Today, I restored the ITM misaligned TEST_OFFSET and TEST_GAINs which I found to be showing as diffs in SDF - Sheila had been playing with them last week but said they should be restored.  Snapshots of the ITMy "was" and "now" (aka good) values are attached.

Turned ITM R0 OPTICALIGN zeroed offsets and inputs off - these are never used and are just confusing when found on.

Cleaned out BIO/COIL/TEST_SW2 and LOCK_SW2 changes like has been ongoing across all SUSes.

Rolf, Jim, Dave:

after removing the LSC RFM receive errors completely last week, I noticed that the ASC is getting receive errors from the ALS models at the end stations at a very low rate.

The rate of error is one every 8 to 10 hours. The return of the errors can be tracked back to the ALS model changes made last Tuesday 07apr2015 during maintenance. A trend of the ALS CPU usage shows a difference in signature (attached plot shows last 2 weeks of trends) starting last tuesday.

To remind everyone, when the end station ISC was split, the ALS is the "slow" component, so RFM errors would not be entirely unexpected from this sender.

An understanding of beam tube motion as a function of wind speed is important in predicting scattering noise from the beam tube baffles and in deciding if we need to re-insulate the beam tube. This data was taken during the windstorm on Apr. 11, 2015 at baffle #96 (near the power maximum for the diffracted light from the pattern on the original optics), near the first double doors towards the CS from MY. Figure 1 is a photo of the accelerometers attached to the beam tube, and Figures 2-4 show the motion along the different axes at 3 different wind speeds (9, 18 and 24 MPH). As a first guess I would expect peak amplitudes to increase roughly with wind speed to the 3/2 power (wind power goes as the cube), and, very roughly, it seems consistent.

Raw data archive

Today's quest to increase the bandwidth of the FSS involved the following steps:

• removing both notches in the EOM path
• re-installing the original notch of 220 uH and 150 pF
• changing out the 150 pF cap for a 120 pF cap
• restoring the TTFSS back to its original configuration

The file NONOTCH1.TIF is the open loop transfer function with no notches installed.

Unfortunately the NPRO noise eater was oscillating for this measurement.

The file NONOTCH2.TIF is without the NPRO noise eater oscillating.  One can see

the ~760 kHz resonance that the notch was installed for originally.

The file 150PF.TIF is the open loop transfer function with just the one notch installed.

That notching being the one with 220 uH and 150 pF.  The unity gain frequency is a little

over 300 kHz here.

The file 120PF.TIF is the open loop transfer function with the 150 pF capacitor replaced by

a 120 pF capacitor.  The rationale being moving the notch frequency higher might still attenuate

the 760 kHz resonance enough and allow the gain to pushed a little higher.  This introduced

either another resonance or did not sufficiently suppress the 760 kHz one.

The file RESTORE1.TIF shows the open loop transfer function with the TTFSS restored to its

original state.

For all measurements the reference cavity transmission was ~1.45 V, the fast gain was 18 dB

and the common gain was 22 dB.

Ed, Peter
 

Many items are skewed since there is currently activity in the PSL this morning.

Laser Status: 
SysStat is good
Front End power is 31.7W (should be around 30 W)
FRONTEND WATCH is RED
HPO WATCH is RED

PMC:
It has been locked 18 minutes (should be days/weeks)
Reflected power is 1.8 Watts  and PowerSum = 24.2 Watts.
(Reflected Power should be <= 10% of PowerSum)

FSS:
It has been locked for 0 h and 18 min (should be days/weeks)
TPD[V] = 1.4V (min 0.9V)

ISS:
The diffracted power is around 11.7% (should be 5-9%)
Last saturation event was 0 h and 19 minutes ago (should be days/weeks)

HAM6 pumping continues, Kyle/Gerardo swapping annulus pumps

Jim working on updates ISI blend filters

CDS synchro/CPS work, vacuum maintenance for 3IFO, fiber installation between corner station and VPW

Investigating the trip on WHAM6 CPS, Friday night.  Mirky for now.  See the output drives fairly high (solid 100s touching 1000 on OUTF mons) with saturaions totalized over the weekend.  Cleared these and saw a few more shortly.

Took ISI and HEPI down to review free hanging position.  HEPI Deltas (Free Hang-Servo position) pretty small except Rz w/ ~24urads.  The ISI too had a larger than we left it Friday shift in Rz.  Of course, HEPI Rz can not affect ISI Rz.  The ISI Rz shifted about 9urads from Friday.  Recall, on Friday we had 70urad shift on Rx.  Maybe further temperature and pressure shifts are settling out.  The Reference Target Positions were reset to match the free hanging location.  The loops close fine and the OUTF drives are as expected, markedly smaller.

If alignment in HAM6 is an issue, use HEPI first for horizontal changes to keep the ISI drive reduced.

Commissioners - SDF shows that someone changed the P and Y Damping gain on Thur April 9th.  I couldn't find why in the alog...  See attached.  This maybe should be restored...?

Sheila, Alexa, Evan

Summary

We have recovered the usual low-noise locking state (except for the outstanding issue of the HAM6 centering scheme) and have turned up the power to 15 W.

Details

• Alexa and Sheila traced the ALS DIFF problems to the wrong coil state in ETMX L1; it was in low range mode, and hence saturating when DIFF engaged.
• IMC-F was saturating every 10 minutes or so. First we increased the UGF for the slow common feedback from 20 mHz to 40 mHz. That helped a little, but we had much more success by adjusting the settings for red tidal feedback. It now engages when each arm's TMS WFS B goes above 50 ct (with a 1 s holdoff), and disengages below 20 ct. This way, the tidal engages during the CARM reduction sequence, right before control of DARM is handed over from ALS DIFF to AS45Q. Excursions of IMC-F are now about 100 kHz pkpk (albeit with low wind).
• ITMX CO₂ laser heating was off. We turned it on to the usual 0.2 W.
• The OLTF gain for the dETM WFS seems to have increased; we had to turn down the filter module gains to 10 for pitch and 5 for yaw (whereas they used to be 14 and 21).
• The DARM optical gain appears to have changed. Our usual DARM ERR offset of 2e-5 ct, which previously gave something like 23 mA total from the two OMC DCPDs at 10 W PSL power, now gives more like 35 mA at 10 W PSL power. When we remeasured the DARM OLTF, we found that we needed a DARM filter module gain of 2000 ct/ct to achieve the same UGF as last week (compared to 1300 ct/ct that was used previously).
• Consequently, the DARM calibration was wrong. Using Kiwamu's script, I reckoned that the optical gain had to be increased from 9.09e-7 m/ct to 1.40e-6 m/ct (see attached plot and xml file). So I changed this value in CAL-CS_DARM_ERR FM5. That made the spectrum more resonable, but it seems that the high-frequency noise floor is still at or above the GWINC 10 W curve, even after bumping up the CARM gain. [Note that the old optical gain produced a phantasmagorical 50 Mpc lock stretch in the summary pages.]
• For 15 W, we closed the POP shutter on ISCT1, since POPAIR_A already has 25 mW dc at 10 W of PSL power.
• For 15 W, the DARM ERR offset was reduced to 1e-5 ct, giving 14.6 mA total dc on the OMC DCPDs. If (IF) the current calibration is correct, the DARM offset is about 12 pm. We could always bump this up to the 15 pm we were using previously.

As before, HV in HAM6 has been turned off for the night.

Here is a measurement of the DARM OLG with the ETMY linearization off.  It isn't changed significantly from when the linearization was on (blue reference). 

Note:  the calibration is off tpnioght, we have changed the DARM offset but haven't corrected the calibration for this so at the moment our sensmon range is nonsense.  More details about todays locking coming later. 

[Evan, Alexa, Koji]

We worked on the ISCT6 yesterday after noon before the vent recovery.

- AS_AIR and OMCR paths were realigned

- AS_AIR path attenuations (ND filter, BS) were removed.

- Power at various points on the table were measured.

- OMC PZT shutter and HAM6 fast shutter were tested. Their shutter speeds are 4usec and 2msec, and fast enough for higher power operation.

Optical path alignment

- The alignment of the beams in the AS_AIR and OMCR paths were revisited.

- The OMCR path had the SRC length setup which had undamped reflections.
  A beam dump formed by two ND filters (OD1.0) and knife-edge dump were inserted in the path.

- Two power-attenuation elements in the AS_AIR path were removed.
  A 50:50 BS common for AS_AIR_A (1f) and AS_AIR_B (2f) were replaced with an HR mirror.
  The other was OD1.0 ND filter fro AS_AIR_A.

Power measurement

The naming of the optics is based on D1201210-v7.
AS-BS2 = 50:50 BS common for the PDs
AS-BS3 = 50:50 BS for the PDs

Before AS-BS2 replacement (but no ND filter for AS_AIR_A)

After the periscope/before the 1st steering mirror: 8.05 uW (@Apr 10 4:52PM local)
After the periscope/after the 1st steering mirror:  8.04 uW
AS-BS2 incident:   4.20uW
AS-BS2 refl:       2.53uW (R(AS_BS2)=60.2%)
AS-BS3 trans:      1.51uW (T(AS_BS3)=59.7%)
AS_AIR_B incident: 0.79uW (R(AS_BS3)=31.2%)
AS_AIR_A incident: 1.31uW (No ND filter)

ND filter (tested at OMCR)

ND incident 1.90 mW
ND trans    0.180mW (T(ND) = 9.5%)

After AS-BS2 replacement

AS-BS2(HR) incident: 4.15uW
AS-BS2(HR) refl:     4.08uW (R(AS-BS2) = 98.3%)
ASAIR_B incident     1.45uW (R(AS-BS3) = 35.5%)
ASAIR_A incident     2.44uW (T(AS-BS3) = 59.8%)
 

Previous
AS_AIR Periscope -> AIR_B thruput:  9.8%
AS_AIR Periscope -> AIR_A thruput:  1.6%

New
AS_AIR Periscope -> AIR_B thruput:  18.0%
AS_AIR Periscope -> AIR_A thruput:  30.3%

=> We decreased the OM1 transmission from 5% to 800ppm = 1/63
     This was compensated by the inclease of the PD incident by
     x1.8 for AS_AIR_B => x35 gain missing
     x19 for AS_AIR_A => x3.3 gain missing

OMCR after the periscope (OMC unlocked): 1.95mW

Shutter test

We performed an end-to-end test of the fast shutter and the OMC PZT shutter.

- Procedure (See also LHO ALOG 16355)

Fast Shutter test:

1) Confirm the shutter threshold level. It was 2V.

2) Thorlabs PDA255 (50MHz) was set on the ICST6 table. The OMCR beam was aligned. Confirmed the OMC was unlocked.

3) Disconnect the analog signal to the trigger logic module, and plug it on a function generator.
    QPD Transimpedance Amp for AS_C (Rack ISC-R5 U18) has QPD SUM1 output at the back side.
    This cable is connected to the trigger logic module on the top of the ISCT6 enclosure.

4) Set the function generator to produce the triangular signal with an offset of 0.7V and the amplitude of 2V. The frequency was 100Hz.
    => The signal ramps from -1.3V to 2.7V.

5) First, the amplitude of the FG was set to be 0. Open the shutter if it is closed.

6) Set an oscilloscope to watch AS_C Sum, Thorlabs PD, and trigger signals. Set the scope to "Single Sequence" mode with "Normal" trigger.

7) Set the amplitude is to be 2V. This already triggers the shutter. Return the amplitude to 0V.

OMC PZT Shutter test:

1) Lock the OMC. Place the PD at the OMC trans (leak) beam.

2) Otherwise the same setup as above.

- Result

- The trigger occured at 2V and almost immediate.
- The fast shutter was closed with the shutter time of 2.3ms between the trigger and the complete close.
- The OMC shutter was closed with the shutter time of 3.9us between the trigger and the complete close.

- Analysis: Are the diodes safe now?

This threshold level shuts off the shutters when the incident power to HAM6 is 1.2W (cf. Dan's LHO ALOG 17787)

According to the lock loss observation at LLO, the AS port power during lock loss is not impulsive but rather slow.

1.2W x 4us = 4.8 uJ
We deliver half of this power on the OMC DCPD. => 4.8/2 = 2.4uJ

This is x400 smaller than the minimum pulse energy level (1mJ) tested in Frank's diode blasting experiment,
which did not show any (or significant) damage of the diode.

Therefore we can conclude that the OMC DCPDs are safe.

OMCR QPDs and OMC QPDs receive 1/20 and 1/250 attenuated pulse with 600 times longer time.
So the below calculation is for the OMCR QPDs.

1.2W x 2.3ms / 20 = 0.14 mJ

This is still x7 smaller than the above 1mJ.
Therefore the OMCR/OMC QPDs are also safe even the beam is on a single element by chance.

report attached.

Alexa, Evan, Sheila, Koji

Koji, Evan and Alexa tested both the OMC PZT shutter and the fast shutter, realinged ISCT6, removed the ND filter from AS AIR and installed ND filters as beam dumps in the OMC refl path.  More details about that are coming in a latter alog.  

After Dave helped us get the h1boot back, we got a late start on trying to recover locking.  We got through the inital alingment procedure, evan set the whitening gain so that the signals out of AS AIR should be roughly similar to what they were before the change of OM1 and removal of the ND filter (he changed it from 42 dB to 15dB).  Some gains still had to be readjusted X arm IR locking gain was changed from 0.04 to 0.12, MICH dark locked gain was changed from -2000 to -1000.

We ran into a few problems along the way, the PSL noise eater was oscillating (and has been for a few days).  After togling the switch the ISS would not lock, we reset the reference level to get it locking again.  We also found a variety of settings which seem to have been wrong, the COMM PLL input was disabled, the DARM input was off, the SRM top stage length was on.  

We have been able to lock ALS COMM, but DIFF is not working.  We can engage it with a low gain, see that the PLL is kept in range, but we cannot turn the gain up. This is behavior similar to what happens when the ESD is not working, but we tried exicting the ESD in angle and watching the OpLev, we can see that it does respond.  

Also, the ETMY ESD has been tripping many times tonight, even though we have not tried to actuate on ETMY at all. 

As Kyle requested, we disabled the HV in HAM6. For the OMC PZT, we disabled the output of the 100 V Kepco supply. For the shutter, we flipped the HV enable/disable switch on the driver chassis next to HAM6.

For Jeff, we are leaving both arms locked in green with neither ALS COMM or DIFF locked.  The wind is not as high as it was forecast to be tonight.  

After the h1boot restart finally got a chance to measure the power output of the X CO2 laser versus the rotation stage angle. The minimum power angle was measured at 38° (a change of 1° from previously). The maximum power output was reduced to 5.01W from 5.44W, measured before the periscope. The configuration was modified to reflect these new measurements. 

The X-arm CO2 laser is now unblocked and running stable. 

J. Kissel, P. Thomas, D. Sigg

Patrick grabbed ten seconds of data from the following Beckhoff ADCs with three different input signals: +/- terminals shorted, +/- terminals connected to a 9 [V] battery, and +/- terminals fed with a 1 [Vrms] amplitude, 10 [Hz] sine wave. 
EL 3004, billed as a 12-bit ADC, "presented as 16-bits" datasheet.
EL 3104, billed as a 16-bit ADC datasheet.
EL 3602, billed as a 24-bit ADC datasheet.

The messages:
- Datasheets are deceiving and misleading. So are the default settings. So is what it means to be a 12-bit or 24-bit ADC, since there's no guarantee that all bits are significant.
- The calibration for these ADCs, regardless of what the ADCs are billed as, is 20 / 2^16 [Vpp / ct] for EL3004 and EL3104, and is 20 / 2^32 [Vpp / ct]
- The noise performance of the EL 3104 16-bit ADC makes the most sense, and is a consistent 30 [uV/rtHz] regardless of input.
- The noise performance of the EL 3004 12-bit ADC is confusing, but makes sense if you consider it to be a 12 bit internal ADC chip running at 50 [Hz] whose [least significant / "last"] bit is not random, sampled by the internal processor at 1 [ms] padded to produce 16 bits. If we treat the injected sine wave's noise performance as the expected noise, 200 [uV/rtHz].
- The noise performance of the EL 3602 24-bit ADC (two channels of which were measured) is non-sense. Further investigation is required.
- The noise performance of the EL 3104 looks like it would be just about good enough to handle the pressure sensor signals as they stand in the HEPI Pump Servo (recall LHO aLOG 16500).
- We should consider looking into some Beckhoff PID modules as well, such that the computation and control is all done inside the Beckhoff system, and EPICs is merely used as a monitor.

We're (as in Patrick) going to get a longer data on Monday set so we can better assess the low-frequency performance, and we'll focus on why the time series of the EL 3602 seems to be pre-averaging the data, causing a huge increase in noise when an AC signal is driven through it.

----
Data, plots, and analysis script live in
/ligo/svncommon/SeiSVN/seismic/Common/Documents/T1500159_BeckhoffADCNoise/