.0 μV. Frequency response of E97.3 looks good. But by the time we come to measure the response of E97.2, we have VA = 1.9 V, suggesting a resistive switch problem, which tends to manifest itself when the device cools down. E98.13 has stopped. Dissect. Note that corner of C2 at edge of epoxy appears to be exposed after removal of silicone and enamel. Battery voltage 0.0 V. Disconnect battery, voltage is 1.9 V and rising, 2.3 V after 50 s. 2.7 V after 10 minutes. Connect external 2.6 V and see 5 mA, increasing to 6 mA. Reception 100%. Current 15 mA after 2 minutes. Turn off and current remains 13 mA. Remove epoxy from C2 and C5. Active current 5 mA. Remove C2. Inactive current 2.7 μA, active current 3 mA. Examining C2, we see a flake of ceramic coming off the top side. It falls off. The capacitor is thinner. Resistance is 0.8 Ω. The fact that the battery recovers so quickly from being drained suggests that the failure happened within the past 24 hrs. Remove C5. Resistance is 50 Ω, capacitance 2 nF. Replace C5. Active current 82 μA. Reconnect battery and transmitter runs well at first, but battery voltage drops to 2.2 V after a few minutes, and we are unable to measure frequency response with 10-mV sinusoid. We attach fresh battery and measure gain versus frequency. Gain is within 2 dB of nominal for the 3.0 V battery. Diagnosis: corroded capacitor failure.

Dissect C96.5. Battery voltage 0.1 V. Disconnect battery, its voltage is 0.2 V. Connect external 2.6 V. Inactive 2.6 μA. Active 53 μA. This transmitter ran for a total of 30 days (1 day burn-in, 29-day poach). Charge consumption should have been 36 mA-hr. Battery capacity is 48 mA-hr. Diagnosis is Unidentified Drain.

We have a new protected-input layout, A302801F. We reduce the length of the board from 565 mil to 550 mil. The width remains 500 mil. The height and width of the A302801D layout, which we have been using for the past several hundred transmitters, are both 500 mil. This new board will be 1.3 mm longer than the one we have been using. The BR1225 mounting holes remain as they were. We remove the holes for BR2330 battery. The 0V tab now has a 50-mil pad on the bottom side only, anchored with a 20-mil via. The VB tab solders to the rim of the BR1225 VB hole. We expand the rim where the BR2330 tab will fold over the circuit board. We move C2, C5, and C4 so they are at least 50 mil from the board edge. Previously C2, C4, and C5 were 25 mil from the edge. All 10-μF capacitors are now 50 mil or farther from the edge. These parts are tall, and had been pressing up against the meniscus of our epoxy. We make octagonal pads for A, C, X, and Y on the top side of the board only, where we solder the leads. Each pad is held in place by a 20-mil via. The identity of each pad is marked on the bottom-side silk screen. We add an isolated pad at the center of the top edge on the top side of the board for a holder wire to be attached. The pad is anchored with a 20-mil via. We reduce the 30-mil vias to 20 mil because we do not plan to use holes for evacuating air. We will be forcing epoxy into the circuit while the circuit is in vacuum. Our A302801D mouse transmitters now have volume 1.3 ml, with the trimmed battery, rotator encapsulation, and top-coat of epoxy. The A302801F will be 1.3 mm longer, producing a ledge beneath the battery 12.5 mm wide and 1.3 mm deep. When covered with 2 mm of epoxy top and bottom, the increase in volume will be only 0.01 ml. We have moved the capacitors in from the edges, so we should now be able to eliminate the top coat, which will save us a 0.5 mm depth on a 12.5 mm square, saving us 0.08 ml.

[28-JAN-16] Yesterday, E97.2 and E97.3 were going strong. We left them out of the oven. Today E97.3 has stopped, but E97.2 is still going strong.

[29-JAN-16] E97.2 still running: 511 samples, VA = 2.40 V, noise = 8.2 μV. We dissect E97.3. VB = 0.9 V. Disconnect, VB = 2.5 V after a few minuts. Connect external 2.6 V. Current is at first 2 mA with the microammeter, switch to millammeter and current drops to 0.08 mA. Switch off. Inactive current 1.6 μA. Switch on, active current 80 μA. Reception perfect, picks up mains hum. This device failed after cool-down to room temperature. The battery recovers to 2.55 V. After 30 days at 80 μA, the battery voltage should still be 2.7 V (58 mA-hr used of 255 mA-hr capacity). Diagnosis is Unidentified Drain.

We assemble our first A3028H, a two-channel 256 SPS, 0-80 Hz EEG/EMG monitor. We replace C9, C10, and C11 on an A3028AV5LF with 2.0 nF. We program one and measure frequency response, see here.

FEB-16

[01-FEB-16] E97.2 still running: 512 samples, VA = 2.29 V, noise = 18.7 μV.

[02-FEB-16] We have batch E100.13-14. Gain versus frequency is E100_G_vs_F. Switching noise less than 5 μV, 20-22 Hz. From a one-second interval, all transmitters together in water we get robust reception and noise less than 10 μV. E97.2 still running: 512 samples, VA = 2.08 V, noise = 9.6 μV;. We put E100.13 and E100.14 in to poach.

[03-FEB-16] E100.13 and E100.14 running. For 1-s interval No13: 512 samples, VA = 2.86 V, noise = 6.2 μV; No14: 512 samples, VA = 2.85 V, noise = 6.1 μV. E97.2 has stopped. We cut open the silicone and it starts running again. For 1-s interval on bench-top: 512 samples, VA = 1.93 V, noise = 15.5 μV. Turns off again. VB = 1.8 V. Disconnect battery, VB recovers to 2.0 V in one minute. Connect external 2.60 V. Inactive current 2.1 μA. Active 84 μA. For 1-s interval 512 samples, VA = 3.05 V, noise = 12.8 μV. No mains hum pick-up. Meanwhile, battery has recovered to 2.35 V. Diagnosis is Unidentified Drain.

[04-FEB-16] For E100.13 and E100.14 we get in 1-s interval in faraday enclosure. No13: 481 samples, VA = 2.84 V, noise = 8.6 μV; No14: 483 samples, VA = 2.82 V, noise = 6.8 μV. We measure their RF spectrum and find its peak at 909 MHz and range 900-918 MHz.

[06-FEB-16] We have a wire stripper capable of displacing a 1-mm section of teflon insulation around a 125-μm Pt-Ir wire. We prepare a new version of the H-Electrode that we first developed last year. We will prepare twelve of these for ION-UCL.


Figure: Electrode H. The Pt-In wire is on the left, EEG pick-up lead enters from the bottom, and a steel guide cannula provides the mounting fixture.

Construction of this version is simpler because we have only two wires to join.

E100.13 and E100.14 running well. No13: 504 samples, VA = 2.86 V, noise = 8.3 μV; No14: 375 samples, VA = 2.85 V, noise = 6.4 μV.

[08-FEB-16] E100.13 and E100.14 running well. No13: 512 samples, VA = 2.86 V, noise = 6.6 μV; No14: 512 samples, VA = 2.85 V, noise = 6.4 μV. Batch B102.4-13 has been soaking in water for 3 days. We measure frequency response with a 5-mV peak to peak waveform and get B102_5mV. In water, VA = 2.51-2.55 V, noise less than 12 μV, reception robust.

[09-FEB-16] E100.13 and E100.14 running well. No13: 512 samples, VA = 2.84 V, noise = 6.5 μV; No14: 512 samples, VA = 2.82 V, noise = 7.0 μV. We have batch H103.1-9 (odd numbers only), six A3028H-AAA dual-channel 0.3-80 Hz transmitters. They have survived 72hr/20C/W/OFF and 6hr/60C/D/ON. Frequency response of all channels recorded in H103_5mV, all good. We have batch C101.1-102.3, seventeen A3028C-AA single-channel 0.3-80 Hz transmitters. They have survived 24hr/60C/ON/D, 72hr/20C/OFF/W, 6hr/60C/ON/D. Frequency response of all devices recorde in C101_5mV_A and C101_5mV_B, all good except C101.11, which shows low gain. During manufacture, we observed cavities beneath the epoxy in devices C101.6, C101.11, and C101.13. We keep as spares H103.1 and B101.13. We put C101.11 (low gain and epoxy cavity), C101.13 (epoxy cavity), and B102.12 in the oven to poach.

[10-FEB-16] C101.11, C101.13, B102.12 all running: No11: 255 samples, VA = 2.76 V, noise = 4.1 μV; No12: 512 samples, VA = 2.74 V, noise = 7.8 μV; No13: 250 samples, VA = 2.76 V, noise = 5.4 μV. E100.13 and E100.14 running well: No13: 512 samples, VA = 2.87 V, noise = 12.9 μV; No14: 510 samples, VA = 2.85 V, noise = 9.7 μV.

[11-FEB-16] C101.11, C101.13, B102.12 running well. Reception 100%. VA = 2.70-2.74 V, noise less than 11 μV. E100.13 and E100.14 running well. Reception 100%. VA = 2.84-2.86 V, noise less than 7 μV.

[18-FEB-16] E100.13 and E100.14 running well. Reception 100%, VA = 2.84-2.86 V, noise less than 7 μV. C101.11 and C101.13 running well. Reception 100%, VA = 2.70-2.70 V, noise less than 7 μV. B102.12 shows 80% reception until we remove it from the hot water, then we get 100% and VA = 2.31 V. We measure gain versus frequency, and it is uniformly 0.8 dB higher than on 08-FEB-16.

[19-FEB-16] E100.13 and E100.14 running well. Reception 100%, VA = 2.86-2.88 V, noise less than 7 μV. C101.11 and C101.13 running well. Reception 100%, VA = 2.76-2.77 V, noise less than 6 μV. B102.12 100% reception, noise 14 μV, VA = 2.26 V. If VA = VB, then yesterday VB was 2.3 V, and today should be drained below 2.0 V by 80-μA consumption of the A3028B. The fact that VA dropped only 0.05 V suggests the resistive switch problem, whereby VA < VB.

[22-FEB=16] E100.13 and E100.14 running well. Reception 100%, VA = 2.87-2.87 V, noise less than 7 μV. C101.11 and C101.13 running well. Reception 100%, VA = 2.74-2.74 V, noise less than 7 μV. No reception from B102.12. Dissect. Battery voltage 2.7 V. Connect external 2.6 V. Inactive current 2.1 μA, active current 57 μA, no reception. VA = 2.1±0.1 V on C5. Spectrometer (A3008) detects no RF power output. Clear epoxy away from C3 and U9. Reception re-starts. Spectrometer sees RF peak at 914 MHz. From X input, VA = 2.39 V, reception 100%, picking up mains hum. Active current 82 μA. Inspect U9, joints look good. Measure VA = 2.4±0.1 V at C5. We remove C3, the 1-nF decoupling capacitor for U9, and obtain poor reception. Restore C3, reception in faraday enclosure is perfect. Diagnosis: Transmit Malfunction.

[23-FEB-16] C101.11 and C101.13 running well. Reception 100%, VA = 2.76-2.67 V, noise less than 7 μV. E100.13 and E100.14 running well. Reception 100%, VA = 2.87-2.88 V, noise less than 7 μV.

[24-FEB-16] We have 20 prototypes of our A3028RV2 circuit, built on the A302801F printed circuit board. This board has a blue glossy solder mask. This is an input-protected circuit, like the A3028R, but with the large capacitors moved farther from the edges of the board, the battery tabs removed, and pads for leads on only the top side. The assembled A3028RV2 is 12.7 mm × 14.0 mm. The A3028R was 13.1 mm × 14.5 mm after removing its battery tabs. The A302801F circuit board provides a dedicated pad for a mounting wire to allow the A3028RV2 to be connected to our encapsulation immersion tool and rotator tool.

During QA we notice flux residue around antenna and lead joints on our new A3028RV2 assemblies. We wash one circuit in hot water and residue is gone. Meanwhile, we replaced U9 on one of these circuits, and two days later inactive current is 10 μA, which suggests water under U3, which is where the dendrites of the resistive switch problem have been forming. We form the following hypothesis: double-washing of the top side of the board is necessary to remove all flux residue before battery loading, and extended baking with current consumption check is required after all washes.

[25-FEB-16] E100.13 and E100.14 running well. Reception 100%, VA = 2.88-2.88 V, noise 6.0-6.1 μV. C101.11 and C101.13 running well. Reception 100%, VA = 2.75-2.77 V, noise 6.0-6.2 μV.

[26-FEB-16] E100.13 and E100.14 running well. Reception 100%, VA = 2.88-2.89 V, noise less than 7 μV. C101.11 and C101.13 running well. Reception 100%, VA = 2.75-2.75 V, noise in C101.11 is 11 μV, and in C101.13 is 6 μV.

[29-FEB-16] C101.11 and C101.13 running well. Reception 100%, VA = 2.75-2.75 V, noise in C101.11 is 20 μV, and in C101.13 is 6 μV. We plot the spectrum of the noise from both transmitters during an eight-second interval here. The first harmonic of the switching noise has amplitude 12 μV. We note that C101.11 failed QC because of irregular gain, see here. E100.13 and E100.14 running well. Reception 100%, VA = 2.85-2.87 V, noise less than 7 μV.

We have batch B104, consisting of B103.13-104.10, following burn-in and soak. Gain versus frequency B104_5mV. Place in water at 25°C. Noise spectra as shown in B104_Noise. Reception robust, VA = 2.51-2.77 V, noise <10 μV.

MAR-16

[01-MAR-16] C101.11 and C101.13 running well. Reception 100%, VA = 2.75-2.75 V. Noise in C101.13 is 6 μV, but in C101.11 it is 35 μV. Gain of C101.13 is 1 dB lower than on 09-FEB-16. Gain of C101.11 is 10 dB lower, with the noise corrupting the applied sinusoid. E100.13 and E101.14 running well. Reception 100%, VA = 2.79-2.82 V, noise 7 μV. We stopped to take a half-hour phone call while these devices were out on the table, and by the time we returned, they had cooled to about 30°C from 60°C.

During transmitter assembly and encapsulation, there are three stages of cleaning. The first takes place at the assembly house, where the components are applied with solder paste and washed. Our AV5LF and RV2 circuits are lead-free and made with a chemical-wash flux. We check the board surfaces with a loop, looking at the reflection of a white over-head light on the solder mask. We never see any film on the boards delivered by the assembly house. The next wash is after we load leads and antenna. We flood the circuit with hot water and scrub with a toothbrush on both sides. We blow dry, wash again, and blow dry again. Now we inspect for a film of flux residue around the leads. If we do not brush vigorously, there will remain a film of flux at this point. But if both washes are vigorous with lots of hot water, the circuits will be clean. If we wash only once, there will often be residue on the board, visible only if we inspect with care the reflection of light from the solder mask. We believe such residue has been present on the top-side of the board in all transmitters before E105.11. The final wash is after we load the battery. This time, we have access only to the bottom side of the board, which we wash and dry twice. We inspect. If we wash only once, we have residue around U3 and other components of the X channel EEG amplifier. We suspect that all batches before E105.11 had at least some transmitters with some residue on the bottom side of the board. The resistive switch problem can be explained only by such residue.

In the IVC Enclosure at ION, we get adequate reception from implanted transmitters. But we are not satisfied. When we tested reception from a rat transmitter on a stick, we obtained 100.00% everywhere in the IVC rack. Now, with the curtain open, transmitter held up between two fingers, we obtain over 95% reception in the entire IVC rack last week. With curtain open, implanted rat transmitter reception drops to 80%, but a transmitter held loosely inside fist gives reception over 95%. With curtain open, transmitter held tightly with antenna pressed against palm, reception around 90%. Transmitter in test tube of water, with antenna folded over the body of the device, door open, we get 80%. It appears that for these implanted transmitters, we get transmission as if the antenna loop were closed up or folded over.

We perform a series of experiments to measure the effect of antenna length, orientation, and environment, which we detail in Transmit Antenna. As a result of these tests, we will suggest to ION that they arrange their antennas in three perpendicular orientations.

We are working on a new transmitter: one with external battery and wires for soldering to a battery, at the request of our Magdeburg users. Here is the circuit before encapsulation photo. After application of epoxy with our rotator system, and two coats of silicone by dipping, we get this.

[02-MAR-16] C101.13 has stopped transmitting. We put it back in hot water and will dissect later. C101.11 VA = 2.74, reception robust, noise 30 μV. E100.13 100% reception, VA = 2.84 V, noise 7 μV. E100.14 77% reception, VA = 2.89 V, noise 7 μV. We cool E100.14 down to 30°C and reception is now 100%.

[04-MAR-16] E100.13 and E100.14 100% reception, VA = 2.84-2.89 V, noise less than 7 μV. C101.11 reception 100%, VA = 2.73 V, noise 120 μV. We dissect C101.13, which failed on 02-MAR-16. Battery voltage −0.09 V. Disconnect battery and measure VB = −0.08 V. Connect external 2.6 V. Inactive current 1.5 μA. Active current 52 μA. Picks up mains hum. Responds to touching. Diagnosis: Unidentified Drain.

[07-MAR-16] E100.13, E100.14, and C101.11 all running. Reception 100%. VA = 2.86 V, 2.89 V, and 2.72 V respectively. Noise for E100.13 and E100.14 less than 7 μV, but for C101.11 is 110 μV.

[08-MAR-16] C101.11 reception 100%, VA = 2.66 V, noise 500 μV. E100.13 and E100.14 are running well. We measure gain versus frequency and find both transmitters are within 0.5 dB of their previous response all the way through the 0.3-160 Hz pass-band.


Figure: Gain versus Frequency for 5-mV Sinusoid, E100.13 and E100.14 Before and After Five-Week Poach.

We have batch R105.11-106.10 after burn-in and soak. This is the first batch made from the A3028RV2 circuit. Frequency response is R106_5mV. Battery voltages 2.62-2.71 V. Switching noise is below 5 μV except in R106.10, 12 μV, and R105.12, 8 μV. We decide to hold these two and ship the rest.

We are wondering if we can measure the location of transmitters in a faraday enclosure by looking at the signal strength received by an array of antennas.


Figure: Received Power Versus Position. In two directions on our bench (X/Y), for 100-mm and 10-mm Diameter Loop Antennas (S/L), and at 0 cm or 10 cm above the bench (0/10).

Only the tiny antenna gives us received power increasing as we come closer, decreasing as we move away.

[09-MAR-16] E100.13, E100.14, C101.11 running well.

[11-MAR-16] C101.11, E100.13, E100.14, have 100% reception, VA = 2.65 V, 2.86 V, and 2.94 V respectively. Noise is 110 μV, 6 μV, and 6 μV respectively.

We have batch E104.11-E105.10, all A3028E-HA, equipped with D-pin for the H-electrode. Frequency responses all look good in E105_5mV. In water at 35°C, switching noise in a 32-s interval is less than 1.2 μV in all fourteen devices. VA = 2.75-2.81 V and noise is less than 10 μV.


Figure: Switching Noise Peaks for E104.11-E104.10 In Water at 35°C. Spectrum of a 32-s interval, 10-30 Hz and 0.4 μV/div.

These are the first AV5LF circuits we have made with two double-wash stages, each stage high-flow, hot water with vigorous brushing, followed by higher-pressure nitrogen blast and loupe surface inspection.

We keep R105.12 and R106.10 to poach. These presented 8 μV and 12 μV switching noise respectively during quality control. Today we put them in water at 34°C and observe less than 0.4 μV noise on both devices. We place in water at 22°C and wait ten minutes. Noise is now 10 μV and 8 μV respectively. We put them in the oven to poach.

[14-MAR-16] Poaching transmitters: R106.10 100%, 2.86 V, 8 μV; C101.11 100%, 2.59 V, 157 μV; R105.12 100%, 2.86 V, 10 μV; E100.13, 100%, 2.63 V, 6 μV, E100.14, 100%, 2.91 V, 7 μV. Transmitter C101.11 has been running for 35 days (34 poaching, 1 burn-in) and battery voltage now appears to be dropping. Transmitter E100.13 has been running for 42 days (41 poaching, 1 burn-in) and battery voltage appears to be dropping.

[15-MAR-16] Poaching, pick-up with one antenna while in 60°C water: R106.10 99.8%, 2.80 V, 8.7 μV; C101.11 99.2%, 2.52 V, 225.8 μV; R105.12 100.0%, 2.81 V, 8.1 μV; E100.13 99.6%, 2.46 V, 7.1 μV; E100.14 99.8%, 2.85 V, 6.7 μV.

[16-MAR-16] Poaching transmitters: R106.10 100.0%, 2.87 V, 8.9 μV; C101.11 100.0%, 2.23 V, 253.2 μV; R105.12 100.0%, 2.88 V, 13.2 μV; E100.13 100.2%, 2.03 V, 8.0 μV; E100.14 100.0%, 3.07 V, 6.7 μV.

[17-MAR-16] Poaching transmitters recorded with 1-s interval with one antenna: R106.10 98.4%, 2.87 V, 7.2 μV; C101.11 99.6%, 1.88 V, 58.7 μV; R105.12 99.2%, 2.87 V, 15.6 μV; E100.13 91.4%, 1.94 V, 6.1 μV; E100.14 100.0%, 3.02 V, 6.3 μV.

[18-MAR-16] Poaching transmitters: R106.10 100.0%, 2.86 V, 9.3 μV; R105.12 100.0%, 2.86 V, 8.4 μV; E100.14 100.0%, 2.97 V, 6.0 μV. Transmitters E100.13 and C101.11 have stopped. Dissect E100.13. VB = 0.6 V. Disconnect battery and VB = 0.7 V. Connect external 2.6 V. Inactive current 1.7 μA, active 83.5 μA. This device ran for 46 days. Expected life was 255 mA-hr ÷ 84 μA = 126 days. Classify as "Unidentified Drain". Dissect C101.11. VB = 0.3 V. Disconnect, VB = 0.3V. Connect external 2.6 V. Inactive current 1.6 μA. Active current 53 μA. Receptin 100%. With 3-V battery, VA = 3.14 V, noise 32.6 μV. C101.11 ran for a total of 39 days. Its expected life is 48 mA-hr ÷ 53 μA = 38 days. Classify as "Full Life".

We have batch E106.11-E107.8. Frequency response E107_5mV. Switching noise in warm water less than 2 μV.

[21-MAR-16] Poaching transmitters: R106.10 100.0%, 2.86 V, 10.9 μV; R105.12 100.0%, 2.88 V, 8.7 μV; E100.14 99.8%, 2.98 V, 8.2 μV. Transmitter E100.14 has been poaching for seven weeks. Its gain versus frequency has the same shape as at weeks 0, but is 1.4 dB lower.

[22-MAR-16] Poaching transmitters: R106.10 99.8%, 2.87 V, 16.1 μV; R105.12 100.0%, 2.88 V, 10.0 μV; E100.14 100.0%, 3.33 V, 5.8 μV. At first, E100.14 is not transmitting. We turn it on. The fact that its battery voltage appears to be higher than we know possible for a fresh BR2330 suggests it has the resistive switch problem. The fact that it turned itself off over-night we classify as an artifact.

[24-MAR-16] Poaching transmitters: R106.10 100.0%, 2.87 V, 13.6 μV; R105.12 96.9%, 2.88 V, 12.0 μV; E100.14 99.8%, 2.94 V, 4.6 μV.

[25-MAR-16] Poaching transmitters: R106.10 99.8%, 2.85 V, 12.8 μV; R105.12 94.1%, 2.86 V, 8.4 μV; E100.14 94.1%, 2.89 V, 5.0 μV.

[28-MAR-16] Poaching transmitters: R106.10 100.0%, 2.86 V, 13.8 μV; R105.12 99.2%, 2.87 V, 18.4 μV; E100.14 99.2%, 2.87 V, 5.3 μV. R106.10 and R105.12 pick up mains hum when we take them out of their water, but E100.14 does not. We suspect that failure of the EEG amplifier occurred around 24-MAR-16, when we first observed noise below 5 μV on E100.14.

[29-MAR-16] We have batch E107.9-E108.5, with E108.8, a total of 12 transmitters. Frequency response E108_5mV is uniform to ±1 dB at all frequencies. Switching noise in 30°C water is less than 4 μV. Battery voltages 2.63-2.77 V. Total noise less than 11 μV except for E108.8, which is 18 μV. We hold back E108.8 for poaching, and E107.9 as well. Twenty minutes later, the noise on E108.8 in 30°C water is only 9 μV. It occurs to us that E108.8 was the last transmitter we turned on, and so ran for the least time before we measured noise, during which its battery voltage would be settling. Poaching transmitters today: E108.8 100.0%, 2.68 V, 9.6 μV; E107.9 100.0%, 2.77 V, 9.7 μV; R106.10 100.0%, 2.83 V, 10.9 μV; R105.12 100.0%, 2.83 V, 9.0 μV; E100.14 100.0%, 2.94 V, 6.3 μV.

APR-16

[01-APR-16] Poaching transmitters: E108.8 87.5%, 2.82 V, 11.3 μV; E107.9 99.8%, 2.81 V, 6.9 μV; R106.10 94.9%, 2.88 V, 8.4 μV; R105.12 94.9%, 2.92 V, 9.2 μV; E100.14 100.0%, 2.94 V, 5.6 μV.

[04-APR-16] Poaching transmitters: E108.8 96.7%, 2.81 V, 11.5 μV; E107.9 98.4%, 2.80 V, 7.3 μV; R106.10 98.4%, 2.87 V, 14.6 μV; R105.12 99.4%, 2.87 V, 18.8 μV; E100.14 93.6%, 2.44 V, 6.3 μV.

[05-APR-16] E108.8 99.0%, 2.84 V, 12.2 μV; E107.9 98.4%, 2.82 V, 7.3 μV; R106.10 99.6%, 2.88 V, 7.9 μV; R105.12 100.0%, 2.52 V, 11.7 μV; E100.14 99.8%, 2.59 V, 5.8 μV.

We have batch B108.9-109.4, all A3028B-AA. Frequency response B109_5mV. We place them in water at 33°C and measure input noise. We see <12 μV on all except B109.3 = 21 μV and B108.14 = 27 μV. Switching noise is ≤5 μV. We will not ship the two noisy transmitters.

[06-APR-16] Poaching transmitters: E108.8 100.0%, 2.78 V, 7.6 μV; E107.9 95.7%, 2.79 V, 8.0 μV; R106.10 98.4%, 2.85 V, 10.7 μV; R105.12 98.8%, 2.59 V, 34.0 μV; E100.14 94.7%, 2.57 V, 5.1 μV. We see baseline shifts in R105.12, sufficient to disturb EEG monitoring. Frequency response is Poaching_5mV_06APR16, and shows failure of EEG amplifier in R106.10. The amplifier in E100.14 failed some time ago, but the circuit is still powered up after 72 days. We turn on B108.14 and B108.3, both noisy but with good gain versus frequency, and put them in to poach.

[07-APR-16] Poaching transmitters: B109.3 98.2%, 2.73 V, 14.8 μV; E108.8 98.6%, 2.83 V, 12.8 μV; E107.9 97.5%, 2.81 V, 7.2 μV; R106.10 97.9%, 2.88 V, 7.2 μV; R105.12 94.1%, 2.61 V, 16.3 μV; B108.14 94.7%, 2.91 V, 6.6 μV, E100.14 100.2%, 2.63 V, 9.1 μV.

[08-APR-16] Poaching transmitters: B109.3 100.0%, 2.77 V, 9.3 μV; E108.8 100.0%, 2.84 V, 10.6 μV; E107.9 100.0%, 2.83 V, 7.0 μV; R106.10 95.3%, 2.88 V, 8.4 μV; R105.12 92.6%, 2.40 V, 14.4 μV; B108.14 94.1%, 2.74 V, 17.0 μV. Transmitter E100.14 was transmitting this morning, but stopped in the afternoon after sixty-six days poaching.

[11-APR-16] Poaching transmitters: B109.3 100.0%, 2.76 V, 9.1 μV; E108.8 100.0%, 2.85 V, 12.4 μV; E107.9 99.8%, 2.83 V, 8.4 μV; R106.10 96.1%, 2.88 V, 12.4 μV; B108.14 98.6%, 2.74 V, 19.8 μV. Transmitter R105.12 has failed. We leave it to poach until we have time to dissect. Dissect E100.14. VB = 0.13 V. Disconnect battery, VB = 0.12. Apply external 2.6 V. Inactive 1.7 μA, active 80 μA. Classify as "Unidentified Drain".

[12-APR-16] Poaching transmitters: B109.3 100.0%, 2.68 V, 26.6 μV; E108.8 95.1%, 2.80 V, 10.6 μV; E107.9 96.7%, 2.79 V, 7.7 μV; R106.10 98.0%, 2.84 V, 35.0 μV; B108.14 100.0%, 2.68 V, 32.9 μV. Dissect R105.12. VB = 0.33 V. We damage the circuit in the neighborhood of R8/R9. Disconnect battery and it recovers to 1.3 V in a few minutes. Connect external 2.6 V. Inactive current 1.6 μA, active 2.4 mA. Remove C5 and C6, current 2.2 mA. Remove C4, current 100 mA. The commencement of baseline shifts a few days ago, combined with battery drain over a few days, suggest a "corroded capacitor".

We have batch H109.5, 7, 8, 11, and 13. We measure frequency response, which we did earlier and now repeat. The gain is now higher, in units of ADC counts, because the battery voltage is now 2.6 V instead of 3.0 V as when we plug in a large auxiliary battery. Gain of the transmitters lies within ±0.5 dB. Switching noise is less than 4 μV. Battery voltages 2.59-2.64 V.

[13-APR-16] Poaching transmitters: B109.3 100.0%, 2.77 V, 10.8 μV; E108.8 95.1%, 2.85 V, 10.7 μV; E107.9 100.0%, 2.83 V, 8.2 μV; R106.10 99.0%, 2.88 V, 24.1 μV; B108.14 96.9%, 2.74 V, 17.6 μV.

[14-APR-16] Poaching transmitters: B109.3 96.9%, 2.76 V, 10.2 μV; E108.8 100.0%, 2.85 V, 10.9 μV; E107.9 100.0%, 2.83 V, 7.1 μV; R106.10 100.0%, 2.88 V, 6.9 μV; B108.14 97.1%, 2.73 V, 20.0 μV.

[15-APR-16] Poaching transmitters: B109.3 94.7%, 2.76 V, 10.6 μV; E108.8 100.0%, 2.85 V, 9.9 μV; E107.9 100.0%, 2.83 V, 6.7 μV; R106.10 100.0%, 2.89 V, 12.4 μV; B108.14 95.7%, 2.72 V, 23.4 μV.

[18-APR-16] Poaching transmitters: B109.3 100.0%, 2.74 V, 12.9 μV; E108.8 99.8%, 2.85 V, 11.6 μV; E107.9 100.2%, 2.83 V, 7.0 μV; R106.10 99.2%, 2.88 V, 8.1 μV; B108.14 99.4%, 2.64 V, 20.2 μV.

[19-APR-16] Poaching transmitters: B109.3 100.0%, 2.73 V, 15.9 μV; E108.8 100.0%, 2.84 V, 11.3 μV; E107.9 99.4%, 2.84 V, 7.0 μV; R106.10 98.4%, 2.88 V, 14.4 μV; B108.14 99.0%, 2.73 V, 20.3 μV.

[20-APR-16] Poaching transmitters: B109.3 100.0%, 2.72 V, 17.4 μV; E108.8 99.8%, 2.84 V, 11.4 μV; E107.9 96.1%, 2.84 V, 7.4 μV; R106.10 99.4%, 2.88 V, 9.0 μV; B108.14 98.6%, 2.73 V, 21.3 μV.

[21-APR-16] Poaching transmitters: B109.3 99.0%, 2.73 V, 13.6 μV; E108.8 94.5%, 2.83 V, 9.3 μV; E107.9 100.0%, 2.83 V, 7.0 μV; R106.10 100.0%, 2.87 V, 12.0 μV; B108.14 93.0%, 2.73 V, 19.3 μV.

[22-APR-16] Poaching transmitters: B109.3 91.0%, 2.70 V, 22.1 μV; E108.8 94.9%, 2.83 V, 11.2 μV; E107.9 97.3%, 2.83 V, 7.0 μV; R106.10 99.4%, 2.87 V, 14.1 μV; B108.14 99.8%, 2.71 V, 27.6 μV.

[25-APR-16] Poaching transmitters: B109.3 97.5%, 2.67 V, 27.7 μV; E108.8 94.3%, 2.80 V, 9.4 μV; E107.9 94.3%, 2.81 V, 7.2 μV; R106.10 99.2%, 2.87 V, 7.7 μV. We have lost B108.14 after 19 days.

[26-APR-16] We have batch R110.1-11 after burn-in and soak. These are made with a mix of RV1 and RV2 circuits. We measure frequency response E100_5mV, all agree to within ±0.4 dB. Noise and battery voltage: R110.1 90.4%, 2.66 V, 8.3 μV; R110.2 90.4%, 2.67 V, 12.4 μV; R110.3 96.1%, 2.68 V, 21.3 μV; R110.4 98.2%, 2.70 V, 11.7 μV; R110.5 91.2%, 2.69 V, 18.4 μV; R110.6 94.1%, 2.63 V, 9.5 μV; R110.7 91.4%, 2.63 V, 23.5 μV; R110.8 94.9%, 2.63 V, 20.2 μV; R110.9 99.6%, 2.67 V, 21.9 μV; R110.10 92.0%, 2.69 V, 8.2 μV; R110.11 93.8%, 2.69 V, 11.5 μV. Switching noise is <6 μV and 20-24 Hz at 34°C.

Poaching transmitters: E108.8 100.0%, 2.66 V, 20.8 μV; E107.9 99.0%, 2.80 V, 7.6 μV; R106.10 97.5%, 2.85 V, 7.0 μV; B109.3 100.0%, 2.56 V, 14.9 μV. All pick up mains hum and respond to movement. We dissect B108.14. VB = 0.06 V. Disconnect battery, VB = 0.11 V. Apply external 2.6 V. Inactive 1.6 μA, Active 78 μA. Analog input does not pick up mains hum. We expose amplifier parts, but break off R5. We replace R5. We notice the X lead is broken. Nevertheless, we cannot stimulate the input with our fingers. The amplifier has lost gain.

[27-APR-16] We have batch C110.12-111.11 freshly-encapsulated. Of these, C110.12-111.3 are made with the RV3 circuit, which is 13.5 mm long, while C111.4-111.11 are made with the AV5LF. The RV3 is made with the A302801F circuit board, which is 14.0 mm × 12.7 mm. The AV5LF is made with the A302801E circuit board, which is 12.7 mm × 12.7 mm. We measure the volume of 6 of each type by water displacement. Each type displaces 8.50±0.1 ml not including the antenna and leads. The transmitter body volume is 1.4±0.02 ml for each. We measure length, breadth, and height across the center of the transmitters and obtain 15.1 mm × 13.7 mm × 8.2 mm for the RV3 and 14.3 mm × 13.7 mm × 8.2 mm for the AV5LV.


Figure: Left: A3028C Made with AV5LF. Right: A3028C Made with RV3.

The extra 0.8 mm length of the encapsulated transmitter is obvious when we look down from above. But the thickness of this extension is less than 4 mm, so its contribution to the total volume of the transmitter should be less than 0.04 ml. Given that our volume measurement gave us the same answer to within ±0.02 ml, we conclude that the RV3 circuit will cause no significant increase in the volume of our mouse-sized transmitters.

[28-APR-16] Poaching transmitters are all running, we measure their frequency response Poaching_5mV_28APR16, which we compare to Poaching_5mV_06APR16.

[29-APR-16] We have batch C110.12-111.11. Frequency response is C111_5mV. Gain for 14 transmitters within ±0.7 dB across entire pass-band. We place in 40°C water. Switching noise ≤3 μV except C110.13, for which it is 6 μV. Reception, battery voltage, and noise: C111.1 97.3%, 2.76 V, 6.2 μV; C111.2 95.3%, 2.72 V, 6.0 μV; C111.3 100.0%, 2.71 V, 5.6 μV; C111.4 94.1%, 2.75 V, 6.3 μV; C111.5 96.9%, 2.65 V, 5.5 μV; C111.6 89.1%, 2.68 V, 7.3 μV; C111.7 91.8%, 2.75 V, 8.0 μV; C111.8 96.5%, 2.70 V, 5.3 μV; C111.9 94.1%, 2.68 V, 5.7 μV; C111.10 98.8%, 2.72 V, 6.8 μV; C111.11 99.6%, 2.69 V, 5.4 μV; C110.12 87.1%, 2.72 V, 5.4 μV; C110.13 95.3%, 2.68 V, 11.1 μV; C110.14 87.5%, 2.65 V, 8.2 μV. Noise is less than 12μV. We keep C110.13 and C110.14 to poach.

Poaching transmitters: B109.3 98.6%, 2.62 V, 27.3 μV; E108.8 100.0%, 2.55 V, 13.6 μV; E107.9 100.0%, 2.76 V, 8.2 μV; R106.10 94.1%, 2.87 V, 18.8 μV; C110.13 99.2%, 2.66 V, 11.1 μV; C110.14 100.0%, 2.62 V, 6.3 μV.

MAY-16

[02-MAY-16] Poaching transmitters: E108.8 97.3%, 2.45 V, 22.4 μV; E107.9 98.4%, 2.73 V, 10.0 μV; R106.10 94.7%, 2.63 V, 22.5 μV; C110.13 96.1%, 2.65 V, 8.2 μV; C110.14 96.5%, 2.62 V, 7.9 μV. We have lost B109.3, which is to be expected after 26 days running. The expected battery life of the A3028B is 600 hrs = 25 days.

[03-MAY-16] Poaching transmitters: E108.8 97.7%, 2.31 V, 20.1 μV; E107.9 98.6%, 2.76 V, 6.6 μV; R106.10 97.7%, 2.08 V, 19.2 μV; C110.13 95.7%, 2.79 V, 4.5 μV; C110.14 100.0%, 2.76 V, 5.6 μV.

[05-MAY-16] Poaching transmitters: E107.9 100.0%, 2.77 V, 7.0 μV; R106.10 99.6%, 2.67 V, 12.5 μV; C110.13 100.0%, 2.75 V, 6.8 μV; C110.14 99.2%, 2.70 V, 7.2 μV. E108.8 has stopped and will not turn on. We see a rust-colored stain underneath the silicone over the positive battery terminal in R106.10. We put C110.14 in a jar of vinegar and place the jar in the oven at 60°C.

Dissect E108.8. Good adhesion between silicone and epoxy, but enamel comes away easily and breaks up. VB = 1.2 V. Disconnect battery, VB = 2.2 V. Apply external 2.6 V. Inactive current 1.9 μA. Active current 530 μA. Device transmits and detects mains hum. Turn off an on again, active current 134 μA. From average X, VA = 2.6 V. We burn off epoxy to get to C4. We replace C4. Current consumption is 85 μA. We clean C4 with solder at 300°C. Its insulation is now >20 MΩ. Diagnosis of failure: corroded capacitor.

[06-MAY-16] Poaching transmitters: E107.9 98.4%, 2.77 V, 6.8 μV; R106.10 98.6%, 2.82 V, 19.4 μV; C110.13 96.9%, 2.80 V, 5.0 μV; C110.14 96.9%, 2.75 V, 6.1 μV.

[09-MAY-16] Poaching transmitters: E107.9 98.8%, 2.69 V, 7.1 μV; R106.10 100.0%, 2.79 V, 19.3 μV; C110.13 98.0%, 2.80 V, 5.0 μV; C110.14 100.0%, 2.75 V, 7.0 μV. Transmitter C110.14 is in vinegar, frequency response is normal, but noise with 20-MΩ source is around 50 μV.

[10-MAY-16] Poaching transmitters: E107.9 99.2%, 2.65 V, 6.5 μV; R106.10 100.0%, 2.79 V, 16.0 μV; C110.13 98.4%, 2.79 V, 5.4 μV; C110.14 100.0%, 2.75 V, 6.7 μV. Transmitter C110.14 shows 6 μV noise with 50-Ω source, 100 μV noise with 20-MΩ source. Frequency response with 50-Ω 5-mV source is perfect.

[11-MAY-16] Poaching transmitters: E107.9 100.0%, 2.67 V, 7.8 μV; R106.10 98.2%, 2.79 V, 7.6 μV; C110.13 100.0%, 2.76 V, 7.1 μV; C110.14 96.9%, 2.69 V, 7.2 μV.

We have our first batch of A3028E with input protection (so they are identical to the A3028R), made from the A3028RV3 circuit board, and encapsulated in epoxy using our new vacuum process and rotation curing stand. We encapsulate six RV3 circuits, one old AV3 circuit, and a blank RV3 circuit board. The six RV3 circuits we remove from glue, allow one large drop of epoxy to fall from the transmitter, invert for thirty seconds, and place on rotator. We angle the rotator upwards to counter a tendency for the glue to accumulate on the far end of the transmitter from the rotator shaft. The AV3 circuit we allow two drops of epoxy to fall from the transmitter before inverting for thirty seconds and rotating.


Figure: Rotator Epoxy Encapsulation. Left: RV3 with One Drop Removed. Right: AV3 with Two Drops Removed. We begin encapsulation with the transmitter in vacuum with the glue. We insert the transmitter into the glue, then let air into the chamber to force the glue into the circuit. We remove and attach to our rotator.

The maximum thickness of the one-drop encapsulation is 9.4 mm, and of the two-drop is 7.8 mm. The width of our hand-applied epoxy is around 8.0 mm after the top-coat. The additional thickness in the one-drop rotated device is part of a spherical, mirror-like convex surface on both sides. The increase in volume is no more than 0.2 ml.

We place the entire E112.12 transmitter in our mouth. We can just tast the battery voltage, which means that the epoxy film does not provide complete insulation of the battery terminals. But the entire body is smooth. We place the AV3 in our mouth. There are many strong points electrical potential, which means the epoxy provides no insulation over capacitors and battery terminals.

We tore the X lead of E113.10 during encapsulation. We re-attach it now, to the bottom side, where the pad is only 25 mils. We knock off a resistor nearby, and note that the space beneath the resistor was entirely filled with epoxy. Frequency response of E113.10 and E112.12 are nominal.

[12-MAY-16] We have batch R111.12-112.10 after burn-in and soak. Frequency response is R112_5mV, all agree to within ±0.74 dB. We place in water at 37°C. Switching noise is ≤4 μV. Noise ≤12 μV. Battery voltages 2.64-2.70 V.

Poaching transmitters: E107.9 99.0%, 2.69 V, 7.2 μV; R106.10 98.8%, 2.77 V, 8.6 μV; C110.13 97.7%, 2.79 V, 5.0 μV; C110.14 98.8%, 2.75 V, 18.8 μV.

[13-MAY-16] Poaching transmitters: E107.9 98.8%, 2.72 V, 7.4 μV; R106.10 100.0%, 2.77 V, 12.2 μV; C110.13 97.7%, 2.77 V, 6.2 μV; C110.14 100.0%, 2.74 V, 75.6 μV. The noise on C110.14 is sufficient to be described as "artifact" at this point.

[16-MAY-16] Poaching transmitters all running, but forgot to cut and paste result string into into this record.

[17-MAY-16] We have batch E112.12-E113.11 encapsulated with our vacuum and rotator, then coated three times with silicone. The silicone has some wrinkles, but coating around all components and battery rim is at least 0.5 mm. Volume of 4 devices in water is 12 ml, making individual volume 3.0 ml, up from previous 2.8 ml.

Poaching transmitters: R106.10 98.8%, 2.74 V, 7.5 μV; C110.13 97.3%, 2.77 V, 31.3 μV; C110.14 99.2%, 2.63 V, 35.3 μV. Transmitter E107.9 has failed. Dissect. Battery voltage 2.2 V. We are able to turn the transmitter on for a few seconds. Disconnect battery, VB = 2.3 V. Connect external 2.6 V. Inactive 2.0 μA. Active 82 μA. Connect to battery. Frequency response within 1 dB of nominal for 3-V battery. Diagnosis: unidentified drain. We add R112.11 to the poach. This transmitter received no top-coat of epoxy, but simply enamel followed by five coats of silicone.

[18-MAY-16] Poaching transmitters: R112.11 100.0%, 2.81 V, 14.2 μV; C110.13 100.0%, 2.75 V, 6.6 μV. Dissect C110.14. VB = 1.1 V. Disconnect battery, VB = 0.9 V. Connect external 2.6 V. Inactive current 3.2 μA. Active 53.4 μA. Reception 100%. Attach external 3.0-V battery. Frequency response C110_14_G_vs_F shows same shape as when transmitter was first made, but lower gain because of higher battery voltage. Noise is around 20 μV. Inactive current is now 1.6 μA. Diagnosis: unidentified drain. Dissect R106.10. VB = 1.3 V. Disconnect battery, VB = 2.2 V. Attach external 2.6 V. Inactive current 2.2 μA. Active 2 mA at first, limited by ammeter. Switch to higher scale and current jumps for a fraction of a second, then drops to 83 μA. Frequency response still wrong as shown here. Diagnosis: corroded capacitor.

[20-MAY-16] Poaching transmitters: R112.11 100.0%, 2.82 V, 7.3 μV; C110.13 100.0%, 2.67 V, 8.2 μV.

[23-MAY-16] Poaching transmitters: R112.11 100.0%, 2.87 V, 15.8 μV; C110.13 99.6%, 1.92 V, 13.9 μV. We have batch E112.12-E113.11 after 24-hour burn-in and 5-day soak. Measure reception, battery voltage and noise: E113.1 98.6%, 2.61 V, 7.0 μV; E113.2 92.6%, 2.65 V, 15.0 μV; E113.3 100.0%, 2.61 V, 7.4 μV; E113.4 100.0%, 2.64 V, 7.4 μV; E113.5 90.4%, 2.67 V, 13.2 μV; E113.6 99.4%, 2.62 V, 8.0 μV; E113.7 98.8%, 2.67 V, 7.6 μV; E113.8 98.0%, 2.65 V, 8.3 μV; E113.9 96.1%, 2.62 V, 12.0 μV; E113.11 94.9%, 2.63 V, 10.2 μV; E112.12 95.7%, 2.66 V, 8.6 μV; E112.13 97.3%, 2.59 V, 9.0 μV; E112.14 93.0%, 2.60 V, 7.6 μV. Frequency response E113_5mV within ±0.5 dB. Switching noise in water at 37°C is ≤5 μV.

[24-MAY-16] Poaching transmitters: R112.11 100.0%, 2.60 V, 11.5 μV. Transmitter C110.13 has stopped. We remove R112.11 from water and find it is generating its own 1-Hz full-scale square wave when its inputs are open-circuit or connected by 20 MΩ. We drive with a 10-mV 50-Ω source and get gain R112_11_10mV. We cannot find an earlier measurement of this device's frequency response. We recall that this particular transmitter was delayed in production because of displaced capacitors on the X amplifier. We replaced one capacitor and re-centered several parts before loading the battery. We put R112.11 back in to poach, to check for battery drain in the long-term.

We dissect C110.13. Battery voltage 1.2 V. Disconnect battery, VB = 2.4 V. Apply external 2.6 V. Inactive current 1.7 μA. Active current 51 μA. This transmitter has run for a one-day burn-in followed by twenty-five days poaching for a total of 624 hrs. We expect it to run for 900 hrs. We re-connect the battery and find it's voltage once again drops to 1.2 V when we turn on the transmitter. Diagnosis: unidentified drain. We connect an external 3.0-V battery and measure frequency response C110_13_5mV. We put rotated transmitter E113.9 in to poach.

Receive four A3028B transmitters back from Edinburgh that drained their batteries. Their encapsulation is in perfect condition, with strong adherence of silicone to epoxy core, no sign of rust, nor any breach of the antenna insulation. Transmitter B104.2 won't turn on. Dissect. Battery VB = 0.3 V. Disconnect battery, VB = 0.3 V. Apply external 2.6 V. Inactive current 1.8 μA. Active 79.7 μA. Reception perfect. Frequency response correct. Transmitter B104.5 won't turn on. Dissect. Battery VB = 0.2 V. Disconnect, VB = 0.2 V. Connect external 2.6 V. Inactive current 1.7 μV. Active 77.6 μA. Reception perfect, picks up mains hum. Frequency response correct. Transmitter B104.3 won't turn on. Dissect. Battery VB = 0.3 V. Disconnect, VB = 0.4 V. Connect external 2.6 V. Inactive current 2.0 μA. Active current 72.3 μA. Reception perfect, picks up mains hum. Frequency response correct. Transmitter B104.9 won't turn on. Dissect. Battery VB = 0.1 V. Disconnect, VB = 0.1 V. Connect external 2.6 V. Inactive current 1.6 μA. Active 78.2 μA. Reception perfect, picks up mains hum. Frequency response correct. The four frequency responses are EU_Ret_1.

[25-MAY-16] Poaching transmitters: E113.9 100.0%, 2.79 V, 7.0 μV; R112.11 100.0%, 2.91 V, 34.2 μV.

[27-MAY-16] Add E113.10 to poaching transmitters. Poaching transmitters: E113.9 99.6%, 2.80 V, 7.1 μV; E113.10 99.0%, 2.77 V, 6.8 μV; R112.11 99.2%, 2.48 V, 19.8 μV. We see R112.11 baseline moving up and down by a few millivolts.

[31-MAY-16] We have batch B113.12-B114.7 after six-hour burn-in and four-day soak. This batch we made with the rotator procedure, allowing 60 s for epoxy drain, 30 s inversion, and applying four coats of silicone. Volume of 8 transmitter bodies is 11 ml, making average body volume a little under 1.4 ml. Inspect all ten devices with loupe and find no thin spots in the silicone, only a couple of small bubbles on the battery tab wall. Frequency response within ±0.6 dBm, shown in B114_5mV. In water at 35°C, switching noise is 5 μV or less. Battery voltage 2.55-2.69 V. Noise 5-20 μV rms one minute after immersion in water.

We turn on B113.13 and B114.4 and add them to our poach test. Poaching transmitters: B114.4 100.0%, 2.71 V, 18.9 μV; E113.9 100.0%, 2.76 V, 7.5 μV; E113.10 99.0%, 2.74 V, 7.4 μV; R112.11 100.0%, 1.94 V, 30.6 μV; B113.13 100.0%, 2.72 V, 11.2 μV.

JUN-16

[01-JUN-16] Poaching transmitters: B114.4 98.2%, 2.74 V, 29.0 μV; E113.9 100.0%, 2.82 V, 7.6 μV; E113.10 98.0%, 2.83 V, 6.8 μV; R112.11 97.5%, 2.13 V, 27.9 μV; B113.13 98.2%, 2.73 V, 13.3 μV. R112.11's noise consists of baseline swings. Measure frequency response. R112.11 stops. We leave it on our desk to dissect later. The other four are fine, as in Poaching_5mV_01JUN16.

[02-JUN-16] Poaching transmitters: B114.4 100.0%, 2.71 V, 14.7 μV; E113.9 98.4%, 2.82 V, 6.9 μV; E113.10 98.8%, 2.82 V, 7.2 μV; B113.13 100.0%, 2.73 V, 9.1 μV. Dissect R112.11. Epoxy top coat is thin we hesitate to cut through the silicone above. When we remove the silicone, we feel moisture underneath, and a coating of enamel crumbles away, revealing the corners of C2, C5, and C6, as well as the top of U3. WE see corrosion across the top side of the three capacitors. This transmitter received no top coat, but was instead covered with enamel.


Figure: Capacitor Corrosion in Transmitter Without Epoxy Top-Coat, Hand-Made. We have removed the silicone and enamel to reveal the epoxy. The thin epoxy coating over the components appears to have been removed by the enamel as the enamel degraded in humidity. We see corrosion on both capacitors.

[03-JUN-16] Transmitter R112.11 is working again. We measure frequency response that shows a peak at around 300 Hz, but good gain at lower frequencies. We have VA = 2.2 V. Poaching transmitters: B114.4 97.3%, 2.66 V, 50.9 μV; E113.9 99.8%, 2.77 V, 7.0 μV; E113.10 99.2%, 2.77 V, 7.4 μV; B113.13 98.0%, 2.67 V, 8.8 μV.

[06-JUN-16] We have batch R114.8-115.7. Frequency response R114_5mV within ±0.7 dB. Switching noise in water at 32°C is less than 5 μV. Total noise less than 12 μV. Battery voltage 2.60-2.66 V. Four have severe cosmetic flaws in the silicone: wrinkles on the battery side. These are R114.9, R115.1, R115.2, and R115.7. One has glue over its label, R114.10. Poaching transmitters: B114.4 93.9%, 2.73 V, 13.8 μV; E113.9 92.4%, 2.83 V, 15.2 μV; E113.10 100.0%, 2.84 V, 8.9 μV; B113.13 96.3%, 2.75 V, 10.9 μV.

[07-JUN-16] Poaching transmitters: B114.4 92.0%, 2.74 V, 20.7 μV; E113.9 96.9%, 2.83 V, 7.4 μV; E113.10 99.4%, 2.85 V, 6.6 μV; B113.13 93.0%, 2.75 V, 12.2 μV.

[08-JUN-16] We have applied another two coats of silicone to R114.8 and R113.2, covering the wrinkles in the third coat so that they are no longer sharp-edges. We deem them worthy of shipment. We will apply another two coats to the rest of the transmitters. Poaching transmitters: B114.4 98.0%, 2.73 V, 20.5 μV; E113.9 100.2%, 2.83 V, 7.1 μV; E113.10 97.7%, 2.85 V, 7.3 μV; B113.13 100.0%, 2.75 V, 9.5 μV.

[09-JUN-16] We have R114.10 and R114.11 with five coats of silicone, second and third coats with wrinkles. We add them to our poaching collection. Poaching transmitters: B114.4 98.6%, 2.73 V, 9.7 μV; E113.9 100.0%, 2.83 V, 7.6 μV; E113.10 98.8%, 2.85 V, 7.5 μV; B113.13 99.8%, 2.74 V, 14.1 μV, R114.10 100.0%, 2.68 V, 10.6 μV; R114.11 100.0%, 2.69 V, 7.6 μV. We have R115.12, which we rotated two days ago. We over-stretched its blue lead while removing it from the rotator chuck. We spent a minute or two handling the transmitter and trying to correct the twist we had introduced into the blue lead. The twelve other transmitters in the same batch were never touched by human hand, only by gloves. After three coats of silicone, each curing for two hours before the next coat, the un-touched twelve have perfect silicone coats. There are no wrinkles anywhere, even around the battery edges. But the one we handled has wrinkles all over half of the battery-side.


Figure: Effect of Handling a Transmitter: Silicone Wrinkles. Three silicone coats, with at least two hours between coats. On the left, E115.10 was not handled between rotation and silicone dipping. On the right, E115.12 we handled for several minutes trying to sort out a problem with its leads after rotation.

[13-JUN-16] Poaching transmitters: B114.4 99.8%, 2.70 V, 9.8 μV; E113.9 100.0%, 2.83 V, 6.6 μV; E113.10 97.3%, 2.86 V, 15.3 μV; B113.13 94.3%, 2.71 V, 13.6 μV; R114.10 94.5%, 2.81 V, 6.6 μV; R114.11 96.5%, 2.81 V, 6.8 μV.

[14-JUN-16] We have batch E115.8-116.5. Rotator curing with three coats of silicone, at least two hours curing between coats. E116.4 we re-soldered the VC lead during production but it's no longer attached. Frequency response of the remainder is E115_5mV, all within ±0.8 dB. Note cavities in silicone around the mounting wire in E116.1 and E116.4. Switching noise in warm water at 32°C is less than 4 μV. Battery voltages 2.59-2.67 V, noise <12 μV, reception perfect. We add E115.12 and E116.4 to our poach, noting that E116.4 has its X− connection missing.

Poaching transmitters: B114.4 97.5%, 2.68 V, 11.3 μV; E113.9 100.0%, 2.83 V, 7.5 μV; E113.10 100.0%, 2.86 V, 7.1 μV; B113.13 96.7%, 2.65 V, 18.2 μV; E116.4 100.0%, 2.72 V, 3.9 μV; R114.10 100.0%, 2.81 V, 7.2 μV; R114.11 100.0%, 2.81 V, 7.0 μV.

[15-JUN-16] Poaching transmitters: B114.4 97.5%, 2.65 V, 11.2 μV; E113.9 93.8%, 2.83 V, 10.0 μV; E113.10 98.2%, 2.86 V, 7.1 μV; B113.13 95.3%, 2.74 V, 10.4 μV; E116.4 100.0%, 2.79 V, 4.5 μV; R114.10 99.6%, 2.81 V, 7.1 μV; R114.11 99.6%, 2.81 V, 6.7 μV; E115.12 100.0%, 2.84 V, 6.4 μV.

[17-JUN-16] Poaching transmitters: B114.4 98.4%, 2.63 V, 8.2 μV; E113.9 99.8%, 2.83 V, 7.6 μV; E113.10 98.4%, 2.86 V, 6.7 μV; B113.13 96.9%, 2.71 V, 7.8 μV; E116.4 99.8%, 2.80 V, 4.4 μV; R114.10 100.0%, 2.82 V, 6.7 μV; R114.11 97.7%, 2.82 V, 6.7 μV; E115.12 97.5%, 2.85 V, 6.6 μV.

[19-JUN-16] Poaching transmitters: B114.4 98.8%, 2.60 V, 12.0 μV; E113.9 99.4%, 2.83 V, 7.0 μV; E113.10 100.0%, 2.86 V, 7.5 μV; B113.13 99.8%, 2.69 V, 14.8 μV; E116.4 92.8%, 2.80 V, 4.8 μV; R114.10 91.0%, 2.82 V, 6.6 μV; R114.11 100.0%, 2.82 V, 7.8 μV; E115.12 98.4%, 2.86 V, 6.4 μV.

[22-JUN-16] We have batch E116.6-117.7, after two-day burn-in and three-day soak. Rotator curing with three coats of silicone. Switching noise in water at 36°C is <10 μV, except for E116.12 and E116.7, where we see 40 μV, and E117.4, where we see 12 μV. Total noise in water is less than 12 μV except for the three noisy ones, which are up to 50 μV. We put E116.12 and E116.7 in cold and warm water. Their switching noise remains 40 μV. We connect to our 20-MΩ signal source, with wires in both possible orientations, and see 5 μV switching noise and 20 μV total noise. With the leads open-circuit in our faraday enclosure we get <5 μV switching noise and 30 μV total noise, dominated by fluctuating 60 Hz. We drop the transmitters into water and we immediately see the switching noise returning. We will ship E116.7 and E116.12 for ISL Test Only to ION, and keep E117.4 here for test. We also have E116.10, which has only a blue lead. Frequency response of all transmitters except E116.10 is within ±1 dB in E116_5mV. Later in the day, we find E116.10 won't turn on. Nor does it emit RF.

Poaching transmitters: B114.4 100.0%, 2.59 V, 8.0 μV; E113.9 97.9%, 2.82 V, 9.0 μV; E113.10 100.0%, 2.87 V, 7.6 μV; B113.13 99.8%, 2.60 V, 8.8 μV; E116.4 95.3%, 2.79 V, 4.8 μV; R114.10 100.0%, 2.81 V, 7.0 μV; R114.11 98.2%, 2.81 V, 6.6 μV; E115.12 99.2%, 2.85 V, 17.2 μV.

We measure E117.4's switching noise and total noise when connected to a 20-MΩ source in water at 35°C, 40°C, 45°C, and 50°C. Switching noise in 20-30 Hz is <5 μV and total noise is around 35 μV. We see 5-20 Hz rumble from the movement of the fluid. We place in 35°C water with leads open. Switching noise is 4 μV and total noise is 9 μV. Earlier in the day, we repeatedly observed switching noise >10 μV with this device.

[24-JUN-16] We dissect E116.10. Battery voltage 0.8 V. Disconnect battery, its voltage is 2.4 V. Connect external 2.6 V, inactive current is 1.6 μA. Active current is 200 mA. U9 is getting hot. We remove U9. Active current is 52 μA. We load another U9. Active current 85 μA immediately after cleaning. Reception 100%. Picks up some 50 μV of mains hum (recall that X+ lead is missing). Poaching transmitters: all running with 100% reception, but did not measure battery voltages.

[27-JUN-16] Poaching transmitters: E113.9 100.0%, 2.82 V, 7.6 μV; E113.10 100.0%, 2.91 V, 102.2 μV, E116.4 100.0%, 2.83 V, 4.7 μV; R114.10 100.0%, 2.83 V, 6.6 μV; R114.11 99.8%, 2.83 V, 6.8 μV; E115.12 99.0%, 2.88 V, 6.8 μV. We have lost B113.13 and B114.4. On Friday their battery voltages were 2.6 V, which means they were roughly 90% of the way through their battery capacity. Friday was 600 hrs operating and today is 672 hrs. We assume they failed at around 650 hrs, which is 27 days poaching. Expected life is 600 hours.

[29-JUN-16] We have batch R117.8-118.6. Frequency response R116_5mV all within ±0.6 dB. Switching noise in 31°C water ≤6 μV, in 37°C ≤5 μV.

Poaching transmitters: E113.9 100.0%, 2.82 V, 7.7 μV; E113.10 100.0%, 2.86 V, 6.7 μV; E116.4 99.4%, 2.83 V, 4.7 μV; R114.10 100.0%, 2.83 V, 6.8 μV; R114.11 99.6%, 2.83 V, 7.2 μV; E115.12 100.0%, 2.88 V, 6.1 μV.

[30-JUN-16] Poaching transmitters: E113.9 99.0%, 2.82 V, 8.1 μV; E113.10 97.7%, 2.86 V, 8.2 μV; E116.4 97.3%, 2.83 V, 4.2 μV; R114.10 99.8%, 2.83 V, 6.4 μV; R114.11 96.1%, 2.83 V, 7.5 μV; E115.12 95.9%, 2.88 V, 6.3 μV.

JUL-16

[01-JUL-16] Poaching transmitters all running 100% reception.

[03-JUL-16] We have batch E119.7-E120.6, excepting E119.12, which we damaged during assembly, all have been through 24-hour burn-in and 72-hour soak. Frequency response is E119_5mV. Gain within ±0.4 dB. Reception perfect. Total noise ≤10 μV. Switching noise in water at 45°C and 40°C is <2 μV, at 37°C is <3 μV.

Poaching transmitters: E113.9 100.0%, 2.82 V, 7.4 μV; E113.10 100.0%, 2.85 V, 7.6 μV; E116.4 93.6%, 2.81 V, 4.2 μV; R114.10 97.5%, 2.82 V, 7.2 μV; R114.11 95.3%, 2.83 V, 10.1 μV; E115.12 97.5%, 2.88 V, 6.4 μV. We remove E113.9 and E113.10 from water and allow them to cool down. As they cool, E113.9 generates 1-mV pulses at 2 Hz and E113.10 produces 1-mV step artifacts. We dry them out in the oven at 60° for fifteen minutes. E113.9 no longer produces any pulses, but its gain is too low from 30-160 Hz. E113.10 produces steps, and its gain is also too low 30-160 Hz.

[05-JUL-16] Poaching transmitters: E113.9 100.2%, 2.80 V, 8.3 μV; E113.10 95.5%, 2.82 V, 34.8 μV; E116.4 96.7%, 2.81 V, 4.1 μV; R114.10 100.2%, 2.82 V, 6.8 μV; R114.11 96.9%, 2.83 V, 6.8 μV; E115.12 99.8%, 2.88 V, 6.6 μV.

[07-JUL-16] Poaching transmitters: all running 100% reception.

[08-JUL-16] We dipped batch E120.7-E121.6 in silicone last night. Today we apply a second coat to four of them. The second coat adheres well to three, but we see immediate wrinkling of the first coat on the transmitter that was not hanging directly over the warm water. We remove the silicone and find that the first coat is still tacky. We re-coat. We have E120.8 that won't turn on after encapsulation. Dissect. VB = 2.83 V. Disconnect battery, VB = 2.87 V. Connect external power supply. Inactive 1.6 μA. Active 78 mA. No reception. We see a weak RF spectrum centered on 913 MHz (−55 dBm peak compared to −35 dBm for functioning transmitter). We remove epoxy around the antenna, which has been forced over the base of the mounting wire during rotator encapsulation. We re-attach the antenna, but still no reception.

Poaching transmitters: E113.9 100.0%, 2.79 V, 7.1 μV; E113.10 100.0%, 2.84 V, 7.3 μV; E116.4 99.6%, 2.83 V, 4.4 μV; R114.10 99.6%, 2.82 V, 9.6 μV; R114.11 100.0%, 2.83 V, 8.0 μV; E115.12 100.0%, 2.88 V, 11.5 μV.

We have E117.4, which we formerly declared as noisy. We place it in warm water. It generates steps and rumble. At times, it generates excessive switching noise. We compare to E120.3, which is quiet with no rumble except during the first minute after insertion in water. We clean the beaker, replace the water, and change the temperature. Eventually, E117.4 starts generating a 1-Hz square wave. We record the entire history in M1468000711.ndf, with notes in the metadata.

[10-JUL-16] Poaching transmitters: reception 100%, battery voltages normal. Measure frequency response and note loss of gain at frequencies greater than 30 Hz for R114.10 and R114.11.

[11-JUL-16] Lot B61534 of 100 A3028RV3 circuits arrived here in June. We have calibrated 72 of them, and of these, 4 did not turn on. We examined these 4 and noted that one or more pads beneath U3 were not soldered, or U3 was off-center, or U3 was rotated. We attempt to replace U3, and succeed temporarily in one case, but later this circuit fails to turn on.

Poaching transmitters: E113.9 99.8%, 2.80 V, 9.0 μV; E113.10 100.0%, 2.83 V, 7.0 μV; E116.4 99.2%, 2.83 V, 4.0 μV; R114.10 99.0%, 2.80 V, 6.8 μV; R114.11 100.0%, 2.83 V, 7.0 μV; E115.12 100.0%, 2.88 V, 7.1 μV.

[15-JUL-16] We have batch B121.7-122.6 after one-day burn-in and three-day soak. These have three coats of silicone on top of rotator encapsulation with top-coat. We allowed 60 s for drip-off after vacuum, and 30 s for invert, then rotate. B121.9 has rust colored residue on the outside of the silicone around the positive battery terminal, where there are wrinkles in the silicone and a 1-mm long crack. After handling the transmitter a few times, we have rubbed off the residue. Two or three others have slight wrinkles, but the rest are perfect. Ten of them displace 12.5 ml, making their average volume 1.25 ml. Frequency response all within ±0.5 dB as in B121_5mV. Noise is ≤12 μV except for B121.9, B121.13, and B121.6. B121.9 has a conduit through the silicone for water to the battery terminal. The other two we find have the battery terminal protruding through the silicone. We resolve to bake the transmitters and apply another two coats of silicone.

We have batch R120.7-R121.6 after one-day burn-in and three-day soak. These have three coats of silicone and rotator encapsulation. R121.1 and R121.2 have wrinkles in their silicone. Four transmitters have volume 12 ml for average volume 3.0 ml. Frequency response within ±0.5 dB as in R120_5mV. R120.9 has rust residue near battery tab. It shows excessive noise. In others, noise is ≤10 μV except R120.12 which is 16 μV. Switching noise in 37°C is ≤4 μV except R120.12, which is 8 μV. We are holding back the two with wrinkles, as well as R120.9, which we will coat twice more.

Poaching transmitters: E113.9 100.0%, 2.79 V, 8.8 μV; E113.10 100.2%, 2.82 V, 10.2 μV; E116.4 100.0%, 2.83 V, 3.8 μV; R114.10 100.0%, 2.79 V, 7.5 μV; R114.11 94.9%, 2.81 V, 6890.9 μV; E115.12 97.1%, 2.88 V, 6.5 μV. R114.11 is producing nearly full-scale oscillations at 175 Hz even when we short the leads together. Measure frequency response with 50-Ω source. R114.10 and E115.12 within 2 dB of nominal. E116.4 has no VC lead. R114.11 oscillating. E113.9 and E113.10 gain within 2 dB of nominal except for 1-Hz 5-mV oscillation on E113.10. E113.9 has been running for 52 days.

[16-JUL-16] We have applied another coat of silicone to R119.11, R119.12, R120.7, R120.7, R120.10-14, R121.3-6, after finding the R120.9 had a breach at the battery terminal. We check noise after curing this coat, and all are less than 15 μV, with switching noise less than 4 μV. We apply two more coats of silicone to B121.7-122.6, so now they have 5 coats. Eight of them displace 11 ml, so their volume is 1.4 ml. We take B121.9, B121.13, and B122.6, which formerly had breaches in silicone around their battery tabs, and place in water at 38°C. Switching noise in B121.9 is 6 μV, in B121.13 is 8 μV, in B122.6 is 6 μV. Overall noise is B121.9 13 μV, B121.13 25 μV, B122.6 14 μV. We will ship B121.9 and keep back B122.6 and B121.13.

AUG-16

[03-AUG-16] Poaching transmitters: E113.9 100.0%, 2.74 V, 8.2 μV; E113.10 99.2%, 2.79 V, 7.6 μV; E116.4 99.2%, 2.80 V, 4.0 μV; R114.10 98.2%, 2.79 V, 7.0 μV; R114.11 100.0%, 2.94 V, 10810.4 μV; E115.12 95.3%, 2.85 V, 9.2 μV. We note that R114.11 continues to generate its own 55 Hz full-scale oscillation.

[08-AUG-16] Poaching transmitters: E113.9 99.0%, 2.72 V, 10.1 μV; E113.10 98.6%, 2.76 V, 7.3 μV; E116.4 98.8%, 2.79 V, 4.1 μV; R114.10 100.0%, 2.80 V, 7.6 μV; R114.11 99.8%, 2.80 V, 7432.8 μV; E115.12 100.0%, 2.85 V, 9.2 μV. Frequency response of E115.12 has lost gain, see here. E113.9 has been poaching for 76 days. Its gain above too low from 10-160 Hz and it generates occasional step artifacts.

We have batch B122.9-B123.5. These have four coats of silicone. Three have battery tabs that we have not filed down. Frequency response is B122_5mV, agree within ±0.6 dB. Noise in water at 37°C is ≤12 μV except B122.11 is 20 μV and B123.5 is 16 μV. Switching noise is ≤6 μV except B122.11 is 8 μV and B123.5 is 10 μV.

[09-AUG-16] We add B121.13 and B122.11 to our poaching transmitters. Both these we rejected for excessive switching noise at 37°C. Poaching transmitters: E113.9 97.7%, 2.66 V, 11.2 μV; E113.10 97.3%, 2.74 V, 11.8 μV; B122.11 100.0%, 2.76 V, 7.9 μV; B121.13 99.4%, 2.74 V, 8.1 μV; E116.4 98.2%, 2.78 V, 4.6 μV; R114.10 100.0%, 2.75 V, 8.4 μV; E115.12 100.0%, 2.84 V, 22.0 μV. We have lost E114.11 after

[10-AUG-16] Dissect E114.11. Silicone is so well adhered to epoxy that we have a hard time exposing epoxy for removal. Battery voltage 0.0 V. Disconnect battery, voltage rises to 2.5 V. Connect external power. Inactive current 2.0 μA. Active current 11±3 mA. We remove C4. Its resistance is 10 MΩ. Active current 11±1 mA. Remove C3, active current 85 μA. Resistance of C3 is 330 Ω, which is consistent with 11 mA Quiescent current from VD.

[11-AUG-16] We have batch D118.13, D121.7, D121.9, three A3028D-DAA transmitters of which we need to ship two. Each of these has a gold-plated D-pin on X, which are channel numbers 7, 9, and 13, and bare steel wires on C and Y input. We place in water at 37°C. Switching noise less than 3 μV. The three X inputs are fluctuating by 2 mVpp, while the Y inputs have noise less than 20 μV, see for eight-second interval.


Figure: Flux Residue Rumble. Three dual-channel transmitters in water. All three C electrodes are bare steel wires. Channels 7, 9, and 13 are gold pins soldered to the tip of the X lead. Channels 8, 10, and 14 are bare wires at the ends of the Y leads. After scrubbing the gold pins in hot water, the rumble decreases to below 100 μV on all channels.

We scrub all electrodes in hot water and return to the bath. We now see only 40 μV rms rumble, and it is present on all channels. When we tap the faraday enclosure, we see matching oscillations on all inputs. After ten minutes, noise is less than 15 μV except in D121.9, which still rumbles at 100 μVpp. Gain versus frequency is within ±0.4 dB, as in D121_5mV.

We have batch A118.7, 9, 11, A119.1, 3, 5, A121.11, 13, A122.1, 3, 5, 7, all dual-channel A3028A-DDC transmitters. Switching noise for all is less than 4 μV in warm water. Frequency response in two plots A119_5mV and A121_5mV.

[17-AUG-16] Poaching transmitters: E116.4 100.0%, 2.74 V, 4.6 μV; R114.10 100.0%, 2.73 V, 8.8 μV; E115.12 100.0%, 2.80 V, 18.0 μV; E113.9 96.3%, 2.65 V, 10.4 μV; E113.10 99.4%, 2.72 V, 16.1 μV; B122.11 98.4%, 2.72 V, 11.6 μV; B121.13 95.7%, 2.71 V, 13.9 μV.

[19-AUG-16] Poaching transmitters: E113.9 99.8%, 2.64 V, 11.4 μV; E113.10 99.8%, 2.70 V, 28.2 μV; B122.11 97.7%, 2.77 V, 10.6 μV; B121.13 100.0%, 2.75 V, 7.8 μV; E116.4 100.0%, 2.77 V, 4.3 μV; R114.10 100.0%, 2.74 V, 8.3 μV; E115.12 100.0%, 2.83 V, 27.2 μV.

[22-AUG-16] We have batch E123.6-E124.5 after burn-in and soak. These drained for 20 s and we added epoxy during rotation, followed by five coats of silicone. Frequency response E123_5mV agrees to ±1 dB. Switching noise in 37°C water is ≤5 μV. Total noise ≤14 μV.

[23-AUG-16] Poaching transmitters: E113.9 100.0%, 2.57 V, 7.5 μV; E113.10 100.0%, 2.73 V, 8.1 μV; B122.11 100.0%, 2.77 V, 10.3 μV; B121.13 100.0%, 2.74 V, 8.7 μV; E116.4 95.3%, 2.69 V, 4.5 μV; R114.10 94.1%, 2.25 V, 12.7 μV; E115.12 99.2%, 2.81 V, 311.5 μV. The E115.12 noise is absent when the device is out of water. There is a rust stain on the positive battery terminal. We can feel with our teeth a hole in the silicone over the battery tab. Frequency response of all devices is Poaching_5mV_23AUG16.png.


Figure: Battery Voltage versus Time for Recent Poached Transmitters.

[26-AUG-16] Poaching transmitters: E113.9 99.8%, 2.50 V, 10.2 μV; E113.10 99.8%, 2.77 V, 7377.0 μV; B122.11 100.0%, 2.55 V, 18.9 μV; B121.13 99.8%, 2.72 V, 8.0 μV; E116.4 100.0%, 2.54 V, 4.9 μV; E115.12 100.0%, 2.83 V, 58.3 μV. We have lost R114.10. Transmitter E113.10 is generating its own square wave.

[29-AUG-16] We have batch of sixteen transmitters with numbers in the range R123.9-R126.10. Gain within ±0.6 dB. In water at 37°C, switching noise <8 μV, total noise <20 μV after five minutes. The two noisiest transmitters were R126.1 and R125.1, both of which had 8 μV switching noise and 19 μV rms total noise.

[30-AUG-16] Dissect R114.10. Silicone well adhered. No discoloration of any parts. VB = 50 mV. Disconnect VB = 2.4 V. Connect external 2.6 V. Inactive current 1.9 μA. Active 2.0 mA falling to 1.0 mA over following few minutes. Reception 100%, picks up mains hum. We remove C6 and C4 but active current remains 1.0 mA. Remove C3, the 1.0 nF decoupling capacitor for U9 and current drops back to normal. Diagnosis: corroded capacitor C6.

Poaching transmitters: E113.9 100.0%, 1.92 V, 14.6 μV; E113.10 99.4%, 2.87 V, 9308.9 μV; B122.11 99.8%, 2.49 V, 13.4 μV; B121.13 99.0%, 2.67 V, 14.9 μV; E116.4 100.0%, 2.74 V, 4.1 μV; E115.12 100.0%, 2.80 V, 66.2 μV. After 97 days poaching, E113.9 appears to be exhausting its battery. E113.10 generates its own 66 Hz oscillation.

We have batch B125.3-125.14, all A3028B-CC. Three have wrinkles in their silicone, but not so severe as to preclude shipping the devices. One has a cavity in the silicone near the drip that should be filled before shipping. These all have 4.5 coats of silicone. There is no sign of corrosion around the battery terminals. We measure frequency response B125_5mV, all agree to ±0.3 dB. In water at 34°C total noise is ≤16 μV and switching noise is ≤8 μV.

SEP-16

[01-SEP-16] Poaching transmitters: E113.9 97.5%, 2.28 V, 8.0 μV; E113.10 98.0%, 2.86 V, 9890.6 μV; B122.11 98.4%, 2.56 V, 16.9 μV; B121.13 98.8%, 2.56 V, 16.1 μV; E116.4 100.0%, 2.73 V, 4.8 μV; E115.12 100.0%, 2.81 V, 12.1 μV.

[05-SEP-16] We have batch E126.3-127.12 consisting of fourteen transmitters. Four of these we dropped during epoxy dipping E127.1, E126.8, E127.4, and E126.13. Their leads are bent and their encapsulation is unattractive. Frequency response E126_5mV within ±0.5 dB. Total noise less than 12 μV. Switching noise in water at 40°C is ≤5 μV. We add to our poach R124.9, left-over from a previous rotator-encapsulated batch, and E126.8, E127.1, E127.11, those with damaged appearance from the most recent batch.

Poaching transmitters: E113.9 99.8%, 1.95 V, 7.5 μV; E113.10 100.0%, 2.86 V, 10173.1 μV; E127.1 98.6%, 2.72 V, 7.8 μV; E116.4 98.4%, 2.65 V, 4.8 μV; E126.8 97.7%, 2.72 V, 8.9 μV; R124.9 96.5%, 2.72 V, 8.4 μV; E127.11 99.6%, 2.71 V, 7.6 μV; E115.12 93.8%, 2.74 V, 255.7 μV. We have lost B121.13 and B122.11.

[09-SEP-16] Poaching transmitters: E113.9 100.0%, 2.51 V, 7.4 μV; E127.1 95.5%, 2.79 V, 13.2 μV; E116.4 100.0%, 2.72 V, 4.3 μV; E126.8 98.0%, 2.79 V, 8.9 μV; R124.9 100.0%, 2.79 V, 12.7 μV; E113.10 95.1%, 2.78 V, 10291.6 μV; E127.11 93.8%, 2.80 V, 8.2 μV; E115.12 94.1%, 2.79 V, 10.2 μV.

[13-SEP-16] Poaching transmitters: E113.9 99.8%, 1.91 V, 7.9 μV; E127.1 98.8%, 2.80 V, 7.0 μV; E116.4 98.6%, 2.70 V, 4.7 μV; E126.8 97.7%, 2.80 V, 7.0 μV; R124.9 98.8%, 2.80 V, 6.7 μV; E113.10 95.5%, 2.85 V, 10365.1 μV; E127.11 97.5%, 2.81 V, 7.0 μV; E115.12 93.9%, 2.78 V, 10.8 μV.

We have batch E127 with epoxy touch-up on a few transmitters, five coats of silicone. Frequency response is E127_5mV. Gain is within ±0.6 dB. Total noise ≤12 μV. Switching noise less than 5 μV in water at 37°C.

We have batch E128 with epoxy touch-up on a few transmitters, five coats of silicone. Frequency response is E128_5mV. Gain is within ±0.4 dB. Total noise ≤12 μV. Switching noise less than 4 μV in water at 32°C.

We have batch B129 with top-coat and 4.5 coats silicone. Frequency response is B129_5mV. Gain is within ±0.3 dB. Total noise ≤12 μV except B130.9, which has total noise 19 μV. Switching noise less than 6 μV in water at 40°C. The silicone on B130.9 and B129.14 is wrinkled. We will hold them back for poaching.

[16-SEP-16] We are experimenting with alternatives to the MED10-6607 silicone, which we use for both dipping and lead-making. We make leads with 4 coats of MED10-6607 to our springs. We dip our mouse transmitters 4.5 times in MED10-6607 and our rat transmitters 5 times. The key features of silicone for our purposes are tensile strength, viscosity, elongation, and hardness.

Silicone Tensile
Strength
(psi)
Viscosity
(cps)
Hardness
(Durometer
Shore A)
Elongation Results
MED10-6607 900 5,500 40 650% Restricted medical grade coating. Produces strong, tough coating
after five dips, cannot cut through with finger-nail, does not yield
to finger-nail, produces flexible leads after four coats, silicone
stretches more than spring.
MED-6607 980 6,500 40 600% As MED10-6607, unrestriced medical grade silicone.
SS-5001 325 35,000 25 325% Produces strong, tough, coating after one dip,
cannot cut with finger-nail, but yields to finger-nail,
insufficient elongation for leads.
SS-5005 145 5,000 5 700% Produces super-flexible leads after three coats,
silicone stretches more than the spring.
HumiSeal IC49 40 9,000 ≈20 100% Breaks up under fingernail. Insufficient elongation for leads.
Table: Comparison of Likely Silicones for Encapsulation and Lead-Making. We include MED-6607 here because we end up using it later as a final coat for all leads and transmitter bodies, it being approved for permanent implants in animal bodies.

We have R129.2 encapsulated with three coats of SS-5001, which is roughly the thickness of eight coats of MED10-6607. The silicone is more flexible than MED10-6607, but still too tough to cut through with a finger-nail. We have two leads made with three coats of SS-5005. They are as flexible as the springs without any silicone. When we stretch them, the silicone extends until the spring is roughly five times its original length, and then tears, which is similar to the performance of the MED10-6607. We will be making batches of transmitters dipped in SS-5001 and equipped with leads coated in SS-5005 and shipping them to our collaborators in the up-coming months.

We add R129.2 to our poach. Its common lead tore off during epoxy encapsulation and we soldered it back on again at the circuit board with the help of acid flux. Poaching transmitters: E113.9 99.8%, 2.34 V, 17.7 μV; E127.1 96.5%, 2.78 V, 7.9 μV; R129.2 94.3%, 2.67 V, 6.4 μV; E116.4 98.2%, 2.68 V, 4.9 μV; E126.8 99.6%, 2.77 V, 7.7 μV; R124.9 100.2%, 2.77 V, 6.9 μV; E113.10 94.5%, 2.81 V, 9844.4 μV; E127.11 100.0%, 2.79 V, 6.9 μV; E115.12 98.8%, 2.75 V, 184.4 μV.

[21-SEP-16] Poaching transmitters: E113.9 100.0%, 2.36 V, 7.9 μV; E127.1 95.3%, 2.82 V, 7.1 μV; R129.2 99.0%, 2.81 V, 7.0 μV; E116.4 97.5%, 2.63 V, 4.8 μV; E126.8 94.7%, 2.81 V, 16.0 μV; R124.9 99.6%, 2.79 V, 37.2 μV; E113.10 92.8%, 2.39 V, 7877.6 μV; E127.11 99.8%, 2.82 V, 7.8 μV; E115.12 99.4%, 2.71 V, 139.0 μV.

[23-SEP-16] We have A132 encapsulated in epoxy and 4.5 coats of MED10-6607. Because of a rush schedule, we have not soaked these for three days. We soak them for an hour until electrode rumble has subsided. Measure noise at 37°C. Total noise ≤12 μV, switching noise ≤4 μV. Frequency response A132_5mV and A132_5mV_2. Gain within ±0.5 dB across the set.

Poaching transmitters: all running with ≥98% reception. Photograph below shows the greater flexibility of leads made with SS-5005 silicone.


Figure: Flexibility Comparison. Blue lead four coats of MED10-6607. Clear lead three coats of SS-5005. The two are the same diameter, but the blue lead has its own shape from the stiffness of the silicone, while the clear lead falls almost vertically with the weight of the spring inside.

[26-SEP-16] Poaching transmitters: all running with ≥98% reception except E113.9, which has stopped. We remove it from water and place it on our work bench. We try to turn it on, but it does not respond.

[27-SEP-16] Dissect transmitter E113.9. No sign of rust except at the tip of the cut-off rotator mounting wire. Silicone well-adhered to epoxy. Battery voltage 1.8 V. Disconnect, battery 2.3 V. Connect external 2.6 V. Inactive current 17.2 μV. Active current 120 μV. Wash and dry. Inactive 12 μA, active 95 μA. Reception 100%. After a while, current rises to 5 mA, then back down to 95 μA. This transmitter ran for a total of 3000 hours. Its expected operating life is 3200 hours. Our diagnosis is "corroded capacitor".

We have batch R129 consisting of R128.5-129.6, encapsulated in silicone on the rotator, dipped five times in MED10-6607. Of these, five have significant wrinkles in their silicone. Frequency response of the R128 devices is R128_5mV, gain variation ±0.5 dB, switching noise in hot water is ≤5 μV, total noise ≤12 μV. Frequency response of the R129 devices is R128_5mV, gain variation ±0.8 dB. Switching noise in hot water is ≤5 μV and total noise ≤12 μV with the exception of R129.7, which has noise 100 μV. Later, the noise drops to 50 μV but we see persistent rumble, even though the leads have only bare wires at the end. We clean and examine all the F-pins and B-screws. We re-solder one screw electrode. Before cleaning, they have some greenish residue on the solder joints. After cleaning, they are shiny.

Poaching transmitters: E127.1 62.3%, 2.82 V, 7.0 μV; R129.2 98.4%, 2.83 V, 7.0 μV; E116.4 65.6%, 1.98 V, 21.4 μV; E126.8 100.0%, 2.81 V, 7.7 μV; R124.9 90.0%, 2.80 V, 7.1 μV; E127.11 99.8%, 2.83 V, 7.3 μV; E115.12 99.4%, 2.75 V, 226.6 μV. Frequency response Poaching_5mV_27SEP16.

We have lost E113.10. Dissect. We peel the silicone off easily. No discoloration or rust. Battery voltage 0.4 V. Disconnect, battery 0.4 V. Apply external 2.6 V. Inactive 6 μA, active 90 μA. This transmitter ran for 3000 hours, compared to a natural life of 3200 hrs. Our diagnosis is "full life".

Device E116.4 no longer transmits. Its VA was earlier 1.98 V and reception 65%. Dissect. Silicone peels off easily. Disconnect battery, voltage is 2.6 V. Connect external 2.6 V. Inactive current 1.7 μA. Active current 85 μA. Diagnosis "resistive switch" failure.

[30-SEP-16] We have the first half of batch E130 consisting of seven A3028E-AA. Frequency response E130_5mV consistent to ±0.4 dB. Total noise in water at 43°C is ⋚12 μV, switching noise <4 μV. We check noise in batch A132, the six that remain here after four-day soak, and find that noise is <14 μV and switching noise is <4 μV.

Poaching transmitters: E127.1 85.5%, 2.79 V, 7.2 μV; R129.2 88.7%, 2.79 V, 6.9 μV; R129.5 91.2%, 2.74 V, 8.6 μV; R129.7 100.0%, 2.72 V, 119.7 μV; E126.8 99.8%, 2.78 V, 8.0 μV; R124.9 93.9%, 2.76 V, 7.0 μV; E127.11 89.5%, 2.79 V, 7.3 μV. B130.9 97.3%, 2.55 V, 75.0 μV; B129.14 100.0%, 2.68 V, 70.8 μV.

We have lost E115.12. Dissect. Wrinkled silicone comes away easily. Battery voltage 0.5 V. Disconnect, 1.1 V. Apply external 2.6 V. Inactive 2.4 μA, active 81 μA. Diagnosis "unidentified drain". We add E129.5 (wrinkles in silicone), E129.7 (noisy), B129.14 (wrinkles), and B130.9 (wrinkles). We note that the A3028Bs both appear noisy just after we put them in their jar.

OCT-16

[03-OCT-16] We have the second half of batch E130 consisting of eight A3028E-AA coated in SS-5001 with leads insulated in SS-5005. Frequency response is E130_5mV. Total noise in water at 45°C <11 μV, switching noise <4 μV.


Figure: A3028E-AA Coated with SS-5001 and equipped with leads insulated with SS-5005.

We had to clip excess silicone off the side opposite the leads, which is the side the silicone drains off when we hang them over hot water. We see no rust around any of the battery terminals or cut-off mounting wire ends after a five-day soak.

[11-OCT-16] We have batch A133.3-136.7, a set of fourteen A3028A-DCC. Bandwidth of Y-inputs is only 80 Hz. We neglected to switch the 2-nF capacitors in the Y-channel EEG amplifier for 1-nF. All of them fail QC for this reason. Four more have wrinkles in silicone. Another was running at the end of the five-day soak. Frequency response A135_5mV for first set of seven, noise ≤12 μV, switching noise ≤ 3 μV at 30°C.

Poaching transmitters: E127.1 90.8%, 2.77 V, 12.3 μV; R129.2 95.3%, 2.84 V, 7.1 μV; R129.5 95.3%, 2.84 V, 8.3 μV; E126.8 92.6%, 2.80 V, 13.1 μV; R124.9 96.9%, 2.79 V, 8.2 μV; E127.11 100.0%, 2.81 V, 8.0 μV; B130.9 100.0%, 2.71 V, 23.6 μV; B129.14 100.0%, 2.73 V, 19.2 μV. We have lost R129.7. There is rust around the base of the antenna. There is a breach in the antenna insulation. After a few minutes drying out, it turns on and we get 100% reception with average value 655 counts. When we touch the EEG leads we see full-scale response. We put R129.7 in the oven to dry out.

[13-OCT-16] Continuing batch A133.3-136.7, second set of seven has frequency response A136_5mV. Noise in A136.1 is 11 μV in the 0.3-160 Hz X but 20 μV in the 0.3-80 Hz Y. Switching noise on X is 1.6 μV, bu on Y is 6 μV. Total noise for the others is less than 12 μV, although in A133.5 we see persistent rumble and twice as much switching noise on Y as X.

[14-OCT-16] Our Quality Assurance (QA) procedure now includes an inspection of the amplifier frequency response. We set our function generator to perform a two-second sweep from 1-1000 Hz, apply to the amplifier with 50-Ω source impedance, and observe the amplitude envelope in the Recorder Instrument. This is a quicker version of the test we used to perform with the Neuroarchiver, see here. We will use this test for checking the frequency response of poaching transmitters also. Our QA now includes: inactive current consumption, active current consumption, transmit clock period, radio frequency spectrum, reception of messages, frequency response, inspection of board surface for flux residue, inspection of wire solder joints, and inspection of electrode solder joints.

Poaching transmitters: E127.1 98.0%, 2.80 V, 6.9 μV; R129.2 99.8%, 2.83 V, 8.5 μV; R129.5 99.8%, 2.85 V, 9.6 μV; R129.7 100.0%, 2.85 V, 6.2 μV; E126.8 95.5%, 2.80 V, 8.0 μV; R124.9 94.3%, 2.78 V, 7.5 μV; E127.11 94.7%, 2.82 V, 8.1 μV; B130.9 100.0%, 2.69 V, 12.8 μV; B129.14 100.0%, 2.72 V, 10.9 μV. We have R129.7 back in the poach after drying out for two days and re-coating with silicone to seal the antenna insulation. Check frequency response of all devices with 50-Ω source, all good.

We have batch E133.8-E134.14. These have five coats of MED10-6607 and leads made with three coats of SS-5005. E133.14 has damage at the end of one lead. We cut back to 10 cm and expose tips. Frequency response E133_5mV, gain within ±0.5 dB. Total noise <12 μV except E134.4, which is 14 μV. Switching noise <7 μV in water at 38°C.

In the P3028A12 firmware, we add support for set set identifiers, as well as for more versions of the A3028.

[18-OCT-16] Poaching transmitters: E127.1 95.9%, 2.80 V, 7.4 μV; R129.2 99.2%, 2.84 V, 6.6 μV; R129.5 98.6%, 2.84 V, 6.8 μV; E126.8 99.0%, 2.80 V, 7.6 μV; R124.9 91.2%, 2.78 V, 6.7 μV; E127.11 98.2%, 2.82 V, 7.0 μV; B130.9 100.0%, 2.48 V, 10.6 μV; B129.14 100.0%, 2.70 V, 19.1 μV. We have lost E129.7 again, despite drying it out and re-coating antenna. Diagnosis: Faulty Encapsulation.

[21-OCT-16] We have transmitters T7 and A134.13 made with one coat of new silicone coating after soaking for three days. Noise is normal, frequency response as we expect. A134.13 is faulty in that its bandwidth is only 80 Hz on the Y input.

Poaching transmitters: E127.1 95.5%, 2.79 V, 7.2 μV; R129.2 100.0%, 2.83 V, 7.2 μV; R129.5 98.4%, 2.84 V, 6.4 μV; E126.8 95.1%, 2.79 V, 7.3 μV; R124.9 95.9%, 2.80 V, 7.5 μV; E127.11 94.1%, 2.81 V, 7.2 μV; B130.9 98.6%, 2.54 V, 16.4 μV; B129.14 98.6%, 2.63 V, 16.4 μV.

[24-OCT-16] Circuit problems for build B63525, 100 of A3028RV3 assemblies: 3 of logic signals not reaching DAC resistors, 3 of U3 missing solder joints.

We have batches E136 and E137, encapsulated with SS-5001 silicone, equipped with leads insulated with three coats of SS-5005 silicone. Batch E136 noise <14 μV, switching <4 μV at 32°C, frequency response E136_5mV, gain within ±0.4 dB.Batch E137 noise ≤11 μV, switching ≤4 μV at 32°C, frequency response E137_5mV, gain within ±0.4 dB.

[27-OCT-16] Poaching transmitters: E127.1 99.8%, 2.79 V, 13.0 μV; R129.2 99.8%, 2.83 V, 8.4 μV; R129.5 99.8%, 2.83 V, 12.2 μV; E126.8 92.2%, 2.78 V, 10.4 μV; R124.9 93.2%, 2.79 V, 6.9 μV; E127.11 95.5%, 2.81 V, 8.0 μV. We have lost B129.14 and B130.9. Last check was 22 days, this is 28 days, so diagnosis is "Full Life". We raise oven temperature to 80°C and continue poaching the above transmitters, but add to them a set of 7 A3028A transmitters. New poaching transmitters: A135.1 98.2%, 2.61 V, 36.2 μV; A135.3 89.1%, 2.59 V, 17.5 μV; A133.5 88.3%, 2.66 V, 47.3 μV; A135.7 95.9%, 2.59 V, 14.7 μV; A135.9 99.4%, 2.60 V, 17.4 μV; A134.11 93.2%, 2.62 V, 16.1 μV; A134.13 99.8%, 2.57 V, 12.5 μV.

[28-OCT-16] Poaching transmitters: E127.1 100.0%, 2.88 V, 6.3 μV; R129.2 96.3%, 2.93 V, 6.4 μV; R129.5 97.5%, 2.93 V, 5.6 μV; E126.8 91.0%, 2.88 V, 6.5 μV; R124.9 92.0%, 3.07 V, 107.0 μV; E127.11 96.5%, 2.91 V, 6.6 μV; A135.1 92.8%, 2.68 V, 42.6 μV; A135.3 93.4%, 2.68 V, 21.7 μV; A133.5 93.9%, 2.74 V, 12.2 μV; A135.7 92.4%, 2.68 V, 11.1 μV; A135.9 93.9%, 2.68 V, 9.3 μV; A134.11 94.5%, 2.72 V, 17.8 μV; A134.13 98.8%, 2.66 V, 12.1 μV. We note that E124.9 is rumbling around. Measure frequency response. E124.9 shows some loss of gain at 130 Hz, in addition to rumble. All others (19 EEG amplifiers in all) give response within 2 dB of nominal with 50-Ω sweep.

[31-OCT-16] Poaching transmitters: E127.1 90.8%, 2.85 V, 12.6 μV; R129.2 82.6%, 2.92 V, 9.8 μV; R129.5 53.7%, 2.93 V, 22.7 μV; E126.8 85.2%, 2.87 V, 12.0 μV; R124.9 20.1%, 2.87 V, 12.0 μV; E127.11 78.7%, 2.90 V, 26.1 μV; A135.1 81.6%, 2.76 V, 12.6 μV; A135.3 85.7%, 2.74 V, 16.0 μV; A133.5 78.3%, 2.79 V, 10.9 μV; A135.7 80.9%, 2.72 V, 14.0 μV; A135.9 87.7%, 2.73 V, 11.2 μV; A134.11 81.4%, 2.78 V, 14.0 μV; A134.13 86.5%, 2.70 V, 16.1 μV. When we take the transmitters out of their 80°C water and let them cool down, reception improves, but noise increases.

NOV-16

[01-NOV-16] We have batch L140, six of the new A3028L-AAA transmitter, 1024 SPS on each channel, 0.3-320 Hz bandwidth, encapsulated with epoxy on our rotator and coated with SS-5001. We made eight but only six made it to QA, two of them failing before encapsulation. Five of them together have volume 33 ml including antennas. Frequency response L140_5mV within ±0.6 dB. Noise in water at 37°C is 9-15 μV for 320-Hz bandwidth, switching noise ≤2 μV.

We have batch A138 consisting of 7 of A3028A-DDC with one coat of SS-5001 silicone. Frequency response is A138_5mV within ±0.6 dB. In water at 37 °C switching noise is ≤4 μV. For twenty minutes we leave the transmitters running and we see persistent rumble on a number of the channels. We take out ans scrub the electrodes. Rumble amplitude is now reduced by a factor of four but persists. In several channels it is 200 μV rms. Apart from the rumble, noise is ⋚12 μV. Batch A139 consists of 7 of A3028A-DDC A139_5mV, gain within ±0.3 dB. In water at 37°C we see rumble of 200 μV even after scrubbing all electrodes. Noise other than rumble is ≤15 μV and switching noise ≤3 μV, except for A137.13, which has noise 25 μV.

Poaching transmitters all running. Apply 50-Ω sweep. All A3028As are running perfectly on both channels after 5 days at 80°C, which is equivalent to at least 300 days worth of corrosion at 37°C. R129.2 has +3 dB gain at around 100 Hz. E124.9, E126.8, and E127.11 have gain dropping from normal at 10 Hz to zero at 160 Hz. E129.5 and E127.1 have normal gain.

[03-NOV-16] Poaching transmitters: E127.1 92.6%, 2.97 V, 10048.9 μV; R129.2 96.1%, 2.91 V, 14.9 μV; R129.5 99.2%, 2.92 V, 13.6 μV; E126.8 99.4%, 2.85 V, 10.8 μV; R124.9 78.5%, 3.23 V, 39.3 μV; E127.11 99.0%, 1.82 V, 17.4 μV; A135.1 91.4%, 2.69 V, 23.2 μV; A135.3 93.8%, 2.70 V, 14.1 μV; A133.5 80.5%, 2.56 V, 11.1 μV; A135.7 97.7%, 2.64 V, 16.9 μV; A135.9 95.1%, 2.41 V, 39.4 μV; A134.11 93.4%, 2.75 V, 10.6 μV; A134.13 89.5%, 2.68 V, 11.4 μV. For all the dual-channel transmitters, both channels have the same average value, including A135.9, which shows 2.41 V today. Transmitter E127.1 is oscillating at 66 Hz full-scale.

[04-NOV-16] Poaching transmitters: E127.1 96.9%, 2.69 V, 48.5 μV; R129.2 100.0%, 2.70 V, 626.9 μV; R129.5 99.4%, 2.79 V, 8.1 μV; R124.9 100.0%, 3.15 V, 124.4 μV; E127.11 100.0%, 101.47 V, 921.6 μV; A135.1 94.3%, 2.47 V, 12.9 μV; A135.3 91.8%, 2.52 V, 46.7 μV; A133.5 100.0%, 1.93 V, 11.8 μV; A135.7 94.3%, 2.40 V, 157.0 μV; A135.9 96.1%, 1.91 V, 9.1 μV; A134.11 95.7%, 2.56 V, 9.4 μV; A134.13 94.9%, 2.49 V, 14.1 μV. We have E134.11 with average X 46k and Y 37k. We have matching rumble on A135.7 both channels. A135.5 and A135.9 appears to be running down their batteries. E129.2 high gain at 100 Hz. We have lost E126.8. E127.11 generating 1-Hz full-scale oscillations.

Dissect E126.8 after leaving it on bench for three hours. Disconnect battery, VB = 2.7 V. Connect external 2.7 V. Inactive 2.2 μA, active 86 μA. Reception 100%, picks up mains hum. Amplifier gain 6 dB too low at 1 Hz and 20 dB too low at 160 Hz. Diagnosis "Temporary Shutdown".

[07-NOV-16] Poaching transmitters: E127.1 94.7%, inf V, 0.0 μV; R129.2 100.0%, 2.82 V, 5.7 μV; R129.5 97.1%, 2.85 V, 19.0 μV; E127.11 100.0%, inf V, 0.0 μV; A135.1 93.2%, 2.31 V, 9.4 μV; A135.3 90.6%, 2.61 V, 41.1 μV; A133.5 98.4%, 2.32 V, 10.8 μV; A135.7 93.9%, 2.37 V, 148.8 μV; A134.11 85.5%, 2.62 V, 11.4 μV; A134.13 95.9%, 2.57 V, 21.7 μV. Transmitter A135.9 started transmitting only when we removed it from its hot water and started to dissect. Silicone looks good, but tears easily. We have 1-Hz pulses on X, mains hum on Y, and VA = 2.25 V for both. Dissect. Disconnect battery, VB = 2.7 V. Connect external 2.6 V. Inactive 2.2 μA, active 138 μA. Diagnosis, "Resistive Switch" failure. We have lost R124.9. Dissect. Silicone still tough, VB = 0.6 V. Disconnect battery VB = 2.5 V. Connect external 2.6 V, inactive 2.2 μA, active 1.0 mA. Reception 100%, X is a full-scale 0.5-Hz square wave. Current decreases to 0.6 mA, then jumps to 1.6 mA. Diagnosis "corroded capacitor". Transmitters E127.1 and E127.11 have X = 0.

[09-NOV-16] Poaching transmitters: R129.2 99.8%, 2.84 V, 67.6 μV; R129.5 99.0%, 2.92 V, 10.2 μV; E127.11 100.0%, inf V, 0.0 μV; A133.5 96.9%, 2.15 V, 50.8 μV. Dissect E127.1. Silicone tough and well-adhered to epoxy. VB = 1.1 V. Disconnect, VB = 2.3 V. Connect external 2.6 V, inactive 2.5 μA, active 1.7 mA. Remove C5, active 46 μA, but not functioning. Diagnosis "corroded capacitor". Dissect A135.7. No discoloration. Silicone tears easily. VB = 1.1 V. Disconnect, VB = 2.1 V. Connect external 2.6 V, inactive 2.2 μA, active 138 μA. Picks up mains hum. Diagnosis "Unidentified Drain". Dissect A134.11. No discoloration, silicone tears easily. VB = 1.0 V. Disconnect, VB = 1.7 V. Connect external 2.6 V, inactive 100 mA. Remove C2, inactive 2.2 μA, active 103 μA, no transmission. Diagnosis "corroded capacitor". Dissect A135.3. No discoloration, silicone tears easily. VB = 0.9 V. Disconnect, VB = 0.9 V. Connect external 2.6 V, inactive 100 mA. Remove C2, inactive 1 mA. Diagnosis "corroded capacitor". A135.1 now running again. Dissect. No discoloration, silicone tears easily. VB = 2.2 V. Diagnosis "Temporary Shutdown". Dissect A134.13. No discoloration, silicone tears easily. VB = 0.9 V. Disconnect VB = 0.9 V. Connect external 2.6 V. Inactive 10 μA. Active 138 μA. Picks up mains hum, reception 100%. Diagnosis "Unidentified Drain".

[10-NOV-16] We have batch E141 consisting of 15 A3028E-AA. We note what appears to be separation of the silicone from the epoxy around the battery tabs on several transmitters. After three days soaking in water, we see no signs of corrosion. In water at 37°C switching noise is ≤4 μV and total noise is ≤12 μV. Battery voltage 2.61-2.63 V. Reception for 14 of them in water over ten minutes with two pick-up antennas 91-98%. Reception versus time for four transmitters in E141_Rx. Frequency response is E141_5mV, within ±0.6 dB.

We have 100 of A3028RV3, build B63806. Some circuits have de-panelization damage. Al have no-clean flux residue on connectors, and all C7 and C12 have the wrong value, perhaps 1.0 nF or 2.0 nF. We replace C7 on one board and it gives correct frequency response. We will replace C7 and C12 by hand during quality assurance for all 100 of these boards.

[11-NOV-16] Poaching transmitters: R129.5 82.8%, 138.08 V, 12.7 μV; E127.11 98.8%, inf V, 0.0 μV. E129.5 has 0.5-Hz full-scale oscillations. E127.11 reports only zeros. A133.5 has stopped. Dissect. VB = 1.8 V. Disconnect VB = 2.4 V. Connect external 2.6 V. Inactive 2.3 μA. Active 139 μA. Reception 100%. Gain 2 to 10 dB too low with increasing frequency. Diagnosis "Full Life". After trying repeatedly to turn on R129.2, we dissect. Battery voltage 3.2 V. The circuit now turns on. Disconnect battery. Inactive 2.3 μA. Active 87 μA. Gain too low by 6-20 dB in pass-band. Diagnosis "Temporary Shutdown".

[15-NOV-16] Poaching transmitters: R129.5 98.4%, full-scale 1 Hz oscillations. We have lost E127.11. Dissect. VB = 0.5 V. Disconnect, VB = 0.5 V. Connect external 2.6 V. Inactive 2.3 μA, active 95 μA. Reception 100%, 1-Hz oscillations on X correspond to fluctuations in quiescent current. Diagnosis "Unidentified Drain".

[18-NOV-16] Poaching transmitters: R129.5 100%, full-scale 1 Hz.

[22-NOV-16] We have batch B142 consisting of 8 A3028B-CC encapsulated with one coat of SS-5001. Frequency response B142_5mV within ±0.2 dB. Two have excessive noise in 37°C water. B142.3 has switching noise 6μV and total noise 20 μV. B142.5 has switching noise 8 μV and total noise 18 μV. B142.5 has waves of milky-white under the silicone on the amplifier side. B142.4 has a little of the same.

Poaching transmitters: we have lost R129.5. Dissect. Silicone comes away easily from epoxy. Epoxy shiny as new. VB = 1.0 V. Disconnect VB = 1.6 V. Connect external 2.7 V. Inactive 3.3 μA. Active 130 μA. Picks up mains hum. No oscillations. Current rises to 160 μA. Gain vs frequency 20 dB too low at 10 Hz, 30 dB too low at 100 Hz. Active current now 108 μV. Diagnosis "Unidentified Drain".

[29-NOV-16] We have batch E144.3, E144.10, and E144.14 after five-day soak. No sign of corrosion. Gain within ±0.1 dB, frequency response E144_5mV. Noise in 37°C water is 9, 20, and 7 μV respectively. Transmitter E144.10 has switching noise 5 μV with harmonics. We see no sign of any breach in the silicone. We have sample batch E145.1 (Set 4), B145.2 (Set 1), L145.3 (Set 2), and B145.5 (set 3). Noise is ≤12 μV except B145.5, which has 15 μV and switching noise 6 μV. Frequency response L145_5mV. We have batch C143 consisting of C143.1-14 after five-day soak. We scrub electrode screws and place in 37°C water for half an hour. We see persistent rumble in all channels. Noise in 2-100 Hz is ≤17 μV. Switching noise is <3 μV except for C143.2 6 μV, C143.10 8 μV and C143.14 8 μV. Frequency response C143_5mV within ±0.4 dB except C143.10 which is 2 dB too low through the pass band.

DEC-16

[02-DEC-16] We have been studying the switching noise in our EEG amplifiers, first observed in May 2011 for the A3019D circuit, which used the A1171 magnetic sensor, and again in September 2013 for the A3028A, which uses the SL353LT magnetic sensor. Today we start with an A3028F circuit board with no encapsulation and a battery plugged into its programming extension. We see no switching noise. We cover the bottom side of the board with epoxy, then the top side. No switching noise. We remove C1 on the programming extension. Switching noise is <1.0 μV. We take transmitter C143.10, which we rejected for is excessive switching noise, and we disconnect its positive battery terminal, and connect an external battery. We heat up the circuit to at least 60°C and place it in a faraday enclosure with its leads in water. Switching noise is <1.0 μV as it cools back to room temperature. Noise in 1-160 Hz is 8 μV rms, where it was previously 20 μV rms. We re-connect its own battery. Switching noise amplitude is 11 μV and noise in 1-160 Hz is 13 μV rms.

We take another A3028F circuit with no input leads, all three input terminal soldered together. Switching noise with battery on programming extension is 0.4 μV rms. Load battery onto circuit board, heat up to 80°C, turn on and allow to cool down in faraday enclosure. Switching noise is ≤0.7 μV in both channels and total noise is <7 μV in 1-400 Hz. Encapsulate with epoxy. Heat to 80°C while epoxy still liquid. Allow to cool down. Switching noise remains <0.7 μV rms. Clip off programming extension. Still no switching noise. Cover in epoxy in a petri dish. Evacuate air. Heat to 80°C and allow to cool down in faraday enclosure. No switching noise. Go back to C143.10, cut off its leads and solder their bases together. Switching noise is 9 μV rms.

We tried to generate switching noise with two other circuits but failed. Most circuits do not generate switching noise even when encapsulated, so we may have to keep trying more circuits. But we stopped switching noise in C143.10 by using an external battery. The switching noise returns if we re-connect the epoxied battery. We wonder how A137.13 can generate 5 μV rms switching noise in X and less than 1.5 μV rms in Y.

[04-DEC-16] We take C143.10 and attach an external BR1225 battery. We see 1.5 μV of switching noise. We re-attach its internal BR1225 and see 8.1 μV. We try another external BR1225 and see 1.3 μV. We return to the first external BR1225 and see 5.1 μV. We are re-arranging the battery wires during these measurements. We apply external BR2477 and see no switching noise. We have E56.2 from two years ago. It's switching noise is 8 μV at 22Hz running off its own BR2330 battery. We remove silicone. Switching noise is the same. We disconnect battery and connect external BR2477. No switching noise. We attach external BR1225 and see 1.5 μV switching noise.

[06-DEC-16] We have a transmitter in our experiment box with channel No13. In water, its switching noise is 12 μV rms. Out of water, but with only the leads in water, switching noise is 12 μV rms. With wires connected in air, the noise is 0.9 μV rms. With external BR2477 and leads in water, 0.6 μV. With external BR2477 in parallel with the internal battery, leads in water, 3.3 μV rms. Rotate transmitter 0.9 μV rms. Rotate again: circuit down on rim of petri dish with leads into water, 15 μV. It turns out that the tip of the antenna is not insulated, and when it enters the same water as the leads, we get 15 μV, otherwise 1.5 μV.

We have E145.10 and E146.10. Both have switching noise less than 1 μV and total noise less than 8 μV. Frequency response included in E144_5mV.

[16-DEC-16] We have batch B146 consisting of 6 transmitters. Frequency response B146_5mV within ±0.3 dB. Switching noise in water at 38°C ≤4.5 μV rms. (Here we are calculating the rms amplitude of the fundamental component of the switching noise by looking at all components within ±1 Hz of the peak.) Total noise ≤16 μV rms.

We have C111.3 back from ION/UCL. They say it did not work. We turn it on. Frequency response is within 0.2 dB of what we recorded during quality control C111_3_5mV. Battery voltage 2.63 V. ION/UCL reports they could not turn the transmitter on after a week implanted and turned off. We turn off and place in oven at 60°C to poach, along with B146.8 and B146.9, all turned off.

[20-DEC-16] Poaching transmitters, C111.3, B146.8, and B146.9. Remove from oven and measure frequency response Poaching_20DEC16. Transmitter C11.3 is generating a square wave.

[23-DEC-16] We have batch N144 consisting of 16 of A3028N-AA, single-channel mouse transmitters with BR1225 battery holder for use as a head fixture. We load batteries into all holders. Frequency response N144_5mV within ±0.5 dB. Noise within faraday enclosure with leads connected together is ≤12 μV. Switching noise is present in many of the sixteen, and is ≤7 μV. We keep back N145.4 and N144.11. These two have switching noise 7 μV.

Poaching transmitters B146.8 and B146.9 have perfect frequency response. C111.3 gives good reception but generates a square wave a few minutes after removing from water. We turn them off and put them back in the oven.

Poaching transmitters B146.8 and B146.9 have perfect frequency response. C111.3 gives good reception but generates a square wave a few minutes after removing from water. We turn them off and put them back in the oven.

2017

JAN-17

[03-JAN-17] Poaching transmitters: B146.8 and B146.9 have perfect frequency response. C111.3 gives good reception but generates a square wave. We turn them off and put them back in the oven.

We have B130.1, B130.4, B125.7, B130.8, B130.10, and B129.12 returned from IIT/Genova. We soak in water for an hour. According to IIT, these devices were implanted 5 weeks after they were shipped, worked upon implantation, were left turned off for ten days implanted, then failed to turn on. We attempt to turn on and off all six devices. Only B129.12 turns on. Reception is 100%, VA = 2.62 V, noise 8 μV. Frequency response is perfect. Silicone and epoxy like new. We place B129.12 in the oven to poach at 60°C turned off. In the remaining five devices, the epoxy coating was thin over corners of U9, U4, and U8. The silicone has pulled the epoxy coating off these corners, leaving an imperfection beneath the silicone. Some devices have wrinkles in the first of four coats of MED10-6607 silicone. By looking at the reflection of our overhead lights in the convex surface of the silicone coating, we confirm that the outer coat of silicone is everywhere intact and unbroken except around the base of the wire we used to hold the transmitters epoxy encapsulation in devices B130.10 and B130.4. On these two devices we see white oxide on the tip of this wire and also on the exposed solder joints of the antenna and leads. Dissect B130.10. Silicone well-adhered to epoxy. Green residue on positive battery tab. Battery voltage = VB = 0.1 V. Disconnect VB = 0.1 V. Connect external 2.6 V. Inactive current 2.0 μA, active 81 μA. Frequency response correct. Dissect B130.1. Silicone well-adhered. VB = 0.2 V. Disconnect, VB = 0.2 V. Apply external 2.6 V. Inactive 2.0 μA. Active 78.4 μA. Frequency response correct, see B130_5mV_2 for comparison of frequency response of these three transmitters just before shipping (1) and today when powered by an external battery (2). We place B130.4, B125.7, and B130.8 in water at 60°C and will dissect at a later date.

[04-JAN-17] Poaching transmitters: B146.8 and B146.9 turn on and obtain 100% reception, pick up mains hum. B129.12 100% reception, VA = 2.8 V. C111.3 turn on and obtain 100% reception with 0.5-Hz square wave.

[06-JAN-17] Poaching transmitters: B129.12, B146.8 and B146.9 remove from water and turn on. All three have 100% reception. After cooling down, B146.9 shows steps changes of several millivolts. These become less common as minutes go by, allowing us to measure frequency response No9_3 in plot. The gain of B129.12 and B146.8 are correct (No12_3 and No8_3 in plot), but B146.9 is 6 dB too low at 100 Hz. When we switch to a 50-Ω 10-mV source, the gain is correct. C111.3 turn on and obtain 100% reception with 0.5-Hz square wave.

Dissect B130.4, no sign of corrosion, silicone in excellent condition. When we peel the silicone away we smell vinegar. VB = −0.2 V. Short the battery briefly, VB = 0.0 V. Disconnect, VB = −0.1 V. Connect external 2.6 V. Inactive 1.7 μA, active 78 μA. Reception 100%. Pick up mains hum. At first we see step artifacts, but after a few minutes these cease and we measure frequency response, which is perfect with 50-Ω source, but 10 dB too low at 100 Hz with 10-MΩ source (No4 in here). Dissect B125.7. No signs of corrosion, silicone is in excellent condition. We smell vinegar. VB = 0.1 V. Disconnect, VB = 0.1 V. Connect external 2.6 V, inactive 1.8 μA, active 78 μA, but fluctuating to 90 μA. Reception 100%. At first, we see a full-scale 0.5-Hz square wave. After a few minutes, this stops. Gain is 20 dB too low with both 50-Ω and 10-MΩ sources. Noise is 3 μV rms total. Dissect B130.8. No sign of corrosion. A few cavities beneath the silicone, but no breaches. No smell of vinegar. VB = −0.05 V. Disconnect, VB = −0.05 V. Connect external 2.6 V. Inactive 1.9 μV, active 78 μV then varying from 100 μA to 140 μA. Reception 100%, full-scale settling to 130 μA rising to 130 μA. We see full-scale oscillations at 110 Hz. We heat up C5 and current consumption stabilizes at 78 μA. Top of U5 is exposed, after cracking off of thin layer of epoxy. This the largest cavity that existed beneath the silicone.

[09-JAN-17] Poaching transmitters: C111.3, B129.12, B146.8 and B146.9 remove from water and turn on. All have 100% reception. C111.3 still generating square wave. B146.9 generating bumps and short-lived oscillations. With 50-Ω 10-mV sweep, frequency response is correct. B129.12 now has gain 10 dB too low at 100 Hz with 20-MΩ source. With 50-Ω source, gain is normal. B129.8 has normal gain with high and low impedance sources.

[10-JAN-17] With a sanding wheel, we grind away the glue from the bottom side of B130.10 to check for penetration of epoxy around and beneath component terminals. So far as we can tell, epoxy penetrated everywhere, beneath and around the pins and bodies of all ICs, and beneath P0402 parts.

[17-JAN-17] Poaching transmitters: B129.12, B146.8, and B146.9 remove from water and turn on, all three have 100% reception and perfect response to 50-Ω sinusoidal sweep. With electrodes open circuit, all three detect mains hum, but B146.0 also produces occasional step and swings of order millivolts. C111.3 has gain 20 dB too low at 100 Hz with 50-Ω source and generates square wave with electrodes open circuit.

[20-JAN-17] We have batch L147 consisting of eight of A3028L-DDA. Frequency response L147_5mV agree to ±0.4 dB. Switching noise less than 2 μV, noise less than 14 μV.

[23-JAN-17] Poaching transmitters: C111.3 generates 1-Hz square wave even when connected to 50-Ω source, but reception is 100%. B129.12 stable input, gain normal for 50-Ω source, reception 100%. B146.8 and B146.9 both generating rumble as they cool down, but gain is normal for 50-Ω source, reception 100%.

[24-JAN-17] We have batch E145 consisting of 14 A3028E-AA. Some of these are particularly thick, with volume as high as 3.9 ml. Three that were so thick we sanded them down on the top side, are now 3.7 ml with a single coat of silicone. In May 2016 the A3028E volume was 3.0 ml. Frequency response E145_5mV within ±0.8 dB. Switching noise in water at 37°C is ≤4 μV, total noise ≤12 μV.

[27-JAN-17] We have batch A148 consisting of 7 A3028A-DDC. Gain versus frequency A148_5mV within 0.5 dB. Switching noise in 36°C water is ⋚4 μV. Noise ≤12 μV after ten minutes in water and after scrubbing some of the electrodes.

[30-JAN-17] We have batch A149 consisting of 4 A3028A-DDC. Gain versus frequency A149_5mV within 0.2 dB. Switching noise in 40°C water is ⋚4 μV. In A147.13 we have 4 μV in channel No14, but ≤ 1 μV in No13. We see rumble and some step artifacts when we first put the transmitters in water. Noise ≤12 μV in band 2-256 Hz after ten minutes.

[31-JAN-17] Poaching transmitters: C111.3 generates 1-Hz square wave even when connected to 50-Ω source, but reception is 100%. B146.9 generating rumble and steps, but frequency response is normal for 50-Ω source, reception 100%. B146.8 and B129.12 no rumble, gain 3 dB too low at 100 Hz for 50-Ω source, reception 100%.

FEB-17

[01-FEB-17] We have our first A3028M-AAA, a dual channel 0.3-640 Hz transmitter with 2048 SPS per channel. Current consumption is 442 μA. We have batch E150 consisting of fourteen A3028E-AA. Volume of four of them together is 12 ml. Another six together are 20 ml. In both cases, we are including the base of the antenna and leads. Our estimate of the volume of the body alone remains 3.0 ml. Frequency response is E150_5mV within ±0.4 dB. Switching noise in water at 36°C is E150_SWN.png all ≤5 μV. Total noise is less than 12 μV in 1-256 Hz. There is no rumble in the signal nor step artifacts from the moment we place the transmitters in water. These transmitters have bare-wire ends. Batch A148/9 had soldered electrodes and we observed large step artifact and rumble when we immersed them in water.

[03-FEB-17] Poaching transmitters C11.3, B146.8, B146.9, B129.12 all turn on and give 100% reception.

[07-FEB-17] We have batch N151 consisting of 7 of A3028M-AA and 1 of A3028M-AAA, our first epoxy-encapsulated dual-channel 0.3-640 Hz transmitter. Frequency response is N151_5mV within ±0.4 dB for the M versions. Variation in the absolute gain is due to variation in battery voltage of the freshly-inserted batteries. We place all the transmitters in their bags, turned on, and within a larger bag, and then in hot water and measure noise. Total noise is ≤12 μV in the Ns, and ≤14 μV in the Ms. Switching noise is ≤3 μV.

[10-FEB-17] Poaching transmitters C11.3, B146.8, B146.9, B129.12 all turn on and give 100% reception.

[17-FEB-17] We have batch B152 consisting of five A3028B-DC. Frequency response B152_5mV within ±0.2 dBm. Switching noise in B152.2 is 6 μV with total noise 16 μV. We reject this one. The other four have switching noise 4 μV or less, total noise 12 μV or less. We have two prototype transmitters, Q154.22, Q154.39 are equipped with the CR-2354/HFN 560 mAhr battery. Switching noise ≤1 μV, total noise 5 μV. Volume is 5.0 ml. Frequency response also in B152_5mV. We have Q154.56 and Q154.73 made with the CR-2450/H1AN battery 620 mAhr. Switching noise also ≤1 μV, total noise 5 μV. Volume 5.5 ml. Frequency response in B152_5mV.

Poaching transmitters C111.3, B146.8, B146.9, B129.12 all turn on and give 100% reception, a one-second interval shows the corrosion artifact in their signals here. B129.12 still functions well enough to record EEG, but its gain is 10 dB too low. We turn them off again and put them back in the oven, but now at 80°C. We put Q154.56 and Q154.73 in to poach at 80°C. We add E144.10 and E150.6, both of which have excessive switching noise.

[22-FEB-17] Poaching transmitters B146.8, B146.9, B129.12 all turn on and give 100% reception. C111.3 won't turn on and we see brown corrosion beneath the silicone around the positive battery tab. Q154.56, Q154.73, E144.10, and E150.6 RF spectrum centered on 900 MHz when first removed from oven at 80°C, and reception with A3027E is 80%. After a few minutes to cool down, 100% reception and correct gain versus frequency for 50-Ω sweep.

[27-FEB-17] Poaching transmitters B146.8, B146.9, B129.12 all turn on and give 100% reception. All three generate a 0.5-Hz full-range square wave. Q154.56, Q154.73, E144.10, and E150.6 reception is 100% after after a few minutes to cool down. Gain versus frequency for 50-Ω sweep is correct.

[28-FEB-17] Poaching transmitters B146.8, B146.9, B129.12 all turn on and give 100% reception. Q154.56, Q154.73, E144.10, and E150.6 reception is 100% after after a few minutes to cool down. Gain versus frequency for 50-Ω sweep is correct in Q154.73, E144.10, and E150.6, but 6 dB too low in Q154.56. Gain versus frequency for 10-MΩ source is too low for all sources at 100 Hz, see here.

MAR-17

[03-MAR-17] Newly-made transmitter E152.13 active current 120 mA after encapsulation in epoxy, before silicone coating, using the multimeter's milliamp range. It has drained its battery. Poaching transmitters Q154.73 gain is 6 dB too low at 100 Hz with 50-Ω source. Q154.56 gain 6 dB too high at 100 Hz. E150.6 gain 1 dB too low at 100 Hz. E144.10 gain normal. Noise ⋚12 μV after allowing to settle. B129.12 gain 10 dB too low at 100 Hz. B146.8 gain 20 dB too low at 100 Hz, B146.9 no gain at 100 Hz, produces 0.5-Hz square wave. Newly-made transmitter E153.11 drains its battery after clipping the extension and loading battery. We disconnect the battery. Active current is 115 mA with the milliamp range and 300 mA with the ampere range. Component U9 is hot. We replace and active current consumption is 78.1 mA. We go back to E152.13, burn off epoxy around U9 and apply power. U9 heats up. We remove U9. Active current is now 41 μA. Before failure, with U9 loaded, active current was 80 μA. We believe U9 is being damaged by shorting A to 0V while loading the battery.

We have batch E155 consisting of fourteen transmitters including some with channel numbers greater than 14. We are sanding down the lump of epoxy that forms over the circuit board during encapsulation on the rotator. Each transmitter has a flat-topped look to it. We find one bubble in silicone that we feel must be filled. One device has been sanded to the point where we can see the tops of two ICs, half of them to the point where we can see one. Frequency response E155_5mV within ±0.5 dB. Total noise 1-250 Hz ≤12 μV for half-second intervals. Switching noise in 37°C water ≤5 μV.

[06-MAR-16] We have batch E153 consisting of thirteen transmitters from set zero. Frequency response E153_5mV within ±0.4 dB. Noise ≤1 μV, switching noise in 38°C water ≤4 μV. Poaching transmitters reception 100% from all devices, gain with 50-Ω source: E150.6 gain correct, E144.10 gain correct, Q154.73 gain 10 dB too low at 100 Hz, Q154.56 gain 10 dB too low at 100 Hz, B129.12 1-Hz square wave, B146.8 0.5-Hz square wave, B146.9 0.5-Hz square wave.

[10-MAR-17] Poaching transmitters reception 100% from all devices. In water, noise ≤16 μV except B129.12 which produces 0.5-Hz square wave. Gain with 50-Omega; source: E150.6 gain 10 dB too low at 100 Hz, E144.10 gain 20 dB too low at 100 Hz and see 0.5-Hz oscillation, Q154.73 gain 10 dB too low at 100 Hz, Q154.56 gain 10 dB too low at 100 Hz, B129.12 gain 20 dB too low at 100 Hz, 0.5-Hz steps, B146.8 gain 20 dB too low at 100 Hz, B146.9 0.3-Hz square wave.

We receive back from Marburg five transmitters that failed prematurely. During e-mail discussions of these failures, we came to some tentative conclusions. "R106.7: Was running while on shelf, exhausted most of its battery before implantation." R106.7 won't turn on. Disconnect battery. Silicone comes away easily, together with enamel coating. VB = 0.5 V. Connect external 2.6 V. Inactive current 1.8 μA, active 83 μA. Reception 100%. Response to 50-Ω sweep correct. With 255 mA-hr battery, operating life should be 128 days.

"R117.14: Broken EEG lead." R117.14 won't turn on. The antenna has been cut 20 mm from the base. Disconnect battery. Silicone hard to remove and in perfect condition. Rotator-made epoxy. VB = 1.9 V. Connect external 2.6 V. Inactive 1.8 μA. Active 78.3 μA. 100% reception with antenna base directly on receiver antenna. Attach external battery. Gain with 50-Ω sweep is normal. Gain with 20-MΩ sweep is normal. Noise with leads in water is 8 μV with switching noise <1 μV at 20°C.

"R115.1: Broken EEG lead." R115.1 won't turn on. Antenna has been pulled away from the EEG leads and is exposed at the tip. Silicone hard to remove and in perfect condition. Rotator-made epoxy. Epoxy so thin around XYC corner that we can see two capacitors and a resistor. VB = 0.3 V. Connect external 2.6 V. Inactive 1.8 μV, active 82 μV. Reception 100%. Attach external battery. Gain with 20-MΩ sweep is correct. Noise with leads in water 15 μV. Noise with leads and antenna tip in water 400 μV. We compare the spectrum of the noise with and without the antenna tip in the water with the spectra below.


Figure: Spectrum of Noise from R115.1 With EEG Leads In Water and Antenna Tip Outside (Left 0.8 μV/div) and Inside (Right 8 μV/div) Water. Note that the vertical scale is ten time greater on the right, in which the fundamental of the switching noise is 60-μV in amplitude. For the appearance of the signal itself see Breached Antenna Noise.

The noise we see with the antenna tip in the water is exactly the noise we see on a currently-implanted transmitter at Marburg, R121.4. This noise has the same spectrum as the noise generated by our hall-effect magnetic sensor, which we call "switching noise". The amplitude of the fundamental component of the switching noise is 60 μV. When we remove the antenna from the water, this amplitude drops to less than 1 μV. We conclude that R117.14 and R115.1 both suffered from noise generated by damaged antenna insulation.

R110.4 we suspect was left running on the shelf before implantation. It failed 80 days after implantation. The device does not turn on. Disconnect battery. Silicone in good condition. Some enamel pulls off. VB = 0.8 V. Connect external 2.6 V. Inactive 1.7 μA, active 83 μA, 100% reception. Gain with 50-Ω sweep is correct. Noise with leads in water 16 μV. With antenna in the same water, noise is 500 μV, with switching noise 120 μV.

R112.6 was implanted immediately upon arrival and failed after 118 days. Antenna has been pulled away from EEG leads. Will not turn on. Disconnect battery. Silicone in good condition. VB = 0.8 V. Connect external 2.6 V. Inactive 1.8 μA. Active 84 μA. Noise with leads in water 15 μV. With antenna in same water, insulation intact, 15 μV. With insulation removed from tip, switching noise is 120 μV. With a 255 mA-hr battery and 84 μA operating current, we expect this transmitter to run for 126 days. It failed after 118 days implanted and a one-day burn-in for 119 days total, which is consistent with battery capacity 240 mA-hr.

We take transmitter B152.2, which failed quality control because of switching noise 6 μV. We immerse in hot water and observe switching noise of 8 μV. We pull the antenna away from the EEG leads and put just the EEG leads in water. Switching noise is now 3 μV. E116.10 when immersed in water shows 4 μV switching noise. We immerse only the leads. Switching noise 5 μV. Pull antenna away from EEG leads. Switching noise 5 μV. Arrange antenna to be far from the leads. Switching noise 3 μV. C143.14 has switching noise 16 μV immersed warm water. We place only the leads in water, 12 μV. Pull antenna away from leads, immerse leads in water, 10 μV. Although the antenna is a potential source of switching noise, it is not the only source, nor even the main source for switching noise arising after encapsulation.

[17-MAR-17] We have batch E154 encapsulated in epoxy on the rotator, sanded flat on top side, coated once with SS-5001, except E154.1 and E153.14, which are coated twice with EE-6001. We measure frequency response E154_5mV within ±0.4 dB. Noise in water at 37°C is ≤13 μV with switching noise ≤4 μV. Silicone coating shows no cavities except for E154.7, which has a 0.5-mm diameter cavity beneath the surface. We take E154.7 and E154.1 to poach.

Poaching transmitters B146.8 and B146.9 will not turn on. Dissect B146.8. Battery 0.3 V. Disconnect 0.3 V. Apply external 2.6 V. Inactive 6 μA at first, dropping to 3 μA, active 90-200 μA fluctuating. Diagnosis "corroded capacitor". B146.9 battery 0.1 V, disconnect 0.1 V. Apply external 2.6 V. Inactive 15 μA dropping to 8 μA. Active 150-2000 μA fluctuating. Diagnosis "corroded capacitor". B129.12 turns on and generates 0.5-Hz oscillation. Q154.56 and Q154.73 are both off. We suspected they were both off last time we checked them, and so were careful to make sure they were on when we put them in the oven and we had no magnet near them when we removed them from the oven. They both turn on and we get 100% reception, gain 20 dB too low at 100 Hz. E150.6 gain 10 db too low at 100 Hz with 50-Ω source and VA = 2.1 V. E144.10 gain normal with 50-Ω source and VA = 2.3 V. Add E154.7 and E154.1.

[20-MAR-17] Poaching transmitters E154.1 and E154.7 100% reception gain normal with 50-Ω source and 20-MΩ source. B119.12 100% reception, no oscillation, gain 20 dB too low at 100 Hz. Q154.56 and Q154.73 are both off. We turn on Q154.73. VA = 3.1 V. Noise 11 μV, gain 20 dB too low at 100 Hz. Q154.56 will stay on only if we leave the magnet resting on the device. It generates a 1-Hz square wave. After a few minutes, it turns off and we cannot turn it on again. E150.6 won't turn on. E144.10 will turn on only if we rest the magnet on the device, and generates 0.3-Hz square wave. We remove !154.56, Q154.73, E150.6 and E144.10 from poach and will dissect tomorrow.

[21-MAR-17] Poached transmitter E144.10 turns on with VA = 1.9 V. Dissect. Silicone appears unaffected by poach. Force to shave it off the epoxy with blade. VB = 1.7±0.2 V fluctuating. Settles to 1.6 V and transmitter is off. Disconnect battery, VB = 2.7 V. Connect external 2.6 V. Inactive current 79 μA, active 320 μA, generates 0.3-Hz sine wave. Active current increases to 500 μA. Remove C6, inactive current 2.0 μA. We have removed the C6 pads as well, so cannot replace it. Diagnosis "corroded capacitor". Q154.56 is now on and won't turn off. Generates 0.5-Hz square wave. Dissect. Shave silicone off epoxy. Note that U2 is part of the epoxy surface. VB = 3.0 V. Disconnect batter. Connect external 2.6 V. Active current 1.9 mA. Apply magnet, transmission continues but current drops to 170 μA. Remove U2 and C4. Inactive 0.3 μA. Turn on by touching U2-2 pad with 2.6V and 0V. Active 82.1 μA. Reception 0%, but see 512 Hz ripple on 2.6 V. Note VA = 0.5 V. Our U2 is cracked on one corner. We load it onto another circuit board, but it does not work. Our symptoms are consistent with a short between U2-2 and the 0V terminal on C4 as well as corrosion in C5 to bring down VA. Diagnosis "corrosion short". Q154.73 will not turn on. Dissect. Shave off silicone. U2 visible. VB = 1.0 V. Disconnect VB = 1.4 V. Connect external 2.6 V. Inactive 2.1 μA. Active 83 μA. Reception 100%, from X, VA = 2.0 V with lots of noise, and 0.3 Hz square wave. We measure VA = 2.6 V, VC = 1.8 V. Expose pins of U5. Oscillations stop. Connect external battery and apply 50-Ω sweep. Gain 20 dB too low at 100 Hz. Gain on U5-7 appears to be okay, but U5-1 has loss of gain at 100 Hz. Remove C9, C10 no change. Now break off C8 with its pads. Diagnosis "unidentified drain". E150.6 won't turn on. Dissect. Shave off silicone. VB = 1.1 V. Disconnect VB = 2.3 V. Connect external 2.6 V. Inactive 3.8 μV, active 52 mA. Remove C4, active 43 mA. Diagnosis "corrosion short".

[29-MAR-17] Poaching transmitter E154.1 100% reception, gain normal for 50-Ω and 100-kΩ sweeps, 14 dB too low at 100 Hz for 20-MΩ sweep, VA = 2.77 V. E154.7 100% reception, gain normal for 50-Ω and 20-MΩ sweeps, VA = 2.80 V. B129.12 will not turn on after 64 days at 60°C and 40 days at 80°C. This poach is equivalent to a total of 10×64 + 60×40 = 3040 days at 37°C.


Figure: Corrosion Beneath Silicone in B129.12. We see white tendrils of some new substance spreading out from the edge of the circuit board, where it is not covered with epoxy. We see brown corrosion at the clipped edge of the circuit board.

The corrosion around the edges of the circuit board suggest that our epoxy coating is too thin on the edges to remain intact when poaching.

We have batch B201_17 consisting of twelve A3028B-AA transmitters with channel numbers in the range 17-33. Gain versus frequency B201_17 within ±0.4 dB. Total noise ≤16 μV except for B201.18 and B201.28 which are 18 μV. Switching noise ≤6 μV except B201.18 and B201.28, which are 8 μV. We reject B201.18 and B201.28 and set them aside for dissection.

APR-17

[04-APR-17] Poaching transmitters E153.7 reception 100%, gain 3dB too high at 100 Hz with 100-kΩ sweep, noise in hot water 15 μV. E154.1 reception 100%, gain 10 dB too low at 100 Hz with 100-kΩ sweep, noise in hot water 15 μV with steps of 1 mV every few hundred milliseconds.

[07-APR-17] Poaching transitters E153.7 reception 100%, gain 20 dB too low at 100 Hz with 50-Ω sweep. E154.1 reception 100%, reports zero-valued samples.

[10-APR-17] Poaching transmitters E153.7 and E154.1 100% reception.

[11-APR-17] We have batch E200_23 consisting of A3028E-FB with channel numbers 23-38. Gain versus frequency E200_23 within ±0.6 dB. Switching noise in 35°C water ≤6 μV. Total noise ≤16 μV. In 40°C water switching noise ≤5 μV, total noise ≤14 μV. Poaching transmitters E153.7 and E154.1 100% reception.

[12-APR-17] Poaching transmitters E153.7 and E154.1 100% reception.

[13-APR-17] Poaching transmitter E153.7 100% reception. E154.1 has stopped. This device has two coats of SS-6001. The silicone has a yellow shade where it is thick, and around the base and tip of the antenna. We bend the base of the leads and the antenna tip pops out of the silicone.


Figure: Discoloration and Cracking of SS-6001 After 27 Days Poaching at 80°C.

We dissect E154.1. VB = 1.0 V. Disconnect VB = 2.2 V. Connect external 2.7V, inactive 2.0 μA, active 1.8 mA. Reception 100%, transmitting all zeros. We remove C5, C2, and C6 but 1.8 mA persists. But VA = 0.2V which suggests a corrosion resistance between VA and 0V. Diagnosis "unidentified drain". We have E153.14, which also has SS-6001 coating. We place it in the oven to poach, along with R129.6, which has MED10-6607 coating.

[21-APR-17] We have a collection of leads that have soaked in acetone at room temperature for a week with a 20-ml lump of dental cement. The acetone is now pink, the color of the dental cement.


Figure: Pink Dental Cement Dissolved in Acetone.

We remove the leads and set them on a piece of paper. After a few minutes they look like this:.


Figure: Residue on Leads After Dissolving Dental Cement.

When we wash the jar with water, we get a sudden appearance of a thick white residue.


Figure: Residue on Inside of Jar After Pouring Out Acetone with Dissolved Dental Cement.

We wash the jar with acetone and wipe it out. A surface discoloration remains, but almost all the residue is gone. We shake the leads in clean acetone, remove, and place on paper to dry. A film remains on the leads, disturbing the reflection of light from the lead surface.


Figure: Residue on Leads After One Acetone Wash.

We soak the leads in acetone for ten minutes, shake them well, and wipe them each on a clean lint-free cloth. We now find that their surfaces stick to gether in the same way they do when they are clean and new. The surfaces are shiny. After a few minutes trying, they have no odor. With tweezers we can make no mark in any film on their surface.


Figure: After Second Acetone Wash and Wipe.

We wash with hot water. The lead surfaces are hydrophobic. We see no sign of the white film we created earlier with water washing in the jar. We conclude that the leads are clean.

Poaching transmitters: E153.7 has stopped. R129.6 100.0%, 2.87 V, 10.6 μV; E153.14 100.0%, 2.82 V, 6.5 μV.

[25-APR-17] Dissect E153.7. VB = 0.6 V. Disconnect, VB = 1.4 V. Connect external 2.7 V. Inactive 4.0 μA, active 37 mA, reception 100%. Remove C3 and C4 but active remains 34 mA. Diagnosis "unidentified drain".

Poaching transmitters: R129.6 100.0%, 2.92 V, 9.0 μV; E153.14 100.0%, 2.89 V, 6.8 μV. Response to 50-Ω and 100-kΩ sweep is correct. Gain of both transmitters is 14 dB too low at 100 Hz for 20-MΩ source and has the same shape.

We have been poaching our selection of silicone leads, after their experiences with acetone and dental cement, for the past four days in water at 80°C. We examine them today. They are clean, shiny and flexible.

We have batch E200_39 consisting of fourteen transmitters. Each has one coat of SS-5001 and an outer coat of MED10-6607. Most of the leads have a squashed point where they were held in spring clamps during dipping. We check all transmitters for breach of insulation at these points and find no breaches. We turn them on and let them run in hot water for an hour. We refresh the water at 37°C. Switching noise ≤5 μV, total noise is ≤13 μV. Gain E200_39 within ±0.4 dBm.

[28-APR-17] Poaching transmitters: R129.6 91.0%, 2.96 V, 13.9 μV; E153.14 100.0%, 2.90 V, 13.4 μV. Response to 50-Ω and 100-kΩ sweep is correct.

We have batch B156 consisting of ten A3028B-CC. Noise ≤14 μV, switching noise <6 μV in warm water. Frequency response within &plumns;0.3 dB B156_5mV.

MAY-17

[02-MAY-17] Poaching transmitters: R129.6 100.0%, 2.84 V, 10.6 μV; E153.14 99.4%, 2.92 V, 15.3 μV. Response to 50-Ω and 100 kΩ sweep is correct for E153.14 and 6 dB too low at 100 Hz in both sweeps for R129.6.

[05-MAY-17] Poaching transmitters: R129.6 100.0%, 2.67 V, 26.7 μV; E153.14 100.0%, 2.75 V, 86.6 μV. Noise on both channels is rumble. Response to 50-Ω and 100 kΩ sweep is is 10 dB too low at 100 Hz for both transmitters.

[09-MAY-17] Poaching transmitters: R129.6 100.0%, 2.49 V, 20.0 μV; E153.14 99.8%, 2.76 V, 15.0 μV. Response to 50-Ω sweep is 10 dB too low at 100 Hz for R153.6 and 20 dB too low at 100 Hz for E153.14. With leads open-circuit in air, E153.14 generates a 1-Hz square wave.

[15-MAY-17] Poaching transmitters: 100% reception from both.

[19-MAY-17] Poaching transmiters: R129.6 100.0%, 2.51 V, 15.7 μV. Gain 20 dB too low at 100 Hz. R153.14 has stopped. We able to turn it on again and it generates a 1-Hz square wave with 100% reception. Dissect. Silicone is yellow and well-adhered to epoxy. Battery voltage 2.9 V. Disconnect, battery voltage 3.0 V. Connect external 2.7 V. Inactive 2.5 μA, active 95-105 μA varying with the square wave. After two minutes, the transmitter turns itself off and inactive current is 2.1 μA. We cannot turn it on again for a few minutes. Now active current is 900 μA for a while, then it turns off. Diagnosis "Corroded Capacitor".

[22-MAY-17] Poaching transmitters: E146.2 98.4%, 2.86 V, 9.9 μV; E154.3 98.4%, 2.86 V, 9.9 μV; E155.21 96.9%, 2.84 V, 11.6 μV; E155.22 96.9%, 2.84 V, 13.0 μV. Response to 50-Ω sweep for all four is correct. R129.6 has stopped. Dissect, VB = 1.6 V, disconnect, VB = 2.9 V. Apply external 2.7 V. Inactive 2.6 μA, active 80-100 μA, jumps to 2200 μA, switch to mA scale and drops back to 300 μA. Later, see 4 mA on mA scale. Diagnosis "Corroded Capactior".

[24-MAY-17] Poaching transmitters: E146.2 96.1%, 2.77 V, 11.7 μV; E154.3 100.0%, 2.73 V, 16.7 μV; E155.21 95.5%, 2.76 V, 13.9 μV; E155.22 100.0%, 2.75 V, 16.6 μV. Response to 20-MΩ sweep is correct for E146.2, E154.3, and E155.21, but too low by 10 dB at 100 Hz in E155.22. Response to 100-kΩ sweep is correct for all four, with 100% reception.

[30-MAY-17] Poaching transmitters: E146.2 98.0%, 2.84 V, 13.6 μV; E154.3 95.1%, 2.85 V, 7.1 μV; E155.21 98.6%, 2.83 V, 7.2 μV; E155.22 100.0%, 2.86 V, 8040.8 μV. E155.22 is generating a full-scale oscillation at around 50 Hz as it cools to room temperature, then settles down to give noise 20 μV. Gain with 100-kΩ source is 3 dB too low at 100 Hz in E155.22, normal in the other three.

JUN-17

[02-JUN-17] We have batch E201.34-51, epoxy rotated, 1 coat SS-5001, 1 coat MED-6607. Frequency response E201_34. Switching noise ⋚4 μV except E201.51, which is 8 μV. We hold back E201.51.

Poaching transmitters E146.2, E154.3, and E155.21 100% and gain for 100-kΩ sweep within 3 dB of nominal. E155.22 has stopped and won't turn on.

[05-JUN-17] Poaching transmitters E146.2 gain normal with 100-kΩ source. E154.3 gain 6 dB too low at 100 Hz with 100-kΩ and 50-Ω sources. E155.21 gain 3 dB too high at 100 Hz with 100-kΩ and 50-Ω sources. E146.2 100.0%, 2.77 V, 10.6 μV; E154.3 100.0%, 2.80 V, 8.4 μV; E155.21 100.0%, 2.85 V, 27.9 μV. E154.3 average value is varying as if VA is going from 2-2.8 V. Dissect E155.22, which failed a few days ago. VB = 1.2 V, disconnect VB = 2.6 V. Connect external 2.7 V. Inactive 2.6 μA, but remains inactive only when magnet is pressed to transmitter. Active 1.6 mA, no transmission. Burn eposy away from U3. Magnetic switch now turns circuit on and off correctly. Inactive 1.9 μA, active 1 mA. Remove C5, C6, and C4. Active 0.5 mA. Remove C3, 0.4 mA. Diagnosis "Unidentified Drain".

[09-JUN-17] Poaching transmitters: E146.2 100.0%, 2.84 V, 7.8 μV; E154.3 100.0%, 2.37 V, 12.5 μV; E155.21 100.0%, 2.62 V, 25.4 μV. With 100-kΩ and 50-Ω sweeps all three show gain normal at 1 Hz and 20 dB too low at 100 Hz.

[13-JUN-17] We have batch E201.71-93, epoxy rotated, 1 coat SS-5001, 1 coat MED-6607. Frequency response L201_71. Switching noise in 37°C water <3 μV. Total noise 2-320 Hz ≤15 μV.

[14-JUN-17] Poaching transmitters. E146.2 100% reception, generating its own square wave when open circuit and in water. Top layer of silicone is coming un-stuck. E154.3 and E155.21 won't turn on. Top layer of silicone coming un-stuck.

[16-JUN-17] Dissect E155.21. VB = 1.1 V, disconnect, VB = 2.4 V. Connect external 2.7 V. Inactive 2.2 μA. Active at first 1.1 mA then drops to 87 μA. Reception 100%. Connect external battery. Frequency respose 20 dB too low at 100 Hz. Diagnosis "Corroded Capacitor". Dissect E154.3. Disconnect battery VB=2.3V. Connect external 2.7 V. Inactive 3.0 μA. Active 47 mA. Remove C4 and C3, active 50 mA. Diagnosis "Unidentified Drain". Poaching transmitter E146.2 100% reception, generating its own 1-Hz square wave in water.

We assemble our first three A3028V-CAC dual-channel EEG/EMG transmitters. Channel X is 0.3-160 Hz, 512 SPS. Channel Y is 30-640 Hz, 16 SPS. We have replaced C12 with 1.0 nF and C14-C16 with 240 pF. We apply a frequency sweep and measure gain versus frequency by looking at the amplitude of the samples we receive.


Figure: A3028V Frequency Response to 10-mV 20-MΩ Sweep. Green: EEG input X 512 SPS. Blue: EMG input Y 16 SPS. Orange: EEG input X when we apply the sweep to EMG input Y.

In the third plot, we connect X to C and apply the sweep to Y while measuring the signal on X. There is no significant cross-talk between the EMG input and the EEG input within the circuit. When we leave all three leads open circuit in air, however, we see the following transmitter-generated noise of amplitude 18 μV.


Figure: A3028V Noise with Inputs In Air. In purple is EEG X 18 μV rms with 16-Hz fundamental, and in pink is EMG Y 56 μV rms.

When we put X and C in water 1 cm apart, so they are connected by roughly 1 MΩ, and leave Y in air, we get the following.


Figure: A3028V Noise with X and C In Water, Y In Air. In purple is EEG X 6 μV rms and in pink is EMG Y 17 μV rms.

[20-JUN-17] Poaching transmitter E146.2 has stopped. Top coat of silicone peeling off. Dissect VB = 0.7V. Disconnect 1.8 V. Connect external 2.7 V. Inactive 5.6 μA, active 2.0-2.5 mA. Diagnosis "Corroded Capacitor".

We have batch E201.52-69. Run in water for an hour, then place in 37°C water. Noise <14 μV, switching noise <4 μV. Some minor ripples in the top-coat of MED-6607. Frequency response E201_52.

[27-JUN-17] Poaching transmitters E201.57 and E201.61 reception 100%, response to 100-kΩ sweep is nominal. Silicone coating in good condition: no discoloration or peeling.

[30-JUN-17] We have batch B202_17 consisting of thirteen A3028B-CC. Frequency response E202_17. Gain ±0.4 dB. Noise in 37°C ≤12 μV, switching noise ≤4 μV. We have batch V202.34 consisting of three A3028V-CAC EEG/EMG monitors. Noise in 35°C water is ≤10 μV for EEG and ≤15 μV for the 640-Hz EMG amplifier. Switching noise is ≤5 μV. Frequency response V202_33, gain ±0.4 dB of nominal.

Poaching transmitters E201.57 and E201.61 reception 100%, response to 100-kΩ sweep is nominal.

JUL-17

[11-JUL-17] Poaching transmitters E201.57 and E201.61 reception 100%. E201.57 gain 6 dB too high at 60 Hz with 50-Ω source, full-scale 60 Hz with 100-kΩ source. E201.61 gain 6 dB too low at 100 Hz for 50-Ω source and 100-kΩ source. Gain for 20-MΩ source is here. Silicone coating intact. No peeling of final MED-6607 layer.

We have batch C201_97 consisting of 13 A3028C-CC. Frequency response C201_97. Noise in 39°C water ≤11 μV, switching noise ≤7 μV, shown here. We reject C201.100 because it occasionaly shows switching noise amplitude up to 9 μV.

[14-JUL-17] Poaching transmitters E201.57 and E201.61 reception 100%. VA = 2.9 V for both.

[21-JUL-17] We have batch M203.17-51, all A3028M-AAA. Frequency response M203_17 and M203_17_More within ±0.35 dB except for M203.51, for which Y gain is 2 dB too low. Noise ≤16 μV in 0.3-640 Hz bandwidth. We observe an intermittent 1-μV peak around 20 Hz in one transmitter, but no sign of swithing noise in any others.

[25-JUL-17] Poaching transmitters E201.57 and E201.61 reception 100%. E201.57 generating its own 60-Hz sine wave. E201.61 VA = 2.9 V.

[28-JUL-17] Poaching transmitters E201.57 and E201.61 reception 100%.

AUG-17

[01-AUG-17] Poaching transmitters E201.57 and E201.61 have stopped.

[04-AUG-17] We have batch E201_113 consisting of fourteen A3028E-AA. Frequency response E201_113 lies within ±0.9 dB, the greatest variation occurring at 130 Hz. Switching noise in 37°C water is <4 μV, total noise ≤12 μV rms except for E201_119, which shows 5 μV and 15 μV respectively. One of them E201_114 has a blue lead 20 mm too short. We keep this one to poach.

We have batch B202_39 consisting of four A3028B-AA. Switching noise in 37°C water <3 μV and noise <10 μV rms. Frequency response E202_39.

[14-AUG-17] Poaching transmitters C201.102 and E201.114 reception 100%, correct response to 100-kΩ sweep.

[15-AUG-17] Dissect E201.57. Outer layer of MED-6607 well-adhered to main coat of SS-5001 and still flexible. We peel most of it off then cut away the SS-5001. VB=1.1 V. Disconnect, VB rises to 2.2 V in one minute. Connect external 2.7 V. Inactive 5 μA, active 1.6 mA dropping occasionally to 90 μA with 100% reception. But after two minutes, rises to 2 mA and no reception, cannot turn off. Burn epoxy away from C2 and C5, connect new battery, get 100% reception, gain with 50-Ω source is 20 dB too low, and we see steps in average value. Diagnosis: Corroded Capacitor.

Dissect E201.61. Peel off outer layer, cut away inner layer, burn away epoxy. VB 1.8 V stepping down to 1 V. No reception. Disconnect, VB=1.8 V. Connect external 2.7 V. Inactive 2.7 μV, active fluctuating 80-90 μA with synchronous steps in X. Damage tracks trying to remove C6. Diagnosis: Unidentified Drain.

[16-AUG-17] We receive recorings from ION/UCL of the last two days of C143.5, 14-AUG-17 after 33 days implanted. This device we shipped 26-NOV-16, so it has been consuming 2.5 μA for 200 days before consuming 50 μA for 33 days, a total of 52 mA-hr from its nominally 48 mA-hr battery.

[18-AUG-17] Poaching transmitter E201.114 reception 100%, VA = 2.8 V, correct response to 100-kΩ sweep, C201.102 has stopped. Dissect, VB = 0.4 V. Disconnect, VB = 0.4 V. Connect external 2.7 V. Inactive 1.8 μA, active 52-57 μA, 100% reception, 0.5-Hz square wave on X 600-65535 counts. Diagnosis "Unidentified Drain".

[22-AUG-17] Poaching transmitter E201.114 reception 100%, VA = 2.9 V, correct response to 100-kΩ sweep.

[25-AUG-17] Poaching transmitter E201.114 reception 100%, VA = 2.9 V, response to 100-kΩ and 50-Ω sweeps have correct shape, but are 4 dB too low.

SEP-17

[01-SEP-17] Poaching transmitter E201.114 reception 100%, 0.5-Hz full-scale square wave. No response to 50-Ω sweep.

[05-SEP-17] We have batch E201_129 consisting of sixteen A3028E-AA. Encapsulated with one coat of SS5005 and one coat MED-6607. Switching noise in 37°C water is ≤4 μV except E201.140 with 5 μV. Total noise is ≤14 μV for all. Frequency response E201_129 within ±0.7 dB.

Poaching transmitter E201.114 reception 100%. Noise 10 μV until place in cold water, then 0.5-Hz oscillations start. We leave the transmitter at room temperature for three hours. It turns itself off. We turn it on again and it oscillates as before.

We have our first transmitters equipped with rechargeable LiPo batteries. All transmit at 512 SPS and 0.3-160 Hz. No1 and No3 have 19-mAhr batteries, No5 has a 190-mAhr battery. We place in warm water. Noise is 4.2 μV, 4.5 μV, and 5.3 μV respectively. Switching noise is 0 μV, 0.7 μV, and 0 μV respectively as viewed with a 32-s interval. Volume of No1 and No3 combined is 3 ml, making each 1.5 ml, slightly more than our A3028B with its 48-mAhr primary lithium cell. Volume of No5 is 6 ml, a little less than our A3028L with its 1000-mAhr primary lithium cell. Looking at the smaller encapsulations, we may have to add more material to round off the corners of the battery pack. Looking at the larger encapsulation, we could use less epoxy and silicone. To the first approximation, the battery capacity per unit volume is 30% of the primary cells devices.

[11-SEP-17] Poaching transmitters with LiPo batteries gain versus frequency normal for 100-kΩ sweep. Total input noise ≤6 μV. VA = 3.76, 3.73, and 4.91 V for No1, No3, and No5. The small No1 and No2 appear unaffected by poach. But No5 looks larger. But we measure its volume to be 6 ml as before. We smell a hint of the sweet odor we associate with old or exhausted LiPo batteries.

[12-SEP-17] Poaching transmitter No5 with the large battery has burst its silicone. The battery is puffed up. It is not running, but when we turn it on, it powers up just fine and we get 100% reception. The other two also give 100% reception, and look unaffected.

[13-SEP-17] Poached transmitter No5 is generating a 0.5-Hz square wave with 100% reception. Battery internal pressure has dropped, but it still bulges with gas. No1 and No3 show battery voltage 3.64 V and 3.60 V respectively and 100% reception. Response to 100-kΩ sweep is normal. Dissect E201.114. VB = 1.4 V. Encapsulation in perfect condition. Disconnect VB = 1.4 V. Connect external 2.7 V. Inactive 3.6 μA. Active 85-125 μV with 0.5-Hz square wave and 512 SPS. Higher current during 0.5-Hz steps. VB where it enters circuit shows 20-mVpp at 512 Hz but no 0.5 Hz. VC shows ≶5 mV noise. Current now jumps to 400 μA when active, only 2.5 μA inactive. Remove C5 and current is varying 86-92 μA. We see no 0.5 Hz on VA. Square wave has more features. Diagnosis of failure "Corroded Capacitor". The 0.5-Hz artifact is some other corrosion.

[15-SEP-17] Poaching transmitters No1 and No3 give 100% reception.

[18-SEP-17] Poaching transmitters No1 and No3 have stopped. This we expect after ten days running with a 19-mAhr LiPo battery. Diagnosis "Full Life".

[22-SEP-17] We prepare mock-up of the A3028E-R, the rechargeable version of the A3028E. Here is how we route connecting wires to the A3028RV3 circuit.


Figure: The A3028RV3 circuit programmed as A3028E, equipped with 190-mAhr LiPo battery.

We cannot recharge the battery through the EEG leads in this mock-up, but the form and battery life will be identical to that of the proposed A3028E-R.

OCT-17

[03-OCT-17] We have the above A3028E-R prototype encapsulated in epoxy and silicone, call it ER.8. Dip in epoxy with thirty second run-off and rotate to cure. Repeat. Paint exposed parts with epoxy. Cover corners with SS-5001 silicone. Dip three times in MED-6607. Maximum dimensions 32 mm × 22 mm × 9 mm. Displaces volume 4.0 ml. We test then put in the oven at 60°C to poach.

[04-OCT-17] Our A3028E-R device ER.8 battery voltage 3.87 V, noise 11 μV rms. Response to 100 kΩ sweep correct. No sign of swelling within encapsulation.

[05-OCT-17] ER.8 reception 100%, battery 3.86 V, noise 6 μV rms. No sign of swelling.

[06-OCT-17] We solder two stainless steel, teflon-insulated wires to a BR1225 coin cell. We need acid flux to solder to the surface. We can apply a solder blob immediately. We let the iron sit on the battery surface for ten seconds. The battery is too hot to touch. We wash in water and attach to an A3028U consuming 150 μA. Battery voltage is 2.5 V. We leave it running.

Poaching transmitter ER.8 battery voltage 3.86 V, reception 100%. Response to 100 kΩ sweep correct. No sign of swelling in ER.8 nor in another non-functioning A3030E we started poaching at the same time.

[10-OCT-17] Poaching transmitter ER.8 battery voltage 3.83 V, reception 100%, response to 100 kΩ sweep correct. No sign of sweeling in ER.8 or A3030E dummy, but both devices have a faint sweet smell, as does the poaching water. Our 150-μA transmitter equipped with over-heated BR1225 is still running with battery voltage 2.6 V.

[11-OCT-17] Poaching transmitter ER.8 battery voltage 3.81 V, reception 100%, response to 100 kΩ sweep correct. Faint sweet smell persists. Our 150-μA transmitter equipped with over-heated BR1225 is still running with battery voltage 2.6 V.

[13-OCT-17] Poaching transmitter ER.8 battery voltage 3.79 V, reception 100%, response to 100 kΩ sweep correct. Faint sweet smell persists but no bulging of battery. Our 150-μA transmitter equipped with over-heated BR1225 is still running with battery voltage 2.6 V. We have a dummy A3028P pup-sized transmitter made with a BR1225 cell, rotated epoxy, silicone on bumps, three coats of MED-6607. Volume 0.8±0.2 ml, length 21 mm, width 13 mm, and height 4.3 mm. With the BR1025 the width will be 11 mm and length 19 mm. With rounded-corner circuit board we can skip silicone on bumbs and apply only two coats of MED-6607 to reduce height to 4.0 mm.

We suspend an A3028RV3 circuit board over a piece of paper 150 mm from a spectrometer loop antenna. With a 50-mm wire we get −34 dBm. With no antenna we get −62 dBm. With the transmitter off we get −67 dBm at the peak of the spectrum. An A3028A in water at same range gives −37 dBm. We load a 63 mm unstretched helical lead and get −45 dBm in air. We cut back the antenna and measure power received in air and water. When in water, the antenna tip is in contact with the water.


Figure: Power Received from Helical Antenna. Same antenna base position and line, but length varies. In water, we have the antenna tip in contact with the water and isolated from the water with hot glue.

We repeat, using hot glue to insulate the end of the antenna each time we cut it shorter. An insulated helical antenna, made with the same spring as our EEG leads, works poorly in water when 25 mm long, but very well when 15 mm long.

[16-OCT-17] Our 150-μA transmitter equipped with over-heated BR1225 is still running with battery voltage 2.5 V. Poaching transmitter ER.8 battery voltage 3.77 V. Noise 6 μV. Response to 100-kΩ sweep is correct.

[17-OCT-17] Our 150-μA transmitter equipped with over-heated BR1225 is running with battery voltage 2.5 V. Poaching transmitter ER.8 battery voltage 3.77 V. Noise 5 μV. Response to 100-kΩ sweep is correct.

We power a 150-μA transmitter with a newly-received ML621 6.8-mm diameter 2.1-mm thick lithium rechargeable battery. Reception is 100%. Battery voltage is 2.7 V. We connect to a 480 μA transmitter, 4096 SPS with a 14-mm helical antenna. Battery voltage 2.5 V, reception 100%. Five hours later, battery voltage is 2.1 V. Connect battery to 2.7 V through ammeter and 400 Ω. See 0.4 mA flowing in. After ten minutes, 0.2 mA. Remove ammeter and leave connected.

[18-OCT-17] Our 150-μA transmitter equipped with over-heated BR1225 is running with battery voltage 2.4 V. Our ML621 battery is drawing no current through 400 Ω from 2.7 V. We connect to 480 μA transmitter and measure battery voltage with a DVM, 2.41 V. We connect to 150 μA transmitter, battery voltage 2.47 V. Disconnect, battery voltage 2.49 V. The source resistance of the ML621 is around 150 Ω. Connect to 35-μA transmitter and leave running. We take our over-heated battery and measure 2.62 V when disconnected, 2.56 V with 150 μA, and 2.46 V with 480 μA. The source resistance of the BR1225 is around 350 Ω.

[19-OCT-17] ML621 voltage 2.50 V. Our over-heated BR1225 is reporting 2.3 V through X. It provided 150 μA for 13 days, a total of 46.8 mA-hr.

[20-OCT-17] ML621 voltage 2.43 V still running our 35-μA transmitter, which reports VA = 2.39 V. Connect ML920S, 9.5 mm diameter, 2.0 mm thick, 11 mA-hr lithium rechargeable battery to 150-μA transmitter. When disconnected, battery voltage is 2.75 V. When connected, the transmitter reports 2.67 V. Poaching transmitter ER.8 still running, battery voltage 3.75 V, noise 6 μV rms, correct response to 100-kΩ sweep. No swelling of battery, sweet smell just detectable. We have our first A3028U, 0.0-160 Hz 200 mV dynamic range dual-channel transmitter, U201.169. Current consumption is 1.7 μA inactive and 143 μA active.

[23-OCT-17] ML621 voltage 1.85 V with DVM, not receiving from 35-μA transmitter. The ML621 powered the transmitter for around 120 hours, but we expect 170 hours. We suspect that our re-charge was insufficient after the first drain. We plug into 2.9V with 400 Ω in series to re-charge. Our ML920S battery still running our 150-μA transmitter, reporting VA = 1.91 V. With DVM we measure 1.89 V. The device has run for around 70 hours, and we expect 70 hours. Poaching ER.8 battery voltage 3.75 V, noise 6 μV rms, reception 100%, correct response to 100-kΩ sweep. No sign of swelling. Feint sweet smell when held to nose.

[24-OCT-17] ML621 voltage 2.85 V with DVM after charging over-night with 2.9 V through 400 Ω. Re-connect to 35-μA transmitter and get VA = 2.81 V. We connect our ML920S to 2.9 V through 400 Ω. ER.8 battery 3.74 V, 100% reception, correct response to 100-kΩ sweep, noise 5 μV.

[26-OCT-17] ML621 voltage 2.51 V with DVM, 35-μA transmitter still running. ML920S voltage 2.89 V with DVM after two days charging from 2.9 V through 400 Ω. Connect to 150-μA transmitter. Both these transmitters are equipped with 1-mm diameter EEG leads for antennas, one 30 mm long, the other 33 mm long. Reception is robust in our faraday enclosure. ER.8 reports VA = 3.73 V, noise 6 μV rms. Response to 100-kΩ sweep correct. Sweet small hardly discernable.

[27-OCT-17] ML621 voltage 2.47 V with DVM, 35-μA transmitter still running, reports VA = 2.43 V. ML920S voltage 2.44 V, 150-μA transmitter still running, reports VA = 2.38 V. ER.8 reception 100%, noise 6 μV, response to 100-kΩ sweep correct.

[30-OCT-17] ML621 voltage 2.33 V with DVM, 35-μA transmitter not running at first. We plug the battery back in and it powers up with VA = 2.25 V. Runs for about ten minutes befor switching off with low battery voltage. The ML621 ran for at most 140 hours when we expect 170 hours. The SL-920S voltage is 1.4 V with DVM, 150-μA transmitter not running. Recharge with 2.9 V through 400 Ω. ER.8 reception 100%, noise 5 μV rms, response to 100-kΩ sweep correct. No swelling.

[31-OCT-17] ML920S voltage 2.85 after charching overnight with 2.9 V through 400 Ω. We connect to our 150-μA transmitter, channel numbers 37 and 38, and place in faraday enclosure with Acquisifier measuring VA every hour. Plug ML621 into 3.0 V wth 400Ω to charge. Three hours later, its voltage is 2.85 V with DVM. Plug into 35-μA transmitter, see 2.80 V. Leave to run while monitoring VA. ER.8 100%, 6 μV rms, correct response to 100-kΩ sweep, reports VA = 3.72 V.

We have eight A3028U-DDK from batch U201.151. We measure volume of all eight by water displacement to be 11 ml, individual volume is 1.4 ml.

NOV-17

[03-NOV-17] We record the battery voltages reported by our 35-μA and 150-μA transmitters when powered by fmanganese-lithium batteries ML621 and ML920S respectively after charging both with 2.9 V through 400 Ω.


Figure: Manganese-Lithium Battery Discharge. We use the average value of X to measure VBAT. The ML621 is discharges at 35 μA. The ML920S discharges at 150 μA through a two-channel transmitter.

The ML621 provides 1.4 mA-hr, far less than its nominal 5.8 mA-hr. The ML920S provides 7.6 mA-hr, less than its rated 11 mA-hr. We find the following re-charge curve, which suggests we should be charging with 3.1 V. Judging by this curve, it looks like 30% of the re-charge energy is delivered above 2.9 V. We charge our ML621 and ML920S with 3.1 V through 400 Ω for 24 hours each.


Figure: Manufacturer's Recharging Curve for ML621. After 24 hours, capacity has reached 4.7 mA-hr.

We have batch U201.147-171 consisting of 12 A3028U-DDB. Our LWDAQ function generator has a DC offset that we remove with a 1-μF capacitor. We connect a 50-Ω 33-mV sweep through the capacitor to each input in turn and measure frequency response from 0.25 Hz to 1000 Hz. We obtain U201_147 within ±0.8 dB. When we place the devices in water, many of them are excessively noisy on either X or Y, but not both. We remove from water and place on foam pad in faraday enclosure. Noise in 2-160 Hz is 35-40 μV rms, or 8 counts rms. Switching noise observable in 8-s intervals is ≤6 μV. Because we still see switching noise, despite the gain being ×10 rather than ×100, the switching noise cannot be introduced at the EEG input. If it is not introduced at the EEG input, it cannot be introduced at the input to the second stage of amplification either. So the noise must be getting into the ADC through its power supplies.

ER.8 reception 100%, VA = 3.70 V, response to 100-kΩ sweep correct, noise 5 μV rms. We put U201.161 and 163 in the oven at 60°C to poach. We see full-scale fluctuations on the inputs. In one interval, we see 161 at 811 counts, 162 at 772 counts, 163 at 45775, and 164 at 65528.

[06-NOV-17] We have batch J204.1-23 consisting of 11 of A3028J-CMC encapsulated with epoxy and three coats of MED-6607 only. We have not yet soldered the silver wire onto the Y lead. The volume of all eleven with antennas included is a 14 ml, making their body volume a little less than 1.3 ml. All parts are well-covered by epoxy and silicone with the exception of U9, which is right next to the edge of the printed circuit board, with corners that push out. But the silicone over these corners is still smooth, even though it protrudes, and there is no metal on the corners. All leads ≤0.8 mm diameter. The Y channels have 0.3-80 Hz, X 0.3-160 Hz. Frequency response is J204_1. Noise is ≤10 μV with switching noise ⋚4 μV.

ER.8 reception 100%, response to 100-kΩ sweep correct, VA = 3.71 V, noise 6 μV rms, no bulges, sweet smell faint. U201.161 and U201.163 reception 100%, response to 70-mV 100-kΩ sweep correct, in faraday enclosure with leads open-circuit, noise is 10 counts rms, 40 μV rms, with VA = 2.46-2.48 V form the four channels. In water, however, we get average input voltages 35%, 46%, 54%, and 51% of full-scale on channels 161-164 respectively.

[07-NOV-16] Having charged our ML-series batteries to 3.1 V we find they last longer. So far the ML621 has provided 3.5 mA-hr, compared to its nominal 5.8 mA-hr.


Figure: Manganese-Lithium Battery Discharges. We use the average value of X to measure VBAT. We use the same ML621 and ML920, recharging them between experiments.

Poaching transmitters U201.161 and U201.163 reception 100%, response to 100-kΩ sweep correct. Input voltages in hot water 37%-85% of full scale. When placed on foam, we measure the input voltages in the range 0.0-270 mV and get 73.3%, 72.4%, 72.4%, 72.8% for channels 161-164 respectively. Noise is around 35 μV rms. ER.8 100% reception, response to 100-kΩ sweep correct. Noise 6 μV rms.

We have batch J204.25-49 consisting of 11 of A3028J-CMC encapsulated with epoxy and three coats of MED-6607 only. Total volume 14 ml for 11 is 1.3 ml each. Frequency response J204_25 correct to ±1 dB. Noise in 37°C water <15 μV rms, switching noise ≤4 μV.

We have B152.5, a transmitter with excessive switching noise. We place in faraday enclosure powered by its own battery and obtain the spectrum on the left, for which total noise 2-160 Hz is 12 μV rms and switching noise fundamental is 8 μV.


Figure: Effect of Internal and External Battery on Switching Noise. Left: internal BR1225 on B152.2. Right: external BR2477 on B152.2.

With external BR2477 battery, noise 2-160 Hz drops to 5 μV rms and switching noise is less than 0.8 μV. We note that switching noise in transmitters made with LiPo batteries is negligible. We note that the A3028U, with ×10 amplifier, still sees switching noise, which means the noise is arising at the input of the ADC. The BR1225 output impedance is around 44 Ω. Its negative side is pressed against the ADC package on the top side of the board.

[09-NOV-17] Our ML621 and ML920S have dropped below 2.0 V after 135 hr (4.7 mA-hr) and 70 hr (11 mA-hr) respectively. Both are consistent with manufacturer's data sheets for a 24-hour charge from 3.1 V.

[10-NOV-17] Poaching ER.8 silicone in perfect condition, slight sweet smell, no bulging, response to 100-kΩ sweep correct, 6 μV rms noise. U201.161 and U201.163 reception 100%, correct response to 100-kΩ source from 0.1-1000 Hz, noise 40 μV rms in air. In water, average values are 1.2%, 0.5%, 86.7%, and 74.4% of full scale for channels 161 to 164. U201.161 has a 10-mm length of exposed 316SS as its VC electrode and two soldered pins for X and Y. We connect its threee input pins together and put it back in water. Inputs are now 57.8% and 57.6% of full scale. We separate the VC lead. We get 1.1% and 1.5% of full scale. We connect C and Y, leaving X separate. Now X is at 0.0% and Y is at 58.3%. Add salt to water, X = 1.1%, Y = 0.4%. Connect all three together in saltwater we get X = 1.1%, Y = 0.4%. But U201.163 in the same saltwater with leads separated gives X = 82.7% and Y = 83.3%. In air again with leads isolate, U201.161 gives X = 71.2% and Y = 70.4%. In saltwater, separate leads, cut VC lead to 1 mm, X = 1.1%, Y = 0.4%. Solder a stainless steel screw to VC, put back in water X = 1.1%, Y = 0.3%. We place leads in water but transmitter body outside of water and get X = 74.0%, Y = 65.5%. This transmitter we rejected because the support wire pad came off and we used the 0V battery pad for the support wire. We tried but failed to cover this cut-off support wire. Furthermore, there is a breech in the silicone near the wire. We can taste the 1.8-V potential of VC when we put the entire transmitter in our mouth.

We connect a BR2477 to B152.2 through a 100-Ω resistor and see <1 μV switching noise. Connect directly to its own battery: 4 μV. Insert 1 kΩ in series with its own battery: 5 μV. Add 100 μF, 4-V P1206 (at 2.7 V bias acts like 55 μF) from VB to 0V on circuit board: <1 μV. Remove 1 kΩ, leaving capacitor in place: 1.5 μV. Repeat above tests and get the same results. We have C143.10 with 4 μV switching noise when powered by its own battery. We add 100 μF to VB: 2 μV. Add 1 kΩ in series with battery: <1 μV. Remove capacitor, leaving 1 kΩ: 5 μV. Repeat tests and get same results. Try 22 μF 16-V P0805 (at 2.7 V bias acts like 11 μF) on B152.2 with 1 kΩ: 2 μV. Remove 22 μF: 5 μV. Add 22 μF 16 V tantalum: 1.5 μV. Remove 1 kΩ and tantalum. We obtain spectrum with direct connection, then with 100 μF P1206 from VB to 0V.


Figure: Effect of 100 μF P1206 4-V Capacitor on Switching Noise. Left: without capacitor, total noise 12 μV rms. Right: with capacitor, total noise 8 μV rms. Note attenuition by capacitor of higher harmonics is more dramatic than that of the fundamental.

[13-NOV-17] Our 150-μA transmitter with ML621 battery stopped after 27 hours, or 4.0 mA-hr. When we disconnect from transmitter, battery voltage is 2.3 V. We reconnect and turn on the transmitter. For ten seconds we see 100% reception of two channels. After ten minutes, VB = 0.65 V. We connect to 3.1 V through 400 Ω to recharge. Our ML920 is charged to 3.09 V. We connect to 35-μA transmitter and leave running.

Poaching transmitter ER.8 reception 100%, noise 6 μA, VB = 3.71 V, no sweet smell after changing the poaching water three days ago, response to 100-kΩ sweep correct. U201.163 X and Y voltages in 20°C water are 83% and 66% of full scale. Reception 100%. Response to 100-kΩ sweep correct. U201.161 reception 100%, response to 100-kΩ sweep correct.

[14-NOV-17] We add J204.45 and J204.49 to our collection of transmitters poaching at 60°C. All poaching transmitters 100% reception, correct response to 100-kΩ sweep. Encapsulation all in good shape.

[15-NOV-17] Our ML621 battery has charged for 48 hours with 3.1 V through 400 Ω. We connect to our 150-μA transmitter. We measure source impedance of various batteries with a known resistor load, either 100 Ω or 1-kΩ. We notice that the output resistance of a fresh battery is greater, so that we might get a voltage of 2.9 V but output resistance double what we see at 2.7 V. Before testing, we exercise the batteries for few minutes with the measurement load.


Figure: Battery Resistance versus Battery Thickness for Various Battery Types.

The BR1225 output resistance is around 140 Ω, while that of the CR1025 is only 70 Ω. The CR2354 resistance is only 10Ω, while the thicker BR2477 resistance is 40 Ω.

[16-NOV-17] Poaching transmitters ER.8, U201.161, U201.163, J204.45, J204.49 all give correct response to 100-kΩ sweep and 100% reception. No sweet smell on ER.8.

[17-NOV-17] Our ML621 is discharged. We re-charge with 400 Ω and 3.3 V. We have B205.1-5 equipped with BR1225 batteries ready for encapsulation. We measure switching noise by putting them in a faraday enclosure with their EEG electrodes in water. We turn them all on within a couple of minutes and record. During 550-750 s we had to re-arrange the transmitters because their leads were slipping out of the water.


Figure: Switching Noise from Five Unencapsulated A3028Bs.

During the 0-900 s interval, the average battery voltage dropped from 2.75 V to 2.55 V. The above drop with time is consistent with our observation of a halving of battery resistance as the battery voltage drops by 0.2 V. We have two A3028Qs equipped with CR2354 batteries ready for encapsulation. We turn them on. Switching noise is <1 μV, with no visible fundamental or harmonics on any of the four available input channels.

[20-NOV-17] Our ML621 is charged to 3.28 V. We connect to our 150-μA transmitter. Poaching transmitters U201.161 and U201.163 have stopped after 14-17 days. Expected life is 340 hours. Diagnosis "Full Life".

[21-NOV-17] Our ML621 voltage is 2.35 V after 24.5 hours providig 150 μA for 3.7 mA-hr. But we did not record intermediate voltages, so start re-charging with 3.3 V and 400 Ω.

[22-NOV-17] Poaching transmitters ER.8, J204.45, and J204.49 correct response to 100-kΩ sweep, reception 100%. ER.8 battery 3.64 V, noise 6 μV. J204.45 and J204.49 battery voltage 2.59-2.62. Switching noise in air at 20°C is up to 8 μV. Place in hot water and switching noise drops to below 3 μV. Total noise 12 μV.

[27-NOV-17] Our ML920 is still running after 336 hours. Its voltage dropped below 2.2 V after 310 hours delivering 35 μA, so its capacity was 11 mA-hr, which is its nominal capacity. Poaching transmitters ER.8, J204.45, and J204.49 reception 100%, noise <12 μV. ER.8 VB = 3.61 V, J204.45 sVB = 2.53 V, J204.49 VB = 2.49 V. Correct response to 100 k-Ω sweep for all five channels.

[29-NOV-17] Poaching transmitters J204.45 and J204.49 have both stopped after 13-15 days. Expected life 15 days. Diagnosis "Full Life". ER.8 still running. We cut the leads off poached U201.161 and weigh on a precise scale: 2.1 g. The three leads and antenna that we cut off weigh 0.2 g.

We have batch B205.1 consisting of five A3028Bs with long, thin leads. Two of them, B205.3 and B205.4, have a poorly-covered stress concentration at the base of the EEG leads, which we will cover with more silicone. Frequency response B205_1. Switching noise in 40°C water ≤6 μV. Total noise ≤12 μV.

DEC-17

[01-DEC-17] We have batch Q201_187 consisting of 8 A3028Q-DDB powerd by CR2354 batteries. Frequency response Q201_187. Noise <15 μV, switching noise in warm water <1 μV. Volume of 4 bodies is 18 ml, 4.0 ml average. We have batch G201_173 consisting of 6 A3028G-DDB powered by BR2330 batteries. Volume is 14 ml for 5 bodies not including leads and antenna, 2.8 ml average. Frequency response G201_173. Noise <17 μV, switching noise in warm water <1 μV.

[05-DEC-17] Poaching transitter ER.8 no sweet smell, response to 50-Ω and 100-kΩ sweeps is the same and correct. We add A3028G transmitters G201.183 and G201.185 to our poach. G201.185 has its mounting wire sticking out of the silicone on one corner. The silicone cover of the positive battery tap appears to be no more than 100 μm on both. Before poaching, with leads resting on the table, G201.183 and G201.185 weigh 5.746 and 5.781 g respectively.

We measure the mass of four A3028J/U transmitters with their leads and get 2.449 g, 2.467 g, 2.461 g, and 2.474 g. We cut the leads and antenna off one and these weigh 0.260 g. The weight of the transmitter bodies is around 2.2 g.

[08-DEC-17] Poaching transmitter ER.8 gain for both 50-Ω and 100-kΩ sweeps is 20 dB too low. Reception is 100%, noise is 5 μV rms. G201.183 and G201.185 response to 100-kΩ sweep is correct.

[11-DEC-17] Poaching transmitter ER.8 reception 95%, VB = 2.17 V, gain for 100-kΩ sweep is correct. G201.183 and G201.185 reception 100%, gain for 100-kΩ sweep correct.

[12-DEC-17] Poaching transmitter ER.8 has stopped. Expose battery terminals. VB = 1.5 V. Connect re-charging leads. After one hour, turns on and reports VB = 4.0 V. Response to 100-kΩ and 50-Ω sweeps is 6 dB too low, but variation with frequency is correct. G201.183 and G201.185 reception 100%, gain for 100-kΩ sweep correct.

[14-DEC-17] Rechargecd ER.8 reception 100%, reports VB = 4.1 V. Response to 30-mV, 100-kΩ sweep correct shape, but 6 db too low. Place in faraday enclosure to monitor battery drain over next ten weeks. Poaching transmitters G201.183 and G201.185 reception 100% after cooling down, response to 100-kΩ sweep correct. Report VB = 2.58 V and 2.62 V.

[15-DEC-17] We have batch E201.211 consisting of seventeen A3028E-AA, coated four times in MED-6607. Gain E201_211 within ±0.4 dB. Gain of E201.212 is 3 dB higher than nominal at 140 Hz. Two lumps on the leads of E211.219. Total noise in 39°C water ≤12 μV. Switching noise <4 μV.

[18-DEC-17] We have batch B206.9 consisting of nine A3028B-AA, coated three times in MED-6607. We measure lead diameters and find them consistent with our new specification 0.7±0.1 mm, with the minimum thickness being 0.6 mm and the maximum 0.79 mm. Mass of transmitter including antenna and leads is 2.37 g with standard deviation 0.03 g. Total noise ≤14 μV in 37°C water, switching noise ≤5.6 μV as shown here. Gain versus frequency A206_9 within ±0.3 dB.

Poaching transmitters G201.183 and G201.185 reception 100% after cooling down, response to 100-kΩ sweep correct. Report VB = 2.62 V and 2.62 V.

[19-DEC-17] Poaching transmitters G201.183 and G201.185 reception 100% after cooling down, response to 100-kΩ sweep correct. Report VB = 2.63 V and 2.63 V.

[22-DEC-17] Poaching transmitters G201.183 and G201.185 reception 100%, response to 100-kΩ sweep correct. Report VB = 2.64 V and 2.62 V.

[27-DEC-17] Poaching transmitters G201.183 and G201.185 reception 100%, response to 100-kΩ sweep correct. Report VB = 2.54 V and 2.51 V.

[29-DEC-17] Poaching transmitters G201.183 and G201.185 reception 100%, response to 100-kΩ sweep correct. Report VB = 2.62 V and 2.60 V.

We have 10 of A3028GV1, first article from assembly house, built to S3028F_1 with the red-masked A302801G printed circuit board. Program all ten as A3028J no problems. Inactive current 1.7±0.1 μA compared to 2.1±0.2 μA for the previous thirty A3028RV3 circuits calibrated. The A3028RV3 contains U1, R2, and R1. With the programming extension in place, roughly 0.3 μA would flow through R2, see S3028C_1. In place of U1, the A3028GV1 uses the logic output of U3 as the power switch for the transmitter circuit. Check frequency response of both inputs at 512 SPS from 1-500 Hz, all correct. Load BR1225 batteries, wash, blow dry and bake. Check noise by connecting all three leads with a clip in a Faraday enclosure. Total noise is 7.4 μV rms on average for all twenty amplifiers. Switching noise is 1.5 μV on average within one minute of turning on with fresh battery, with maximum 2.2 μV and minimum 0.9 μV. The second harmonic of switching noise is half the amplitude of the first harmonic, and the third and higher harmonics are too small to see. Compare to switching noise in the first minute for A3028RV3s equipped with the BR1225 here in which noise was 2-11 μV. The A3028GV1 is equipped with a total of 40 μF decoupling on VB (C1, C2, C19, and C20, assuming VB and VD are well-connected through U3) compared to 10 μF on the A3028GV1.

2018

JAN-18

[02-JAN-18] Poaching transmitters G201.183 and G201.185 reception 100%. For channels 183, 185, 186 response to 100-kΩ sweep correct. For channel 183, response has correct shape but gain is 4 dB lower than 184. Report VB = 2.62 V and 2.60 V.

[08-JAN-18] Poaching transmitters G201.183 and G201.185 reception 100%. For channels 183, 185, 186 response to 100-kΩ sweep correct. For channel 183, response is 10 dB too low at 100 Hz. Report VB = 2.62 V and 2.53 V.

[09-JAN-18] We have batch B203_53 consisting of 8 of A3028B-H-AA, formerly known as the A3028N, head-fixture transmitter for mice. Frequency response B203_53 within &plumn;0.3 dB.

[11-JAN-18] Poaching transmitters G201.183 and G201.185 reception 100%. For channels 183, 185, 186 response to 100-kΩ sweep correct. For channel 183, response is 10 dB too low at 100 Hz for 50-Ω and 100-kΩ sweeps. Report VB = 2.60 V and 2.44 V respectively.

[12-JAN-18] Poaching transmitters G201.183 and G201.185 reception 100%. For channels 183, 185, 186 response to 100-kΩ sweep correct. Channel 183 response10 dB too low at 100 Hz. Report VB = 2.58 V and 2.43 V respectively.

We have 10 of A3028J-AAA made with the A3028GV1 assembly, unmodified to give the 160-Hz bandwidth on X and 80-Hz on Y.

[16-JAN-18] Poaching transmitter G201.183 still running, 100% reception, VB = 2.33 V. Response of channel 183 to 50-Ω sweep correct, for 184 it's 10 dB too low at 100 Hz. Transmitter G20-1.185 has stopped running. We start it again, VB = 2.1 V. Response to 50-Ω sweep of 5 mV is correct. It is 42 days since we started poaching, suppose 185 failed at 41 days, and suppose 183 would fail before tomorrow, or 43 days. Typical operating life for the A3028G is 42 days. These two devices we burned in for one day in the dry oven before poaching, so they delivered their full life. We end our test of both, diagnosis "Full Life". Silicone in perfect condition. Serial number label ink now dark gray rather than black. No corrosion around mounting wire stub. No separation of silicone from epoxy. Minimal corrosion of solder joints on electrode pins.

Our test batch of 10 of A3028J-AAA made with the A3028GV1 assembly have been soaking in water. No sign of rust. Switching noise less than 2 μV for all devices. Turn on and place in oven at 80°C to poach.

[19-JAN-18] Poaching transmitters all running with 100% reception. Response to 50-Ω sweep is correct for all except No13, which reports VA = 1.9V and cannot amplify its input.

[22-JAN-18] Poaching transmitters all running except No3 and No13, which have stopped and won't turn on.

[23-JAN-18] Eight poaching transmitters all running 100% reception and reponse to 100-kΩ sweep correct. Dissect A3028J No13. VB = 0.4 V. Disconnect VB = 0.4 V. Connect external 2.7 V. Inactive 1.4 μA, active 270 μA, nominal is 150 μA. Response to 100-kΩ sweep correct. VA = 2.58 V with VB = 2.73 V. Active 240 μA. Remove epoxy from C2 and C5. Active current 180 μA. Replace C5, active 150 μA. Heat up C4, active 140 μA. Replace C4, 140 μA. Replace C6, 160 μA. Remove C1, 170 μA. Remove C3, 160 μA. Remove R4 and R3, 130 μA. Remove U9, 60 μA. Replace U9 160 μA. Diagnosis "Corroded Capacitor", C5 in particular. We recall that No13 reported VA = 1.9 V, which is consistent with VB = 2.5 V and current 600 μA, which would drain the battery in three days, which is what happened. Dissect No3. VB = 0.4 V. Disconnect 0.5 V. Connect external 2.7 V. Inactive 1.9 μA. Active 139 μA. Reception 100%. Response to 100-kΩ sweep is a 1-Hz full-scale square wave on both channels. VC is stable. VA flucuates by 25 mV or so. The oscillation we associate with corrosion, but because we have no excess current at the moment, we won't be able to identify the source of the current drain. Diagnosis "Unidentified Drain".

We have batch A206_23 consisting of ten A2038A-DDC. We apply a 6.3-mV sweep A206_23. Switching noise in 37°C water ≤4 μV. Total noise in 2-160 Hz ≤15 μV rms after scrubbing the pins and screws twice, but up to 40 μV before due to rumble.

[24-JAN-18] Poaching transmitters No1, 5, 7, 9, 11, 17, 19, 21 still running. Report VA = 2.62-2.67 V except for No21, which reports 2.50 V. Gain for 100-kΩ sweep correct except for No7 X which is 6 dB too low at 100 Hz. We have B206.78 failed after encapsulation. Dissect. VB = 2.6 V. It turns on and off irregularly. Discard.

[25-JAN-18] Eight poaching transmitters still running with 100% reception.

[26-JAN-18] All eight poaching transmitters still running. Apply 100-kΩ sweep. No7 X gain 6 dB too low at 100 Hz, Y gain 6 dB too high at 60 Hz. No5 X channel gain 3 dB too low at 100 Hz. No21 gain of both channels 3 dB too high at 10 Hz. All other channels within 3 dB of nominal.

We have batch C206_88 consisting of 12 A3028C-CC. Apply 5-mV 20-MΩ sweep. Frequency response C206_88 within ±0.3 dB. Total noise <9 μV. Switching noise <8 μV in 37°C water.

[29-JAN-18] Poaching transmitter No19 VA = 1.9 V, No1 VA = 2.1 V response to 10-kΩ sweep attenuates above 60 Hz on both channels, No7 VA = 1.9 V seen on X with square wave on Y, No11 VA = 2.2 V but response of X is still correct for 0.3-160 Hz bandwidth, while response of Y is 6 dB too low at 80 Hz. No9 VA = 2.5 V, response of X 6dB too low at 120 Hz and of Y 6dB too low at 60 Hz, No17 VA = 1.9 V but still see some response on X and Y to sweep, No5 and No21 have stopped and won't turn on. They have run for 310 hours. Their expected life is 340 hours.

[30-JAN-18] Dissect poached No5. VB = 0.6 V. Disconnect, VB = 0.6 V. Connect external 2.7 V. Inactive 2.3 μA, active 138 μA. Response to 100-kΩ sweep is 6 dB too low at the high end of the pass band for both X and Y. Diagnosis "Unidentified Drain". Dissect No21. VB = 0.7 V. Disconnect, VB = 0.7 V. Connect external 2.7 V. Inactive 3.2 μA. Active 141 μA. Response to 100-kΩ sweep 6 dB too low at top end in X but perfect in Y. Diagnosis "Unidentified Drain".

Poaching transmitters No1 VA = 1.9 V, reception 60%. No7 turns off when it cools to room temperature. No9, No 11, No17, and No19 have all stopped. Dissect No1, VB = 2.0 V. Disconnect, VB = 2.6 V. Inactive 2.4 μA, active 140 μA. Response to 100-kΩ sweep 10 dB too low at high end of X and Y. Diagnosis "Full Life". Diossect No7, VB = 2.6 V, now turns on. Disconnect VB = 2.6 V. Connect external 2.7 V, inactive 2.4 μA, active 135 μA. 100-kΩ sweep response 6 dB too low at high end. Diagnosis "Full Life". We don't bother dissecting No19, No11, No9, and No17 because all have run for 340 hours, and typical life is 48 mA-hr / 140 μA = 340 hours.

We have batch B206_75 consisting of ten A3028B-AA. Gain within ±0.3 dB see B206_75. B206.87 has a sharp-edged breech in epoxy on battery rim with inadequate silicone cover. Reject. Total noise in 40°C water is <9 μV. Switching noise <5 μV.

FEB-18

[02-FEB-18] We have C206.102 a delayed member of an earlier batch. Gain versus frequeny matches the rest of the batch and switching noise in 37°C is 4 μV.

[09-FEB-18] We have batch B206_45 consisting of 9 of A3028A-DDC. Volume of five of them we measure in a beaker to be 6 ml, or 1.2 ml each. We measure the volume of a single transmitter by surface reflection with a precision of ±0.05 ml (one drop) and get 1.2 ml. Frequency respose A206_45 within ±0.3 dB. Switching noise in warm water is <4 μV.

[13-FEB-18] We have EEG/EMG recordings from Edinburge. The EEG electrodes are bare wires held in place with screws, one over cortex, one over cerabellum. The EMG electrode is bare wire in muscle with silicone cap screwed back on to secure. In twenty-five hours of recording there is not a single step artifact in the EEG. Here is the spectrum of an ewight-second interval.


Figure: EEG (Green, No1) and EMG (Blue, No2). Both channels 512 SPS. EEG amplifier 0.3-160 Hz. EMG amplifier 0.3-80 Hz.

Judging by the frequency response, we are guessing that this is an A3028J-AAA. We see switching noise of amplitude 5 μV rms.

[15-FEB-18] We receive M1513090524.ndf from Edinburgh University, a recording made of intercostal muscles using an A3028B. We see heartbeat and respiration.


Figure: Heartbeat and Respiration in 0-20 Hz Portion of Intercostal EMG. Taken from M1513090524.ndf 16-s interval starting at 48 s.

The heartbeat fundamental is at 6.3 Hz, with harmonics at 13 Hz and 19 Hz. Respiration fundamental is at 2.1 Hz with harmonic at 4.2 Hz.