Why is AMP GLOW even an issue?

[oymd]

I keep reading every now and then that a specific camera has this advantage of ZERO amp glow, or forum members concerned about the amount of amp glow in specific camera models they are researching for their next purchase etc....even camera manufacturers advertise on their spec sheets that a certain new sensor has ZERO amp glow etc?

I'd like anyone to explain to me the benefit of that? What is the point, really?

As an example, my 294MC Pro has obvious amp glow on exposures at and above 5 minutes and -5C. The glow is exclusively at the edge of the frame.

It has ZERO effects on my imaging experience, and calibrates out completely with some darks. And we all know that we take darks just ONCE, or maybe yearly for the fastidious.

Similarly, a 2600MC Pro or 533MC Pro has ZERO amp glow, but almost everyone using those cameras will still take some darks to make sure they have well calibrated images, and make sure they have got rid of any hot/cold pixels etc. Some might say that they do not take darks with those cameras at all, but surely, if one would be aiming for an APOD or AB's IOTD, they most certainly will take darks, even with the IMX571.

So, what is the point of LACK AMP GLOW in the first place. Why is it even an issue that keeps being mentioned?

Full well depth, QE, Dynamic Range, sensitivity, cooling etc...all those parameters make a real difference, but amp glow, I find it strange that its even mentioned.

To my mind, its comparable to that pesky dust particle stuck on your filter, that will completely calibrated out with your flats?

To summarise, under what condition would AMP GLOW be detrimental to a final image?

Edited November 27, 2021 by oymd

[JamesF]

I have always assumed that amp glow limits the dynamic range available for the data you actually want.  Depending on the well depth and integration time that may perhaps not be a problem for some, whereas it might for people who are getting into double figures of minutes per sub.

James

[vlaiv]

It is one of those "scary" words young imagers are being frightened with

If handled properly - amp glow is non issue. Even amount of signal is very small so associated noise is not very big either (comparable to read noise in normal exposures).

I guess problem is because people are thought the wrong way of doing things when using DSLR. Often they are told - don't do darks, do just bias and you'll be fine. It turns out that doing just bias is not proper way to calibrate and if you have amp glow - it won't go away.

 

[Allinthehead]

My understanding is that the signal you actually want in the area that has amp glow is swamped by the amp glow. Remove the glow and that area now has a lower signal to noise ratio than the rest of the image.

[Allinthehead]

I should add as Vlaiv has stated that it's not that much of an issue. I did however notice with my Asi1600 that those glow areas had less contrast. I'd always prefer to have a sensor without those issues as I don't bother with darks with my Imx571.

[vlaiv]

Here is an example of how much noise there is in amp glow areas.

This is single dark from my ASI1600:

You can see two areas on the right side and some of it on top left. I'm now going to subtract two dark subs (that will eliminate offset, dark current and amp glow signal) and measure noise in center of the sub and to the right - say bottom part where there is amp glow.

StdDev column shows noise. This is 240s dark exposure, and above noise is read noise + dark current noise of a single sub (I divided with square root of two because I subtracted two subs).

In one case, noise is slightly above read noise of 1.7e being about 1.93, while in corner it went up to 2.85e

Both of these are smaller then read noise only from very good CCD (which would be around 4-5e).

This is signal in dark part and brightest part of amp glow:

Offset is not removed here (was set to 64), but you can see that increase is about 3.5e per 4 minute exposure in amp glow area.

[DaveS]

Sure amp glow will calibrate out if done correctly, but isn't it better not to have it there in the first place?

I consider it to be an inherent "fault" or perhaps a weakness of certain CMOS sensors.

[MartinB]

One small benefit of zero amp glow is that you can optimise darks.  This means you don't need your dark exposures to match your lights.  This can be convenient sometimes.  

I have found that bias frames with my 2600 are stable but I definitely have to use flat darks with my 294.  I don't know whether or not this is true of all zero amp glow cmos chips.

I don't think zero amp glow is a game changer but it is agood thing!

[vlaiv]

3 hours ago, MartinB said:

I have found that bias frames with my 2600 are stable but I definitely have to use flat darks with my 294.  I don't know whether or not this is true of all zero amp glow cmos chips.

Have you tried double exposure dark trick?

I spent some time thinking about it and even if bias is not stable you can try to optimize darks by following approach:

Take dark of 60s and take dark of 30s.

When you subtract them you should be left 30s pure dark signal (this of course works if bias is stable in long exposures and only causes issue when taken on its own. We know it is stable on same exposure length as that makes dark subtraction possible).

You can then use 30s dark and remove pure 30s dark signal to get bias (which you would not be able to record otherwise - like using regular bias with 0 exposure).

[MartinB]

12 minutes ago, vlaiv said:

Have you tried double exposure dark trick?

I spent some time thinking about it and even if bias is not stable you can try to optimize darks by following approach:

Take dark of 60s and take dark of 30s.

When you subtract them you should be left 30s pure dark signal (this of course works if bias is stable in long exposures and only causes issue when taken on its own. We know it is stable on same exposure length as that makes dark subtraction possible).

You can then use 30s dark and remove pure 30s dark signal to get bias (which you would not be able to record otherwise - like using regular bias with 0 exposure).

Thanks Vlaiv, I haven't tried this and it makes perfect sense.