With the closing of most of the chambers recently, we are in a state where we can relieve Alarm Levels on some Dust Monitors (note: HAM6 is still open & we have exposed parts in the West Bay Test Stand area). At any rate, I went through Dust Monitors 1, 2, & 3 & raised their alarm levels (on the order of 10^2 counts for 0.3 & 0.5um). Could do this for other Dust Monitors which are away from clean hardware.
I notified Jeff Bartlett about this. Will chat with Calum about whether we want to establish long term values for these levels when we are all buttoned up.
Took a quick look at the FSS MEDM screen and noticed that the RFPD DC value was fluctuating wildly. Pulled up the filter for this channel and saw that the 2 Hz low-pass filter was not engaged and that the gain value was set to 1.000. The 2 Hz low pass filter is now engaged, as it should be, and the gain value was set to 0.150 which gives readings consistent with those obtained with a digital multimeter plugged into the DC output of the RFPD.
no restarts reported
Alexa, Dan, Koji, Sheila, Stefan, Kiwamu,
Today we spent some more time on alignment of the IMC with the use of the WFSs loops. After some struggling, we eventually could close all the loops stably. However, the ASC loops tend to drift on a time scale of 30 min.
In parallel to it, we made a first attempt of the ring down measurement. The data did not make sense this time. Ongoing.
IMC alignment:
Our original hope was that once the WFSs were centered, they should be able to servo the IMC to a good alignment. However, it turned out that stably engaging the ASC loops were very difficult for some reason. The first issue we noticed was that the MC2 trans QPD did not show a reasonable signal. It looked all noise and indeed changing the whitening settings did not change the shape of the noise. This indicated some issues in the analog circuits. Stefan and Sheila went to the rack and found that a number of cables were unplugged. Also, they found that the whitening electronics were turned off. So they plugged the cables back to the whitening box and powered up the whitening box. This recovered signals on the MC2 trans QPD. This then allowed us to close the 'DOF3' loop which is the centering servo for the MC2 QPD.
However, we were still having a difficulty in closing the rest of two loops in the IMC ASC. They use the signals derived from the WFSs. The symptom is that the error signal looked too big even though the loops are closed. It looked as if the loop is not suppressing the signals. Acutually, keep running the ASC loops made the power buildup and visibility worse on a time scale of 30 sec or so. We made sure that the dark offsets were successfully removed and centering on these WFSs were fine. But, no success.
After people left, I did some more investigation on the ASC loops. Since I did not like the beam shape in the reflection camera, I went to the table and checked out the beam. I found that there was a ghost beam not being dumped and going into the WFSs. I narrowed the aparture of the iris that was infront of the REFL RFPD such that it catches the ghost beam. Then I digitally removed the dark offset from all the WFS segments and tried closing the ASC loops again. The loops stayed stable for approximately 10 min. However then it started degrading. See the attached. I am not sure why, but I am worried about RFAM. I am leaving the ASC loops running for the night to see the long term stability.
Ring down measurements:
We also spent some time trying to measure the cavity decay time (storage time) in order to estimate losses in the cavity. We tried various ways of shutting the control or light off, but ended up with switching the polarity of the fast signal in the IMC board. This is the same technique as Livingston did recently (LLO 13748). We were expecting to see something like 17 usec 1/e decay time, but the measurement tended to give us about 35 usec which is twice bigger than the expectation. At this point, it is unclear why the decay time is such long. The attached is a picture of the raw data displaced on an oscilloscope. The PD is Thorlabs, PDA100A on the IOT2L table. The bandwidh is 2.4 MHz according to the data sheet which should be fast enough for this measurement.
It maybe independent, but we know that this transmitted light is largely clipped (seemingly more than 20% of its total intensity) somewhere in chamber before it comes out to the IOT2L table. So we are interested in repeating this measurement at a different port. Maybe at REFL or IOT2R. The measurement is therefore ongoing.
The attached is a trend from the last night.
You can see that the alignment degrades on a time scale of ~ 2 hours. Every time it hit the lower threshold of the ASC trigger, the alignment was forced to be back to the starting point due to the 'clear history' operation in the ASC loops.
G. Moreno, K. Ryan, J. Worden, R. Weiss A preliminary report of the procedure is given here, a more thorough report will be made once the post injection charge measurements have made. We introduced both positive and negative ions of N2 into the ETMY chamber. The input gas was boil off from liquid nitrogen which was additionally passed through a liquid nitrogen trap before entering the ionizer. The conditions in the ionizer: pressure on the needles 210 to 303 torr, pressure after the apertures was a maximum 45 torr when the chamber was at 42 torr. We stopped the process after the ionizer had become too cold to be comfortable with a elastomer O ring at the needles. (We should have a second heat exchanger to warm the gas back to room temperature before entering the ionizer to avoid the cooling.) The various times: start to stop of the ion injection 29 minutes, roughing pump down time after the injection to reach 1 torr about 3 hours, pressure now 3.5 x 10^-7 torr after 5 hours of turbo pumping. A disappointment of the procedure was that we were never able to achieve equal injection of positive and negative ions, the sampled ion current was: i+ = 4 x 10^-9 ampere, i- = -1 x 10^-9 amps. The injected currents are about 100 times larger. It requires tuning of the pressure on the needles and the flow rate to achieve equal positive and negative currents. If the total gas flow is too large one cannot get to the proper tuning. This was the case today. RW made a mistake in fully opening the gate valve between the chamber and the ionizer before tuning the system in an auxiliary flow to a pump and then gently transferring the flow between the pump and the chamber. There is a reasonable chance that enough ions did enter the chamber to neutralize the etmy. The success or failure should be known shortly after the morning meeting. If the neutralization has failed we could try again today. The times are shorter than estimated.
J. Kissel, K. Venkateswara After gathering a "first light" performance measurement of the H1 EX Beam Rotation Sensor (see first attachment -- the system is at air, poorly thermal shielded, and auto collimator is not well-secured to structure), we've spent nost of the day characterizing the separation between suspension point and center of mass, or "d". The goal is to have this separation be as small as possible, to maximize the rejection of translation noise. We're running one final 1 [mHz] data point over night to confirm, but we believe the C.o.M. is currently 20+/-4 [um] above the suspension point. We think we should be able to get this down to 2 [um], with an overall mass adjustment, which we'll do tomorrow. The second attachment shows the process -- drive the platform at X [mHz], measure the ASD, and since we do not change the amplitude of excitation, this is a direct map of the Beam Tilt / Ground Tilt transfer function. In reality, we make a fit to a sine wave to gather the amplitude with more precision and to have an uncertainty estimate. The first page shows a single measurement to demonstrate our signal is well above the noise floor, the second shows the collection of measurements in the form of an ASD, and the final page shows the modeled transfer function against our measurement points.
Turned off purge air at X-End station. Metal vent/purge valve is closed.
Rai was able to inject ionized gas into the YEND station chambers today. It took about 15 minutes to reach 42 torr where we ended the experiment.
The chamber is roughing down again and will transition to turbo this afternoon.
Pirani gauge plot attached.
Y-End is now on turbo, CC is ON, pressure is at 1.34x10-06 torr.
Cooling lines open for the turbo.
It actually took 29 minutes to get to 42 torr, faster than I had calculated.
The LVEA pumpdown has stalled at 10^-4 torr. There are many possible suspects with vented annuli on 8 chambers, and many new viewports and electrical feedthroughs. We will be busy restoring these systems in the upcoming weeks and probably spraying helium.
Day's Activities
This morning I adjusted the I-Q balancing of all the demodulators at the AS port (i.e. the ones in ISC R3 rack).
I did the same technique as described here. I did not turn on any of the whitening stages. The gain of the whitening filters were set 21 dB (although I am not sure if this is close enough for the future operational condition.) except for ASAIR_B_RF18 which I set the gain to be 45 dB. The beatnote were ajusted such that they are in between 40-50 Hz. Since the amplitude imbalance were typically smaller than 0.5 % or so, I did not try to correct the amplitude imbalance. Therefore only parameters I adjusted were the phase difference btween the I and Q signals. Here are the results:
= = =
At the very bottom of the rack, there is a quad demodulator which takes care of 18, 90 and 45 MHz LSC signals. From channel 1 to 4, the order of the demodulation frequencies should be 18, 90 and 45 MHz in order to satisfy the planned cable layout (see for example D1200666). However, this quad demodulator (S1001003) has these signals in order of 45, NAN, 18 and 90 MHz. The point is that this is a special mod version, where it has a fancy diplexer in it. So you can not easily swap the modulation frequencies between the channels unless one opens up the box and physically allocates the position of the boards. Because of that, the signal at the ADC were screwed up. For example, I obtained a signal from 18 MHz demodulator at a digital channel which was dedicated for 45 MHz (i.e. LSC-AS_A_RF45). In order to quickly fix it, I decided to swap the two ADC cables that were connected to the back side of the whitening filter. This pf course lead to another modification -- I had to swap the BIO cables.
RichardM, PeterK A few pictures were taken of ETMY in order to see if any dust particles could be spotted prior to Rai's vacuum test. The pictures were taken from the viewport under the beam tube as this appeared to offer the best view and allow some manner of support for a camera. A number of shots were taken with a 30 sec exposure. The illuminator was on for these photos. Even when zooming in on the images, it is not clear if there are any dust particles present because of the images becoming pixelated. One of the images shows a black spot near the centre of the optic but I think this is an artifact of using a smaller aperture.
(Borja)
Several issues has not allowed me to drive the ETMY ESD until late afternoon today. At this point I was able for the first time to test, with real data, the automation code for the ESD charge measurements develped at Livingston. I did have previously adapted it for Hanford's slightly different configuration but this was the first time I was able to test its results. Unfortunately the automation on the injection, data request and analysis is not robust, not allowing for the whole process to finish several times. Also the code does not take into consideration conversion factors on the V BIAS from Voltage to counts and viceversa. I assume this is taken care in Livingston outside of the code but certainly that solution does not make it universal.
Looking at the procedure with Rai I realized that we have to be careful on the level of the driving signal amplitude to be below the minimum V BIAS used in the analysis otherwise linear approximation assumptions in the methodology are no longer valid.
Rai is leaving on Saturday and we need to apply his discharging technique before then (optimally tomorrow). Before this takes place we need to have some ESD charge measurement data so that we can compare with data taken after the discharge and see the effects observed. This time constrains has made me decide to do the measurements manually tonight. I may be able to run the automation code afterwards and compare it tomorrow with the manual measurements but this may not be possible. See manual measurements in the attached document.
We are going to use the manual measurements as a basis for comparison before and after the discharge procedure which we hope to do today. I cannot comment on the automation program but urge several things: 1) The absolute value of the bias voltage needs to be greater than the absolute control voltage. 2) The phase and amplitude of the motion needs to be used. 3) The linearization routines should NOT be used. If these simple considerations are observed the algebra to fit for a charge dependent force coefficient is straightforward and linear. Vbias + Vcharge = deflection angle*a where a is an unimportant constant
The feeling of the LLO charging people is "Yes!" to all these points. Regarding the automation scripts, it *should* be a simple matter of setting up the desired biases and measurement parameters and hitting go. The scripts take the measurement in exactly the same way one would if doing the measurement manually, with all the amplitudes and bias offsets user-configurable and the signals are injected directly at the individual ESD quadrants (i.e. not through any linearization).
There is an important point which although I did not mention explicitly in the aLog entry is highlighted in the document attached to it. This is that the excitation driving each of the 4 quadrants provided a good SNR (of between 4 and 15 depending on the BIAS Voltage) of the oplev deflection values at the injection frequency with the exception of the LL quadrant which only showed noise (evidenced by the random values of the measurement points in magnitude and phase and by the low coherence of the between the excitation and the oplev deflection at the injection frequency). This issue, I have found, has a long and complex history which I will deal with in another aLog entry on the 9th August. I have also added to this comment the plots of the measurement results, an updated version of the measurements pdf and a table with final values of slope and Veff for each measured quadrant.
UL | UR | LR | |
Veff PITCH [urad] | 122 | 52 | 123 |
PITCH slope [V] | 2.6e-7 | 1.9e-7 | -2.65e-7 |
Veff YAW [urad] | 125 | 103 | 144 |
YAW slope [V] | -2.2e-7 | 2.34e-7 | 2.3e-7 |
model restarts logged for Tue 05/Aug/2014
2014_08_05 10:16 h1hpietmy
2014_08_05 10:16 h1iopseiey
2014_08_05 10:16 h1isietmy
2014_08_05 10:20 h1iopsusey
2014_08_05 10:20 h1susetmy
2014_08_05 10:20 h1sustmsy
2014_08_05 13:02 h1fw1
Unexpected freeze of h1seiey, unexpected dolphin glitch of h1susey during recovery, unexpected h1fw1 restart.
sorry, wrong date. Should be Tuesday 5th August.
model restarts logged for Mon 04/Aug/2014
2014_08_04 12:39 h1lsc
2014_08_04 12:42 h1lsc
2014_08_04 12:45 h1broadcast0
2014_08_04 12:45 h1dc0
2014_08_04 12:45 h1fw0
2014_08_04 12:45 h1fw1
2014_08_04 12:45 h1nds0
2014_08_04 12:45 h1nds1
no unexpected restarts. LSC ipc work plus related DAQ restart.
Sorry, got the dates wrong and skipped Sunday:
Sunday 3rd August: no restarts reported
Monday is actually the 4th of August.