405CC flats issue - 294C Sensor

[andyofastro]

Hello,

viewtopic.php?t=3855

I stumbled across the post above about creating flats. With my 294C 405CC camera - which have been impossibly difficult to calibrate, I have researched the flats issue and I normally take data at the default gain and offsets in NINA. I have read a great deal about making sure the flats are longer than 3second duration. However my flats still didn't calibrate the data.

I then retook some flats with a higher gain of 200. The data however was take take with a gain of 150 and offset of 99. However the data did calibrate and gave a flat field to work with.

In the post above, showing the histogram, I am not 100% sure what the graph data is telling me, can someone explain how to read the graph? The graph has lots of little peaks and the a large peak at the end - are the little peaks demonstrating that the pixels are unsaturated?

Thanks
regards
Andy

[admin]

Hi Andy,

the one to look at is the histogram in this post : viewtopic.php?t=3855#p20347

That histogram was captured with the camera massively over-exposed. When over-exposed, the histogram should just be a single spike on the right hand side (as shown below). Instead, with the 294/492 sensor and the incorrect gain selected, the histogram still displays 'humps' near to the right hand side when significantly over-exposed.

Screenshot 2024-02-27 150739.jpg (34.05 KiB) Viewed 458 times

Because the 294/492 sensors do not give a fully 100% white output image when over-exposed, it's very easy to create bad flats for them - you think you have exposed the flats correctly, but actually they are over-exposed.

On affected cameras, the best thing to do is to avoid the problem gain range (120 to ~190 on ZWO cameras, probably the same for SVBony). Altair cameras do not show the issue because SharpCap delays the switch into HCG mode until gain 900 (equivalent to 190 on ZWO), avoiding the problem zone.

cheers,

Robin

[andyofastro]

Hi Robin,

Thank you for the reply. That is much clearer. Do you know if all 294/492 sensors do this? Or is it inconsistencies in the manufacture of these sensors - some do it, some don't for example?

In addition is the long 3 second duration upwards for flats a myth - or again, some do it, some don't for example?

Thanks
regards
Andy

[admin]

Hi Andy,

all the sensors have this peculiarity - it's mentioned in the documentation for the sensor, which specifies the range of settings under which the problem will occur. The implication of the documentation is that camera manufacturers using the sensor should stay away from those settings - some do, some don't. I have heard that the ZWO 294 colour has a workaround of some sort for this, but have never used that model myself, so can't be sure.

cheers,

Robin

[andyofastro]

Thanks Robin,

I would really like to see the documentation for the sensor. Is there a link from Sony? I've looked for the IMX294 documentation but I can't find it currently.

Thanks again for your help, I think this information changes the camera from a paper weight to something incredibly powerful.

Regards
Andy

[admin]

Hi Andy,

it's meant to be confidential, but someone else has posted a copy online, which you can find linked here : https://www.eoshd.com/comments/topic/28 ... m-e2-gh5s/

The bit of interest is on page 52 - restrictions on setting analog gain when MCOVGAIN=1

cheers,

Robin

[andyofastro]

Thanks Robin,

Extremely interesting. Just so I fully understand.
The Hexadecimal values for analogue gain:
0h to 7A5h - 0 to 27db (in low gain conversion)
4D8h to 7A5h - 8.07 to 27db (in high gain conversion 14bit)
4A5h to 7A5h - 7.54 to 27db (in high gain conversion 12bit)
6O8h to 7A5h - 12.17 - 27db (in high gain conversion 10bit)

Should be 'restricted' from use. Ie they are saying are a bit questionable, so don't use them - therefore logic dictates go above or below these ranges.
Is that correct?

edit: I've also just seen that the value of 7A5h appears to be the peak of the analogue gain of 27db. Does that basically mean that the analogue gain is basically not worth using at all? In essence, are they saying by restricting the whole gain range at 14bit from 6O8h to 7A5h you are pretty much removing the analogue pre amp range to either be (pretty much off to fully on), and therefore we are only using the digital gain? Is the pre amp at issue?

regards
Andy

[admin]

Hi Andy,

my reading was that the ranges given there were the usable ones - that is you can use the whole range in LCG, but in HCG (14 bit) you need to avoid the range from 0.0 to 8.07db and only use 8.07 to 27db. Unfortunately many camera makes are allowing access to that gain range, leading to the problems associated with certain settings on these cameras.

cheers,

Robin

[andyofastro]

Thank again for your help. That is really helpful.
regards
Andy