QHY 268-C sensor analysis delivers false SystemGain (-e/ADU) and Fullwell Cap.

Andreas
Posts: 12
Joined: Thu Jun 10, 2021 9:04 am

Re: QHY 268-C sensor analysis delivers false SystemGain (-e/ADU) and Fullwell Cap.

#11

Post by Andreas »

Hi,

I´m not really sure if I´ve understood your advice regarding the colour balance right. I tested it with the figures you find attached. With these levels the red, green and blue curve are at the same level.

The result was a little higher system gain, but a lower fulwell - but still far away from the QHY datas on their homepage.

If I´ve done it wrong please let me know in simple english words ;) I´m only a german...

Thanks
Andreas
Attachments
QHY 268-C - Highgain Mode- SharpCap 4.0 beta - Camera Control panel 2.JPG
QHY 268-C - Highgain Mode- SharpCap 4.0 beta - Camera Control panel 2.JPG (133.67 KiB) Viewed 2033 times
User avatar
admin
Site Admin
Posts: 13177
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: QHY 268-C sensor analysis delivers false SystemGain (-e/ADU) and Fullwell Cap.

#12

Post by admin »

Hi,

no, that's not quite what I meant.
Capture.JPG
Capture.JPG (114.41 KiB) Viewed 2026 times
Here you can see what I was trying to describe. I have deliberately set up the camera to overexpose the image (big areas of max brightness).

In that situation, the histogram normally goes all the way to the right, but because I have turned down the WB(R) to 14, the red channel does not reach the right hand side. WB(G) is set to 16 - this is the smallest value that gets the histogram to the right hand side for this camera and is therefore the correct 'neutral' white balance value. WB(B) is set to 18 - you can see that the increase from 16 to 18 doesn't really do much to the histogram.

This happens because the white balance adjustments are applied after the data comes off the camera. If you set the WB(R) to 14 then all red pixel values are multiplied by 14/16 (0.875). If you set it to 18 then all red pixel values are multiplied by 18/16 (1.125). When the multiplication factor is less than 1, the histogram limit doesn't reach the right hand side of the graph, so you can find the 'neutral' WB levels by finding the smallest values that lets an overexposed histogram reach the right hand side of the graph.

cheers,

Robin
Andreas
Posts: 12
Joined: Thu Jun 10, 2021 9:04 am

Re: QHY 268-C sensor analysis delivers false SystemGain (-e/ADU) and Fullwell Cap.

#13

Post by Andreas »

Thanks Robin, just one more question before starting the test at the weekend again.

When I have detected these special WB values for R/G/B in overexposure, I have to write them down. That´s important, because after pressing START for the sensor analysis, all values are set back to a standard and equal level (15 for my 268-C) and I have to readjust them at this time. All other levels, like gain, offset or exposure time are also set back to another level, because the system wants to reach a histogramm level of 65% with exposure times between 0,1 and 1 sec (remark: with my ASI 120 the max exp.time can be choosen a double to 2 sec). Now I press proceed and let the analysis work.

Is there any other advice from you?

Thanks
Andreas
User avatar
admin
Site Admin
Posts: 13177
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: QHY 268-C sensor analysis delivers false SystemGain (-e/ADU) and Fullwell Cap.

#14

Post by admin »

Hi Andreas,

you should be safe to leave the other controls (gain, exposure, offset, etc) alone as the analysis will handle those correctly. If the neutral white balance values do come out to be different to those that SharpCap is using as defaults then please let me know and I will update the code.

The only other thing I can think of is don't be tempted to adjust the illumination brightness during the analysis run - some people do this thinking they are helping, but it usually leads to incorrect results.

cheers,

Robin
Andreas
Posts: 12
Joined: Thu Jun 10, 2021 9:04 am

Re: QHY 268-C sensor analysis delivers false SystemGain (-e/ADU) and Fullwell Cap.

#15

Post by Andreas »

Hi Robin,

I´ve got still the same results, nothing changed (Systemgain and Fullwell are at 50%). I´ve adjusted the WB as you can check in the picture below. The sensor analysis starts automatically with a WB of 15 for each chanel, but with this there is a huge differenz between the the curves of red/green/blue. So I put in the WB data for each channel according to the picture.

I´m now bored doing any other analysis again - the question is still, if the data of the diagrams shown on the webpage of QHY for this camera are false, or if sharpcap is calculating the results wrong.

I think its also in your interest that this is working fine - please do me a favour and test it on your own and let the community know what we can do to reach the same results.

Thank you,
Andreas
Attachments
Sensor analyses QHY 268-C 19.6.21.JPG
Sensor analyses QHY 268-C 19.6.21.JPG (145.46 KiB) Viewed 1972 times
didiflei
Posts: 20
Joined: Wed Feb 17, 2021 4:23 pm

Re: QHY 268-C sensor analysis delivers false SystemGain (-e/ADU) and Fullwell Cap.

#16

Post by didiflei »

hello together
maybe the following statement of Dr.Qiu Hongyun from QHYCCD helps.
He says the following on "cloudynights" about this problem:

Maybe need to check the RGB gain setting. 16=digital Gain 1.0 If exceed 16, the digital gain will be added and it will cause the fullwell reduced (because the value * dgain, when dgain>=1.0, it will exceed 65535 and will be forced = 65535)
Best regards,
Qiu Hongyun


unfortunately, I could not do anything with this statement. On inquiry on my part I have unfortunately received no answer.

best regards
didiflei
User avatar
admin
Site Admin
Posts: 13177
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: QHY 268-C sensor analysis delivers false SystemGain (-e/ADU) and Fullwell Cap.

#17

Post by admin »

Hi,

I would interpret Dr. Qiu's comment to be the same point that I am making - that if the R/G/B white balance settings are set too high it will affect the results. He is suggesting that 16 is the correct 'neutral' value, but I notice from the sensor analysis reports I have that SharpCap seems to be setting them to 15. That would tend to have the opposite effect though of increasing the reported FWD.

All the reports I have for the 268C seem to give similar answers - the FWD is about 70% of the expected numbers. Reports from Altair/ZWO cameras with the 571 sensor give FWD of ~50K and read noise of ~3.5, which is as expected for non-extended-FWD mode.

I suspect that there is some additional gain in the imaging system somewhere that is causing the wrong results - this could be down to the colour balance controls or some other setting. Somehow the gain at 'minimum' gain settings is about 1.4x...

Unfortunately I don't have this model of camera to test with, so cannot investigate myself.

I am still not sure about the R/G/B values in the screenshot above - I would expect the neutral values to all be the same as determined by the procedure I described of adjustment on a strongly over-exposed image.

cheers,

Robin
Andreas
Posts: 12
Joined: Thu Jun 10, 2021 9:04 am

Re: QHY 268-C sensor analysis delivers false SystemGain (-e/ADU) and Fullwell Cap.

#18

Post by Andreas »

Thank you for all your efforts and explanations .... in the end we unfortunately still have no clear idea why the QHY measured values of the 268-C ​​in the official diagrams for System Gain and Fulwell deviate so significantly from the sensor analysis in SharpCap, while the values ​​for readout noise and Dyn. range fit to 100% to these diagrams.

Therefore, I ask you to understand that me and other users of the QHY 268-C continue to be more than unsettled when there is scientific speculation about the causes instead of an official attempt by SharpCap (or QHY), so that the sensor analysis module can be trusted with every approved camera.

For me, that would be the final point that needs to be clarified, before we can close this chapter because of the growing boredom ;)))

Thank you, Andreas
didiflei
Posts: 20
Joined: Wed Feb 17, 2021 4:23 pm

Re: QHY 268-C sensor analysis delivers false SystemGain (-e/ADU) and Fullwell Cap.

#19

Post by didiflei »

Hi Robin,

is it so difficult to get a camera from QHY so they can test this extensively. This should be in the interest of both parties.
I can well understand that some owners (including me) of this camera are very uncertain. After all, this part costs 2300 €.
And that's a hell of a lot of money.

best regards
didiflei
User avatar
admin
Site Admin
Posts: 13177
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: QHY 268-C sensor analysis delivers false SystemGain (-e/ADU) and Fullwell Cap.

#20

Post by admin »

Hi,

I have just spent some more time looking through analysis results that have been performed with this camera and with the 268M.

All of the 268M results are correct (~50ke , ~85ke in exteneded full well mode).

The 268C results vary, and there is a strong correlation between the results and the White Balance (Green) setting used (SharpCap measures the green channel during sensor analysis - the results don't depend on which channel you use, but mixing them up would cause problems).

When WB(G) is set to 15/16 (or more), the e/ADU and full well come out lower than expected.
When WB(G) is set to 11/12 they are much closer to the expected values.

I am relatively sure that the problem is the WB R/G/B values are the source of the extra gain that is throwing off the measurements.

Please can someone do the following test for me:

* Camera in RAW16 Mode
* Show histogram
* Set gain to medium/high
* Set exposure to 0.5-1s
* Illuminate camera well (At this point the whole image should be solid white - strongly over exposed).

* Now turn down all the WB(R/G/B) controls to 10 - look at the histogram. See if the big spike is at the 100% point on the right, or further left (like in the image below)
Capture.JPG
Capture.JPG (115.61 KiB) Viewed 1934 times
If the big spike is at 100%, try again with all the WB(R/G/B) values reduced by 1 (ie 9/9/9). If the big spike is to the left of 100% as shown in the image, try again with all the WB(R/G/B) values increased by 1 (ie 11/11/11).

We are not aiming to get a nicely colour balanced image here, just find the values where the histogram spike for an over exposed image is at the far right. In fact the result I need to know is the smallest value of WB(R/G/B) that moves that spike to the far right. It might be 11 or 12 or 15 or 16.

For my 294 it happens with all the WB values set to 16
capture2.JPG
capture2.JPG (54.21 KiB) Viewed 1934 times
If that value turns out to be 11 or 12 then we have found the cause of the problem.

If it turns out to be 15 or 16 then there is more head scratching to do!!

thanks,

Robin
Post Reply