[PIPE2D-377] Measure PFS cross-talk coefficients Created: 26/Feb/19  Updated: 05/Jan/21  Resolved: 27/Jun/20

Status: Done
Project: DRP 2-D Pipeline
Component/s: None
Affects Version/s: None
Fix Version/s: None

Type: Bug Priority: Normal
Reporter: rhl Assignee: price
Resolution: Done Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Relates
relates to PIPE2D-376 Enable crosstalk corrections for PFS ... Open
Story Points: 3
Sprint: 2DDRP-2021 A
Reviewers: hassan

 Description   

Measure the cross-talk coefficients using very saturated arc lines.  We can also do this using enough darks (as was done for HSC using Yagi-san's cosmic ray-based algorithm), but I suspect that the saturation trails are good enough.



 Comments   
Comment by hassan [ 27/Feb/19 ]

Neven: To accurately measure the coefficients we need to request more focused data from LAM with sufficient exposures to reveal the saturation trails. Neven will look into this.

Comment by cloomis [ 27/Feb/19 ]

Just an FYI: when I looked at this a long time ago, the crosstalk was bi-polar. The per-amp signals from the preamp to the ADC are differential twisted pairs, so that crosstalk can go either way. Should be stable for assembled cryostats....

Comment by ncaplar [ 01/Mar/19 ]

The data that are available for this ticket:

12922 - 12960 the set of Neon data:
unsaturated data, exptime=15 sec with attenuator at 240
saturated data, exptime=15 sec with no attenuator
saturated data, exptime=100 sec with no attenuator
saturated data, exptime=300 sec with no attenuator

Robert recommended to work with saturated, ``bleeding'' data to easily identify which fibers crosstalk to where and to get good signal. Jim notes that this is only valuable if you know what is the flux in overexposed lines. In order to do that one would need to estimate flux from unsaturated exposures and then multiply by ratio of exposure time and take into account suppression by attenuator. Attenuator values have been deduced in PIPE2D-323 (see the coefficients in the Figures there).

Comment by price [ 17/Jun/20 ]

Using the LSST tool, measureCrosstalk.py on visit=12922..12951 (the unsaturated data, which gives the best S/N), I get 2-3 sigma detections of the some of the crosstalk coefficients.

measureCrosstalk INFO: Coefficients:
[[ 0.00000000e+00  1.44343314e-04 -1.94648110e-05 -3.16576757e-06  1.42585166e-04  8.77068742e-05  2.74841033e-04  1.67585129e-03]
 [ 1.25909573e-04  0.00000000e+00  1.67928265e-05  3.45523513e-05  1.35091439e-04  9.84909784e-05  1.21150915e-04  1.19224918e-04]
 [-3.17470112e-05 -2.45970878e-05  0.00000000e+00  3.19747371e-04  8.55176186e-05  7.86037490e-05  5.40274705e-05  7.14990747e-05]
 [ 2.50621597e-05 -6.37933866e-08  2.38304929e-04  0.00000000e+00  1.27087420e-04  2.85813672e-04  5.63632930e-05  9.70012043e-05]
 [ 1.15817631e-04  1.68083658e-04  1.64138619e-04  2.31510116e-04  0.00000000e+00  8.75362020e-05  4.00934223e-05 -1.93203887e-05]
 [ 1.18028176e-04  1.32240646e-04  1.25063758e-04  1.97395566e-04  1.74431858e-04  0.00000000e+00  1.02643426e-04 -1.42276876e-05]
 [ 1.72953616e-04  2.18197252e-04  3.94008151e-04  1.52665016e-04  1.58990340e-04  1.40468474e-04  0.00000000e+00  1.64068726e-04]
 [ 1.21380063e-03  1.64559446e-04  6.32526717e-05  8.29588462e-05 -3.57388853e-05 -2.21135579e-05  1.80654883e-04  0.00000000e+00]]
measureCrosstalk INFO: Errors:
[[           nan 9.91449488e-05 9.59764293e-05 9.13887925e-05  8.65858747e-05 9.88380634e-05 1.06215986e-04 5.50728524e-04]
 [9.51092006e-05            nan 9.34346172e-05 8.84959009e-05  8.87283604e-05 9.10020608e-05 8.01342321e-05 8.08069526e-05]
 [1.01827194e-04 1.00470643e-04            nan 1.04412786e-04  1.12072987e-04 1.05188803e-04 8.17734108e-05 8.59935826e-05]
 [9.88750107e-05 9.80467812e-05 9.92097339e-05            nan  1.03731429e-04 5.95550635e-04 7.89797268e-05 8.79871077e-05]
 [7.20556418e-05 8.66447008e-05 1.05128216e-04 9.71880945e-05             nan 9.30918904e-05 8.16232277e-05 7.48949533e-05]
 [7.40533360e-05 8.87360584e-05 8.47910414e-05 2.74526799e-04  9.22506442e-05            nan 8.18406988e-05 8.51190489e-05]
 [9.79577817e-05 1.00221820e-04 9.54559218e-05 9.06347123e-05  1.00276600e-04 1.07225656e-04            nan 1.23933671e-04]
 [1.95971646e-04 9.59757526e-05 8.57987689e-05 9.65587824e-05  1.04003899e-04 1.05910200e-04 1.08329492e-04            nan]]
Comment by price [ 25/Jun/20 ]

I ended up using quartz exposures, as that provides a lot of target pixels (the full height of the image, rather than just a few discrete points). I'm not perfectly happy with the result (there are some small residuals remaining), but it does a decent job and this provides a tool to make solid measurements in the future. Some thought needs to go into selecting a particular set of fibers that minimises the overlap between traces and crosstalk for the various amplifiers.

This is based on PIPE2D-472, so that needs to be reviewed first.

Comment by price [ 27/Jun/20 ]

Merged to master.

Generated at Sat Feb 10 15:52:28 JST 2024 using Jira 8.3.4#803005-sha1:1f96e09b3c60279a408a2ae47be3c745f571388b.