Test on SANS instrument

[Tom-Willemsen]

LOQ could be a useful candidate as a SANS beamline which also does continuous-motion scanning (for the laser) and also isn't usually 100% scheduled as far as I know so may be easier to get beamtime on.

Acceptance criteria

• Make sure we have tickets for any requirements needed by that test, which we haven't yet covered
• Make sure the above tickets get proposed in good time before the relevant test
• Arrange the testing time
• Go to the beamline and do the test
• Write up any relevant issues/feedback in further tickets

[Tom-Willemsen]

Split this ticket into 2, keeping this one as sans test and muon test in #85

[Tom-Willemsen]

According to RD:

• Should be able to schedule time on LOQ to test with neutrons during cycle 2024/5, but no definitive date yet
• Want to test sample changer scans, which will give a top-hat or trapezoid style fit

[rerpha]

script used on LOQ:
bluesky_test.zip

screenshots of scans:
\\isis\shares\ISIS_Experiment_Controls\data for tickets\ibex_bluesky_core_80\screenshots

[Tom-Willemsen]

Test on LOQ with L. Ca. (SANS group):

• Generally went very smoothly, test was mostly run by @rerpha and @jackbdoughty with minimal input from me. Feedback from scientists included:
  • "very happy" (in relation to making continuous scans get decent statistics much faster)
  • "that's very useful for many different types of scan" (in relation to adaptive scans)
• Tested:
  • Continuous-motion laser scan - worked well, comparing to a 6-7 minute continuous motion scan previously, we can get equivalent statistics in about 30-40 seconds using bluesky (mostly due to more efficient CA & matplotlib usage, meaning we sample more points from the galil). Tested this on sample changer and aperture axes, worked well in both cases using a trapezoid fit.
  • Step scans with neutrons - worked fine
  • Adaptive scans with neutrons - worked fine
• Issues spotted:
  • Bluesky CoM callback didn't work well with one of the continuous scans - may have either been getting confused with non-constant point spacing, or with negative intensities or similar. Need to debug this offline and either fix upstream or write our own (prefer to fix upstream if possible)
  • Logging re-appeared again, different cause to last time, but we might need logger.propagate = False which is bad practice generally but 🤷 might be the best we can do...
  • DAE on LOQ seemed much slower than CRISP to pause/resume/change periods, not sure if this is just a DAE2 thing. Means that neutron step scans are quite painful even counting for relatively few frames.

Will show to a couple of other members of SANS group tomorrow - but general feedback from this test is that this is covering SANS use-cases well so far (excluding the spin-echo/polarization use-cases, mostly on larmor)