Aug 16-18 (Camilla, Keita)
We used the Thorlabs power meter (59603) to make a power budget along the beam path. The Thorlabs sensor has ~+-0.2% systematic position dependence for the central +-2mm or so aperture and ~+-0.15% angle dependence for within ~2degrees of AOI, understand that there should be about +-0.25%-ish error for each measurement point, which is NOT included in the numbers presented.
All data points are in this google spreadsheet.
Beam was too high by 2mm at B:L2. No change in the throughput after correction.
Though we couldn't see a clear sign of clipping by looking at the Wincam image in the IFO path last week, and couldn't see any IR scattering from the aperture beamdump on B:L2, we measured the beam height at B:L2 using a metal ruler and IR-sensitive camera.
It was 65.5mm, full 2mm higher than the nominal 2.5"=63.5mm. The aperture beamdump attached to the lens has a diameter of 5mm so the distance from the edge is ~0.5mm IF the aperture is exactly at the right height and the Gaussian radius of the beam is ~700um, so this could have been a major source of the loss.
We changed the beam height and saw no significant change.
B-path throughput before the change was 97.4+-0.1%, 97.5+-0.4% after.
We measured the beam height for two other lens apertures.
For A:L2 it was a bit high but OK (~64.2mm, ~0.7mm too high, ~1.8mm from the edge for ~500um beam radius, 1.8mm/500um=3.6 radius).
For B:L1 it was right at the nominal 63.5mm height.
Throughput as of Aug 18
It's hard to make a power budget of A path because our reference photodiode to measure the power coming through OPO is in the FC path. The procedure is to stick the power meter in the path, measure the power level (and the dark level of reference PD), move it out of the path, measure the reference PD level. If the power changes between these measurements by X%, there'll be that much systematic in the result. Let's call this an A-path type measurement.
In the B-path, reference PD level and the power meter level are measured at the same time. Let's call this B-path type.
Also we were afraid that the reflection from the power meter surface goes back to OPO and hit unknown spot when we measure the light directly out of OPO, so we used the power measured between A:DC1 and A:M1 as the OPO output power. From many data points we made, we don't think that there's a meaningful loss (transmission, scattering etc.) on A:DC1.
A-path single path (FC/OPO):
| Throughput | Sanity check | |
| FC/OPO i.e. A-path single trip (both data points are Apath-type) |
0.993+-0.005 | |
| (A-path Single trip)^2 = 0.987+-0.010 |
||
|
from OPO to post-TFP of SFI1 |
0.986+-0.005 |
|
| From post-TFP of SFI1 to IFO i.e. B path single trip (both data points are Bpath type) 99:1 BS is not excluded from the number. |
0.975+-0.004 |
|
| (A-path RT)*B-path = 0.961+-0.006 |
||
|
Overall |
0.961+-0.005 |
99:1 BS was measured to be 98.97:1.03.
B-path potential high-loss points
Unlike in the above table, numbers below are just single numbers (no repeated measurements). And remember again about up to ~0.25% error in each measurement. Take them with a grain of salt.
B:L1 0.35%, B:M3 0.46%, B:M4 0.35%.
Replacement of B:M4
After these findings, I hit B:M4 with a corner of the power meter and replaced it with a new one though I saw no obvious damage (but it should be checked under a super bright white light).
What was in: E1900392, SN80, HR1064+532 AOI 35-45 degree. Actual AOI is ~55 degrees, this is supposed to be an OK optic (vendor design is C1900351).
New one: E1000595, SN111, HR1064, AOI 45+-5deg.
We're checking other optics. Details will follow.
Aug 19, optical characterization in the lab is complete (Camilla, Keita)
New B:M4 is better. We gain +0.5% +- 0.4%
The old number had a larger error bar so it's still possible that it was about 98% before, but I'd like to think that there was a real improvement.
Out of 2% remaining loss, 1.0% (measured) comes from 99:1 BS. Faraday is supposed to be ~0.5%. The last 0.5% or so is likely from B:M3, B:L1 and/or B:L2.
Again the data is in the same spreadsheet as before (tab named "data5, B path").
We didn't repeat the overall throughput measurement because nothing changed in A path. The following numbers are still the newest and valid.
Note about loss budget for A path
We actually made some measurements to make a detailed loss budget of A path that's as nice as B-path, but they didn't make much sense (some losses looked big both positive and negative). See my comments about Apath-type measurement points in the alog above. No ISS means that we needed to repeat many measurements and hope that things will average out in the end. We didn't have time to do that, so we just repeated measurements for A path single and round-trip, and once we knew that they are consistent with each other without excessive losses we were happy.
Note about the beam spot on B:M4
Since we lowered the beam height on B:L2, I eyeballed the beam spot position on B:M4 using a viewer card, using various edges of the Siskiyou mount as reference points.
It was ~2mm off to the side (closer to B:BS1) when viewed face-on, and was about 3.7mm too low in the worst case estimate. Even if I take into account the relatively large 55deg AOI, it seems that the distance between the center of the beam axis and the edge of the mirror is about 5.8mm. I'm not concerned.
Inspected B:M3, B:L1 and B:L2, nothing obvious was found
We didn't replace any of these.
Beam profile measurement in the IFO path
We made a final measurement of the mode profile in the IFO path as the beam spot position on B:L2 is lower than it used to. This is just to check that nothing crazy happened. No analysis yet but no reason to believe that something is seriously wrong.
Did somebody damage Wincam?
We used a nanoscan instead of Wincam because Wincam showed a strange vertical thing (picture 1) that nanoscan didn't show even when rotating the head. When I blocked bottom or top half of the beam using a sensor card, only the beam is blocked but the vertical thing stayed on Wincam until the entire beam is blocked. I remember seeing this when I and Georgia was trying to find any sign of clipping in the IFO path (last week), but this wasn't the case a few weeks ago. Something is wrong with Wincam but I don't know what. Lee wanted to use Wincam in chamber, so we need to investigate.
Forward-rejected beam by A:P2 isn't clipped by F:L1 lens holder
Since F:L1 lens was very close to the main path of the SQZ beam going through SFI1, there was a concern that the forward-rejected SQZ beam in a wrong polarization (P-pol) might fall on F:L1 holder. We made a crane contraption to insert QWP between TFP and A:F1 so a significant power is rejected by SFI1 and confirmed that, though the rejected beam is very close to F:L1 (picture 2), it actually goes to the aperture plate on A:L2 (picture 3). Using a viewer, we couldn't find any scattering on the aluminum surface of F:L1 holder (picture 4).
Installation of an aperture plate between B:F1 and B:P1
We installed a single (nominally double), conical side faces B:F1. This change is the result of a discussion among SQZ team.
Alignment procedure: Horizontally set the position using an IR-sensitive camera and a viewer card to avoid parallax problem as good as we can. (It's a bit hard to see in picture 5, but there's a beam spot on the viewer card and the beam line is almost exactly under the edge of one of the aluminum brackets on SFI2 . Everything looks right horizontally.) Next measure the height of the beam using a ruler and the viewer card (accuracy +-0.5mm). It was 63.5mm+-0.5mm above the OPO platform (nominally 63.5mm), almost perfect.
We also set up an external retro reflection mirror in IFO path and inserted a QWP between the IFO path retro mirror and B:M4 to obtain wrong polarization propagating back to see where the rejected beam fall on the newly installed aperture plate.
It was pretty close to the flat part of the aperture, but picture 6 shows that the beam is almost entirely on the conical surface. Good.
Installation of an aperture plate between A:F1 and A:P1
As per design, installed a single, conical side facing A:F1.
Same alignment procedure as the above. Picture 7 shows the horizontal alignment. The beam height was measured to be 53.5mm+-0.5mm above the SFI sled, nominally it's 53.975mm, so it's a bit low but it's still good.
QWP was inserted between the retro mirror in IFO path and the external 90:10 BS so the back-propagating beam has a wrong polarization component. The rejected beam fell on the conical surface of the newly installed aperture plate.
All temporary irises we installed as the alignment helper were removed.
Aug/20, prep for balancing (Camilla, Keita)
Attached are the pictures of the fibers in storage pan.
F path (CLF) fiber is in a deep pan (picture 1) with two fibers inside. The one we've been using has a long ribbon of aluminum foil at the end that goes to the collimator. The other end of the fiber is attached to a temporary barrel.
G path (pump) fiber is in a relatively shallow pan (picture 2).
H path fiber is in another shallow pan with two fibers inside (picture 3). My fingers are pointing at the fiber ends we'll be using. The other fiber is broken.
Also attached are the microscope pictures of the fiber connector that would go on the collimator. You can see some dusts but they all look great around the core.
Attached are the photos of the fibers we will install in HAM7 and inspection photos for G and H.
Serial numbers we are using:
G:COL1 = DIAMOND9708433--00-2135431V003
H:COL1 = DIAMOND9708433--00-2135431V005 In the photo Keita is pointing to which fiber to use.
(DIAMOND9708433--00-2135431V001 is broken but in the box with H fiber)
I inspected one in-vac IR fiber that's supposed to be a spare, and it seems to have a damage around the core at one end.
Here's my understanding of long in-vac IR fibers we have at LHO (E1700235 is not up to date).
These are all D1700396.
Microscope image shows that there's some kind of damage at one end, the other end looks good.
See attached.
Looks good.