OK, so I decided to have another look myself. For pictures see the
original thread.
I identified the following components in the amplifier module:
Voltage regulators
L7808 +8V (screwed to case)
L7908 -8V (screwed to case)
9L05A -5V (above the keyboard connector)
78L05A +5V (lower left, close to EEPROM)
Amplifiers
3x LM7171AIM - 200MHz difference amplifier
2x OP27G - OpAmp
1x INA132 - Difference amplifier for current measurement (upper right)
Other
MC68HC705P6A - taken from the original thread, there's a sticker ("SONDE 24") on mine that I don't want to remove
DG442DY - SPST CMOS analog switch, normal open (lower right)
24c02 - I2C EEPROM (2K, lower left)
MAX504 - 10bit SPI DAC (below µC)
First of all I tried to investigate the serial communication that I observed on pins 1/2 during my original investigation in 2015 or so. To my surprise, I couldn't find any communication there (anymore).
I traced back connector pin 1 to CPU pin 5 and connector pin 2 to CPU pin 3. There are resistors in series, so a continuity tested won't find the connection from pin to pin.
Honestly, I'm not sure why the communication is gone. Maybe it was only there while I had the supply issues and was some kind of debug message?
Anyway, my original assumption was that the EEPROM could be connected to pin 1/2 for calibration. This is however not the case. Indeed, the SCL pin is connected to pin 1 of the µC and SCA to Pin23 (6th from the left in the upper row). Since there is an EEPROM and a DAC, I still assumed the EEPROM could store calibration data, so I tried to measure the communication on the SCA/SCL pins. However, from what I can tell, it reads the bytes 0x35 0x30 0x30 0x30 0x30 0x30 0x31 0x31 0x31 0x31 from address 0 once after reset. This doesn't look like calibration data, indeed every byte seems to be an ASCII number, so I guess the EEPROM start stores the string "5000001111". So, well, a "50" maybe for the i50s/PR50 followed by "0000" and "1111". While it's beyond me y why somebody would add an EEPROM just to store something as useless as this, I guess we can rule out that the EEPROM stores calibration data.
So I had a look at the DAC (MAX504) - which in my original theory would be used to apply the calibration data. It's an SPi device, so I traced the communication as well and measured its output.
Note: Bipoff is connected to Refout/Refin and RFB to Vout, so it bipolar configuration (-2.048V .. +2.048V) - which, well, already hints that it's not used for gain control.
Directly after reset, a 0x800 (0V) is sent and after the degaussing there's a sequence of 7 communications with a period of ~51ms (and 30.5kHz clock).
0x800 : 0V
0x150 : -2.048V*(512-(0x150>>2))/512= -1.712V
0x60 : -2.048V*(512-( 0x60>>2))/512 = -1.952V
0x800 : 0V
0x4B0 : -2.048V*(512-(0x4B0>>2))/512 = -0.848V
0x470 : -2.048V*(512-(0x470>>2))/512 = -0.912V
0x470 : -2.048V*(512-(0x470>>2))/512 = -0.912V
My scope measurement on Vout showed the same behavior.
So IMHO this looks like the auto-zero functionality in work. Maybe in two cycles for the two ranges.
I just wonder if should concern me that the offset are a bit on the high side given that the DAC is limited to -2.048V.
So yeah, so much about that electronic calibration idea.
Unfortunately, I's a bit difficult to access the other side where the amplifiers and trimpots are located (either desolder two TO220 or remove the sticker on the back of the case).
From the pictures I took in 2015 (and the better one in the thread), I assume the three LM7171 form an instrumentation amplifier. I would think that the trimpot in the middle of the three LM7171 might be used to trim the gain of the instrumentation amplifier. Looks like the feedback resistors of the two inputs OpAmps are 390Ohm and the trimmer is maybe connected in parallel with two 1k2 resistors between the feedback loops of the two LM7171.
I don't really have a clue yet what the other trimmer is for. It doesn't really seem to be connected to the instrumentation amplifier. Maybe it's related to the INA132 on the other side of the PCB. Unfortunately, I wasn't really able to follow the traces. Seems to be at least a four layer PCB. Actually I'm also not sure what the INA132 is used for. Maybe for current measurement during degaussing or something.
Also the two OP27 can't be really part of the output circuit as their bandwidth seems to be too low. I assume they are used for overcurrent protection.
Last but not least, I didn't really understand yet how the range switch works. I wonder if the analog switch (DG442DY) could be used for that.
Then again, my main interest is to calibrate the i50s to reduce the gain error. Currently the readings are only ~92% of what they should be in the 1A/1V range. My best guess at this point is to use the trimmer in the middle of the three LM7171 for that. But it's a drag to get there and I'm still a bit puzzled if this is really the official approach.