The +-1.5% sawtooth pattern is probably due temperature variations on the Belden 213 signal cables used to bring the BCM5,6 signals upstairs. It's great 50 Ohm signal cable, but it ain't heliax: at 1.5 MHz the voltage attenuation is probably one order of magnitude. (see attached) The specs don't provide a temperature dependence for the attenuation, but if you guessed a modest 0.1%/C it would probably explain what we see here.
I'd like to think that slow gain variations don't matter in an asymmetry measurement since the gain will cancel anyway and the pedestal probably doesn't change much. But we shouldn't use BCM5,6 for yields until the new cables are pulled. The BCM5,6 electronics were moved upstairs to stop degradation due to rad damage. The temperature variations will be much smaller when the signals are switched to 1/2" heliax (assuming we had enough money and the order was placed).
My heart was in my throat for a few minutes wondering if this tempco could have a significant effect on the apparent long term stability of our various CM detectors. After all, our MD signals come up on an even longer run of the same Belden 213 cable, plus we have another 100' or so of lossier RG58. But the attenuation at the 0-25 KHz of our CM signals must be vanishingly small. If we have a nonvanishing tempco for CM signals, it won't be due to the cable run.
Figure 1 cable specs
Belden RG213 Date 8267.pdf