Subcutaneous Transmitter Development

© 2013-2021, Kevan Hashemi, Open Source Instruments Inc.

Contents

Introduction
Accelerated Ageing
Circuit Versions
2013
2014
2015
2016
2017
2018
2019
2020
JAN-20FEB-20MAR-20APR-20
MAY-20JUL-20AUG-20SEP-20
OCT-20NOV-20DEC-20
2021
JAN-21FEB-21MAR-21APR-21
MAY-21JUN-21JUL-21AUG-21
SEP-21

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 ageing 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 Ageing

[29-MAR-17] We started accelerated ageing 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 ageing 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 ageing 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 ageing 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.

We no longer keep track of all our accelerated ageing tests in this document. Each transmitter we poach has its own record sheet. We have our test results up to FEB-21 in our Development Archive. At the end of each poach, we write down our diagnosis of failure. The circuit problems are identified by two-letter codes. FL is "Full Life", CC is "Corroded Capacitor", CS is "Corrosion Short", UD is "Unidentified Drain", TM is "Transmit Malfunction", TS is "Temporary Shutdown", OG is "Out Gassing", FE is "Faulty Encapsulation", CW is "Corroded Wire". If the poached transmitter had artifact problems at the beginning of the test, we enter "NA" for the first artifact time.

Circuit Versions

[25-JAN-21] 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.
PV1 A302801Q 10μF 6.3V X5R TDK lead-free chemical X 40 Hz, Y omitted, smaller logic chip, no mosfet power switch, side-mounted battery, leads attach on side facing the programming extension, and extension used to hold circuit during encapsulation.
HV1 A302801H 4×10μF 6.3V X5R TDK lead-free chemical X 160 Hz, Y 80 Hz, mounting fixture extension eliminates mounting wire, temporary antenna included in extension board, pads for lipo batteries on one neck.
Table: Summary of Circuit Board Versions. Links are to bills of materials.

The long-running GV1, and the new HV1 assemblies come ready-made from the assembly house for single-channel 160-Hz and single-channel 80-Hz transmitters, both 1.3-ml and 2.8-ml sizes. We call these our "standard transmitter circuits". The GV1 eliminated 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 ageing. The GV1 provided a total of 40 μF of decoupling on the battery voltage, which reduced switching noise down to below 1 μV rms in encapsulated transmitters. The GV1 offered 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. The HV1 circuit replaces resistive input protection with diode clamps whose second purpose is to allow a rechargeable battery to be recharged through the EEG leads. The HV1 uses a newer, more compact logic chip with slighly lower quiescent current consumption, replaces the magnetic sensor with the same lower-power, lower-noise version we use in our pup transmitter circuit, so as to reduce inactive current consumption to 0.8 μA.

The PV1 is a miniaturized circuit with only one input channel, intended for use at 0.3-40 Hz with a side-mounted CR1025 battery to make the A3028P series of transmitters. We call it our "pup transmitter circuit". We can side-load it also with a ML621 rechargeable battery for the A3028T series, or a BR1225 for the A3028S series. This circuit consumes only 0.8 μA on the shelf.

2013

See Development Archive.

2014

See Development Archive.

2015

See Development Archive.

2016

See Development Archive.

2017

See Development Archive.

2018

See Development Archive.

2019

See Development Archive.

2020

JAN-20

[03-JAN-20] Active poaching transmitters L213.43 and L213.51 100% reception and correct response to 10-MΩ sweep. We have batch P214.6 consisting of three A3028P1 circuits equipped with ML920 batteries as IIS prototypes, 128 SPS. Frequency response P214_6. Total noise 1-40 Hz in 41°C water 4-5 μV. Switching noise <0.4 μV. We have batch S213_53 consisting of fourteen of A3028S2-DA equipped with BR1225 batteries. Ten of them displace 9.75 ml, making their average volume 0.98 ml. Frequency response P213_53 within ±1 db. Total noise 2-80 Hz in 39°C up to 16 μV due to switching noise, which consists of 10-ms spikes at 5 Hz, height 80 μV. We reject S213.65 and S213.68, leaving twelve with switching noise pulses less than 40 μV and total noise ≤12 μV, see M1578071676.ndf. We resolve to equip the S-series devices with CR1225 batteries in future, with no tabs, soldered to the circuit board directly on one edge, and with a wire to the negative tab. Both joints require acid flux, and both are best done quickly with the iron at 350°C (650F). Add S213.65 and S213.68 to active poach.

In our tests, Panasonic CR-series batteries delivered full capacity after soldering with an iron at 350°C (650F). We do not have to be quick when soldering CR-series batteries: five seconds of heating causes no degradation in battery capacity. But ML-series batteries suffer damage with a one-second application of 350°C (650F). With a 260°C (500F) iron we are able to solder directly to ML-series batteries, provided we are quick about it. If we take more than a three seconds, the battery is likely to fail. The same goes for the BR-series: a 350°C iron damages a BR-series battery, but 260°C it can tolerate for short periods. If we are not quick enough with the soldering, we will see the plastic seal bulging out between the terminals, which is a sure sign of damage and reduced capacity. We resolve never to solder ML or BR-series batteries, but to get them equipped with tabs, which we bend and shape so as to mate with pads on our circuits. When we use CR-series, we choose the battery that is most convenient: we can solder directly or use tabs. When we solder, we do so quickly with acid flux and our iron at 350°C (650F).

[06-JAN-20] Active poaching transmitters L213.43, L213.51, S213.65, and S213.68 100% reception, response to 20-MΩ sweep correct.

[07-JAN-20] Active poaching transmitters L213.43, L213.51, S213.65, and S213.68 100% reception.

[08-JAN-20] Dissect E201.124 returned from ION, made from A3028RV2 circuit. Silicone and epoxy in good condition. Inactive current 1.8 μA. Active current 3 mA rising to 5 mA over several minutes. Reception 100%. Remove C3 and C4, no change. Diagnosis: "Unidentified Drain".

[10-JAN-20] We have batch E213_67 consisting of 14 of A3028E-AA. Response to 20-MΩ sweep E213_67.gif within ±0.7 dB. Total noise 2-160 Hz in 39°C water is ≤6 μV and switching noise ≤2 μV except for E213.74, which has total noise 10 μA and switching noise 3 μV. Active poaching transmitters L213.43, L213.51, S213.65, and S213.68 100% reception, response to 100-kΩ sweep correct. Add E213.74 to poach.

[13-JAN-20] Active poaching transmitters L213.43, L213.51, S213.65, S213.68, and E213.74 100% reception.

[16-JAN-20] Active poaching transmitters L213.43, L213.51, S213.65, S213.68, and E213.74 100% reception. Response to 20-MΩ sweep correct except for S213.68, which shows some step noise during sweep. Response to 100-kΩ sweep correct for S213.68.

[18-JAN-20] We have five A3028GV1 circuits that passed calibration but all failed before encapsulation with the same symptom: 16 mA current consumption, does not turn on or off. One failed quality control. Four passed quality control and failed after battery loading. On all five boards, removing U9 MAX2624 stops the excessive current consumption. We replace U9 on four boards and get correct current consumption and 100% reception. We are unable to salvage two boards because we have already removed U3, and find ourselves unable to \replace U3. We salvage three board, check their new RF spectra, and find them centered in 913-918 MHz. We recall that the MAX2624 is sensitive to large metal objects touching its antenna output while it is turned on. We hypothesize that U9 is getting damage in this way during assembly and calibration. Active poaching transmitters L213.43, L213.51, S213.65, S213.68, and E213.74 100% reception.

[20-JAN-20] We have the following report from dissections of transmitters returned from ION. As noted above, E201.124 has excessive current consumption and won't transmit. We are unable to identify a point of corrosion in the circuit. We note black dots and bubbles in the silicone at the base of the leads.


Figure: Results of Dissecting Fourteen Transmitters Returned from ION. We check reception (Rx %), recording of mains hum (Hum), as well as measure current consumption active and inactive, and inspect silicone for cuts.

Two of the transmitters are fully-functional upon return. With their original batteries connected, we get 100% reception, full RF power observed with Spectrometer, RF center frequency correct, and response to 20-MΩ sweep is correct. We have batch A213.69 consisting of six A3028A-CCC. Response to 20-MΩ sweep A213_69 within ±0.4 dB. Total noise 2-160 Hz in 40°C water ≤12 μV, switching noise ≤4 μV.

Active poaching transmitters L213.43, L213.51, S213.65, and E213.74 100% reception. But S213.68 has stopped. Dissect. VB = 3.0 V because battery is hot. Disconnect. Notice that the 0V wire comes away from the board easily. Connect external 2.7 V. Inactive 0.8 μA, active 45 μA, 100% reception. Diagnosis "Corroded Wire".

[24-JAN-20] We have batch E213_121 consisting of 8 of A3028E-AA. This batch has suffered from five or six failures of U9 during encapsulation. Three of them, 126, 129, and 133 have U9 replaced. Frequency response E213_121 within ±0.4 dB. Total noise 2-160 Hz in 41°C water is ≤8 μV, switching noise ≤3 μV. We have batch C213_85 consisting of eleven of A3028C-CC. Frequency response C213_85 within ±XX dB. Total noise 2-80 Hz in 37°C water is ≤8 μV, switching noise ≤4 μV, except C213_94 12 μV total and 8 μV switching. Reserve C213_94 for poaching.

[27-JAN-20] Active poaching transmitters L213.43, L213.51, S213.65, and E213.74 100% reception and response to 20-MΩ sweep correct.

[28-JAN-20] Active poaching transmitters L213.43, L213.51, S213.65, and E213.74 100% reception. Add C213.93 to poach.

[29-JAN-20] Active poaching transmitters L213.43, L213.51, S213.65, E213.74, and C213.93 100% reception.

[31-JAN-20] We have batch E213_183 consisting of five A3028E-AA. Frequency response E213_183 within ±0.4 dB. Total noise 2-160 Hz in 37°C water is ≤8 μV, switching noise ≤4 μV. Active poaching transmitters L213.43, L213.51, S213.65, E213.74, and C213.93 100% reception. Response to 20-MΩ sweep correct.

FEB-20

[03-FEB-20] Active poaching transmitters L213.43, L213.51, S213.65, E213.74, and C213.93 100% reception.

[04-FEB-20] Active poaching transmitters L213.43, L213.51, S213.65, E213.74, and C213.93 100% reception.

[07-FEB-20] Active poaching transmitters L213.43, L213.51, S213.65, E213.74, and C213.93 100% reception.

[09-FEB-20] Active poaching transmitters L213.43, L213.51, S213.65, E213.74, and C213.93 100% reception, response to 20-MΩ sweep correct except S213.65, which shows −10 dB gain at 80 Hz and occasional step artifacts.

[11-FEB-20] We have batch C213_98 consisting of 9 of A3038C-CC. Frequency response C213_98 within ±0.7 dB. Total noise 2-80 Hz in 40°C water is ≤7 μV, switching noise ≤2 μV. Active poaching transmitters L213.43, L213.51, S213.65, E213.74, and C213.93 100% reception.

[12-FEB-20] Active poaching transmitters L213.43, L213.51, C213.93, E213.74, 100% reception. Device S213.65 has stopped. Diagnosis "Full Life".

[14-FEB-20] We have C213.189 and C213.190 we add to our earlier batch C213_98 of A3028C-CC. Add frequency response to C213_98. Total noise 2-80 Hz in 37°C water ≤7 μV, switching noise ≤2 μV. We have batch Q213_135 consisting of 8 of A3028Q-NDB. Frequency response Q213_135 within ± 0.3 dB. Total noise 2-320 Hz in 41°C water ≤12 μV, switching noise <0.4 μV.

[18-FEB-20] Active poaching transmitters L213.43, L213.51, C213.93, E213.74, 100% reception.

[21-FEB-20] Active poaching transmitters L213.43, L213.51, C213.93, E213.74, 100% reception. Response to 100-kΩ sweep correct.

[24-FEB-20] Active poaching transmitters L213.43, L213.51, C213.93, E213.74, 100% reception.

[28-FEB-20] Active poaching transmitters L213.43, L213.51, C213.93, E213.74, 100% reception.

MAR-20

[03-MAR-20] Active poaching transmitters L213.43, L213.51, C213.93, E213.74, 100% reception. Response to 100-kΩ sweep correct except for L213.43 has gain 6 dB too low at 60 Hz on X input.

[06-MAR-20] Active poaching transmitters L213.43, L213.51, E213.74, 100% reception. C213.93 has stopped. Diagnosis "Full Life".

[09-MAR-20] Active poaching transmitters L213.43, L213.51, E213.74, 100% reception, noise ≤8 μV.

[10-MAR-20] We have batch Q213_153 consisting of 12 of A3028Q-NDB. Frequency response Q213_153 within ± 0.3 dB. Total noise 2-320 Hz in 37°C water ≤10 μV, switching noise <0.4 μV.

[13-MAR-20] We have batch B211_43 consisting of 14 of A3028B-AA with 55 mm leads. Frequency response Q211_43 within ± 0.6 dB. Total noise 2-160 Hz in 37°C water ≤10 μV, switching noise <4 μV. Active poaching transmitters L213.43, L213.51, E213.74, 100% reception, noise ≤8 μV. Response to 100-kΩ sweep correct for E213.74, but L213.43 gain 6 dB too low at 60 Hz on X input and L213.51 reads 25±2 counts.

[20-MAR-20] Active poaching transmitters L213.43, L213.51, E213.74, 100% reception, noise ≤8 μV.

[23-MAR-20] Active poaching transmitters L213.43, L213.51, E213.74, 100% reception, noise ≤8 μV. We have batch C213_193 consisting of 21 of A3028C-DC. Frequency response C213_193 within ±0.7 dB. Noise 2-80 Hz in 37°C water ≤10 7 μV, switching noise ≤4 μV, except C213_210, switching noise 8 μV, don't ship.

[26-MAR-20] We have batch C207.169 consisting of 17 of A3028C-AA. Frequency response C207_169 within ±0.9 dB. Noise 2-80 Hz in 37°C water ≤6 μV, switching noise ≤4 μV.

[27-MAR-20] We have batch C207.188 consisting of 17 of A3028C-AA. Frequency response C207_188 within ±1.0 dB. Noise 2-80 Hz in 37°C water ≤6 μV, switching noise ≤4 μV. We hold back No189 because of a bubble in silicone.

[27-MAR-20] Active poaching transmitters L213.43, L213.51, E213.74, 100% reception, noise ≤8 μV. Response to 100-kΩ sweep correct for E213.74, but L213.43 X-gain 6 dB too low at 60 Hz and L213.51 X reads 25±2 counts. Add C213.197 (replaced yellow lead) and C213.210 (noisy) to poach.

[31-MAR-20] Active poaching transmitters L213.43, L213.51, E213.74, C213.197, and C213.210 100% reception, noise ≤8 μV. Response to 100-kΩ sweep correct for E213.74, E213.74, C213.197, and C213.210.

APR-20

[01-APR-20] We have batch B200_200 consisting of 17 of A3028B-AA. Frequency response B200_200 within ±0.4 dB. Noise 2-160 Hz in 37°C water ≤10 μV, switching noise ≤4 μV. Active poaching transmitters L213.43, L213.51, E213.74, C213.197, and C213.210 100% reception, noise ≤8 μV

[07-APR-20] Active poaching transmitters L213.43, L213.51, E213.74, C213.197, and C213.210 100% reception, noise ≤8 μV.

[10-APR-20] Active poaching transmitters L213.43, L213.51, E213.74, C213.197, and C213.210 100% reception, noise ≤8 μV.

[13-APR-20] Active poaching transmitters L213.43, L213.51, E213.74, C213.197, and C213.210 100% reception, noise ≤8 μV. Respose to 100-kΩ sweep correct for channel numbers 44, 74, 197, 210, incorrect for 43, 51, and 52.

[15-APR-20] Active poaching transmitters L213.43, L213.51, E213.74, C213.197, and C213.210 100% reception, noise ≤8 μV.

[18-APR-20] Active poaching transmitters L213.43, L213.51, E213.74, C213.197, and C213.210 100% reception, noise ≤8 μV.

[21-APR-20] Active poaching transmitters L213.43, L213.51, C213.197, and C213.210 100% reception, noise ≤8 μV. E213.74 has stopped.

[23-APR-20] Active poaching transmitters L213.43, L213.51, C213.197, and C213.210 100% reception, noise ≤8 μV.

[24-APR-20] We have batch B200_220 consisting of 16 of A3028B-AA. Frequency response B200_200 within ±0.4 dB. Noise 2-160 Hz in 37°C water ≤10 μV, switching noise ≤4 μV, except P200_202 noise ≤12 μV. We dissect E213.74. VB = 1.0 V. Disconnect battery, use external 2.7 V. Inactive current 1.7 μA, active 90 μA for a few seconds, then 1.6 mA. Increase supply to 4.0 V and current increases to 3.0 mA. Diagnosis "Corroded Capacitor", probably C1 or C3.

[27-APR-20] Active poaching transmitters L213.43, L213.51, C213.197, and C213.210 100% reception, noise ≤8 μV.

[28-APR-20] Active poaching transmitters L213.43, L213.51, C213.197, and C213.210 100% reception, noise ≤8 μV.

MAY-20

[01-MAY-20] Active poaching transmitters L213.43, L213.51, C213.197, and C213.210 100% reception, noise ≤8 μV. Respose to 100-kΩ sweep correct or C213.197. L213.51, C213.210 VA near 2.2 V, no dynamic range for sinusoidal response. L213.43 has stopped. Add B216.2, noisy device, to poach.

[05-MAY-20] Active poaching transmitters L213.51, C213.197, B216.2, and C213.210 100% reception, noise ≤8 μV, but C213.197 and C213.210 VA near 2.2 V.

[06-MAY-20] Active poaching transmitters L213.51 and B216.2 100% reception, noise ≤8 μV. Devices C213.197 and C213.210 have stopped. Diagnosis "Full Life".

[15-MAY-20] We have distributed the chore of checking the poaching transmitters, so will be reporting only diagnoses here. On 12-MAY-20 L213.51 stopped working after being left out for several hours. Today it is again working, but we dissect all the same. VB = 2.7 V. Attach external 2.7 V. Active 259 μA, inactive 2 μA. Diagnosis "Temporary Shutdown". Dissect L213.43. VB = 0.1 V. External 2.7 V, active 268 μA, inactive 3 μA. Diagnosis "Unidentified Drain".

JUL-20

[29-JUL-20] We have batch U217_9 consisting of 7 of A3028U-AAA. We measure frequency response U227_9. Our function generator has a DC offset of +1.5V or thereabouts, which takes the A3028U input too close to the top for us to measure a good-sized sinusoid, so we connect the input leads the other way around from usual, and the input moves down, in which direction it has more room to move. Gain within ±1.5 dB, which is at the limit of our specification, and occurs in the range 10-100 Hz. Switching noise <8 μV, total noise 1-160 Hz ≤25 μV in 37°C water. We have batch P204_54 consisting of 13 of A3028P1-AA. Frequency response P204_54 within ±0.7 dB except for No68 which has wandering baseline. No54 has visible white corrosion around the battery. The rest have negligible switching noise and total noise 1-80 Hz of <4 μV.

[17-AUG-20] Dissect U217.9 failed at 12 days poach 60°C. VB=0.5V, disconnect, VB=1.5V. Connect external 2.7 V. Active current 145 μA reception 100%, inactive 1 μA. Diagnosis: "Full Life".

AUG-20

[27-AUG-20] Dissect P204.57. Visible corrosion around battery edge. Remove battery, VB=0.2V. Connect external 2.7 V, active current 35 μA, device is fully-functional. Diagnosis "Corrosion Short".

SEP-20

[04-SEP-20] We have batch E203_81 consisting of 17 of A3028E-DA. Response to 100-kΩ sweep within ±0.7 dBm is E203_81. Switching noise <3 μV, total noise 1-160 Hz ≤12 μV in 37°C water.

[07-SEP-20] Poaching transmitter P204.54 has stopped. Remove silicone. Thin layer of epoxy around battery has crumbled, but see no sign of corrosion. Beneath the label, the interface between the epoxy and the battery is tacky. When we touch with iron, it generates acrid smoke. Remove battery and connect external 2.6 V. Active current 35 μA, inactive 0.8 μA. Diagnosis "Full Life". We put E203.98 and E203.99 in to poach.

OCT-20

[01-OCT-20] We have batch B205_44 consisting of 12 of A3028B-DD. Response to 20-MΩ sweep within ±0.5 dBm is B205_44. Switching noise <4 μV, total noise 1-160 Hz ≤12 μV in 37°C water. Rust on No45 and No49 due to imperfect encapsulation of battery tab corners precludes shipping. Two others 50 and 55 have traces of rust, but now tabs are well-covered and we decide to ship them.

[02-OCT-20] We have batch B213_217 consisting of 14 of A3028C-CC. Response to 20-MΩ sweep within ±1.0 dBm is B213_217. Switching noise in 37°C water <4 μV, total noise 1-80 Hz ≤10 μV.

[12-OCT-20] We have four A3028T1-R made two years ago that have been sitting on the shelf. They will not turn on. We charge with A3033B, which provides 4.2 V in series with 1 kΩ and wait four hours, still no reception, but when we connect 4.2 V through an ammeter, with no series resistor, we see 3 mA flowing in, and we have good reception. We turn off with magnet, see 0.7 mA flowing in. Disconnect 4.2 V and get reception for a few minutes. We see the same thing for the other three devices. After two years draining the battery, none of them draw more than 0.8 mA from 4.2 V.

[19-OCT-20] We have batch H207_209 consisting of 17 of A3028H-AAA. Frequency response H207_209 within ±1.2 dB. Noise 1-80 Hz in 37°C water is <12 μVrms, with most channels being around 7 μV rms. Switching noise ≤4 μV.

[26-OCT-20] We have two A3028C-CC that won't turn on after epoxy encapsulation. Batteries are both drained. Current consumption from external 2.6 V is 17 mA for both, no signal received, no change when we apply magnet. Replace U9, MAX2624, on one, and now we see 52 μA active current consumption and 100% reception. These failures appear to be identical to those we began to encounter in January.

[30-OCT-20] We have batch C217_25 consisting of ten of A3028C-CC. Frequency response C217_25. Total noise in 37°C water ≤10 μV, switching noise ≤2 μV.

NOV-20

[09-NOV-20] We have batch H218_23 consisting of 18 of A3028H-AAA. Frequency response H207_23_A and H207_23_B within ±1.0 dB. Noise 1-80 Hz in 37°C water is <9 μVrms. Switching noise ≤3 μV.

DEC-20

[14-DEC-20] We have batch C217_39 consisting of 14 of A3028C-CC. Frequency response C217_39 within ±0.8 dB. Total noise in 37°C water ≤10 μV, switching noise ≤3 μV.

[16-DEC-20] We begin layout of the A302801H printed circuit board, which will be the basis of a new two-channel A3028H assembly, schematic S3028H_1. Changes with respect to the previous A3028GV1 are as follows. Replace U8 LC4064ZC in 6-mm × 6-mm BGA-56 with LC4064ZE in 5-mm × 5-mm BGA-64 as used on A3028PV1. Replace obsolete ASH7KW oscillator with new ASTMTXK in ASTM package, as used on A3028PV1. Replace unreliable and impossible to hand-replace SN74AUP1G80 in UDFN-6 with same device in SOT-353. Replace obsolete SL353LT in SOT-23 with S5712A in smaller SNT-4 package. Remove series resistors R25 and R26, as well as RF filtering capacitors C17 and C18. Add recharging diodes D1 and D2 in P0402 packages, as on A3028PV1.

[26-DEC-20] We solder 10-KΩ resistors to both terminals of two CR1225 coin cells using acid flux and a 650F iron. One battery is by Celewell, the other by Wama. With 300 μA flowing into the resistor we expect the 48-mAhr batteries to last for 7.4 days. Two days later we find them exhausted. The seal between the battery tabs of both batteries is bulging and the battery surface tastes like soap. We mount two more batteries in holders and solder the resistors to the holders. It is 4 pm 26-DEC. We solder BR1225 with side tabs (Digi-Key P290) to pup transmitter circuit boards by bending the tabs. They come out fine, and we are not heating the battery.

[28-DEC-20] The two CR1225 batteries in holders with 10-kΩ loads are at 2.8 V after 2 days at 300 μA.

[29-DEC-20] We have batch C217_81 consisting of 11 of A3028C-CC. Frequency response C217_81 within ±0.7 dB. Noise 1-80 Hz in 37°C water is <12 μVrms. Switching noise ≤4 μV.

2021

JAN-21

[13-JAN-21] We have MAX2624 made by Rochester Electronics. We load one onto an A3014MT modulating transmitter and re-wire the tuning input to the !SHDN pin on the chip. We trim the TUNE input to give 915 MHz and mix the output through −10 dBm with +7 dBm of 900 MHz. We view IF terminated with 50 Ω on our scope along with a 3.3-V shutdown signal, to see if the duplicate part can turn on and off fast enough.


Figure: Enable Signal (Yellow) and RF Output (Red) for Rochester Electronics MAX2624. The RF output is attenuated by 10 dB then mixed with 900 MHz in a ZAD-11.

We do the same with the MAX2624 made by Maxim Integrated Circuits.


Figure: Enable Signal (Yellow) and RF Output (Red) for Maxim Integrated Circuits MAX2624. The RF output is attenuated by 10 dB then mixed with 900 MHz in a ZAD-11.

We are confident the Rochester part will be a drop-in replacement for the Maxim part we have been using for the past fifteen years.

[24-JAN-21] We receive first article of five pieces of A3028HV1 from our assembly house. The boards function perfectly, with the exception of one lifted pin, shown below, which we fix easily.


Figure: Lifted Pin U7-6 on A3028HV1.

Inactive current consumption is 0.8 μA. Active current consumption at 512 SPS is 72 μA, compared to 79 μA for the A3028GV1. We can expect a 10% increase in operating life for all SCTs made with this circuit, and a doubling of shelf life. We have batch B205_62 consisting of 12 of A3028B-DD. Frequency response B205_62 within ±0.5 dB. Noise 1-160 Hz in 37°C water is <12 μVrms. Switching noise ≤4 μV.

[26-JAN-21] We have batch S217_56 consisting of twelve of A3028S2-DA with 45-mm x 0.5-mm leads and 13-mm helical antenna. The D-pin termination is missing on three devices, No62, 66, and 70. They are loaded with BR1225 batteries by bending and soldering tabs, rather than by soldering directly to the battery. We decided against soldering directly to BR-series batteries because they are easy to damage with the iron. We purchased CR1225 batteris with no tabs from two non-Panasonic manufacturers, and these proved to be vulnerable to soldering as well, unlike the CR-series batteries made by Panasonic. Frequency response S217_56 within ±0.7 dB, total noise in 37°C water is ≤12 μV in 1-80 Hz, but we see significant switching noise because of the high source resistance of the BR-series batteries. See M1611697994.ndf for today's recording at roughly 37°C from the entire batch. We reject S217_70 for 8 μV amplitude, 5-Hz switching noise. The rest have switching noise 0-5 μV and are ready to ship. We order these CR1225 batteries from Renata via Digi-Key in the hope they may prove as resistant to soldering as the Panasonic CR-series batteries.

[27-JAN-21] We place batch S217_56 in water again, this time starting with 48°C, and record. Switching noise in the worste offender, S212_70, is now zero. At 40°C, No70's switching noise has risen to 2 μV, and appears as 50-μV pulses at 5 Hz. We allow water to cool to 38°C and record for five minutes, by which time the temperature has cooled to 36°C, see M1611772233.ndf. We note that the pulses in No66 are opposite to those in the rest of the devices. We are struck by the absence of low-frequency noise in the three devices with soldered pins missing, compared to continuous rumble from those with pins.

[29-JAN-21] We have batch S219_74 consisting of 11 of A3028S2-DA. Frequency response is 0.3-40 Hz instead of 0.3-80 Hz, so we must reject them all, see S219_74. Noise in 37°C water is ≤12 μV and switching noise is ≤5 μV, except S217_99, which is 8 μV and we reject for that reason.

FEB-21

[03-FEB-21] We have batch B205_76 consisting of 9 of A3028B-DD. Frequency response B205_76 within ±0.3 dB. Noise in 37°C water is ≤5 μV rms, except No83 which is 8 μV because of switching noise fundamental of 4 μV. Switching noise of remainder is <2 μV.

[04-FEB-21] We have CR1225 from Renata Batteries, via Digi-Key. We use 650F iron and acid flux to put two solder blobs on one sample, clean it up, put it in battery holder with 10 kΩ load. We put an un-altered battery in another holder. Both have voltage 3.1 V. It is 14:03. We expect 178 hrs battery life.

[12-FEB-21] We have two more boards with the symptoms recorded on 18-JAN-20, "16 mA current consumption, does not turn on or off".

[15-FEB-21] We have batch H218_65 consisting of 15 of A3038H-AAA-B45-B made with the last of our A3028GV1. Volume of 10 is 12.0 ml. Thickness 7.7 mm, same as two earlier devices we have left from a previous batch. Frequency response H218_65 within ±0.5 dB. Noise in 37°C water 2-80 Hz is ≤7 μV, switching noise <2 μV, except for H218_89, for which noise is 11 μV and switching noise is 4 μV, hold it back for poaching, 500-s archive is M1613410322. We have batch B216_14 consisting of 15 of A3028B-AA-B45-B. Volume of 10 is 12.2 ml. Frequency response B216_14 within ±0.5 dB. Noise in 37°C water 2-160 Hz is ≤8 μV and switching noise ≤2 μV except B216.23, for which noise is 12 μV and switching noise is 5 μV. Our two CR1225 batteries have run down. Both batteries were >2.3 V after 170 hrs. The un-heated battery was 2.3 V after 190. The heated battery had dropped to 1.9 V. At 300 hrs, both are at 20 mV. We conclude the battery capacity is near 48 mAhr.

[17-FEB-21] We load three CR1225 batteries onto three A3028PV1 circuit boards programmed as A3028S5, 2048 SPS, current consumption 230±2 μA. We use acid flux and 650F iron, see CR1225_Soldering video. Place in Faraday enclosure, unencapsulated, noise in 2-160 Hz is 5-6 μVrms, switching noise less than 0.4 μV. We leave running in an enclosure, recording battery voltage.

[19-FEB-21] We have batch C218_99 consisting of 17 of A3028C-AA-B45-B. This is the first batch of transmitters made with the new A3038HV1 assemblies, which are 0.5 mm shorter, and have the battery 0.2 mm closer to the circuit baord. Volume of 10 pieces is 10.6 ml. Outer dimensions 13.3 mm × 13.3 mm × 7.5 mm. Previous A3028C made with A3028GV1 was 13.5 mm × 14.5 mm × 7.5 mm. Noise in 37°C water is dominated by 5-Hz switching noise with amplitude up to 6 μV and total noise 2-80 Hz is up to 15 μV rms. Three have total noise >12 μV rms, see first eight-second interval in M1613781261.ndf. Minimum noise is only 4 μV rms. We reject No101 (15 μV rms) and No104 (13 μV rms). At the end of the archive, the water has cooled to around 30°C and the noise in No116 has increased from 10 μV rms to 16 μV rms. See original study of switching noise versus temperature here. We order 100 of CR1223AH by Renata, looks like this, with mounting pins suitable for hand-soldering, but "not suitable for reflow soldering", which makes them perfect for us, they are a drop-in replacement for the BR1225AHB.

[22-FEB-21] Of three soldered CR1225 we left running, one has dropped to 2.2 V and the other two are still going strong with 2.6 V after 115 hours, when we expect 209 hrs operating life.

[26-FEB-21] S217.59 failed after 24 days poaching. Battery 1.0 V. Disconnect battery. Connect external 2.7 V. Active current 46 μV, inactive 0.8 μV. Diagnosis "Unidentified Drain". We have batch S218_119 consisting of 13 of A3028C-AA-B45-B. We place in water at 40°C and record as water cools to 36°C, recording is M1614369326.ndf. Switching noise pulses vary from <1 μV to 10 μV. We reject the three with the greatest noise, leaving eight with pulses <1 μV and two with pulses of 5 μV.

MAR-21

[01-MAR-21] We have the completed battery drain for our three soldered CR1225 batteries from Renata. Plot of battery voltage versus time, as deduced from transmitters with 230 μA active current is here. Expected life 208 hrs. Actual live 110, 170, and 200 hrs.

[03-MAR-21] We acquire a Sunkko 788H spot-welding machine. We cut thin nickle strips and weld them to Renata CR1225 batteries. We use 40 V, 2 pulses, and 50% power. We are pressing the nickle and battery up against the two welding electrodes attached to the welding unit. The nickle strips are 100 μm thick. We solder the nickle strips to our three A3028S5 circuits. Switching noise pulses 2 μV in No226, not visible in the other two.


Figure: Nickle Tabs Spot-Welded to Renata CR1225 Batteries.

We place A3028S5s in Faraday enclosure to run down their batteries at 230 μA.

[04-MAR-04] We spend some more time spot-welding, this time with the extension leads to hand-held electrodes. We have 100-μm thick nickle strips. We cannot get strong welds with 40 V, 2 pulses. But we do get strong welds with 80 V, 1 pulse. One in five welds does not produce strong joints. We switch to 2 pulses and every weld is strong. When we switch to 150-μm nickle, all bonds are weak, no matter how hard we press. We cannot weld 150-μm nickle to 150-μm nickle, but we can weld 100-μm nickle to 150-μm nickle.

[05-MAR-21] We have fifteen unencapsulated A3028C-AA-B45-B loaded with CR1225AH batteries. No12 is GV1 circuit, the rest are HV1. Turn them all on and put in 60°C oven for ten minutes, remove to Faraday enclosure and start recording for ten minutes as they cool down, see M1614972869.ndf. Througout recording, switching noise fundamental is <2 μV. Total noise 2-80 Hz ≤5 μV rms.

[10-MAR-21] We have batch B212_12 consisting of 15 of encapsulated A3028B-AA-B45-B made with A3028HV1 and CR1225AH battery. No22 shows rust stain in water and we find a breach in the silicone on the battery rim. The silicone on all rims is 200 μm over epoxy of 100 μm, so far as we can tell. We turn all on and place in 37°C water. No visible switching noise, except in No22, which shows sharp 5-Hz spikes of 30 μV when we immerse the breach in the water with the leads. Volume is 1.0 ml. We are going to apply more silicone.

[11-MAR-21] Our spot-welded Renata CR1225 have run down their batteries after 220 hours, where expected life is 208 hrs.


Figure: Life of Renata's 48 mA-hr CR1225 Coin Cell with 230 μA Drain and Spot-Welded Tabs.

[14-MAR-21] After one additional coat of MED-6607, batch B212_12 encapsulation looks rugged and thick. Volume is 16.5 ml for 15 = 1.10 ml, up from 1.0 ml prior to the final coat. We see antenna noise on No22 at first, but the noise does not persist. No19 has deformed silicone on battery rim due to previous breach. Frequency response B212_12 within 0.25 dB. Noise 2-160 Hz is 4-5 μVrms. Switching noise in 37°C water is <0.4 μV, see M1615748047.ndf.

[15-MAR-21] We have batch C217_110 consisting of 15 of A3028C-CC-B45-B, made from A3028HV1 with CR1225AH batteries, one coat of MED10-6607 plus three coats of MED-6607. Volume of 12 is 13 ml, individual volume 1.08 ml. Outermost dimensions 13.5 × 12.9 × 7.3 mm3. Frequency response C217_110, switching noise in 37°C water is <0.4 μV and total noise 2-80 Hz is ≤9 μV rms (note: soldered screw electrodes).

[18-MAR-21] We have C218.124 encapsulated in epoxy with one coat of SS5001 and one coat of MED-6607. Volume 1.3 ml, looks rugged. We have a batch of 12 of A3028C-AA-B45-B encapsulated in epoxy with three coats of MED-6607. Volume of 12 is 12.8 ml, 1.07 ml each. After comparing these devices to older ones made with the A3038GV1 circuit, we decide to coat them once more.

[22-MAR-21] Batch B212_29 consists of twelve of A3028B-AA-B45-B with four coats of MED-6607. Volume 1.14 ml. Frequency response B212_29, total noise 1-160 Hz in 37°C water is 5-6 μV rms, switching noise cannot be detected at all, even in 16-s intervals.

[25-MAR-21] Batch H205_89 consists of eight of A3028H-DAA-B90-B with four coats of MED-6607. Mass 20.20 g ÷ 8 = 2.52 g. Volume 9.9 ml ÷ 8 = 1.23 ml. Frequency response H205_89. Total noise in 37°C water is 5-6 μVrms for Y with bare wire and ≤12 μV for X with soldered D-pin, no sign of switching noise.

[31-MAR-21] We have batch E203_100 consists of 18 of A3038E-DA-B150-A equipped with BR2330 batteries and built with A3038HV1 circuit boards. We have some trouble constraining the leads during rotation and cutting the programming extension after epoxy encapsulation. Mass 5.9 g for one, volume 31 ml for ten, making each 3.1 ml. Frequency response E203_100, total noise in 37°C water 2-160 Hz is ≤10 μV, switching noise ≤4 μV.

We construct the first A3038J prototype, 3-200 Hz passband with differential amplifier input and a Y− lead. We connect 100-kΩ, 30-mVpp sweep to Y+/Y− and see 7000 counts rms amplitude, or 28 mVpp with the prototype sensitivity of 1.4 μV/cnt. Frequency response is roughly 3-200 Hz, with a single-pole filter on either end. We connect the same signal as common-mode to both Y+ and Y−, while connecting the signal ground to C, and see 71 counts rms amplitude. Our common mode rejection ratio is 100, or 40 dB.

APR-21

[05-APR-21] We have two A3028J prototypes with all four leads and antennas loaded. We connect X+, Y+, Y− to a 10-Hz sinusoid and X− to signal ground. In No1 we see 12162 counts rms on X, meaning 5.0 mV rms, and 67 counts rms on Y, meaning 93 μV rms. Common mode rejection ratio is 54. In No3 we see 5 mV and 70 μV for CMRR 71.

[07-APR-21] Batch H205_107 consists of eight of A3028H-DAA-B45-B made with A3028HV1 and CR1225AH. No switching noise, total noise less than 10 μV when we ignore chemical artifact generated by the D-pins. Frequency response H205_107.

[07-APR-21] We have an A3028E-DA-B150-A made with CR2430AH 285-mAhr battery, E203_119. Volume is 3.2 ml, mass 6.7 g. We have three of the same type made with BR2330 255 mAhr batteries, volume 3.2 ml and mass 5.8 g. Frequency response E203_119. No sign of switching noise in E203_119 or E203_122, but E203_120 and E203_121 have harmonics of amplitude 2 μV. Total noise 2-160 Hz is ≤10 μV rms, dominated by the low-frequency rumble we associate with soldered D-pins.

[12-APR-21] We have two A3028J-AAAA-B45-B made with CR1225AH and A3028HV1. Response to 16 mVpp sweep is J203_1.

[15-APR-21] We have batch C217_129 consisting of eight A3028C-CC-B45-B made with CR1225AH and A3028HV1. Frequency response C217_129.gif. Noise in 1-80 Hz in 37°C water 5-6 μV rms. No sign of switching noise.

[28-APR-21] We have batch C217_137 consisting of twelve A3028C-CC-B45-B made with CR1225AH and A3028HV1. After three coats of silicone, volume was 1.0 ml. No140 had a breach in silicone over battery tab, which causes sinusoidal waves in Y. No142 had fiberglass spike. After a fourth coat, volume is 1.12 ml. Frequency response C217_137.gif. Noise in 1-80 Hz in 37°C water 5-6 μV rms. No sign of switching noise.

MAY-21

[02-MAY-21] We have batch C217_152 consisting of 13 of A3028C-CC-B45-B. After three coats of MED-6607 they are 1.04 ml with sharp corners on the battery rims and tabs. After four coats they are 1.13 ml, all surfaces are rounded and silicone looks thick and tough. Frequency response C217_152. Total noise 2-80 Hz 4-7 μV rms, no sign of switching noise.

[06-MAY-21] We have batch L217.183 consisting of 12 of A3028L-AAA-B150-A. Volume of 6.0 ml. Response to 20-MΩ sweep L217_183. Noise in 37°C water 2-320 Hz is 6-8 μV rms. No sign of switching noise. Place all twelve in an FE3A with four antennas, spread them out on shelf, and see 10% loss due to collisions.

JUN-21

[04-JUN-21] We equip three A3028S2 transmitters with 25-mm, 35-mm, and 50-mm helical lead loop antennas. We are using our 0.7-mm diameter helical lead with clear silicone and we are holding the tip of the loop to the transmitter body opposite the input leads using a rubber band. We use an A3038A tracker and an A3027E receiver to compare reception in air and water for these three and two A3028S2 with the established 13-mm helical straight antenna. We are operating out on our bench in the presence of interference. In air, the 35-mm loop does well, reception twice as good as 13-mm straight and slightly better than 50-mm loop. In water, the 13-mm straight appears to be best, but only marginally so.

[05-JUN-21] We move our A3028S2 reception test into our Faraday canopy (FE5A). We have five transmitters. We hold each in hand with antenna in air and move over the entire extent of the A3038A platform from height 1 cm to 10 cm and estimate reception by watching the ALT's reception light. We move slowly so as to allow the light to turn off in a reception dead spot. We repeat with the transmitter in the bottom of a beaker, its antenna resting upon the glass. All antennas are made of our 450-μm diameter 316SS springs, MDC13867A. No74, 13-mm straight, 95% in air, 100% in water. No55, 13-mm straight, 50% in air, 98% in water. No92, 25-mm loop, 100% in air, 98% in water. No75, 35-mm looop, 75% in air, 98% in water. No97, 50-mm loop, 100% in air, 95% in water.

[11-JUN-21] We test reception from S217.55 (equipped with C-antenna) and C217.152 (equipped with B-antenna) over an A3038A tracker platform. We place each transmitter on its own in a 50-ml beaker and move the beaker around over the platform, rotating at random. We start with no water, then add water in increments and repeat. We are estimating reception with the Receiver Instrument.


Figure: Reception versus Volume of Water in Beaker. Transmitters with 30-mm stranded loop (B) and 13-mm straight helix (C) sit in the bottom of a 50-ml beaker with some water and we move the beaker over an ALT platform.

We perform the above experiments in an FE3A enclosure with the door open. When we encounter poor reception, we close the door to check if the problem is interference. Poor reception arising from water volume 10-20 ml does not improve when we close the door. We perform the above experiment in front of an A3015L antenna connected to an A3027E and measure reception with the C-antenna moving 5-30 cm away. For 0 ml of water we get 80% reception, for 15 ml we get 50%, and for 40 ml 95%. According to Oregan State University, dissolved salt greatly increases the loss factor of a water dielectric. In our Faraday canopy we start with S217.55 in 40 ml of 1% saline, watching the ALT reception indicator. We estimate 50% reception. As we reduce the volume of saline down to 0 ml, reception rises to 100%.

[18-JUN-21] We have batch C217_211 consisting of thirteen A3028C-CC-B45-B. Silicone is discolored and patchy, but otherwise robust and tough. Frequency response C217_211. Total noise in 37°C water ≤8 μV, no sign of switching noise.

[24-JUN-21] We have batch C219_2 consisting of fourteen A3028C-CC-B45-B. Silicone is perfect. Frequency response C219_2 within ±0.8 dB. Total noise in 37°C water ≤8 μV, no sign of switching noise. Volume 1.09 ml.

[28-JUN-21] Complete the A302801K layout, Revision 1 for the A3028KV1 assembly, schematic S3028K. The new assembly has the same outline as the A3028HV1, but uses the LC4064ZC-75M56C in BGA-56, the same chip we used in the A3028GV1 and earlier circuits. We just obtained 3400 pieces at a good price. The BGA-56 is 6 mm × 6 mm, compared to 5 mm × 5 mm for the BGA-64 we use on the A3028HV1. To accommodate the extra chip area, we eliminate two 10-μF capacitors C19 and C20. Our use of the CR1225 reduces switching noise below our ability to detect, and we deem that two fewer capacitors will improve our corrosion resistance. We eliminate the hole in the center of the board that we have been using for the negative battery tab. We are going to spot-weld our own tabs onto batteries from now on, wrap the tabs around the diagnoal corners of the circuit, and solder to bottom-side pads. We add C17 and C18, which will not be loaded in the A3028KV1 assembly, but which permit us to modify the circuit for EMG or EKG monitoring with a differential amplifier, as in A3028J.

JUL-21

[02-JUL-21] We receive 100 of A3028HV1, build B103097. We pass five A3028C versions through QC1. One of them has gain 6 dB to high at 1 Hz and decreasing to 80 Hz when driven by 20-MΩ source, but correct response when driven by 50-Ω source. We replace R12, C12, and wash. After washing, sweep response is correct for 20-MΩ source.

[21-JUL-21] We have our first batch of A3028WZ consisting of four A3028WZ-AAA-B45-D, equipped with the new slender D-Antenna, designed for use with two depth electrodes and a cerabellum reference electrode to measure cortical spreading depressions. Bandwidth 0.0-40 Hz, gain ×10, 128 SPS on each of two channels. Frequency response WZ215_13. Sensitivity is 4.1 μV/cnt. Volume 1.0 ml. Total noise 1-40 Hz at 37°C is 18-20 μV rms. Switching noise ≤4 μV, see spectrum.

AUG-21

[15-AUG-21] We have first article of A3028KV1, using LC4064ZC. Code is P3028C. Program and calibrate as A3028B, quiescent current 79 μA, inactive 0.7 μA. As A3028H, current is 50 μA, as A3028C: 50 μA. More of them as A3028C active current (μA): 50, 49, 50. As A3028B: 78, 77, 77, 76, 76. All have correct frequency response. With fck_divisor = 28 and frequency_low = 8 we obtain robust reception from all nine devices. Loading the custom tabbed CR1225 is straightforward, although we would like both tabs to have larger pads on the top side for this arrangement. We wash three times then remove battery and see flux residue on top side. Matt solder mask makes it harder to see the flux.

We have five A3028PV1 circuits from build B99023 of January 2021 that are drawing excessive current. In the three that draw 15 mA, we cannot program the logic chip, but when we remove the logic chip (U8) the current drops to 30 μA. If we remove U7 or U9 first, we see no reduction. The other two we find have intermittent behavior as we press the logic chip package. By reflowing the logic chip balls we can stop the intermittency, but we make excessive current permanent. We program another A3028PV1 with the P3028B firmware, whose correct target is the LC4064ZE on the A3028HV1 assembly, not the LC4064ZE on the A3028PV1. The result is 15 mA current consumption and we cannot re-program the board because 1V8 is at 1.0 V. We connect external 1.8 V to the 1V8 and we are able to re-program, active current 32 μA. We program 23 A3028PV1 with P3028P firwmare as A3028P1. Current consumption 31-34 μA except one, which is 102 μA no matter how we try to erase and re-program. We wash this one and its current drops to 32 μA. So we have 24 out of 24 passing program and current consumption.

[26-AUG-21] Changes to A302801K circuit board. To make Revision 2. Move OSI and version scripts to bottom side and put large white writing area on top side. Increase size of top-side battery pads. Mask antenna and lead pads on bottom side. Decrease width of antenna pad and move its edge away from the enlarged negative batery pad. Specify glossy solder mask. Move P2 away from P4 or rotate it so that we can get access to P4 with a 2-way socket. Changes to A302801Q circuit board. To make Revision 2. Add an isolated pad for antenna loop. For antenna and lead connections we decide upon a standardized pad and spacing: 35 mil pad with 10-mil hole, spaced 55 mils.

[29-AUG-21] C219.51 with custom battery tabs ran from 15-JUL-21 until today, 46 days.

[30-AUG-21] We add an antenna to the neck of the A302801QR2 layout. A 5-mil track running 10 mil below the ground plane to the neck, then extending 18 mm with zig-sags along the neck with ground and power planes cleared above and below. On its way to the neck the distance is 6 mm. Impedance is 100 Ω and wavelength 170 mm, so 0.035 wavelengths. When cut at the neck, the stub looks like −400i Ω, which we can ignore. The 18-mm antenna will be adequate during calibration for robust reception.

SEP-21

[04-SEP-21] We put three ugly A3028C in to poach. We shipped six other ugly ones in a batch of twenty to ION, and we are confirming that the ugly transmitters do not corrode any faster than good-looking transmitters. Photo here. Prior to noise measurement we note poor silicone coating over rim of battery of No82, and spot of rust on battery rim. Noise in 37°C recorded in M1630769873.ndf, with one transmitter No82 showing more jumps than others.

[17-SEP-21] In the near future, we are moving over to a one-wash transmitter assembly process. We calibrate and program, then load leads, antenna, and battery tabs. Wash two or three times until perfectly clean. After that we perform Quality Control One (QC1) with the battery tabs sticking straight up. We must avoid breaking the tabs off while handling the circuits.

  1. Examine circuits for flux residue.
  2. Examine circuit P0402 modifications if present.
  3. Examine battery tab solder joints.
  4. Examine leads for kinks and breaches.
  5. Leads should be requested length ±2 mm.
  6. Diameter of leads: B-Leads 0.6-0.8 mm, C-leads 0.4-0.6 mm.
  7. Length of antenna: A-Antenna 50±3 mm, B-Antenna 30±2 mm.
  8. Examine electrodes. A-Electrode 3-mm of wire, no silicone residue inside, all others check solder joints.
  9. Current consumption, active and inactive.
  10. Transmit clock period should be 195-218 ns.
  11. Microwave center frequency should be 913-918 MHz.
  12. Sensor frequency response: apply sweep and watch.

We clip off the programming extensions, sand the cut thoroughly, blow off particles of copper and fiberglass with compressed air. Load battery at spot-welding station. Measure and record battery voltage in QC sheet, should be 2.8-3.1 V. Encapsulate. In Quality Control Two (QC2) we do the following.

  1. Inspect encapsulation for defects.
  2. Examine circuit board cuts through silicone, looking for loose copper particles.
  3. Examine electrodes.
  4. Measure average volume by placing several in graduated cylinder.
  5. Measure response to 10-MΩ frequency sweep.
  6. Place in 37°C water and measure noise, in particular look for switching noise. Fundamental harmonic of switching noise must be ≤6 μV in Neuroplayer's spectrum view. Total noise above 2 Hz must be ≤10 μV rms.

Batch E212_70 showed switching noise up to 6 μV (No81, held back for poaching). These A3028E are powered by BR2330, 255 mAhr, 3.2 g batteries. The CR2330 has capacity 265 mAhr and weighs 3.8 g. The total weight of the existing A3028E is 5.8 g. If we switch to the CR2330, we will gain 4% battery life, eliminate switching noise, and add 10% to the weight. We are ordering some CR2330 coin cells to which we can spot-weld.

[20-SEP-21] A3028S2-AA number S218.142 is one of our earlier devices with spot-welded battery tabs. We started it on 02-AUG-21 it stops some time 17-SEP-21 to 20-SEP-21, making its operating life at least 46 days, where our prediction is 40 days.

[27-SEP-21] We are looking at running the ispLever compiler and ispVM programmer from the DOS command line or a TclTk console. We enter the following command at the DOS prompt and are able to program an A3028KV1:

c:\Kevan\A3028\P3028C> \ispLEVER_Classic1_7\ispvmsystem\ispVM -infile P3028C.xcf

Or from Lattice Tcl 8.3.2 console or LWDAQ 10.3.2 console:

exec [file join / ispLEVER_Classic1_7 ispvmsystem ispVM] -infile P3028C.xcf

We use the Lattice Tcl recorder to record the Tcl commands required to compile the P3028C firmware. We save to Compile.tcl. From the Lattice Tcl 8.3.2 console we can compile the P3028C01.abl file and write the JEDEC file with:

source Compile.tcl

The same thing works from our LWDAQ console, but errors in compiling are not detected, and the process continues despite a failure to compile the ABEL code. We note that the Tcl in LWDAQ is 8.7.3.

OCT-21

[15-OCT-21] We receive 100 of A3028KV2, same as KV1 but larger battery pads, more space around battery pads, glossy solder mask, and a few silk screen changes and connectors spaced better. We measure active current versus sample rate of four of them and get intercept 20.7 μA and slope 0.11 μA/SPS. What's remarkable is the consistency of the active current: standard deviation from the straight line fit is 1.2 μA.

[19-OCT-21] We try out the MAX2623 on an A3028KV2 assembly. The 23 version operates roughly 50 MHz below the 24. We estimate that the value of frequency_low (f_low) in P3028C01 should be twelve counts higher than for the 24, based upon a historical slope of 4 MHz/step. We start with U9 = MAX2624 then change to MAX2623.

U9f_lowf_center (MHz)
MAX26248909
MAX26249914
MAX262410918
MAX262310870
MAX262317895
MAX262321911
MAX262322915
MAX262323920
Table: Transmit Power Center Frequency for Various Oscillators and Tuning Voltages.

We obtain 100% reception with the MAX2623 and f_low 21-23. We will ship MAX2623 to our assembly house for a run of 200 of our new A3028PV2 circuit. The MAX2622 operates 50 MHz below the MAX2623, which would require our f_low to be 34, out of the range of our five-bit tuning DAC. The MAX2623 if more available than the 24, and less expensive.

We have 9 of A3028E made with CR2330 coin cell, soldering tabs onto circuit, single wash, spot-weld tabs to battery with help of kapton tape insulator, and encapsulate. Volume is 2.8 ml, mass 6.1-6.2 g. We have 2 of A3028Q made with CR2450 in the same way, volume 4.5 ml, mass 8.8-8.8 g.