35W beam. The ISS gain was changed from -2.1 to 7dB which seemed to quiet the noise at high frequencies.
Thomas V. Mark D.(Apollo) John W. Gerardo M. We have finished installing the enclosures for ITMY and ETMY. During installation, ETMY was misaligned and I mistakenly re-aligned to the reaction mass ghost beam, will fix to reflect off the ETM. ITMY is fully operational, some internal focusing of the telescope may need to take place since the beam diameter is approximately 8 mm, spec calls for 2-3 mm. However, it is up to the commissioners of the OAT whether or not this adjustment is worth doing. Deviations from design: 1) Due to the tightness between the ITMY transmitter pier and the viewport, we were only able to get the viewport protector in between them and couple the enclosure to the protector via sheet metal and tape. 2) It was also pretty tight between the viewport protector and the enclosure on the ETMY receiver pier, we were not able to fit a full sized bellow to couple the connection between the two. We ended up cutting the bellow to a smaller size and using that. Also, we adjusted the level of the receiver pier in order to not have the enclosure touch the beam tube (the corner of the enclosure was butting up against the side before).
Editorial note:
I assume this entry refers to the enclosures around the optical lever beams for ITMY and ETMY, though nowhere in the entry is the term 'optical lever' (or any of its derivatives) found. This makes the entry rather abstruse, and it also means that if one searched the log for optical levers, oplevs, optlevs, etc, one would not find this record.
Technical note:
If the design calls for the beam size to be adjusted to be 2-3 mm (radius? diameter?) on the detector, it should be adjusted to that. The optical lever isn't really considered to be installed until that is true.
For the evening shift, we set out to check the VCO calibration, and get some new calibrated spectra. This work starts where entry 3400 left off...
After relocking the cavity we found that he SR560 path (which was described in 3400 as having a gain of 10 at the SR560, and another gain of 10 at the TTIF) saturated due to low frequency motion (below 1Hz). With a gain of 1 on the SR560 it worked, and the 2 signal paths (SR560 and FMON) showed essentially perfect coherence. This also showed that the FMON calibration is not -20dB at DC and 20dB at HF, but rather 0dB at DC and 40dB at HF (see comment to 3400). With this direct measurement of FMON in hand, we removed the SR560 path.
Discussions with Bram brought out a missing factor of 2 in my previous calculation: since the PLL is in the 1064nm output of the laser, and we are using the 532 for locking, we get twice as much frequency change in the light seen by the cavity as the light seen by the PLL. (The same result can be achieved by ignoring the fact that green light is used at all, such that the conversion from frequency to length becomes L_cavity / f_laser = 3995m 1064e-9m / 3e8m/s = 14.2pm/Hz.)
In summary:
614pm/count using VCO, which is awesome. 0.8nm/count using OSEM, which is 30% off of the above.
Yet people keep saying that OSEM calibration should be fine at 20% level.
For people outside of the ISC:
1. Frequency offset between the refcav in the corner station lab and the EY ALS red beam is kept at the VCO frequency by a PLL loop.
2. Transfer function from VCO frequency to the red laser frequency is equal to G_PLL/(1+G_PLL) where G_PLL is the OLTF of PLL. Since UGF of G_PLL is about 20kHz, this is almost equal to 1 for f<1 kHz or so.
3. Electronics transfer function from VCO tuning input to the frequency of the VCO was measured to be H_VCO=139kHz/Volt *pole(1.6Hz)*zero(40Hz).
4. The green laser is the second harmonic of the red, therefore the transfer function from VCO tuning input to the green frequency is 2*H_VCO.
5. To lock the laser to the arm, we control the frequency of the green laser by feeding PDH signal via CM board back to VCO tuning input.
6. When the laser is locked to the arm, VCO input is equal to
v= (1+G_PDH)/G_PDH / (2*H_VCO) * df
where G_PDH is the OLTF of PDH loop and df is the equivalent frequency fluctuatoin of the cavity length, refcav and pll loop noise combined.
Since UGF of G_PDH UGF is 8kHz or so, again (1+G_PDH)/G_PDH is almost 1 for f<1kHz or so, therefore
v ~ df/(2*V_VCO)
7. Frequency to length conversion is 4000m/ 563.5THz (green light).
8. PDH CM board output is connected to PDH FMON channel, which has the transfer function of 2^15 counts/10 volt * zero(10, 10)*pole(100, 100).
Therefore, all in all, VCO input measured by FMON is calibrated back to length using
dL = FMON / (2^15 cts/10 V) *pole(10;10) * zero(100, 100) * 2*H_VCO * 4000m/563.5THz
= 602 pm/cts * FMON *pole(1.6;10;10) * zero(40;100;100).
Small difference in number (Matt's is 614 pm/V) came from the fact that I used 2^15/10 cts/Volt for ADC while Matt used the ADC calibration he did with his table top diode box as an amplifier.
Attached are plots of dust counts > .5 microns in particles per cubic foot.
The H1 workstations in the control room now have a controls account. Use caution, the controls account is the same home directory for H1 and H2, but the environment will be set up for H1 on the H1 workstations and H2 for the H2 workstations. So, even though the controls account looks the same on H2 and H1 workstations, you cannot do H2 work on an H1 workstation, or H1 work on an H2 workstation. Confused? The H1 workstations are labeled as such with a tag on the upper left corner of the display. The background of the controls account screen will no longer indicate it is "H2".
HSTS MC1 has been moved to the testing bench and cabled to the test stand. The offsets and gains have been entered into the MEDM screens and the OSEMs have been positioned to 50% light. We will recheck and adjust the 50% light positions in the morning, if necessary. This suspension should be ready for Phase 1 testing. M1 BOSEMs T1 S/N 578 OL = 25960.0 Offset = -12980 Gain = 1.156 T2 S/N 436 OL = 29277.9 Offset = -14639 Gain = 1.025 T3 S/N 440 OL = 27858.1 Offset = -13929 Gain = 1.077 LF S/N 433 OL = 27758.7 Offset = -13879 Gain = 1.081 RT S/N 424 OL = 24529.4 Offset = -12265 Gain = 1.223 SD S/N 658 OL = 25139.5 Offset = -12570 Gain = 1.193 M2 AOSEM UL S/N 453 OL = 27359.6 Offset = -13680 Gain = 1.097 LL S/N 320 OL = 24780.7 Offset = -12390 Gain = 1.211 UR S/N 228 OL = 25117.9 Offset = -12559 Gain = 1.194 LR S/N 438 OL = 25521.6 Offset = -12761 Gain = 1.175 M3 AOSEM UL S/N 227 OL = 28354.2 Offset = -14177 Gain = 1.058 LL S/N 353 OL = 26509.5 Offset = -13255 Gain = 1.132 UR S/N 276 OL = 27920.9 Offset = -13960 Gain = 1.074 LR S/N 299 OL = 28435.2 Offset = -14218 Gain = 1.055
The login scripts have been modified for the controls account to allow use of the account for either H1 or H2. This should have minimal impact on H2 users. Let me know if there are problems with the environment for controls.
Jim & Greg adjusted the level of the HAM3 ISI Table using HEPI Springs. It was a little slower than expected. We are ready for IAS to take another look before we continue with the HEPI Actuator installation.
As of around 4pm, sounds like the Gate Valve (GV18) will be opened tonight for OAT work.
John W. observed that actuating GV18 for 4 minutes and 18 seconds did not hard-close the valve, i.e. is a good value to use for soft-closing GV18. History: The individual actuation times resulting in a hard closure of the beam-tube valves (electric) was observed and recorded early on. This value is used to determine the actuation time to soft-close a given (electric) beam-tube valve. We are now wanting to record the actuation times of all of the electric valves.
Rebooted h2pemeyaux at 17:55 PDT July 10 to set IPMI address to correct value after reassigning the IP address of the computer, following dcuid value changes to the models running on the machine.
I started the evening shift with calibration of the VCO + Frequency divider used for the ALS PLL (S1200570 and S1000748). The objective was to measure the response of the VCO output to its Tune input, which is used for locking the arm cavity and thus provides a good calibration point for the arm cavity signal.
I set up the DC calibration measurement with a signal generator (output displayed on a scope to be sure), and a frequency counter. I took 6 samples between -1V and 1V, which resulted in an excellent linear relationship: f = 39.6MHz + 139 kHz/V * V_tune. For example, at V_tune = 1V, I measured f = 39.739 MHz. This is not the expected result of 50kHz/V given the nominal 100kHz/V from the VCO and the division by 2 which takes us from 80MHz to 40MHz.
From there I moved on to measure the AC response. I did this by rigging up a frequency to voltage converter with some bubble gum and duct tape I found nearby. The output of the converter was 725mV / V_tune. With the help of an SR785 and a lot of patience I turned this into a response measurment between 0.5Hz and 10kHz. I saved the curves on the SR785 floppy, which is a lot like throwing them into a black hole, so I also wrote down a few points: {-30.7dB, 178dg} at high frequency, {-27.7dB, 138dg} at 40Hz, {-16.8dB 112dg} at 8Hz, {-5.8dB, 136dg} at 1.6Hz and {-3.26dB, 136dg} at 5.62Hz. These are well fit by the measured DC response and the expected 1.6Hz pole and 40Hz zero.
Lastly, in preparation for using this calibration for the arm cavity, I cabled up an SR560 to the IR_PWR_MON channel. The SR560 is AC coupled with a gain of 10, and the tabletop interface gain is set to 10. From the output of the SR560 to counts of IR_PWR_MON, I measured 32060 counts/V. Just to be sure, I checked the gain of the SR560 and appears to be very close to the stated value of 10, so we should expect 320.6 counts/mV refered to the SR560 input, which is attached to V_tune. Assuming the above VCO calibration is correct, this converts to 0.434Hz/count as seen at IR_PWR_MON. The cavity should lock should convert length to frequency as L_cavity / f_laser = 3995m / 5.635e14Hz = 7.09 pm/Hz, so we will be left with 3.07pm/count as our length calibration. (To relate this to the number found with the OSEMS in 3363, we should remove the gain of the SR560, the TTIF box, and another factor of 10 for the FMON signal filtering. This would give 3nm/count instead of the 1nm/count found with the BOSEMS, so somewhere we're missing a factor of 3, not unlike the 139kHz/V vs. 50kHz/V discrepence... to be investigated tomorrow in the light of day.)
On a side note: I found the PLL unlocked and no beatnote evident. On closer inspection there was no light coming from the fiber, so I went to the optics lab and found the laser in Standby. I have no idea why it would be left in this state, since this clearly makes cavity work impossible. That should have been enough to make me leave it as I found it, but instead I turned it back on and relocked the reference cavity. From the trend data, it looks like it unlocked about 15 hours ago (at 16:20 UTC, see figure).
A harder look at the VCO schematic lead Keita and I to believe that 139kHz/V is plausible. Also, the FMON path has a DC gain of 0dB and gain at HF of 40dB (2 zeros at 10s, 2 poles at 100Hz), so the relation to the OSEM calibration in entry 3363 is 0.3nm/count from the VCO calibration (above) vs. 0.8nm/count from from OSEMs. Still a factor of ~3 is missing.
Short story: I created a manual burt backup and burt restore script to back up the PLL and PDH common mode board settings in the EY Beckhoff system. You can find them in /ligo/home/controls/bram.slagmolen/burt/h2etcatey and are called h2ecateyBurtbackup and h2ecateyBurtrestore. They use a local autoBurt.req and safe.snap file.
Long story: Every time the beckhoff system in EY breaks all settings are losts, this greatly inconvience the the locking of the PLL and the PDH system.
Patrick T. generated an autoBurt.req file from Beckhoff system h2ecatey at EY with all the epics channels.
With help from Dave we created and place the autoBurt.req file in the target directory and created a safe.snap file in the burt/ direcotry (burtrb -f autoBurt.req -o safe.snap).
> controls@opsws5:burt 0$ pwd
> /opt/rtcds/lho/h2/target/h2ecatey/h2ecateyepics/
I edited the autoBurt.req file adding 'RO' infront of some channel names, this prevented some errors while creating the .snap file.
When trying to restore from the .snap file, not all settings are restored correctly. With some digging around, it seems that when we added and set '.ZNAM Off ' and '.ONAM On' to the variable names it seemed to work. In addition in the autoBurt.req file I added .VAL to the same channel names. I only did this to few required channels in the PLL and PDH CM path. After a burtrb the burtwb worked correctly.
This is seemed only required for the toggle swithces. Why this is required I don't know, although this doesn't happen with the QPD whitening channels. I think the OPC database may need some work.
When the Beckhoff system is rebooted all the .ZNAM and .ONAM addition are lost. I made a little script set_ZNAM_ONAM_records to add these again (not tested but should work).
A listing of the directory follows:
controls@opsws5:h2etcatey 0$ pwd
/ligo/home/controls/bram.slagmolen/burt/h2etcatey
controls@opsws5:h2etcatey 0$ ls -al
total 156
drwxrwxr-x 2 controls controls 4096 2012-07-10 18:44 .
drwxrwxr-x 3 controls controls 4096 2012-07-10 15:53 ..
-rw-rw-r-- 1 controls controls 28005 2012-07-10 17:39 autoBurt.req
-rw-rw-r-- 1 controls controls 27971 2012-07-10 16:17 autoBurt.req~
-rwxrwxr-x 1 controls controls 241 2012-07-10 18:20 h2ecateyBurtbackup
-rw-rw-r-- 1 controls controls 38 2012-07-10 16:55 h2ecateyBurtbackup~
-rwxrwxr-x 1 controls controls 225 2012-07-10 18:21 h2ecateyBurtrestore
-rw-rw-r-- 1 controls controls 34666 2012-07-10 18:23 safe.snap
-rw-rw-r-- 1 controls controls 31283 2012-07-10 17:01 safe.snap~
-rwxrwxr-x 1 controls controls 716 2012-07-10 18:44 set_ZNAM_ONAM_records
-rw-rw-r-- 1 controls controls 703 2012-07-10 18:09 set_ZNAM_ONAM_records.txt
-rw-rw-r-- 1 controls controls 498 2012-07-10 18:07 set_ZNAM_ONAM_records.txt~
controls@opsws5:h2etcatey 0$
Patrick updated the OPC database. He rebooted the Beckhoff and it recovered its settings.
He generated a new autoBurt.req file, which was copied to the target directory.
Doing my local burtback up and burt restore without any modifications to the channel names or burt files seemed to work.
An hourly burt back up is performed automatically and the .snap files can be found at /ligo/cds/lho/h2/data/burt/{year}/{month}/{day}/{date-time}/
Attached are plots of dust counts > .5 microns in particles per cubic foot.
Work resumed following yesterday's "Stop Work" and today's subsequent "Stop Work Release"
The first thing I found in the morning was that Beckhoff was dead.
On Windows computer that runs Beckhoff, both of two PLCs showed an error message dialog box ("connection error" or some such) and it was not talking to the hardware.
Alexa did several attempts to make it work without rebooting Windows, but eventually Patrick rebooted Windows.
This did the trick, but it had a side effect of wiping the settings for OPC variables (again).
Bram is working on a sad implementation of burt restore.
Jim Jason & Hugh The Initial Alignment Subsystem(IAS) crew gave us the bad news late morning and we translated East 5mm shortly after lunch. Given all the initial movement we saw on the Dial Indicators when we first floated & further when we tweaked the level yesterday, I'm not surprised by this shift requirement. And you all have heard my spiel about from where our initial position originated. Anyway, after the East shift we were almost bang on N-S & rotationally with ~.5mm & 300urad to correct. That done, East-West checked again and still good. Now back to level and elevation: our runout on the table level is 0.3mm and we are ~0.4mm low in elevation. Tomorrow we will bring this back in and attempt to equalize the load cell readings while we are at that and finally followed by a revisit of the table position by IAS.
