I was trying to look at why we seemed to have a fast lockloss after ~3.5 hours earlier this afternoon.  So far I don't have a conclusion on that, but I did find that there is perhaps a new freqency of interest that we should be watching in our fast DCPD channel.

Attached is the spectrum of the fast DCPD A channel, at a few different times before the lockloss.  Blue is "well" before the lockloss, green is about a minute beforehand, brown is 24 seconds before lockloss, and pink is ~1 second before lockloss.  The peak at 20,863 Hz increases by 2 orders of magnitude in this time.  However, the maximum peak value is a little more than 10 counts, while the peak at 10.4kHz is sitting quiescently at about 2000 counts.  So, I don't think that this peak is itself causing the lockloss, unless it really rings up *fast* in the last second or so. 

The RMS of the 1 second before lockloss (measured at 200 Hz) is 2799 counts.  The RMS well before lockloss (at the time of the blue curve) is 2218 counts.  The new 20.8 kHz peak does increase the RMS (as shown in the second attachment), but it still (in these plots) doesn't look like it's going to cause a saturation of the DCPDs. So, no real conclusion on the lockloss, but certainly a frequency to keep an eye on.

I believe that our ring heaters are in the same situation as they were when we were able to stay locked for ~5 hours a few weeks ago, but we now have more annular CO2. 

Edit: The time series leading up to the time of lockloss doesn't reveal anything to me.  The dominant frequency even the moment before lockloss in the 64kHz channel is the 10.4 kHz line, so I'm not finding evidence of the 20.8 kHz peak ringing up, unless it's growing really really fast.