Touptek ATR2600M Sensor Analysis in High full well mode

[admin]

Well, I had a good think about this and came up with a sort of theoretical understanding of what is (probably) going on with HFW mode...

The sensor manufacturer - Sony - sets a minimum gain that you should use the camera at. This value is chosen to ensure that all pixels can saturate properly at that gain value. In order to saturate, the full well depth of the pixel must be greater than the maximum ADU (65535 in 16 bit mode) multiplied by the e/ADU value.

Now, in spite of highly sophisticated manufacturing, the pixels on a camera will not all have the same full well depth - there will be some sort of range. There will also probably be a bit of a range between different sensors too. Sony will choose a minimum gain value that is fairly conservative - ensuring that the image properly saturates even for the worst case scenario of the smallest expected full well depth value. There's almost certainly some sort of margin too, meaning that the actual full well depth of all pixels on the camera will be somewhat above the maximum FWD that we usually use.

My belief is that the 'High Full Well' modes offered by various astro camera manufacturers basically allows you to select gains below the minimum recommended by the camera manufacturer. To begin with, this eats into the margin of full well depth and we still see fully saturating images with extended full well depth. Beyond that point, we get to the situation where a few of the 'worst' pixels fail to fully saturate, because they have the lowest full well depths and they can't ever reach the target full well value needed for saturation.

Next, we start to get to the point that the electron count needed for saturation is beyond the full well depth of *any* of the pixels on the camera - that's the sort of situation I was seeing at gain=100 in HFW mode with the over-saturated histogram peaking at ~70 to 80%.

On my specific ATR2600M, if I set a gain value below about 152 then some pixels fail to saturate - starting with just a few and increasing in number as I go to lower gains. If the gain is kept above ~154 to 155 then all is fine. I suspect that the value of 177 suggested by ToupTek is based on a wider range of cameras, where some will be like mine and cope with lower gains and some won't. Of course we don't know how common those cameras that need 177 gain are compared to the average sort...

So, I adjusted the code to set the minimum gain in HFW to 154, and got a new plot...

Screenshot 2025-11-18 161406.png (136.53 KiB) Viewed 1354 times

Here, the HCG mode enables at gain=465 and we see that the read noise is not much increased at 1.65 but we do get some additional full well capacity - up to 21557, giving a DR of 13.67 stops.

I will release the code in this state for the next update and we can see how it goes.

cheers,

Robin

[giraud]

Thanks Robin,

The test I performed with my camera in HGC HFW ULN, with the Ascom driver, allowed me to obtain results with a gain value of 100. I didn't check the behavior of individual pixels, but the only thing I can say is that the test went through without any error message. Does that mean my camera can operate at such a gain and that all pixels can still saturate (so I would have a great individual sensor)?

Do you believe Player One or QHY are better at controling the IMX571, or are ordering IMX571 selected by Sony from the best production lots, are that they are pushing their marketing...?

Cheers,
Eric.

[admin]

Hi,

I personally wouldn't use the camera in a mode that does not fully saturate the pixels - not only is the expected dynamic range not really there (because you can't hit full pixel values), but also you can be badly misled. One problem that happens is that you can think that flat frames are properly exposed (because the image isn't saturated), but actually you are overexposed and the flats fail to work at all.

There are different grades of sensor - I managed to get hold of an industrial camera IMX571 for a very good price and the sensor is nowhere near as good as the astro cameras (lots of noise). However, I don't know if there is any finer grading of sensors or how different manufacturers specify the sensor grades.

I suspect that QHY, ToupTek, PlayerOne, etc are all tinkering with the same settings on the camera, just taking different positions on how far they adjust those settings. ToupTek have basically passed that decision on to the user - allowing the full range to be accessed (for good or ill...).

cheers,

Robin