On the layout for IOT2L, there's acritical alignment just before the REFL steering mirror, where one beam dump, IO_MCR_BD5, dumps a second reflection, and an optic, IO_MCR_M14, picks off the P-pol beam and sending it to a beam dump, IO_MCR_BD6.
The alignment of the IMC REFL beam between the beam dump and pick-off optic is critical to allowing only the IMC REFL beam through to the REFL PD, the IMC WFS, and the IMC REFL camera.
When the camera image shows more than one beam, the alignment of the IMC REFL beam through that critical aperture between the beam dump and pick-off mirror is so significantly misaligned, that one or both beams are getting through.
The beam that is picked off by the optic can have up to 1.4W at 50W into the IMC, so the power in that beam is significant. I haven't seen it reach the REFL sensor, but the potential for scattering onto the sensor does exist.
An alignment on the table that allows more than one beam through to the IMC REFL path has a potential to also allow the beams going to the high power dumps to have moved enough to be clipping on the dumps, and possibly fall off. Both high power beam dumps have black metal shields behind them, to prevent any stray beam from hitting the enclosure, but the metal sheets are not made to dump beams at 25W (2 Trap-It beam dumps rated to 50W each, and each gets 1/2 the power of IMC REFL).
Peter F and I went to IOT2L after a lockloss to look at the beams on the MC refl path.
We did not find the origin of the ghost beams on the camera; they're co-propagating with the main beam from at least as far back as MCR-BS1 (see diagram linked in Cheryl's alog). We put an iris just before L2, to block the beams visible on the camera and another, more spatially-separated, ghost beam, that was previously propagating through L2 and being blocked by the mount of L3.
We didn't see evidence of clipping, or stray beams getting through the aperture between BD5 and M14. The p-polarised beam going into BD6 was possibly hitting the dump a bit low, but it was hard to get a clear view of the front-face of the beam dump.
There was a ghost beam hitting the edge of M3, and generating a spray of light, however someone must have already known about this, as there is a beam razor blade beam dump on the air-side of the table for this "beam".
It's possible that the beam is clipping in vacuum, or that a REFL Periscope mirror shifted (though this was considered in the meeting today, and doesn't seem to fit), or that the IMC REFL beam is very misaligned through the IOT2L REFL beam path. More than one beam on the GigE camera has in the past been associated with a IOT2L table beam misalignment. Finding a beam that's scattering onto the table from IO_MCR_M3 is a clue that the table is misaligned, starting from the top periscope mirror, because the design of the optical layout and beam path does not place any beams near or over the edge of that optic. That condition increases my concern about the alignment into the High Power Beam Dumps, since IO_MCR_M3 is the optic that launches almost the full power IMC REFL beam to the two 50W Trap-It beam dumps. Unless the location of the beam on IO_MCR_M3 was intentional, the 50W beam dumps absolutely need to be checked. On the REFL camera, there is significant extra light, and they extra light appears to be extra beams, and there should only be one beam coming through IO_MCR_BS1, so finding multiple beams coming through IO_MCR_BS1 is an indication that there are beams (whatever the source) getting past the beam dumps and pick-off mirror in front of that optic.
From the aolg I see that I realigned the REFL path in January (alog 40313), and it was checked by Sheila and Gabriele in June (alog 42362), though searching for these events is not a guarentee of finding all of them.
I looked at the alignment changes for the IMC from June until today, and found changes in the tens of urad in pitch for MC1, MC2, and MC3, and in yaw for MC3.
Changes: June 6th to November 9th: