ASI2600MM Sensor Analysis and Binning

[rkymtnbiker]

Hi, Robin.

Long time SharpCap user primarily for polar alignment and planetary imaging. But, now I'm trying to use Sensor Analysis to not only determine the optimal subframe exposure, but to determine SNR given a target's brightness. And, I'm particularly interested in how CMOS binning for the ASI2600MM affects SNR.

I just completed Sensor Analysis for my new ASI2600MM camera. It seemed to complete everything including the binning test. I expected to see separate graphs/tables for the different binning modes; I didn't see anything like that. I expected Sensor Analysis to tell me how the camera deals with binning; but I didn't see that in the report/graphs/table either.

Can I access the detailed findings of Sensor Analysis including how the camera deals with binning?

And, if Sensor Analysis didn't produce graphs/tables for each of the binning modes, is there a relatively straightforward way I can produce them from the 1x1 case that I did get? All I really need for my calculations are tables for the 2x2 and 3x3 cases that have the same info as was provided for the 1x1 case.

Attached is the output of my Sensor Analysis.

Thanks,

Ted Stokes

[admin]

Hi Ted,

Binning for CCD cameras was a useful tool, since in CCD sensors, binning is handled by the electrons from more than one pixel all getting lumped together and fed through the A to D converter in one go. That meant that you only got one lot of read noise, but you had more signal, so it considerably boosted the S/N ration.

In CMOS, binning is a much weaker tool, since it *always* happens after A to D conversion (even when labelled as 'hardware binning' by the camera manufacturers). This means that for CMOS, binning has exactly the same effect on image quality and noise as taking the unbinned image as a FITS/PNG file and reducing the resolution by a factor of 2 (or whatever your binning factor is). There is no gain in quality due to being able to process the electrons from more than one pixel with only a single set of read noise...

This means that 2x2 binning on a CMOS camera will always double the SNR (four times the singal, twice the noise for additive binning or same signal, half the noise for averaging binning). All that SharpCap is trying to work out in the binning test is which of these two approaches the camera uses by seeing if the image brightens when binning is enabled. In theory, SharpCap ought to try to work out if the camera is a CCD or CMOS at this point, but that would add complexity and more time to every analysis, and I know from usage statistics that just guessing CMOS is right in about 99.9+% of cases...

So, the reason you don't see any interesting information about binning is that for CMOS cameras there really isn't any interesting information to show. Once you know how the camera performs in unbinned mode, you can work out the noise performance in any binned mode from simple statistics, and SharpCap does that when the smart histogram/brain function is enabled in a binned mode.

One last note, you shouldn't assume that binning has no useful effect for CMOS cameras - there are at least two...

1) Reduced output file sizes
2) For 'hardware' binning where the calculations are done on the camera, you can get higher frame rates due to less data needing to be transferred.

cheers,

Robin

[rkymtnbiker]

Thanks for that explanation, Robin. I definitely agree that reduced file size can be a benefit.

For me, I think there will be 4 benefits:

1) I'll be using this camera at 2350mm often; so binning 2x2 will give me an image scale of 0.66"/pixel and 3x3 will give me 0.99". These values are much more reasonable than 0.33" which is what 1x1 would give me.
2) I'm interested in getting that doubling or more of SNR. I've done this by using 4x4 binning with my 294mm with the same telescope and I seem to get significantly better images.
3) I'm interested in the larger FWC and corresponding Dynamic Range. According to my calculations, I should be able to get 14.4 stops with 2x2 and Gain of 100. 3x3 could give me 15 stops. 1x1 would be about 13.4 stops.
4) Finally, as you mentioned, smaller file sizes help. I used the 1x1 48 Megapixel mode on the 294mm recently and those photos were SLOOOOW!

I did PixInsight BasicCCDParameters for this camera for ZWO Gain of 100 for 1x1, 2x2, and 3x3. e/ADU gain seems to go up linearly per the number of pixels binned. Read Noise seems to go up by the square root of the number of pixels binned. And, FWC seems to go up linearly per the number of pixels binned.

Is that correct?

BTW, in case you haven't seen this, this is how ZWO explains their binning:

https://astronomy-imaging-camera.com/tu ... ntals.html

Thanks,

Ted

[admin]

Hi Ted,

yes, that sounds correct from PixInsight - the signal strength is proportional to the number of pixels binned and the noise is proportional to the square root, so the SNR improves proportional to the square root.

I hadn't spotted the ZWO web page previously - everything is pretty much as I expected, except that I had not seen the bit about 'hardware binning' really being skipping pixels. If that's the case then best to avoid it except for high speed imaging where the frame rate is the most important thing.

BTW, the 294MM is 'special'. SharpCap presents the 1x1 high resolution mode as a '47 Megapixel' read mode and the 2x2 mode as a '11 Megapixel' read mode. This is because the read noise doesn't quite behave as you would expect if the binning was just summing or averaging, so having the separate read modes allows SharpCap to maintain separate analysis data for both.

cheers,

Robin

[rkymtnbiker]

Yes, I also have the 294MC Pro that does not have the ability to "unlock" and use the 2.315 micron pixel mode. So, it's confusing even between the color and mono versions of this camera. But, I appreciate your implementation for the 294MM in SharpCap with the Read Mode that helps make sense of it. When binning the Mono camera, it's almost better to just think of it as 2 cameras: 1) a Micro 4/3 sized ASI183MM that has 2.315 micron pixels that can be binned only 3x3 and 2) an ASI294 with 4.63 micron pixels that can be binned 2x2.

Thanks for your continual improvements to SharpCap and for answering these questions so quickly!

Ted

[bwells]

Ted, Robin:

I have been using a QHY165C for many years and I just purchased the ASI2600MC to use with my Edge800, which is similar to the longer focal length OTA Ted spoke of. I see this this thread is a year old, so would like to hear from you Ted on how you are making out using your 2600MC with your longer focal length OTA and Robin, I would like to revisit the sensor analysis in SharpCap once I get the new ASI2600MC. I would love to know how well things are working out for you both when it comes to using the ASI2600MC and SharpCap.

Thanks
Bryan

[admin]

Hi Bryan,

Hope that you get some great images with the ASI2600 camera - I don't have that particular model, but I do have cameras with the equivalent sensor from a couple of other manufacturers and I must say that I find the IMX571 (in the ASI2600, Player One Poseidon, Altair 26C/M, etc) to be a really good sensor (easy to work with). For those looking for the same easy to use sensor at a lower price, cameras based on the smaller IMX533 sensor are worth checking out.

Anyway, you should find that the sensor data for the ASI2600 is pre-loaded into SharpCap, but you can of course run your own. My experience is that there is remarkably little variation between different samples of the same camera, so I suspect you will get figures very similar to the built in ones.

cheers,

Robin