IRiS ADC Hat - Assembled & Working
The first IRiS ADC Hat is soldered and fully working! It uses two XADC78H90 ADCs for 16-channel simultaneous readout, which means 8 channels per eye, and this is the first time I've gotten the pcb's and assembled them.
Current Progress
Good news first: both XADC78H90 chips show up and have good readings that arent jumbled garbage.
The tests
Below is the output from my first test to check if the channels are detected. You can both chips are detected aswell as some noise which I can only presume is from the chips themselves or poor pcb design (or maybe that pesky capacitor)
$ python3 adc_test.py
Detecting ADCADC78H90 chips ...
ADC1 (CS=GP15): DETECTED
ADC2 (CS=GP17): DETECTED
----------------------------------------
ADC1 (CS=GP15)
--------------------
CH0: 0 (0.000 V)
CH1: 0 (0.000 V)
CH2: 0 (0.000 V)
CH3: 0 (0.000 V)
CH4: 0 (0.000 V)
CH5: 0 (0.000 V)
CH6: 0 (0.000 V)
CH7: 0 (0.000 V)
ADC2 (CS=GP17)
--------------------
CH0: 44 (0.035 V)
CH1: 0 (0.000 V)
CH2: 0 (0.000 V)
CH3: 12 (0.010 V)
CH4: 45 (0.036 V)
CH5: 0 (0.000 V)
CH6: 1 (0.001 V)
CH7: 0 (0.000 V)Next is my output from the speed test after some tinking, base on the current state of the pcb id say this speed is very acceptable.
$ python3 adc_bench.py
IRiS ADC Bench
==========================================
ADC1: OK
ADC2: OK
----- Speed -----
PIO-ADC1 (8 ch) 22472 rd/s ( 2809 Hz/ch 356.0 us/scan)
PIO-ADC2 (8 ch) 22443 rd/s ( 2805 Hz/ch 356.5 us/scan)
The Dumb Issues
Of course, assembling the actual board surfaced a few dumb issues I only noticed once it was in my hands, stuff that should've been incredibly obvious looking at the pcb but just happened to slip my mind.
Image of the dumb bypassed decoupling cap, it made sense when I was routing it i swear

None of these are hard to fix, just annoying ones I shouldve caught from the start.
Next Steps
- Actually route the 3.3V rail through the decoupling cap this time
- Add the current limiting resistor I forgot
- Give each XADC78H90 its own SPI bus so both can be read at full speed simultaneously
My expectations
Given the current speed of the boards im expecting the fixed versions to easily reach 5khz per channel, possibly even more, while being safe and easier to assemble (oops).