esp8266lighthouse

Project log as of 5/28/17

So, general summary / log for esp8266lighthouse: It works. Surprisingly well.

1. If the base station is too close, it the signal quality gets pretty rough. Too close = 0.8 meters or closer. This may not be resolvable without a different opamp. I don't know yet.
2. There are some corner cases that will cause the ESP to freeze. This is sad-making, however shouldn't be hard to troubleshoot.
3. The algorithm I'm using to determine frequency seems SLIGHTLY dependent on strength. When pointed directly at the lighthouse and close, it seems to read slightly high (by about 10-15 kHz) When at glancing angle or barely visible, it reads spot on.
4. The noise issue due to the ESP is really confusing to me. Arbitrary moving something or switching power plugs around drastically changes the performance of the system.
5. The ESP8266 will draw ~14mA even in modem-sleep mode all the time if it is monitoring for a lighthouse. This is sad. I want to move to an FPGA, but because of the pinout on the ICE40 FPGAs and the fact they never put them in a reasonable package, PCBs for them cost about 10x what normal PCBs cost.
6. Though I can precisely tell frequencies, I cannot simultaneously decode pulses. Each pulse must have about 2uS between them in order to be treated as different pulses. If we run lighthouses asynchronously, there will be some collisions. They will need to be ignored.
7. The base stations CAN be run asynchronously and at different rates. It seems to me that if the lighthouses are operating at the same frequency, they cannot disambiguate asynchronous pulses, but, we should be able to.

Internal notes:

NOTES for transimpedence2/3
TESTS Performed with no C1, TSV992
Making R4 220k instead of 47k seeems to have little effect, but does make it worse.
Switching R1 for 47k just makes it appear drunk. There's some weird feedback thing going on.
    Like it tuned wrong or something.   Making it 10k quiets the signal a lot.
REMOVING C4 no apparent impact.

R5=10k, C5=10p -> R5=4.7k, C5=18p =>> Improvement.
Adding C1=1pF decreases amplitude a little bit but also cleans up edges.
Removing C6 seems to reduce amplitude slightly.

Far peak:
C1 none C6 none => .1238V noise, .998V Far Light
C1 none C6 1pF    => .1238V noise, .498V Far Light
C1 1pF  C6 none => .1135V noise, .358V Far Light
C1 1pF C6 1pF     => .087V Noise, .202V Far Light

Recommendation: Remove C6, C1.

Out of the following op amps...
TSH82, TSV992, AD8052, TSH22, TSH462,
TSV992 demolished.


Useful photodiodes:
PART / NOISE / FAR / Cost
PD70-01C/TR7:  .159 / .418 / .47
PD438C: .237 / .550 / .39
QSD2030F: .216 / .5363 / .30
BPW34/S-Z: .283 / .6345 /.497
PD15-216/TR8: UNUSABLE
VBPW34FAS: .175 / .815 / .778
VBPW34FASR: .2631 / .644 / .778
BPW34: .237 / .577 / .62
PD15-22C/TR8: Unworkable.
PD15-22B/TR8: Unworkable.
PD70-TR10: .2063 / .4024 / .41
BPW46: .247 / .968 / .61

Winners: BPW46, BPW34/S-Z, BPW34FAS + VBPW34FASR
Major Winners: BPW46/VBPW34FAS