[INSTRM-1708] Update the PfiTransform code to work with the correct fiducial positions Created: 17/Sep/22  Updated: 01/Dec/22  Resolved: 01/Dec/22

Status: Done
Project: Instrument control development
Component/s: None
Affects Version/s: None
Fix Version/s: None

Type: Story Priority: Normal
Reporter: rhl Assignee: rhl
Resolution: Done Votes: 0
Labels: EngRun
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Blocks
is blocked by INSTRM-1746 Update write of mcs_pfi_transform table Done
Story Points: 2
Sprint: preEngRun08Nov

 Description   

The PfiTransform code is used to map the measured fiducial positions onto the known positions, thereby putting the cobra positions onto that coordinate system.

With the corrected fiducial positions in use for the September 2022 run, the initial guess for this transform will be incorrect, so please change it.  The new values should be close to 0.0, but we should check that this is true.

We probably don't really need to do this, as the code should converge even starting at the old positions with suitable clipping parameters, but it'll be more robust if we make this change.

It is probably worth noting that the functional form of the best fit model could change once we've taken out this gross misalignment;  we should monitor this during the next few engineering runs.



 Comments   
Comment by hassan [ 02/Oct/22 ]

Following discussions with Yuki Moritani, agreed that no work was needed for this ticket for the Sep 2022 Eng Run

Comment by rhl [ 13/Oct/22 ]

Not for September, but it still needs fixing.  We would be able to do this now, except that the DB table that records the values being used isn't being correctly loaded.  Once that is done, we can confirm that the true value of the rotation is indeed nearly 0 and close this ticket.

Comment by hassan [ 13/Oct/22 ]

rhl Kiyoto Yabe Is the problematic opDB table the fiducial_fiber_geometry one? I'll file a separate blocking ticket for that.

Comment by rhl [ 13/Oct/22 ]

There's a discussion on slack (#opdb-targetdb-queuedb, 2022-10-05), so I assumed that ASIAA had converted it to a ticket once they'd confirmed that I wasn't crazy.  But if not, yes, please file a ticket.

Comment by yuki.moritani [ 01/Dec/22 ]

I quickly checked the coefficients (x0, y0, dscale, scale2, theta) in November run.

  • Average of (x0,y0) is ~0, and make circle with radius of ~1mm on xy plane. 
  • dscale ~ -0.32
  • scacle2 ~ 0.006
  • abs(theta) < 1e-8

 

Generated at Sat Feb 10 16:38:34 JST 2024 using Jira 8.3.4#803005-sha1:1f96e09b3c60279a408a2ae47be3c745f571388b.