Chiller 2 looks much quiter at least from an environmental point of view. As well the temperature stability is recovered at the level of how it was before Chiller 1 started misbehaving. This is described in the following plots:
Figure 1, shows the transition between Chiller 1 and Chiller 2. Magnetometer in CEB see much less glitches (and none correlated to the chiller) and also the chiller itself vibrates much less (the displayed accelerometer is attached to the chiller chassis)
Figure 2 is a zoom of the signal for Chiller 1: see a magnetic glitch coincides with the chiller increasing vibration (something is starting which has significant inrush current... as shown in previus entries these glitches are also seen in the mains probes). Waht seems strange is that at this time the temperature of the chiller water starts to increase(!) by some tenth of deg. in a few seconds (it seems a lot).
Figur 3 is a zoom of the signal for Chiller 2: see there are no correlated magnetic spikes, and the chassis vibration is much less, we can barely appreciate a modulation that seems in-phase with that of the water temperature. The temperature is much more stable, a few hundredth of deg.
Figures 4 and 5 compare water temperature now with Chiller 2 and how it was with Chiller 1 back in June: they are similar.
We should conclude that the misbehaving of the Chiller 1 is not due to a change in the environment but an actual failure of the device.
It can be interesting to check if glitches disappeared also in INJ signals (a good probe for those was SSFS signal, Antonino and Camilla suggest also the pump diodes: PSL_SL_PUMP1(2)_AC ) and of course the final target is DARM