Need help to understand sensor analysis (Moravian C3-61000 Pro

[julien.t]

Hi,

I tried to do a sensor analysis of my Moravian C3-61000 Pro, but result looks strange.

Here is the result

result.png (133.04 KiB) Viewed 2438 times

My questions

- I expected the read noise to become lower with gain increasing, why it is increasing ?
- The Moravian's gain can be set with a value from 0 to 4030. So why the test only use 0 to 30 ?
- The documentation says to use constant illumination but to avoid artificial light. I have done the test at the end of an overcast day. In front of the scope I put a quite thick plexiglass sheet. The light was slowly lowering, but I don't know how to do otherwise, in the middle of the day the sky was to bright.

Regards,

Julien

[admin]

Hi Julien,

yes, that does look odd. I have sometimes seen the read noise creep up a bit at the very highest gains, but normally it drops with increasing gain and enough people use the IMX455 sensor in cameras from other brands that we would have heard about any oddity like this if it was genuine.

My suspicion is that it may be linked to the odd gain range... SharpCap tries to display the gain value for Moravian cameras in decibels of gain, rather than the raw values (displaying in dB gives a more physical value - increasing by 20 is supposed to be 10x brighter). The conversion between the two comes from the Moravian SDK, and I am wondering if the SDK that SharpCap is using doesn't have support for the C3-61000 camera and is therefore giving strange results.

I can see that there is a version 4.17 now available (https://www.gxccd.com/cat?id=164&lang=409) , but SharpCap is currently using 4.16. If you want you could try copying the files from the new SDK download over the ones that SharpCap installs to see if that helps - just be sure to get the x86 ones for 32-bit SharpCap and the x64 ones for 64-bit SharpCap.

I will try to put the new version in SharpCap for the next SharpCap update.

If the new SDK doesn't help then we will need to try a few other things to understand what is going on.

cheers,

Robin

[julien.t]

Hi,

Updating the Moravian lib doesn't help.

Maybe this is simply me who made a mistake while doing the measures ?

Julien

[admin]

Hi Julien,

no, something is not right with the gain I think - when you adjust the gain in SharpCap, does the image on screen get consistently brighter as you increase the gain from minimum to higher values? The reason that I ask is that the read noise seems to change in the opposite direction to usual...

Another way to potentially debug what is going on with the gain control to see if it is behaving is to use the option to record 'Camera hardware communications' in the SharpCap log (see https://docs.sharpcap.co.uk/4.1/#Troubleshooting) - once this is on, open the camera and make various adjustments to the gain - increasing from minimum to maximum probably. Then close the camera, access the SharpCap log and send it to me - it will record all the communication between SharpCap and the Moravian SDK, including queries regarding gain values. Hopefully that will reveal something of interest.

cheers,

Robin

[julien.t]

Hi,

I have do a few exposures. First using the ASCOM driver with gains 0, 2000 and 4030. Then I have done the same with the native driver with gains 0, 5.82 and 36 which is equivalent to the value used with ASCOM according to the camera documentation.
In both cases I get similar brightness values (I checked the max value using Pixinsight).

An other observation, I the sensor analysis I have done, the value for gain 0 is very close to the theoretical values mentioned in the camera documentation.

[julien.t]

Hi,

I have done a new attempt. This is much better.

The difference with the first one :

- I have done the analysis in the morning before the sun is to high
- I used the ASCOM driver.

Capture d'écran 2024-01-20 102847.png (146.85 KiB) Viewed 1664 times

There still is a strange small drop of the read noise, just before the big drop at 2750.
Edit : I just understood that what I took for a drop of the read noise is the label of the graph

If you want I have logs of all the process.

[admin]

Hi,

thanks for sharing - that looks like a much better result and certainly shows that something is not quite right with the way that camera is working under direct SDK control. I suspect that I am missing something regarding the setup or configuration of the camera, but not sure what.

It's interesting that the gain range of the camera is still rather small - the maximum relative gain showing is only 3.74x brighter than minimum gain. That seems odd - I would expect a range of at least 100x and probably 300x to 1000x based on other manufacturer cameras with the same sensor. Although you may not want to use those high gains for imaging, they are very useful for framing, focusing, plate solving, etc.

I don't think there is an early drop in the read noise before the main drop - I think that is just the legend of the graph unfortunately being close to the read noise points in this case.

cheers,

Robin

[julien.t]

It's interesting that the gain range of the camera is still rather small - the maximum relative gain showing is only 3.74x brighter than minimum gain. That seems odd - I would expect a range of at least 100x and probably 300x to 1000x based on other manufacturer cameras with the same sensor. Although you may not want to use those high gains for imaging, they are very useful for framing, focusing, plate solving, etc.

Yes, but this match the values of the Moravian documentation :

Gain number Gain in dB Gain multiply
0 0.00 1.00×
500 1.13 1.14×
1000 2.43 1.32×
1500 3.96 1.58×
2000 5.82 1.95×
2500 8.19 2.57×
3000 11.46 3.74×
3500 16.75 6.88×
4000 32.69 43.11×
4030 35.99 63.00×

I don't think there is an early drop in the read noise before the main drop - I think that is just the legend of the graph unfortunately being close to the read noise points in this case.

Yes, that is what I noticed later.

Julien