It is from HP 4280A CV plotter, I do have full schematics. Yes, looks like some sort of a hybrid module. It is not 100% dead, but gain is around 0.9 while it should be 1. This is messing up readings as reference and measurement channels have different gain now, and difference is way more than calibration pots can compensate for.
What in a world could have caused a unity gain buffer OA to develop a gain of 0.9?