2x2 binned flats overcorrecting ? Oh, and a first light !

[glowingturnip]

Hi all,

Long time no post, it's been a while since I've managed to do any imaging, but I have been on a major splurge this year - proud new owner of a Moravian G2-8300 mono CCD and a Skywatcher 200PDS all  sitting on an AZ-EQ6GT - happy bunny 

Anyway, I've noticed that the flats I use for the 2x2 binned R, G and B frames seem to over-correct, whilst the 1x1 Ha and L frames don't get over-corrected.  See below:

 

Bottom-left is the red channel, 2x2 180s before flats correction, and bottom right is after flats correction, over-correcting (strong stretch on both to show it).  Top are the 1x1 Ha and L after flats correction, which look good to me. I get similar on the B and G, and on another target with similar set of data too.

All flats taken at the same binning as the lights they're applied to, at 30k average ADU, 20s delay between each flat.  Flats were calibrated with flat darks taken at same exposure time, binning and delay as the flats they're applied to.  Software used is SGP for capture and Pixinsight for processing.

Any idea what's going on ?

(I probably won't use 2x2 binning for my RGB going forward actually, looks a little under-sampled to me)

 

Oh, and  I guess this is my first light for my new set-up, can you guess what it is yet ?

 

Cheers, 

[glowingturnip]

just ran a bunch of experiments with the BPP script - with/without bias, with/without flatdark, with/without dark-optimisation, bias+longerdark+optimisation, but all give the same result, over-correction.

Going to experiment with the manual calibration tools next

[Rudeviewer]

Try your average ADU for 23k to 25k and see if that doesn't take care of it.  30k for the 8300 sensors seems a bit high.

[swag72]

11 minutes ago, Rudeviewer said:

Try your average ADU for 23k to 25k and see if that doesn't take care of it.  30k for the 8300 sensors seems a bit high.

According to the QSI website information on recommended ADU for flats for the QSI683 (KAF8300 sensor) it should be 35,700K for 1x bin and 35,000k for any other binning.... I'd have no reason to expect another companies KAF8300 sensor to be VERY different.... unless I'm missing something?

[Rudeviewer]

Just now, swag72 said:

According to the QSI website information on recommended ADU for flats for the QSI683 (KAF8300 sensor) it should be 35,700K for 1x bin and 35,000k for any other binning.... I'd have no reason to expect another companies KAF8300 sensor to be VERY different.... unless I'm missing something?

I'm sure you are correct if that is what they state for that sensor.  I may be incorrect in thinking that the 8300 sensor is the same as the one in my 383l+ which my flats need to be 23-25k in order to work correctly for me.  Atik has provided me those numbers so it seems there may be different answers out there from different companies.  

[glowingturnip]

SOLVED

solved it - after lots of experimentation, I've  found it's actually the BPP script that seems to be over-correcting.  The master dark, bias and flat files it produces all actually seem to be ok, but when you put all the masters together with the lights and run BPP it's then over-correcting.

However, if I run them through the ImageCalibration script, then it works nicely, here's one of my calibrated reds now (super-stretched):

 

For anyone with the same problems as me, once you have your master dark, superbias and master flat, put them all in the ImageCalibration tool, bias should be calibration disabled, dark should be optimise enabled and calibration enabled,and flats calibration disabled.

 

Cheers,

Stuart

 

[Droogie 2001]

On ‎24‎/‎08‎/‎2017 at 17:05, swag72 said:

According to the QSI website information on recommended ADU for flats for the QSI683 (KAF8300 sensor) it should be 35,700K for 1x bin and 35,000k for any other binning.... I'd have no reason to expect another companies KAF8300 sensor to be VERY different.... unless I'm missing something?

Very interesting. I have a G2-8300 Moravian for the KAF8300 chip and have been using flats of 25k ADU. This has given me good results however after seeing the QSI link I am now wondering if increasing this could be beneficial?
The Moravian site does not seem to have suggested Flat ADU's (at least I have not seen any) but I can confirm the Full Well depth is 25,000 on their G2-8300 CCD, which correlates with the QSI page of a Flat ADU of 35700 Sara mentioned.