Fixes/lamni during beamtime 2 #347

Closed
holler wants to merge 0 commits from fixes/lamni_during_beamtime_2 into main
pull from: fixes/lamni_during_beamtime_2
Member
No description provided.
holler added 26 commits 2026-11-03 08:49:16 +01:00
feat(lamni): print the base shift in tomo_scan_projection
CI for csaxs_bec / test (push) Successful in 3m5s
9740c32f6e
Next to the X-ray eye correction and the constant shift, also print
the base shift (tomo_fovx/y_offset, from center finding / first X-ray
eye click) at 0 deg and rotated to the projection angle, as applied by
lamni_fermat_scan.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LJWYXx4SS8NgFWog81xjyf
These manuals are for all users.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ALDnH16dC6iyHjhe5Ey4w9
fix(lamni): check filters after leye_out in tomo_alignment_scan
CI for csaxs_bec / test (push) Successful in 3m7s
a30bc8e5c5
leye_out() retracts the filters via csaxs.fil_trans(1), so checking
before it warned about filters that were about to be removed anyway.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ALDnH16dC6iyHjhe5Ey4w9
Log analysis of the flOMNI beamtime 2026-10-07..12: time budget, per-point
dead time, step-by-step scan-to-scan timeline mapped to code, analysis
method/pitfalls, and ranked improvement ideas. Add TODO pointer.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The file written before the alignment scan predicted first_scan + k per
angle, which is wrong with stitching ((2*sx+1)*(2*sy+1) scans per angle).
Predict with that stride, and rewrite the file after every angle so an
interrupted run still leaves correct numbers.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ALDnH16dC6iyHjhe5Ey4w9
fix(webpage): downscale large full-size ptycho PNGs before upload
CI for csaxs_bec / test (push) Failing after 2m19s
e5c5d64307
omny.psi.ch/upload.php rejected the full-size probe/spectrum PNGs of
stitched reconstructions with HTTP 413. Full-size copies above 900 kB are
now saved as downscaled PNGs until they fit.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ALDnH16dC6iyHjhe5Ey4w9
The alignment scan now measures 13 angles, 0 to 360 inclusive; 360 repeats
0 as a check that nothing drifted during the run. The angle list lives in
alignment_scan_angles(), shared with write_alignment_scan_numbers().
lsamrot soft limits set to -2/362 (were 0/0, i.e. none).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ALDnH16dC6iyHjhe5Ey4w9
fix(lamni): scan-number file lines 3/4 are xeye x and y in um
CI for csaxs_bec / test (push) Successful in 3m2s
f680f9dd18
BEC_ptycho_align reads line 3 as scan_xpos and line 4 as scan_ypos (um).
lamni wrote the x correction twice, in mm (line 3 rounded to 10 um).
Write x and y X-ray-eye corrections in um instead.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ALDnH16dC6iyHjhe5Ey4w9
laser_tracker_check_signalstrength() slept 1 s on a too-low signal and
then compared the same stale value again. Re-read it.

Add docs/plans/flomni-batched-single-point-tomo.md: one armed scan per
batch of angles for single-point tomography (not implemented).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ALDnH16dC6iyHjhe5Ey4w9
The RT controller triggers only once the position is truly reached and
works independently of beamline trigger hardware; BEC software triggering
stays as fallback.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ALDnH16dC6iyHjhe5Ey4w9
feat(lamni): golden ratio bunch size range 30..200, default 100
CI for csaxs_bec / test (push) Successful in 3m1s
45609b6ef0
Setter, tomo_parameters() prompt and the tomo params GUI spinbox (lamni
profile) now allow 30 to 200 (was 20 to 100); default 100 (was 20).
flomni's GUI range is unchanged.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ALDnH16dC6iyHjhe5Ey4w9
reset_correction() now also clears the iteration-2 lookup table (as on
FlOMNI), so it is reset at a new X-ray eye alignment and when a new
rotation center is applied. _existing_alignment() reports it too.

tomo_params: spin and combo boxes in the scrollable panel ignore the
mouse wheel and pass it to the scroll area, so scrolling cannot change
values unnoticed.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0129zz6QGiURiV2agXouBv4T
The idle detection treated any gap between projections longer than
max(60 s, 5 x per-point exposure) as idle. A ptycho projection takes
minutes, so most of every projection was subtracted from the elapsed
time: a LamNI scan accumulated 6.8 h of "idle" in 9 h and showed an ETA
of 2 h 43 min instead of ~10.5 h.

New shared OMNY_shared/idle_time.account_projection_gap() learns the
typical gap from the scan itself (median of the last 10 normal gaps) and
only counts gaps beyond max(3 x typical, typical + 300 s) as idle.
recent_projection_gaps is reset with the other timing fields on every
fresh tomo scan (LamNI _reset_progress_for_new_scan, FlOMNI progress
defaults, queue reacquire) and kept on resume.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0129zz6QGiURiV2agXouBv4T
feat(lamni,flomni): optics-Z shift for queued loptz/foptz moves
CI for csaxs_bec / test (push) Successful in 4m21s
411a9ff1aa
- New loptz_shift_x/y (lamni, mm) and foptz_shift_x/y (flomni, um):
  non-rotating shifts added next to manual_shift_x/y, but not snapshotted
  by tomo queue jobs, so a queued optics-Z move's shift survives the
  following scan jobs. Zeroed at x-ray eye alignment (flomni also on
  sample change); the optics-Z position is stored as
  <optz>_shift_reference for a future calibration.
- lamni: new tomo-queue action move_loptz_with_shift (limits checked,
  refused on OSA collision at current losaz); flomni's
  move_foptz_with_shift now sets foptz_shift_x/y.
- lamni RT controller no longer locks loptz read-only with feedback on.
- Tomo-queue GUI: live OSA-collision warning for the loptz target.
- Command-job summaries (tomo_queue_show, GUI table/tooltip/builder)
  now show units via format_command_step(); shift prints carry units.
- Docs + omny TODO note.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017NrH3poJsYBCTVq2Y9fuHZ
feat(lamni): number_of_projections argument for tomo_alignment_scan
CI for csaxs_bec / test (push) Successful in 4m16s
0830a71759
Optional, default 13 (every 30 deg, 0 and 360 inclusive). The angles
actually used are written to ptychotomoalign_scannum.txt. Values that
are not an integer >= 2 raise ValueError before anything moves.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- csaxs.slits_calibration_report(): with all slits centered at a known
  opening, computes the offsets that make centers read 0 and sizes read
  target_width (EPICS user_offset for sl1-3, virtual-slit config offset
  for sl4-6); optional full dump of all machine/frontend/optics/endstation
  devices (readings + config signals) and save to text file
- csaxs.slits_apply_epics_offsets(): writes userParameter 'epics_offset'
  to the EPICS user_offset where configured and different; called at
  cSAXS client startup since the IOC does not keep the offsets reliably
- slits_show_all() labels values with the motor names

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
config(csaxs): calibrated slit offsets for sl1-sl6
CI for csaxs_bec / test (push) Successful in 4m20s
1ff8496dab
EPICS slits sl1-3: userParameter epics_offset; SmarAct slits sl4-6:
virtual slit deviceConfig offset. From slits_calibration_report with all
slits centered and opened to 1 mm (2026-10-21).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Rename xrayeye_rotation_center_calibration_smear_experimental() to
xrayeye_rotation_center_calibration_smear() and
find_rotation_center_smear_experimental() to find_rotation_center_smear(),
and drop the EXPERIMENTAL markers from code, tests and docs. The feature
has been tested and is in regular use.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017NrH3poJsYBCTVq2Y9fuHZ
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017NrH3poJsYBCTVq2Y9fuHZ
Stitch parameter n used to give 2n+1 tiles per direction (offsets -n..+n).
It now gives n+1 tiles placed symmetrically around the scan center
(offsets i - n/2), so stitch 1 is two tiles at +-half a step and stitch 2
is the old three-tile layout. Stored values change meaning.

- shared stitch_tile_offsets() helper in OMNY_shared
- LamNI alignment scan-number file steps by (sx+1)*(sy+1)
- tomo_parameters prompts/printout show the tile count
- tomo-params GUI lamni point estimate uses the outermost tile (n/2)
- user docs updated; omny port noted as TODO in omny/AI_docs

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017NrH3poJsYBCTVq2Y9fuHZ
With a circular FOV and stitching, corner tiles can hold far fewer Fermat
points than the center tile. A tile below the scan minimum aborts the
projection, and the retry then stops the tomogram. The point-count check
now uses the smallest tile (new lamni_smallest_tile_position_count()):
the GUI refuses submit / add-to-queue, and tomo_parameters on the command
line skips the confirm question and repeats the scan-geometry prompts
until the count is sufficient.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017NrH3poJsYBCTVq2Y9fuHZ
Clicking into a spin box left the cursor at the end, so typed digits were
appended and silently rejected when above the maximum (e.g. golden-ratio
bunch size 100 + '5' -> 1005 > 200) or the decimal count, leaving only the
arrows to change the value. Applies to all spin boxes in the parameter panel.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017NrH3poJsYBCTVq2Y9fuHZ
config(csaxs): adjust sl2yc/sl2ys epics offsets
CI for csaxs_bec / test (push) Successful in 4m28s
c4a23cc5b6
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017NrH3poJsYBCTVq2Y9fuHZ
The XBox2 BPM amplifier outputs a negative current, so more beam showed as
more negative blade values. invert: true in the device config negates the
input signals; the BPM position/diagonal ratios are unchanged and the
intensity becomes positive. Enabled for bpm_xbox2_slowrb.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017NrH3poJsYBCTVq2Y9fuHZ
Galil readback and gain devices are now named after the beamline BPM
number, matching the stages (xbpm2x/y, xbpm3x/y) and xbpm1; the hutch and
XBox go into the description:
- bpm_xbox2_slowrb -> xbpm3_slowrb, gain_bpm_xbox2 -> xbpm3_gain
- bpm_xbox1_slowrb -> xbpm2_slowrb, gain_bpm_xbox1 -> xbpm2_gain (commented out)
- bim_xbox3_slowrb -> xbim_slowrb, gain_bim_xbox3 -> xbim_gain
- csaxs.diagnostics.bpm_xbox1/.bpm_xbox2/.bim -> .xbpm2/.xbpm3/.xbim
- bpm1 removed (duplicate of xbpm1); xbpm1c1-c4 kept

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017NrH3poJsYBCTVq2Y9fuHZ
feat(csaxs): csaxs.diagnostics.show_plots() opens beam diagnostics waveform window
CI for csaxs_bec / test (push) Successful in 4m17s
CI for csaxs_bec / test (pull_request) Successful in 4m15s
602eac82ba
Opens a BEC GUI window with waveform plots of the XBPM3 blade signals and
the XBIM signal by default. Further plots by name (xbpm3_position, xbpm2,
beamstop, polarization), or only the named ones with defaults=False.
Devices not in the config are skipped; calling it again rebuilds the plots
in the open window.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017NrH3poJsYBCTVq2Y9fuHZ
holler closed this pull request 2026-11-03 09:24:25 +01:00

Pull request closed

This pull request cannot be reopened because the branch was deleted.
Sign in to join this conversation.
No Reviewers
No labels
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: bec/csaxs_bec#347