add hotfix for OpAmp input offset

README.md

@@ -130,7 +130,10 @@ Errata for V1.0
     have an offset of 1V, we will send 19 mA through the ESD diodes (limited by the series
     resistor), which isn't great but should be harmless.
 2. Silk text for SWD connector is swapped.
-  - **Hotfix**: At correct pinout with a sticker.
+  - // **Hotfix**: At correct pinout with a sticker.
+  - **Actual Hotfix**: Paint over the wrong text with a white marker. Add correct pinout
+    in other documentation, e.g. we will have documentation for the pinout of the main
+    connector in each box anyway.
 
 ### Hotfix is necessary to avoid degraded function
 
@@ -146,6 +149,29 @@ Errata for V1.0
     method to make WS2812 work with lower voltage signals.
   - **Hotfix**: Replace by BSS138. They are a known-good choice and fxk8y has some at hand.
   - Fix is needed for Q7 but it would also be useful for Q3 and Q4.
+3. Input offset of OpAmp can flush small input signals to zero.
+  - The input offset is larger than some of our small signals, e.g. 24 V / 0 ohms in our 2 kOhm
+    measure mode have an amplitude of 0.7 mV at the input of the amplifier while its offset can
+    be as high as 4.5 mV. This is as planned and we can compensate the offset because the input
+    signal will be zero when both MOSFETs are off.
+  - The common mode range includes negative voltages until -0.3 V below GND. Therefore, -4.5 mV
+    at the input (zero plus worst case offset in negative direction) is not an issue in itself.
+    However, this would be amplified to -250 mV at the output and the output cannot go below the
+    negative rail. Therefore, we won't see the real offset when we try to measure it.
+  - The offset can be high enough to cancel all or almost all of our 0.7 mV signal. This has
+    happened for 7 or our 60 channels (5 boards with one channel, 1 board with both channels).
+  - **Hotfix**: Add 1 MOhm between 3.3V and the input of the amplifier to add 3.3 mV positive offset.
+    - The lower resistor of this voltage divider is not the 56 kOhm but 1 kOhm (plus 0.1 ohms).
+    - We only do that for channels that have below 50 counts on the ADC when the input is zero.
+      We usually will have 200 to 300 counts after the fix.
+    - I have used 2.2 MOhm in a few cases when the offset wasn't so much below 50 counts.
+    - The resistors are soldered to JP6 and then connected to pins 3 and/ or 5 of U12.
+  - We have nominally 2.8 V for 500 mA of input current. Therefore, 0.25 V of additional offset
+    pushes us over 3.0 V, i.e. we might not be able to measure up to 500 mA. We only need 300 mA
+    so that is fine. We can consider reducing the gain (e.g. 47x with 47k instead of 56x) if we
+    want to support more current.
+  - **Proper fix for new revision**: Add the resistors to the PCB. Maybe add the full 4.5 mV to
+    be on the safe side and consider adjusting gain.
 
 ### Planned changes before deploying the boards