Pedestal measurement protocol: how much does g1, g2 duty cycle matter?

[graeme-winter]

@ndevenish asks

"How badly does not having dead-time in the dark runs sabotage the results - it looked like we had anecdotal evidence for this but we should quantify"

[graeme-winter]

Can use e.g.

ws448 hewl5a :) $ ls -l 
total 6147073
-rw-rw-rw-+ 1 xfk23773    cm33852_3       1701 Jun 26 15:24 hewl_chipmap.txt
-rw-r--r--+ 1 i24detector cm33852_3 1048645856 Jun 26 15:22 hewl_dark_G0_0_0.h5
-rw-r--r--+ 1 i24detector cm33852_3 1048645856 Jun 26 15:22 hewl_dark_G0_1_0.h5
-rw-r--r--+ 1 i24detector cm33852_3 1048645856 Jun 26 15:22 hewl_dark_G1_0_0.h5
-rw-r--r--+ 1 i24detector cm33852_3 1048645856 Jun 26 15:22 hewl_dark_G1_1_0.h5
-rw-r--r--+ 1 i24detector cm33852_3 1048645856 Jun 26 15:22 hewl_dark_G2_0_0.h5
-rw-r--r--+ 1 i24detector cm33852_3 1048645856 Jun 26 15:22 hewl_dark_G2_1_0.h5
-rw-rw----+ 1 xfk23773    cm33852_3        343 Jun 26 15:24 hewl_parameters.txt
ws448 hewl5a :) $ pwd
/dls/i24/data/2023/cm33852-3/jungfrau_ssx/260623/hewl5a

compared with

2023-06-26-14-19-33_darks

which are presumably similar exposure times?

[graeme-winter]

ws448 hewl5a :) $ h5ls -rvd hewl_dark_G2_0_0.h5/exptime
Opened "hewl_dark_G2_0_0.h5" with sec2 driver.
exptime                  Dataset {SCALAR}
    Location:  1:1400
    Links:     1
    Storage:   8 logical bytes, 8 allocated bytes, 100.00% utilization
    Type:      native double
    Data:
        (0) 0.001

vs.

ws448 2023-06-26-14-19-33_darks :) $ h5ls -rdv G2_0_0.h5/exptime
Opened "G2_0_0.h5" with sec2 driver.
exptime                  Dataset {SCALAR}
    Location:  1:1400
    Links:     1
    Storage:   8 logical bytes, 8 allocated bytes, 100.00% utilization
    Type:      native double
    Data:
        (0) 0.001

so, comparable

[graeme-winter]

It matters rather a lot

('G0', 0) 0.7475251846313475
('G1', 0) 564.6115133285523
('G2', 0) 183.60439851951602
('G0', 1) 0.2640929626847251
('G1', 1) 648.1113066253662
('G2', 1) 166.43095762825016

[graeme-winter]

Now I need to work out a way of visualising this in a useful way, but the gain maps are certainly [negatively] impacted by the high duty cycle effect

[graeme-winter]

Looking for a "null" reading by comparing side-by-side "proper" pedestal runs... there is stull some variation

ws448 jungfrau_commissioning :( $ dials.python $OLDPWD/issue11_compare_pedestals.py 2023-06-26-15-38-36_darks 2023-06-26-15-37-42_darks
('G0', 0) 243.53963269405293
('G1', 0) 77.80792124176025
('G2', 0) -3.30109093093872
('G0', 1) 239.7622150937608
('G1', 1) 46.37543894386293
('G2', 1) -10.63802938079834

Oh, hang on. I need to check that these have the same / right masks 🤔

[graeme-winter]

Anyway, the differences show some clear structure (this is the 100/10% duty difference)