Thank you all for your replies! We continue to have trouble with our MAT253 dual-inlet stability. I have swapped the accelerating voltage board, the emission controller board, the 55V power supply, the power distribution board, and the datalogger board, as well as cleaning and resetting every resistor in the amplifier housing. I also checked the magnet control cable and it seemed fine but I did not replace the pins with "a luster terminal". Scanning all software available parameters (i.e. trap, box) shows nice stable white noise. After all this we continue to have signal instability. Two more noteworthy details:
1) The within-sample d18O vs d13C slope is 1 or greater. The historic slope varied from -0.5 to 0.5. I start seeing the presence of this slope in the data at about the same time as the initiation of instability. Within a single sample, d13C or d18O do not change with time so this covariation is being created with individual measurements being relatively light or heavy but without a trend in time. All deltas seem to show a slope when historically, they did not (e.g. d46 vs d45, d47 vs d45, d48 vs d45, etc). What causes within sample delta covariation when no change with time is detectable?
2) the amplifier housing and boards seem to be more vibration sensitive than previously. That is, if I move the vacuum pumps around I can make the source-off amplifier intensity signal vary by many mV. If I take care to center the pumps and pad the tubing, this noise reduces to tenths of mVs. Furthermore, turning the pumps off momentarily has the least amount of noise at a few tenths of a mV. The other 253s in the lab do not appear to be this sensitive to pump vibration.
Any other thoughts out there? Thank you.