This pressure change is too small to account for the vetical drift. In log 15887, ETMY is measured to sag by 120 microns during pumpdown, a pressure change of 760 torr. For a pressure change on the order of 1e-8 torr, as seen here, you would therefore expect a vertical change of 

(1e-8 / 760) * 120 microns = 1.6e-9 mincrons or 1.6e-15 meters. This is well into the noise of the OSEMs.

I think it is more likely the in-vacuum temperature is different from the VEA temperature sensor. Are the any in-chamber temperature sensors on the ISI table?

A consistency check you can do is see if it is just the suspension, or really a change in the whole chamber is to see if the ISI is sagging as well; or rather trying to sag since it has active position control. So, if the vertical ISI actuators are pushing up harder proportional to how much the quad is sagging, then there is at least some consistency throughout the chamber.

Since the suspension pitch isn't changing (which is great) that suggests that either the suspension is super well balanced, or it is just the top springs at the ISI table that are sagging. The temperature of these springs likely follows the temperature of the ISI much more closely than the ones lower down.

J. Kissel, B. Weaver

I'm the curious. I did a bunch more starring at these channels, trying to come up with a theory for all of this ETMX oscillation business. My best theory at this point: this wiggling is a mixing of tidal + length to angle coupling of the QUADs against power fluctuations arising from the ISS 2nd loop's feedback to the 1st loop, which is then fed back to the AOM's signal that we stare at for drift. Check out my attachment. All of the channels in the attachment, representing DHARD, Tidal, the IMC, the ISS... are all wiggling at roughly the same frequency.

Here're some things I've notice while staring:
(1) Notice that the Tidal Error signal (as shown by H1:IMC-F_OUT16 in DARK Green) is perfectly anti-correlated with the DHARD Pitch control signal (H1:ASC-DHARD_P_OUT16 in Purple). This is no surprise to me because I know that Tidal is fed to the UIM stages of both QUADs, which are known to have plenty of UIM-length-to-TST-angle imperfections. You could also claim some correlation between the DHARD yaw control signal (H1:ASC-DHARD_Y_OUT16) as well, but it's not so obvious, so I segregated it.

(2) Along the bottom are the DC component of 1st / Inner loop ISS PDs (H1:PSL-ISS_PD[A/B]_CALI_DC_OUT16), these lag just a smidge (5-10 [sec]), but perfectly behind the IMC TRANS PD (H1:IMC-TRANS_OUT16). This PD is watching the same light the 2nd ISS array is watching, whose control signal is fed back to the -- you guessed it -- the error point of 1st loop of the ISS, whose actuator -- you guessed it -- is the AOM. The same AOM who's diffracted power we monitor for drift (which is why Betsy had it up). When you actuate on the diffracted power, you're directly changing the input power into the IFO. Not so stable at low frequency, this intensity stabilization loop...

(3) The cavity powers (PRC and ARM, as measured by H1:LSC-POP_A_LF_OUTPUT and H1:LSC-TR_[X/Y]_NORM_INMON, respectively), also show the same "oscillation" as the ISS Inner Loop PDs, but they're look like a mix between the IMC TRANS / ISS Inner Loop PDs's fluctuations (i.e. input power fluctions into the PRC) and the Tidal / DHARD Pitch fluctuations (i.e. arm power fluctuations because of inadvertent pitch from tidal control). Some times they follow one, other times they follow the other. I suspect this is why the POP_LF trace (shown up on the front wall in bright yellow, so very obvious) seems to show slow oscillation, with and even slower envelope.

I have a feeling that when you add a gain peaky-very-low-frequency blend filter into the equation (ETMX, when in the 45 [mHz] blends), it exacerbates the whole problem. (Note, I attach the correct configuration of the BSC ISIs's blend filters)

Hurumph! It seems like the source of all of these symptoms appears to be the IMC -- both tidal control and PRC input power fluctuations. Maybe we should take a serious look at the low-frequency performance of the IMC? My deepest darkest fear is that it all comes from the HAM3 0.6 [Hz] problems, but I think that's too fast this issue.

Anyways as I've claimed all along, I think at least these sets of traces convince me that this is a particularly complex interaction of *many* plants and anima*ahem* control loops. I think we might be at the point in LIGO's life time that we can't keep getting away with just assuming that everything below 100 [mHz] won't interact with each other, and I expect this will only get worse with higher power.

Since we'll shortly have the oppurtunity to play again, I suggest we try a few things and see if it helps:
(1) Change / improve the isolation configuration on HAM2/HAM3 
(2) Try improving the UIM L to P 
(3) Change the UGF of the DHARD loops
(4) Rejigger the ISS loop (or at least characterize it better) such that the low-frequency loop shape is different.

P.S. The template for this StripTool lives in 
/ligo/home/jeffrey.kissel/2016-01-03/PowerFluctionations.stp

Betsy's pulling up DTT on this, check for coherence, but just by looking at the time-series, I'd have a tough time mathematically proving that this is a linear coherence. We'll post if we get any clues there.

Note the coherence of many things with the HAM3 0.6Hz...  (spectra taken from this lock stretch).

Dear DETCHAR - Since we can't seem to get the data from 10 days ago in our own controlroom, can you please look into coherence with IFO channels during that time when the 0.6Hz SEI HAM3 peak wasn't around?

Thanks-

Betsy, Kissel

Sad that I have to tag SEI on this... Perhaps in the O1/O2 interim we need a concerted campaign of replacing electronics in HAM3.

The low frequency gain of the ISS should be large enough: in the normal configuration there is a low frequency boost and an integrator. However, in the past it was clear that the low frequency (below ~20 Hz) error signal was spolied by some additional noise, most likely scatetred light. Indeed, when the beam on the external table was not properly dumped, we saw a lot of increased noise due to back scattering.

So, it's also possible that the ISS loop has enough gain at low frequency, but the error signal is polluted by some scattered light, which might be correlated with the motion of some suspended object.

Reset H1SUSETMY timing error.

   It is currently snowing at the site, a dry light snow with about .25" accumulation. The roads and parking lot do not appear to be slippery. However visibility is somewhat limited. 

The last lock loss on 1 Jan was due to PSL issues (see alog). The low frequency noise reported here seems to be correlated with unusual noise in PSL ISS and also shows up in IMC frequency noise. The attached pdf shows apparently normal behavior on 1 Jan for PSL ISS, IMC frequency noise, and strain followed by excess noise in all three on 2 Jan. While things appear to have calmed down on 3 Jan (so far), there appears to be a PSL glitch that shows up in the strain channel.