While we were down due to vacuum issues, I checked the filters on the rms-to-DC cards for BCM1/2 and found them both set to the high bandwidth option (R10 resistor = 2kOhms) with 3dB points at 100KHz+-25KHz. That's exactly where we wanted them to be, but it may leave us with no hypothesis for why BCM2 looks funny.
Next: The input filters on the rms-to-DC cards are nominally centered at 100 KHz but their central frequency still needs to be briefly rechecked when beam comes back. Later, I will opportunistically sweep the rms-to-DC converter inputs from 50KHz to 150 KHz with a function generator and measure the response of the filter network.
Since the MD preamps have a low pass cutoff (3dB) at 25 KHz, I think that means the BCMs have almost exactly twice the desired bandwidth. If BCM2 didn't look funny this would be great news, because it means we just have to filter the BCM signals down until we've optimized it somehow (eg, Take data on a Carbon target. FFT the beam current. Identify several strong high frequency peaks. FFT the normalized detector yield. Keep adjusting the filter until the peaks in the beam current spectrum are beaten down without digging divots in the white noise.)