With Enzo, Hirata-san, Kozu-kun and Terrence.

Today:

• Yesterday's F0 data showed considerable hysteresis, so we looked for sources of unrepeatability and found three suspects:
  • The screws fixing the blade clamps to the filter top plate were not being tightened to any particular standard torque, so we decided to use a torque wrench and settle on a torque value. At first Hirata-san tried 25.5 N.m, which was recommended for general-use M10 screws, but it was obviously inadequate and did not close the gap between the clamp block and the top plate. He settled on 50 N.m, which was (i) the limit of the torque wrench, (ii) recommended for construction-grade screws, and (iii) about what he had been applying manually.
  • The screws holding the tips of the blades to the keystone had been tight originally but had worked loose, and one could be rotated by hand. We tightened them.
  • The collar on the tip of the shaft of the magic wand which held the flexure from the keystone was not properly clamped to the tip of the blade and wobbled. We tightened them.
• After implementing all these "refinements", Hirata-san and Kozu-kun recharacterized the filter for compressions of 20, 22 and 25 mm, but unfortunately the load capacity and frequency were worse than before! See forthcoming klog by Kozu-kun.
• We immobilized the BF with the EQ stops on the L-brackets, and measured its bounce frequency and Q: 0.417 Hz, 22.28.
• We locked the BF keystone and unlocked the body, and measured the SF bounce frequency and Q: 0.34 Hz, 27.03.
• These frequencies and Qs are quite good, and suggest there is no problem with rubbing or overly tight cables. 
• Tomura-san and Shimode-san reported that the HPCD we thought we had damaged yesterday (S1807813) was not in fact permanently damaged, but had just had a watchdog trip, and could have been reset by a power cycle or via the BIO. We tried applying steadily larger offsets to the replacement one (S1807810) and found that it tripped somewhere between 25000 and 28000 counts and did in fact recover with a power cycle.
• Unfortunately, a bit later, while applying a 20000 ct offset as part of the bounce frequency test, the actuation failed again and did not recover after cycling the power. So we swapped in a third HPCD.

We are running out of ideas for what to do about the top filter and suggestions are welcome. One possibility we are considering is to put shims under the inside or outside edges of the blade clamps, so as to change the launch angle. Hirata-san recalls that in early testing there was sometimes a very small gap at the outside edges of the blade clamps, and when we got more careful it went away but the performance got worse. So perhaps by deliberately simulating the gap in a consistent way we can claw back some load capacity.