Closed JochiSt closed 1 year ago
The dac step fit for version 2 is the same than the one for version 1. There is 1 LSB difference.
In this channel the wiggles of the low gain are not observable. I suspect an issue with the gain switching amplifier in channel 0.
To exclude the problems with the gain switching amplifier, I reduced the resistors by a factor of 10, so that R47 = 47k and R51 = 5k6. Same for R46 = 33k and R50 = 5k6.
By using the lower resistors, this effect is gone. There is now a gain mismatch, but this might be wrong population of the resistors. If I get this channel working, I'll update the other one as well.
The gain mismatch was a swap of the population, I did not mirror the board. Thus the U gain and I gain has been mixed up.
It seems, that there is a slight voltage and current bias in the low gain with respect to the high gain. (see attached image)
This can be calibrated out (not working 100% in the image above) and by this the measured resistances match much better.
The offsets have been implemented in b254d9760f564fae4d8d9865fb7bdbdeee8afc4e. The default values are somewhat arbitrary, they have to be tuned to match the exact board.
after tuning the offsets, the resistance at the saturation border of the high gain is rather smooth between HG and LG.
The PID starts working better and better. Also the gain switching is working.
The usage of different gains for voltage and current needs some further investigations.
Since v3 will have a offset voltage included to get a higher resolution at higher currents, I close this issue here and in case it is needed, I'll open a new one.
Channel 0
High Gain
The current fits the measured current with the multimeter quite well (99.5%). There seems to be an issue with the low gain channel in the low current regime. (see the gray / black oscillations)
The residuals also look promising: