Alignment into HAM2

Locked JAC with RF at 1W. IMC WFS was centered in IOT2L and IMC started working. As we steered JM3 MC2 TRANS responded as expected (i.e. JM3 PIT -> MC2 YAW and vice versa though there's a significant cross coupling). Sheila had to recenter the WFS once again at some point as the WFS started to get off-centered as we turned JM3, but as far as the WFSs were centered, things worked.

In the end, MC2 TRANS was centered reasonably well (right before the cursor in the attached), JM3 DAC output was O(1e7) while a 28 bit DAC's range is +-134e6.

Apparently we made the last step of JM3 PIT adjustment in the wrong direction at the time (we didn't notice because the WFS is very slow) and overshot. I steered it back later, now H1:SUS-JM3_M1_OPTICALIGN_P_OFFSET is -39 instead of -19 (second screen shot).

Anyway, this means that the alignment into HAM2 is good with more than a comfortable range left for JM3. No need to further refine.

One caveat is that the IM4 trans is totally off in PIT. But that's a downstream problem which we'll have to deal with later in HAM2. That should not prevent us from moving forward to close down the chamber.

The tasks listed below are described in different alogs.

JAC TRANS PD calibration

Last beam dump (alog 89249)

ALS beam path check

We installed JM1, balanced it and aligned it. A beam dump was placed behind it though we could not see the transmission with 1W input.

After this, JAC locked with RF without any problem though the input was wobbly when the purge was up.

We searched for unexpected ghost beams (also with 1W input) and didn't find any.

We uninstalled many (but not all) temporary dog clamps and irises.

We revisited the IMC alignment because it's been off in PIT since Thursday or Friday. We locked JAC using dither (because we wanted to turn down the purge air). We enabled the IMC WFS just for MC optics and steered JM3, but weren't able to center the MC2 trans. Steering JM3 just made the IMC transmission worse while making not much impact on the desired degree of freedon (JM3 PIT -> MC2 trans YAW, JAM3 YAW -> MC2 trans PIT).

Tomorrow, we'll revisit the IMC alignment. We'll also measure the power coming into JAC TRANS PD as well as the actual transmission of JAC while locking it with RF so we can use JAC trans PD as the measure of the power into HAM1.

Modulation index measurement for 45M and 9M, they're healthy.

1W into JAC, no WFS for IMC, ITMY single bounce. Scanned OMC with both 45MHz minimal power (~4dBm) and 9MHz full (~27dBm), then both in full power, then only 45MHz in full (see screen shot).

Jennie will post the modulation index later, but the peaks look healthy to me.

The alignment into IMC was rather off in PIT and the guardian had a hard time locking IMC (I changed the lock threshold in the guardian). We didn't move MC, didn't bother to do any thorough investigation, but either JAC or MC mirrors moved. Maybe we have to lock IMC, use WFS for MC, and steer JM3 and see if the MC mirrors will be driven far from where they are. If they are we might have to touch up in-chamber alignment again.

JAC TRANS BS was swapped with the real one (laser window with AR only on one surface).

We replaced the JAC TRANS BS (which was a temporary high reflector until today) with the real one that is just a bare glass (the reflectivity is 7% or so depending on AOI). 

We locked JAC with RF, was able to see the beam coming to the TRANS PD but wasn't able to see the reflection of PD, so we raised the power to 1W.

Initially the beam was missing the PD because the new optic is thinner than the temporary one. We realigned the optic to steer the beam back to PD. Since we could not directly see the PD surface, we just steered the beam up and down, left and right until the PD output starts to drop, and put the beam roughly in the middle.

We were able to see the PD reflection using an IR card and an IR viewer. There were two reflection blobs (which was the case before, too),  and I moved the JAC TRANS BD in +X direction by 1/4~3/8" to catch both.

Beam dump for HAM1-HAM2 septum window AR reflection.

This was installed. 

Beam dump for the -Y door viewport reflection.

Jason and Betsy installed it. There seems to be a discussion as to whether or not something else needs to be done.

EOM and JAC power budget again.

We measured the power at various places. While Jason held the power meter head still, I ran the statistics function of the power meter for a few seconds. I only list the mean and the standard deviation.

Wrong pol is king of large, ~0.2% of the main beam power coming out of EOM. That's 200mW when 100W goes through the EOM. 

JAC throughput of 92% is not great, but Jason says the alignment and the matching are not really optimised.

We locked JAC with 1W input to find ghost beams. Details will come after we're done with the septum window reflection, but anyway here is the list:

1. EOM AR reflection. Properly dumped today.
2. L1 reflections (front and back surface) are reflected by JAC output coupler and go through L1. Properly dumped today.
3. L2 reflections (front and back). Properly dumped today.
4. L3 reflections (front and back). Properly dumped today.
5. EOM wrong pol forward going beam. Changed the beam dump position today. This was brighter than I thought, we will measure the power again.
6. JM2 transmission. Already dumped, confirmed it's still there.
7. JM3 transmission. Already dumped, confirmed it's still there.
8. Septum window reflection. This was coming back horizontally in -Y direction, was about half inch off from the main beam at the top periscope mirror, and hitting the metal part at the top edge of the bottom peri mirror. Not dumped yet. 
   • We'll have to improvise something so the beam is dumped at the top periscope position before entering into the periscope structure. Something like a variation of the beam dump behind the top mirror of the  JAC input periscope. 

None of the new beam dumps are interfering with the main beam and JAC refl/trans.

We refined the alignment from JAC to IMC by iterating small amount between JM2 and JM3. It was hard due to purge air and the suspended JM3, instead of relying on the IMC scan and minimize 1st order modes, we locked IMC and minimized the IMC REFL DC on average. After that the beam was still centered on JM3 well.

We closed the chamber and turned down the purge all the way and the alignment was indeed good. There was almost no 01 (PIT) mode power, 10 (YAW) mode power was less than 1% of 00 mode power, and the 20/02 mode power was 0.23 to 0.24%. See MM.png.

(To identify which mode is what, I intentionally misaligned JM3 in YAW (that causes PIT misalignment for IMC. See  mode_identification.png.)

This morning we have pushed the JAC EOM by about 0.6mm (using ~25 thou thick washers) in -Y direction, following the finding of last Friday (alog 89158). 

After that the beam was good on the input side plate (the beam is offset in +Y direction by 0.1mm) and was OK on the output side plate (0.5mm offset in -Y direction).

The beam position on the crystal itself should be ~0.13mm in +Y direction on the input face and ~0.36mm in +Y direction on the output face. The angle between the nominal path and the actual path outside of the crystal is about 0.6 degrees. See pictures and cartoon.

Calculation depends on the refractive index, I assumed n~1+deflection/wedge=1+2.35/2.85~1.85, but using 1.85+-0.5 instead won't change anything in a meaningful manner.

This is acceptable, the beam is more than 1.5mm away from the side face of the crystal, cannot remember the beam radius but it should be smaller than 600um if FDR is still valid, so it might be 2.5 beam radius or maybe more.

T. Sanchez, J. Oberling

This afternoon we took a look at the JAC Refl path and positioning of IOT1.  We ended up pushing IOT1 in the +X direction by ~1"; new marks were made on the floor to represent this position.

From here we used JACR-M1 and JACR-M2 to align the beam through the viewport simulator and onto the top periscope mirror of IOT1.  This only took a couple of iterations before things looked mostly OK.  The beam is currently a little low in the viewport simulator, maybe 0.5" to 1" (rough measurement by eye) and is hitting the top periscope mirror.  We finished by measuring the beam height at 2 places on the ISI table roughly 28" apart, to give us an idea of how the beam is pitched.

• Point 1: 109.0 mm
• Point 2: 108.25 mm
• Pitch: ~1.05 mrad down

This seems OK for now, but on some reflection while sitting in my office typing this alog we may want to briefly revisit this.  The beam is slightly pitched down and is high on JACR-M2 (~0.5" on a 2" diameter optic), so I think there is some room to improve this, to get the beam closer to center on JACR-M2 while still passing through the viewport simulator and onto the top periscope mirror of IOT1.  More to come.

Following alog 89115, we found that the old batch crystal from that alog (S/N10252003) had a big chip at one corner. It is pretty bad we don't want to use that.

Betsy found another old batch (S/N10252007, "inspected 12/21/11" and UF tag dated 4/21/09), so we A-B-ed that one with the spare new batch (S/N B1913108).

The beam path was made as level as possible at 3" height using a beam leveling tool (a black metal thing with a tiny aperture at each inch of height).

We put the crystal on a platform that is roughly 2" 29/32 (which is about 2.4mm lower than 3"). The crystal is 4x4x40mm so that's about the right height.

We spent some time to make YAW alignment as good as we can for each of the crystals.

We scanned the beam in PIT from top to bottom (or bottom to top), each extreme is where the beam is almost clipped (but not actually clipped) by the top or the bottom face of the crystal.

Look at the attached, the new batch (left column) clearly shows multiple beams even though the focus is not as sharp as the old batch photos. As we misalign in PIT, the dark place moves relative to the main beam and the contrast changes too, but multiple ghost never went away. At the extrema (very close to the top or bottom edge) it looked as if the beam is better but I'm not sure it actually was.

The old batch (right) didn't show such a behavior. The beam shows something like a diffraction pattern but no separate ghost beams. Everything moved with the main beam.  Not sure if the diffraction pattern came from the aluminum surface or EOM, but clearly this is MUCH better than the new batch.

Note, due to the apparatus (the steering mirror is 20" upstream of the EOM), we haven't searched in a huge PIT angle space, it's actually roughly +-4mrad or so, the angle is not negligible but it's more parallel displacement scan than an angle scan.

Also note, when the crystal was put in place it seems that there's some vertical deflection which was different for the old and the new. On the top two pictures, there's no change in the input alignment into the crystal. 

Based on this observation, I'd say using the old batch makes sense. LHO people (Jennie, Rahul, Betsy and myself) had a brief conversation with Masayuki and MichaelL and we all agreed that that's the way to go.

It took much longer than expected but we set up the beam path for the RTP test in the OSB optics lab.

Since more power makes it easier to see the ghost beams, I removed the beam dump that used to receive most of the red power (~530mW) and directed the beam to the front of the table (red path in the attached). I stole the steering mirror that used to be used for the low power P-pol path (circled in red). The low power p-pol path is now simply blocked. No other change was made to low power S-pol path (orange) as well as green path (green), but the beams are blocked by beam dumps. If you want to use these, simply unblock.

The beam radius will be 300~400 um or so at the location we plan to put the RTP (represented by a green rectangle in the second attachment). Elenna will post the plot of the beam size measurement.

The third picture shows the containers for different RTP. Left is the one for the crystal in HAM1. The middle seems to be from the same batch. Right looks different, on the bottom of the container there's a label saying "I/O something something 2017" so this is likely the old one.

We didn't have time to actually test the crystals, wait for tomorrow's udpate.

Jennie W, Jason O, Keita K.

As reported in this alog (#89073) from Masayuki and Keita, after we turned the power in HAM1 up to 1W we found a series of vertically spread ghost beams aroubnd the main beam after the EOM and before JM3.

These could not be removed by translating, yawing or pitching the EOM position relative to the beam. It was decided in a larger meeting with EOM design personnel that we would first check if the crystal was cracked or damaged anywhere in case this is the cause.

First photo shows the EOM from above, using a green torch to illuminate the beam path. I can't see any scatter from defects or cracks in the crystal.

Second photo shows possibly a chip at the corner, but this should not affect the beam as its right at the edge.

Third and fourth show side view with illumination from the top at an angle.

In summary we did not see any 'smoking gun' to cause these ghost beams.

Summary of investigation into the vertically split beam from the EOM

• First, we confirmed that the vertical beam splitting observed yesterday originates from the EOM itself. To check the possibility of multiple reflections from lenses, we inspected the back-reflection port of the JAC output mirror. Two reflected beams were observed, most likely originating from the planar and curved surfaces of the lens, and they were mainly separated in yaw. Since the space between the EOM and the lens was blocked during this test, these reflections were conclusively identified as lens reflections. No vertically split beams were observed from this source.

• A knife-edge–like test was performed by slowly lowering a metal ruler from the top at both the EOM input and output. At the EOM input, the entire beam disappeared simultaneously, whereas at the output the beam disappeared gradually from the top. This behavior confirmed that the vertical splitting is generated inside the EOM.

• To accurately determine the beam positions at the EOM input and output, beam positions were measured from photographs. Taking refraction and geometry into account, it was found that the beam is slightly displaced in the horizontal direction. Details of this analysis will be documented in Keita’s alog.

• Based on this result, the EOM was rotated in yaw. Dog clamps were placed at the ±y projection points of the EOM input and output, in contact with the base plate. One 0.5-mm shim was inserted between the base plate and each dog clamp to rotate the EOM counterclockwise. However, the beam pattern did not change. Additional shims were tested, but no significant change was observed. The EOM was fixed with one 0.5-mm shim at each position.

• Next, pitch adjustments were explored. The original shim configuration (two shims on the +y side and one on the −y side) was changed by moving shims to the +x side (one location at center) and the −x side (two locations, upper and lower). Each location initially had two shims, and by adding or removing shims, it was observed that the vertical positions of the split beams changed. When the +x (downstream) side was raised, the vertically split beams appeared in the upper part of the beam profile (approximately 5–7 beams; Keita will upload photos). Conversely, when shims were added to raise the −x side, the split beams moved toward the lower part of the beam. With the downstream side lowered by approximately 0.25 mm, about two beams were observed in the upper part and one in the lower part.

• 1W input power shows 6 or more beams, but 2-3 beams can be observed even with 100mW.

• With the last configuration, we proceeded the IMC scan measurement after alignment. The 2nd order mode peak was the same level as we observed when we sim up the EOM first time.

• In summary, the EOM shows highly questionable and nontrivial behavior. Possible causes include diffraction due to crystal defects or multiple reflections at the AR-coated surfaces. However, identifying the exact mechanism is challenging at this stage.

J. Kissel
- D2500175 :: S3228003
   . D2100573
   . E2300345
- D2400281

I'm finally back in the optics lab, taking the next steps towards assembling the SPI. 

Specific to this entry -- I'm assembling the Class A clean components of the fiber-coupled seed laser light. In other words, I'm integrating 
    (1) S3228003*** -- One of the 4.5" (inch) 1064nm V-FT Optical Fiber Feedthrough (Feedthru) Conflat Flange (DIAMOND) (D2500175) -- which includes the "off the shelf" (OTS) feedthru and integrated 3 [m] patch cord (the industry jargon for fiber optic cable), and its MIT-designed strain relief assembly D2100573 -- all delivered to LHO after class-A cleaning and assembly at Caltech per E2400159.
    (2) the ANU-designed Fiber Storage Spool assembly, D2400281, of which we have two.

Sina characterized the power transmission of all three of the feedthrus at Stanford before the clean-n-bake process, and identified that S3228003 had 100% transmission, so I've chosen that to be assigned to the MEAS path, where we need the most input power (as it's distributed through the most number of beam splitters). I plan to further integrate the patch cord into the SuK fiber collimator S0272503 for no better reason that to "pair the S[...]03 feedthru with the S[...]03 fiber collimator." 

***The serial numbers for the OTS feedthrus (D2500175) are of the alpha-numeric form 2153228V00n, where n = 1, 2, 3. That full version of the serial number is indicated in their ICS record, but in order to conform to the mold of the DCC S-numbers, I truncated the format to be numeric, S322800n, i.e. removing the identical leading 215 and misleading/unnecessary V character in the middle. 

Pictured here is 
    - The pre-assembly components of the fiber storage spool (First) 
    - The completed assembly of the spool with S3228003's patch cord wound up within it (Second and Third)

The feedthru's patch cord still has a Thor Lab Narrow-Key Mating sleeve (but NOT polarization maintaining) ADAFCB3 that is not intended to be a part of the final assembly, just there for fiber storage during shipment. I'd yet to detach it in these pictures.

Commentary:
     - Coiling the fiber within the spool was nerve racking. It feels like you're trying to coerce dry spaghetti into a curve without it snapping. If you let it go, it "sproings" into a wild relatively straight mess. In the end, holding it all mid-air with both hands, I used the weight of the mating sleeve to slowly pull the coil tighter as I rotated the coil nudging the rest of the coil into the newer smaller circle, until I met the radius of the storage spool. I had the goal of coiling it with one end "on top" and the other "on the bottom" of the stack, but I gave up on that. Once to the desired radius and no smaller, I used the securing cross, resting loosely across only 1/4 of the spool to hold the bulk of the coil of fiber in place while I tucked the rest of the length into the guiding channel. This is doable with a chair and patience in the open space of the optics lab, but I'm not looking forward to ding this in chamber.
     - Thinking through the install, my current plan is as follows :: WHAM3 D5 is currently a 12 inch blank with no 4.5 inch flange adapters. So it *needs* to be replaced by a 12 inch to 3x 4.5 inch flange adapter. So let's create the full 12 inch flange assembly with the 2x, MEAS and REF, fiber feedthroughs and 1x 4.5 blank -- and spool the fibers -- in the optics lab. Then we bring and install the whole 12 inch assembly on to HAM3 as a whole. 

(Randy Thompson, Corey Gray)

In preparation for the upcoming Bigger BeamSplitter Suspension (BBSS) Installation at BSC2 in a few weeks (~midMarch), a new type of cleanroom infrastructure was needed too allow for more vertical space for the cartridge to be craned in/out of BSC2.  Instead of using our standard large mobile cleanroom (which are used for HAM & BSC chamber work), a smaller clean space will be attached to the elevated work platform for BSC2.  There will be a "cleanroom tent" on this platform (here's the platform [white] with 2-sets of 4 "tent poles") and a new array of (qty-4) HEPA Filters Fans (same type as the ones on top of the large mobile cleanrooms which blow down) will be on the side of this cleanroom tent to push air flow horizontally for this elevated workspace for the BBSS install.

Randy designed a unistrut framework for this 4-HEPA fan Array (weighs 377lbs).  This morning this assembly was carefully craned into place on top of 3 unistrut brackets attached to the BSC work platform (in between BSC1 & BSC2).  For added support, this afternoon, Randy added a "leg support" on both ends of this HEPA Filter Fan Array (here is a photo of one of these legs that is in the Beer Garden).

1. EOM was put in chamber and moved to its final location. One temporary iris, a cable bracket for the picomotors and one dog clamp for the last lens for JAC needed to be relocated as they interfered with the EOM pivot base.
2. In-vac cables were identified and were connected to the EOM. Minor annoyance about the cable strain relief.
3. EOM S11 coefficients were measured and were acceptable though the dip frequencies were not the same as in the lab, so we haven't retuned it.
4. EOM was connected to the RF sources (now we're modulating!).
5. The light goes into and comes out of the EOM.
6. We forgot that the EOM wedge has a big effect to deflect the beam. We started working on the alignment downstream of the EOM and this will continue tomorrow.

Details to follow.

Nothing wrong with our previous reflection dip measurements with alumina.

We (MichaelL, StephenA, MattH, Gabriele, Elenna and myself) had a meeting in the morning. 

Looking at the "with the alumina" reflection screen shots, Michael didn't see any serious problem so we decided that the electrode-crystal-face plate capacitance is OK. We won't worry about that, we'll just make sure that there's no visible gap.

Third mounting method ("in-between") was proposed and tested.

Stephen proposed an in-between method where we use washers between the input side plate and the front plate (bottom of three_methods.png cartoon). After the washer contacts the face plate AND the input side plate, screws are gently tightened in a balanced manner like in Appert method.  (In a retrospect this is not that different from Laxen method except the way screws are tightened and that the input side plate contacts with the face plate at two points.) 

We first tried to use a presicion thickness shim washer for 1/4-20 screws but I didn't like that they're too big. We ended up using smaller stamped washers that is 0.039" or 0.99mm thick (according to the caliper). That's not flat but seems OK to me.

This in-between method worked in that it was doable and gave us a reasonable reflection dips.

Mechanical stability test of Appert method and in-between method. The latter is better, we'll use that for the real RTP.

My original concern for the original Appert method (middle of three_methods.png cartoon) was that somehow the screw gets loose during transport or after a large change in the in-chamber temperature and the tuning will be off. Upon hearing this, Stephen proposed a test to loosen one screw and see if the tuning changes. We performed this test for both Stephen method and the in-between method. (Spoiler: yes.) We measured all four dip frequencies right after the alumina piece was mounted but only tracked the frequency change of 118MHz peak.

Loosening one screw changes the frequency, but the frequency change for the in-between method is an order of magnitude smaller (10 to 30kHz) than the original Appert method (300kHz) when the FWHM (or rather the width between the points where reflection is 6dB larger than the bottom of the dip) is about 70kHz or so. This is just one trial but I'm convinced that in-between method (or maybe Laxen method too though we haven't tried) is better, so that's what we'll do for the real crystal.

We only loosened the screws on the output side for both mounting methods. 

Shake and it will shift, we need to measure it again in chamber?

It's good to know that the in-between method can somewhat withstand the loosening of the second screw (because the tighter screws still support the face plate). However, it's disappointing to find that the assembly is susceptible to shaking.

In the in-between method, we couldn't record the initial 118MHz dip because it jumped up by 10kHz or so in front of our eyes when we were moving around the table. Not sure what happened but I assumed that it was some kind of shaking. However, I gently tapped the front and side plates and couldn't cause another shift.

In the original Appert method, since we knew something could happen, I tapped the front and side plates and there was a huge jump. See the difference between initial_118.jpg and taptap_118.jpg.

Even though we'll use in-between mounting method, it's plausible that the frequency shifts during transport or when the EOM lands on the ISI surface. I'm thinking we'll have to measure it in situ after everything is tuned in the lab.

What's to come tomorrow.

We'll install RTP and tune. Before doing that, though, I'll discuss inserting indium foil between the crystal and the front plate with Masayuki. Michael suggested that (and even between the crystal and the electrode, though that would be tricky) today, Masayuki and I talked about the possibility briefly last week, it just sounds like a good thing as a buffer to absorb gaps here and there.

Other things.

In the previous alog (88886) in one of the pictures (gap.jpg), there was a time when it looked as if the circuit board was slightly bowed. We took a picture of the electrode today (electrode_contactpoint.jpg) and it looks as if the electrode is more abraded close to the outside edge of the crystal, so maybe the board bowing is real.

10-32x0.375" SHCS that was blocking the access to one 1/4-20 screw was replaced with a low profile 10-32 SHCS.

"Issue 2" in alog 88862 was solved.

See picture, Mitch found a 10-32x0.5" SHCS with a low profile head. 0.5" seemed to be OK in that it's not too long, but we used two washers to make sure that the scrwe doesn't bottom out.

We have intalled the POP periscope stiffener.

Some dog clamps in the REFL path as well as the cable bracket for PM1 were relocated to accomodate. 1st and 2nd picture are "before" photo. 3rd one is after PM1 cable bracket relocation but before installing the stiffener.  There are also three "after" photos showing how things look on the table.

POP beam clearance:

I took the picture of the bottom periscope mirror through dichroic (HR for IR, transmission for green) to see the beam clearance. In the first such photo (POP_beam_clearance.jpg), the camera is close to the center of the dichroic and the short stiffener beam is close to the edge of the optic but not occulting the mirror, so we're OK. Just to make sure that we're absolutely safe, I moved the camera closer to the -Y edge (right on the picture) of the dichroic (POP_beam_clearance_extreme.jpg) and it still looks OK.

If it's hard to understand what was done, look at the annotated photo (the last attachment After_stiffener_installed_annotated.jpg), the cellphone camera was inserted to "Camera" position.

REFL beam path:

I confirmed that the long stiffener beam doesn't interfere with the motion of the REFL beam diverter. Also, when the REFL beam diverter is open, I looked into the last steering mirror for the REFL air path from the viewport position to make sure that the short stiffener beam won't occult the REFL path. 

Some hiccups:

We used D2500433 -11 variant S/N 4 and -1 variant S/N1 even though page 11 of T2500339 suggests it should be -10 and -2 variant, respectively.　We didn't have -10 variant, and -2 variant was absolutely too short.

B&K

We performed B&K hammer measurements before/after the stiffener installation for POP periscope. Before, there was a 70Hz-ish peak. After, it was pushed higher up in frequency. The transducer was attached to the ISI table and Jim hammered the top of the periscope.

Likewise we did B&K test for the input periscope of the JAC even though it was not absolutely necessary.

We haven't done B&K for the JAC output periscope because it's not even fully clamped down (we will move it).

Jim will post the data.

Since nobody seems to have made an alog, here it is.

We've steered the JAC transmission beam into HAM2 and removed the viewport cover on HAM2 on the +Y side to look inside. At first we had a hard time seeing anything. We steered the beam in YAW and PIT and still nothing.

After a while we found that if we position the IR viewer at a specific position and look into the baffle hole of the MC refl periscope (circled in yellow in the 1st attachment), we can see some kind of ugly IR that definitely comes from JAC, but no beam seemed to be coming out of the baffle hole.

2nd attachment shows the picture shot by an IR sensitive camra when we focused on the IR, and the 3rd attachment shows the same picture shot from the same position but focused on the  baffle.

The distance from the sensor to the subject according to the lens' indicator was something like 4m for IR and 1.5-2m for the baffle. The indicator is only good for visible light and not for IR, but empirically the scale is not a factor of 2 off for IR, so we're looking at something that is far from the baffle (i.e. we're looking at the image of the source reflected by the periscope mirrors).

Another possibility that Masayuki points out is that it could be some IR beam (probably not the main beam) hitting the vertical metal pillar of the periscope behind the bottom periscope mirror and we're looking at that through the space between the baffle hole and the periscope mirror (see the 4th attachment). I think that unlikely because the pillar is merely inches away from the baffle and the distance indicator of the lens doesn't agree. But we'll see.

Tomorrow we intend to remove the septum window cover for IFO REFL and POP and look into HAM2 from there, that way hopefully it's easier to find where the JAC beam lands in HAM2.