Here are results on a study of using iwave to control closely spaced doublets of sinusoids. This study was conducted on test data at 16384Hz sampling rate, of 100 second duration. The dataset consisted of two sine waves at 27.8 and 27.9Hz, each of amplitude 1, on a background of Gaussian white noise of RMS 0.1. Two iwave configurations were used to try and subtract the two components. In both schemes, iwave was running with a tau parameter of 5 seconds and initial guess frequencies at exactly the line frequencies. Iwave was running in phaselocked mode. The loops on the lower and upper frequency lines were set to close 5 and 6 seconds into the dataset, respectively, although these time delays are unimportant; they could both have been set to zero just as well and I mention them for completeness.

CONFIGURATION 1: the raw data is injected into iwave1 set to track the lower frequency (27.8Hz). The D phase output of iwave1 is subtracted from the raw data, and the result is injected into iwave2, set to track the upper frequency (27.9Hz).  The D phase output of iwave2 is subtracted from the input to iwave2, and the difference is the final output. In other words, the two iwave filters are cascaded.

CONFIGURATION 2: the raw data minus the D phase output of iwave2, set to track the upper frequency line (27.9Hz) is used as the input for iwave1, set to track the lower frequency line (27.8Hz). The raw data minus the D phase output of iwave1 is used as the input for iwave2. In other words, for each iwave, the input is the result of the raw data with the D phase output of the other iwave subtracted. 

The first attached pdf shows the results of studies of the the two configurations; the first two pages are for configuration 1 and the pages 3 and 4 are for configuration 2. In each case, there are three plots - the extracted line amplitudes and frequencies (together on the same page), and amplitude spectral densities of the raw data, the data after the subtraction of the first line, and the the output data after both lines are subtracted. 

The 'cross-subtraction' technique is superior, having much better line attenuation and markedly less sidebands introduced. This is because iwave uses the product of the input data and the Q phase output to sense phase shifts, but when the input data is polluted by a close by line, this causes residual beats between this line and the q phase output in the phase measurement. You can see this on page 1, where both the inferred frequency of the line and its amplitude oscillate at the difference frequency between the two lines, which is 0.1Hz, or a 10 second beat. When each iwave input has the inferred line from iwave outputs on other loud lines removed before input, this distortion goes away. 

I have attached a schematic of the cross subtraction configuration, 2, as a second attached figure, in case the written description above is not clear enough. One detail is, the output of iwave2 which is then subtracted from the input data before injection into iwave1 must be derived from the previous sample of iwave2 output, and hence has to be phase shifted forward by one sample period to accurately subtract the line. This is easy to do using both the Q and the D phase outputs of iwave2, and the cos and sin of the phase shift per sample for the frequency of the line tracked by iwave2.

1315 -1400 hrs. local -> To and from Y-mid (then X-mid, unrelated)

Opened exhaust check-valve bypass-valve.  Opened LLCV bypass-valve 1/2 turn -> LN2 @ exhaust in 55 seconds -> Restored valves to as found configuration.  

Next CP3 overfill to be Monday, July 11th.  

The IFO top node appears to be not monitoring some SEI BS and ALS nodes.  The IFO top node is reporting OK, even though multiple nodes are not reporting OK:

This is an indication of two issue:

• IFO top node is not monitoring SEI BS nodes, as well as some of the ALS nodes.
• If the SEI BS and ALS nodes are in their desired states, then their NOMINAL states are not set correctly.

The USERAPPS/h1/guardian/IFO_NODE_LIST.py that is in the SVN shows that those nodes should be managed, so my guess is that what's currently running has not been committed to the SVN.

We have operated for over 10 hours in "nominal low noise, observation, intent bit set" condition.  This was a temporary configuration to allow locking, after the laser diode box failure, for the purposes of testing ER9 pipelines. We are returning to 40W operation.  

H1 is in Observe at 25W.

State of H1: locked at 25W in ENGAGE_ISS_2ND_LOOP

Activities / Events:

• H1 lost lock at 15:12UTC - I haven't looked into this yet, but lock broke suddenly, no ring up before lock loss
• H1 relocked with some arm alignment for green, no trouble locking DRMI

Upon my arrival (at 7:45 am PT) in the control room I found the IFO locked and in "Observation" mode with the intent bit set.  The Lock Clock showed 8:50 hours and counting.  The input power was 24.3 W.  Camera images looked stable and as expected. (Because the video5 computer had frozen! argh).  DARM ASD looked terrible, but not unreasonable for the operating condition of the IFO.  ISC_Lock Guardian was in Nominal Low Noise.

This is a better start of the day then we have had for awhile. This should allow the ER9 testing of the hwinj, cal, and data pipelines and, with the intent bit set, allow the analysis streams to test their readiness.

15:13 UTC (8:13 PT) - just lost lock. Morning activities?

Patrick, Sheila, Carl and Ross damping PIs

After Keita's ISS fixes we had a few hours of bad weather and earthquakes that contributed to dificulty locking.  Once things calmed down we locked and sat at several stages to see if the lock was stable, DC readout, 25 Watts, ISS 2nd loop engaged, and now we have made it to some kind of "nominal low noise" state.  Patrick hit the intent bit.  

There are several things wrong with this spectrum, some of which could be solved by going back to 40 Watts in the morning:

• We skipped the coil drivers state, since we don't expect it to help our noise and we lost lock in this state earlier. We don't see any reason not to try doing this tomorow though. 
• There is a lot of intensity noise in DARM.  (1st attached screenshot shows 2nd loop PDs before ISS was on 6:40 UTC July 8th, and after the 2nd loop is engaged at 7:30 UTC. The 2nd attachement shows that there is still a lot of coherence of intensity noise in DARM.)
• The CO2 guardians were in preheating mode for the first bit of time we had the intent bit set (this is a result of not thinking through all the consequences of changing the nominal power down to 25 Watts).  We set the CO2 guardians to NOMINAL_LOCKED_CO2 level, which means that they are now at the settings that are appropriate for 40 Watts I think, which is probably closer to the right settings for 25 Watts than preheating at least. 
• Our low noise hard ASC loops were designed for 40 Watts, so the decision to go to 25 Watts means that we would need to do some measurements and loop design work to have a low noise state for 25 Watts. 
• We probably have different spot positions on our optics now than we would at 40Watts, so we would need to run a2l for 25 Watts if we want low noise.  
• DIAG main is complaining that the POP beam diverter is not closed, this is because we are still locking on the POPX WFS, this will be open throughout ER9.
• The OMC alingment is clearly off, on the camera it is on the left half of the screen. 
• The DCPD whitening is off, which makes our high frequency noise worse.

The last attached screenshot is just to show how many locking attempts everyone made in the last 14 hours. The range displayed by the DMT viewer in the control room seems to have a problem, it should be stable at some low range but on the DMT viewer it looks like we lost lock in the last half hour.  

There has been something wrong with the ETMY oplev for quite a while now, it says that the optic is moving a lot more than the other test masses, swinging around by more than half a micro radian.  This must be false, and needs some investigation.  

The lockloss from the state REFL_POP_WFS which happened at 23:56:29  UTC July 07 2016 is very similar to what is described in alog 26840 and comments.  I've attached the guardian logs for both ISC_DRMI and ISC_LOCK, and a plot of the lockloss.  

The first thing that happens is that the ISC_LOCK gaurdian starts to transition our front end triggering for the LSC from POP18 to POPLF, by lowering the thresholds, sleeping 0.1 seconds, and then setting the trigger matrix elements to 0.  

Started and ran the purge-air skid, Turbo, QDP80 and newly added scroll pump.  Tested the functionality of the Turbo's Safety Valve with the control cable connected to the scroll pump's relay box - demonstrating that the Safety Valve closes upon the loss of scroll pump motor AC and thus mimicking the "QDP80 Running" 
signal.

Note: The purge-air skid has developed some problems from lack of use over these past few years, namely the radiator fan never came on while running the compressors, the drying tower never cycled and the low pressure alarm never sounded.  These features worked fine when last this unit was run.  Now they don't.  Also, the Turbo spun-up normally even with the locally mounted scroll pump running (new vibration source) but tripped into EMERGENCY OPERATION upon BRAKING and when at ~1/3 NORMAL rpm.  This behavior is seen on other Turbos (XBM for example) and might be an age related issue?  

I am leaving the Turbo energized overnight to ensure that the rotor is completely stopped.  I will de-energize it tomorrow.  

With the measured sensing function (28123) in hand, we did an MCMC-based numerical fitting for the sensing function (see E. Hall, T1500553).

The fitting results are

• Optical gain = 8.980393e+05 +/- 7.965381e+02 [cnts/m]
• Cavity pole = 3.294194e+02 +/- 5.626792e-01 [Hz]
• Time delay = 1.973719e+02 +/- 3.386483e-01 [usec]
• Spring frequency = 9.825078e+00 +/- 5.453538e-02 [Hz]

[New sensing function form]

As reported by Evan (27675), it seems that the extra roll off in the DARM response at low frequencies are due to an SRC detuning (or something equivalent). While such detuning should be suppressed by some control loop ideally, we decided to include the detuning-induced functional form in addition to the ordinary single-pole response. We use the following approximated form for the DARM response

S(f) = H / (1 + i f / fc ) * exp( -2 * pi * f * tau) * f^2/(f^2 + fs^2)

where H, fc, tau and fs are the optical gain, DARM cavity pole, time delay and spring frequency. Some details of the derivation will be reported later. Apparently, we now have an additional quantity (i.e., fs) to fit.

Note that since H1 seems to be in (unintentionally) an anti-spring detuning, fs should be a real number whereas it should be an imaginary number for a pro-spring case. Obviously, Q is set to infinity for simplicity.

[MCMC-based numerical fitting]

Following the work by Evan (T1500553), we adapted his code to include the spring frequency as well. In short, it is a Bayesian analysis to obtain posteriors for the parameters that we want to estimate. We gave a simple set of priors as follows for this particular analysis.

• 0 < H < infinity
• 0 < fc < infinity
• 0 < tau < infinity
• 0 < fs  < infinity (real number)

The quad plot below shows the fitting result. The estimated parameters are obtained by taking the mean values of the resutling probability distribution. As usual, the two plots on the left hand side show a bode plot of the measured and fitted data. The two plots on the right hand side show the residual of the fit. As shown in the residuals, there are several points which are as big as 20% in magnitude and 6 deg in phase below 10 Hz. Otherwise, the residual data points seem to be within 10-ish % and 4 deg. The code is attached in pdf format.

EDIT: I am attaching the actual code as well.