revision 2 of that ADC module I was posting about the other week. changes and lessons learned from the original:
- hand-soldering the TI ADS7056 ADC is impossible. this reduces yield, so I've switched to the Analog AD7276, which is worse in every conceivable way (fewer bits, single-supply, no external reference, and costs more) but it comes in SOT-23-6 so I might be able to do better than 1 working device out of 6
- clamping the output on the 1st stage amplifiers (both halves of the LMH6658, on the left) is not optional because it's inverting, so at 0 volts on the input (which will happen during power-up) the output will fly all the way up to 10 volts and make the 2nd stage amp pretty unhappy, and it's very easy to mis-adjust the 1st stage offset to get voltages exceeding 3.3V there
- zener diode probably isn't the best way to go about that due to leakage current and soft switching, but I can calibrate out any effects related to the leakage current and it's much easier to implement compared to a precision rectifier or something. I think the 1k resistor on that op-amp input makes it Technically Okay per the datasheet, since it says if you don't exceed 10 mA when over-driving an input it might be fine but doing literally anything here should take it from "might be fine" to "definitely fine"
- using a 1-turn pot for the 2nd stage DC offset was hubris, it's extremely touchy and I will gladly pay the $3 for a 11-turn pot to not have to go through tuning that in again
- putting test points (J2) on the thing designed to be calibrated in using an oscilloscope might have been a good idea
that aside, though, the analog stage on the rev1 board does work, I've currently got it wired in to the single working motherboard ADC that was suffering for lack of a frontend and it's much happier. so I'm very happy about that
