Happy with my New ZWO ASI6200mm Pro.... but have a few techniccal issues - Forums

Topic starter

Sep 17, 2021

I am very happy with the camera, the purchase of which was instigated by looking at the images of Jerry Macon...

First 10s of light:

https://www.astrobin.com/zdcl0w/

Anyway, despite the help I have gotten a few issues remain and people need the same set up as I to help, but maybe here we will find the right person:

1. What is the right gain setting for the ASI174... my guide camera... have heard 90-100?
2. Has anyone gotten the camera to work properly with Maxim DL? It is capturing but not capturing the correct DEC RA info in the FITS header. I have the very latest driver. Doug says it needs to be connected to the Mount in Maxim but it never was for mt SBIG 8300M.
3. Getting first flats tonight and have darks, but understand bias are not needed?

Cheers

Ruediger

Sep 18, 2021 · Edited: Sep 18, 2021

Hi Jerryyyyy,

I can only reply to 1 and 3 since I use the same combi.

1. gain as low as possible since you get more noise and the risk increases a guide star is not detected in the noise. Also SNR drops. With raising gain also fulwell drops and you risk to get fully saturated stars. So coming to you question: it depends on the star region, guide scope, exposure time and seeing. But keep away from fully saturated stars.

3. no Bias with CMOS cameras. Take darks, Flats and Flats-Darks. Bias is contraproductive and leads to very weird effects. Reason is, that CMOS chips are not working strictly linear at their shortest exposures possible. And when using Darks and Flat-Darks the Bias noise is also contained there. CMOS requires very strict matching between light (also Flat) and darks exposure times. There is only very little tolerance. Take care to switch off dark optimization in PI, when using.

What I also have noticed: the 6200 has so little noise, that Darks sometimes make it worse and I use only Flats and Flat Darks.

cs
Rüdiger

Helpful

Colin

Sep 18, 2021

For 2/3
2: You should need to be connected to a mount within MaximDL otherwise it doesn’t know where it is pointing (positional data comes from being connected to the telescope).
3: Taking darks the same as your flats (flat darks) is the best way. Bias frames don’t work well with some CMOS sensors, not sure with the ASI6200 but it definitely didn’t go well with the ASI1600.

Ruediger

Sep 18, 2021

Colin:
not sure with the ASI6200 but it definitely didn’t go well with the ASI1600.

You are completely right: Bias destroyed all images I had tried. Bias frames are basically a CCD topic only.

Jerry Yesavage

Topic starter

Sep 28, 2021

Hi,

Have gotten the guiding solved with a lot of mechanical work and 2x binning. There was a protective spacer on the guide camera that had to be removed.

I am still having challenges with the flats. Sometimes it seems to work others not… have had a lot of advice, including just use the Bias without Darks. I was advised to increase the average ADU but that I think has compounded the problem…. The flats are not horrible… but I need to fine tune and need to do it systematically. I am using Sky Flats.

I have to go back and look at this… despite bad weather tonight I have flats with 35k and 30k ADU collected and can process this with Darks versus Bias and see how things look. I have a complete collection of Darks. I need to do this systematically, but it takes forever to process these huge images…. I probably should just work on the worst flats, which has surprising been the Green.

Cheers

Luca Marinelli

Sep 28, 2021

Hi Jerry,

Congratulations on the new camera! For flat correction make sure you are matching dark duration to the sky flat exposure:

calibrated light = (light - dark)/(flat - flat dark)

Dark should have the same duration as light, flat dark should have the same duration as flat. If you do this methodically, there should be no over/under-compensation of your flats.

–Luca

Well Written Helpful Concise Supportive

Linwood Ferguson

Sep 28, 2021 · Edited: Sep 28, 2021

The way to think of bias is this: A Bias is a dark frame with no significant exposure time.

If you use bias frames the signal from a light looks like this:

S = L - B - (D - B)

In other words, both Light and Dark would have bias subtracted. Basic algebra shows:

S = L - D

It's really a bit worse than useless though, since the B and D numbers are so awfully tiny that you often truncate the result in trying to do the (D - B).

Bias become necessary if you try to scale ("optimize" or proportionally apply) dark frames, i.e. apply a 600s dark frame to a 300s image by dividing it in half. For that math to work you have to subtract off the bias before the prorate. HOWEVER, my (and others' ) experience is that prorating darks makes a royal mess on low noise CMOS cameras. Just don't do it. Better off not applying a dark I think.

In fact there is a growing number of people who claim that on cameras like the 6200 the darks are so small anyway that you are better off not using darks either, just lights and flats. I remain unconvinced but the argument is fairly strong that the lack of amp glow or other gradient artifacts in a dark means it just becomes part of the background much like sky glow and gets handled the same way. I still shoot darks and flat-darks, but I'm starting to come around when I look how even the darks are.

CCD traditions will still be with us I think even when the replacement for CMOS comes along.

Helpful Insightful

Blue

Sep 29, 2021

To (Possibly) go against the first reply, I would definitely at least look at shooting bias frames with the 6200 and testing. The common wisdom about bias frames being no good with CMOS cameras seems to me to be rooted in earlier sensors which really did produce unusable bias frames. I used to have an ASI294MC Pro and bias frames were unusable from that camera, the read noise changed at low exposure times so it was different to what light or flat frames contained. It meant that bias was no good with lights and the only way you could make good flats was with a fixed exposure time and a light panel so you could shoot matching dark flats.

I replaced that camera with an ASI2600MC and the bias frames that produces (And the ASI2600MM I bought since) actually calibrate lights and flats properly. It makes gathering flat frames a lot easier as you can shoot anywhere from about .25 to 30 seconds and calibrate with a master bias. The sensor in the 6200 seems to be the bigger brother to that in the 2600 so it will hopefully produce good bias frames too.

Given the similarity between cameras, I would suggest shooting flats at 0 gain. I chased some calibration issues round in circles recently which turned out to be from using gain 100 for narrowband flats, at 100 gain the camera runs at approx 0.2e/ADC so the quantisation precision just didn't cut it. My Oiii flats shot carefully at 100 gain produced brightened corners and quick, rough flats shot at 0 gain worked perfectly!

I now calibrate all my flats with a master bias, and my lights with the resulting master flat, and with an appropriate master dark (Which if generated in Astro Pixel Processor do not have bias subtracted, so the master dark calibrates for bias as well)

Helpful Insightful Engaging

Linwood Ferguson

Sep 29, 2021

To me it's more relevant that they are not needed than that they don't work.

Optimizing (prorating) darks definitely do not work well, on a 6200, I tried it (with pixinsight). You need bias for optimizing darks. Otherwise, you really don't.

Blue

Sep 29, 2021 · Edited: Sep 29, 2021

In my experience of both ASI2600 flavours (OSC and mono) master bias calibrated master flats in turn calibrate lights better than uncalibrated flats (I have not bothered with dark-flats since buying the 2600 and finding that they calibrate well with a master bias) and at very high stretch levels, lights calibrated with flats produce a better looking result when also calibrated with either a master bias or master dark (Containing bias information) I use the master dark as it also takes out predictable hot pixels (Not that either camera has many) leaving less for outlier rejection to do.

Having said that, for both of my 2600 cameras you could get away without using bias or darks if you wanted to and I know a few people doing that and getting nice results.

I have never tried scaling darks as I tend to use a range of fixed exposure times so a master dark library is not onerous to build.

Helpful Insightful Respectful

Linwood Ferguson

Sep 29, 2021

I agree with you, because at these low noise levels a bias and a dark flat are basically the same image. So whether you use a dark flat or bias you get the same basic result (which frankly is almost the same as nothing at all for flats given the brief exposure, but I still do it).

The noise at (say) 1 second (dark flat) and 0.00001 second (bias) is pretty much the same.

Helpful Insightful Respectful

Jerry Yesavage

Topic starter

Sep 29, 2021

Everyone, thanks for this lively discussion. I am weighing every word...

As a research center director (not astronomy!) in my day job I decided to obtain some data on the issue and collected flats with an average 25k ADU and 30k ADU and then processed both with either bias or darks. Hard for me to tell, but I seem to prefer the bias corrections on the 25k... perhaps somebody could suggest a quantified method to declare a victor? The images/screen shots are Revisions located behind my "First Light" image:

First Light ZWO ASI 6200M The Cave (Sh2 155 ) Tests of Flats with Bias versus Dark Subtraction

Jon Talbot

Sep 29, 2021

My two cents.
I’ve used the ASI 6200 for over a year now. Previous to the 6200 I only used ccd’s (20yrs). At first I took flat darks to match with my flats. A few months back I tested calibrating the flats with just a master bias. To my surprise a master bias worked just fine.

I tested calibration using both a flat dark and a master bias. I could not tell the difference. So, something is different about the chip and electronics in this camera compared to other CMOs cameras. It behaves much more like a ccd camera.

As far as darks, I don’t optimize darks. I use master darks at the same exposure time as the main subframes. All my calibration frames are also taken at the same gain as my subs.

Helpful Insightful Respectful

Jerry Yesavage

Topic starter

Sep 29, 2021

Yes, this is my experience and squinting at these images I do not see a difference between using bias versus darks… I have to try the suggestion about of nothing but the flats… there might be something to be said for that if all these "control" manipulations just contribute to noise or diminish the signal.

I have to look into the gain = 0 issue when I get my Chroma O-III filter, which is backordered. That was always the worst with my SBIG and Astrodons.

Jerry Yesavage

Topic starter

Oct 31, 2021

Guys,

Hard to believe I started this thread a month ago. I have made significant progress thanks to folks like Jon, but I am still having some flats problems which is probably some sort of light leak.

Here is the progress:

The Skull Nebula PLN 118-74.1 (Comparision of SBIG 8300M versus ASI 6200M Images)

Here is the problem:

I am collecting ideas as I tear the system apart. This thread on CNs has images:

https://www.cloudynights.com/topic/793398-the-asi6200mm-the-lord-of-light-leaks/?p=11464136

Blue

Nov 1, 2021

One thing I have worked out since I last posted in this thread. Flats on very short exposure time lights are tricky. With both my ASI2600MC and MM I have found that they very faintest of coloured halos are present in the flats, it is a different shape between my refractor and my Newtonian (Which has a Paracorr so there is a refractive element to deal with) but it is visibly about the same for each scope regardless of the camera used (For the mono camera I combined R-G-B flats as a colour image using Astro Pixel Processor)

When correcting short exposure lights a colour cast in the background was introduced that matched the pattern seen in the flats. A lot of head scratching later it seems to be the colour cast in the flats (Which was present no matter how I shot them, uncalibrated, calibrated with a master bias or fixed length flats calibrated with dark flats) produced matching casts on the lights in areas of very dark background, looking at unstretched 60 second lights, the background areas are only 3 to 10 ADC above the levels of a master dark (I am under Bortle 3 to 4 skies) and it seems that the read noise is dominating the background. I reshot the same stars at 120 seconds and the cast in the background effectively disappeared. It was actually still just visible if you stretched the image insanely hard, but invisible at a useful stretch level.

I was aiming for the shortest subs I could get away with to get the best star colours, so I wanted as few saturated pixels as possible but it seems that 60 seconds introduces too much measurement uncertainty in very dark areas where the read noise is a significant portion of the read out levels.

The long and the short of it, there really is such as thing as subs that are too short.

Helpful Insightful Respectful

Jerry Yesavage

Topic starter

Nov 1, 2021

Blue:
One thing I have worked out since I last posted in this thread. Flats on very short exposure time lights are tricky. With both my ASI2600MC and MM I have found that they very faintest of coloured halos are present in the flats, it is a different shape between my refractor and my Newtonian (Which has a Paracorr so there is a refractive element to deal with) but it is visibly about the same for each scope regardless of the camera used (For the mono camera I combined R-G-B flats as a colour image using Astro Pixel Processor)

When correcting short exposure lights a colour cast in the background was introduced that matched the pattern seen in the flats. A lot of head scratching later it seems to be the colour cast in the flats (Which was present no matter how I shot them, uncalibrated, calibrated with a master bias or fixed length flats calibrated with dark flats) produced matching casts on the lights in areas of very dark background, looking at unstretched 60 second lights, the background areas are only 3 to 10 ADC above the levels of a master dark (I am under Bortle 3 to 4 skies) and it seems that the read noise is dominating the background. I reshot the same stars at 120 seconds and the cast in the background effectively disappeared. It was actually still just visible if you stretched the image insanely hard, but invisible at a useful stretch level.

I was aiming for the shortest subs I could get away with to get the best star colours, so I wanted as few saturated pixels as possible but it seems that 60 seconds introduces too much measurement uncertainty in very dark areas where the read noise is a significant portion of the read out levels.

The long and the short of it, there really is such as thing as subs that are too short.

So you are shooting your lights at 120s not 60s. I do not see what scope you are using (F). I am using 60s for a Stellarvue F6.9.

Blue

Nov 2, 2021 · Edited: Nov 2, 2021

Yes, I am considering "Short" exposures to be 120 seconds (At 0 gain anyway) at least on the refractor and using the 2600MC. I had been using 120 seconds with the 2600MM camera on the newtonian as well, I was basing it off having no more than a couple of hundred saturated pixels per sub per filter to avoid skewing star colours or saturating them to white. I suspect that differences in sky conditions could change the minimum exposure time before it became evident. A friend images from a light polluted city location and I doubt the issue would even be visible from his location as the sky background brightness would overcome the read noise very quickly. I am hoping he can come out to image at my place at some stage as I would like to see if a similar effect appears in his gear (Different scope, different camera)

The scopes in my case are a Stellarvue SVX80T at it's native F6 and a 10"F4 newtonian.

Helpful Insightful Respectful Engaging

Jerry Yesavage

Topic starter

Nov 2, 2021

Thanks, I have the SVX130T and Jon Talbot has the SVX152T. The 130 barely fits in my ROR observatory… by millimeters.

I started with 120s but I did some calculations suggested by Gary I and it was clear that was way too long for my sky noise…. probably like your friend. I did a good deal of testing and basically I am better with 10x60 for RGB than 5x120. Also gives me more frames and I can be selective with what I actually use. I am Bortle 6 at least.

Helpful

Jon Talbot

Nov 2, 2021

With the 152 I'm at F8 and use Gain 100 now almost exclusively. I calculated my exposure time using Dr Robin Glovers (Sharpcap developer) technique of calculating the sky electron rate and his noise calculation method which can be found on YouTube. It works great from home (Bortle 5) and dark sites. My home lights are 180s Lum and 300s color using the 6200. You definitely do not want to expose short enough that your mean background is less than about 550 adu if using offset 50 (default). Brighter skies will probably require shorter exposures. If you haven't, check out Glovers video about noise.

Helpful

Michele Campini

Nov 3, 2021

Ruediger:
Hi Jerryyyyy,

I can only reply to 1 and 3 since I use the same combi.

1. gain as low as possible since you get more noise and the risk increases a guide star is not detected in the noise. Also SNR drops. With raising gain also fulwell drops and you risk to get fully saturated stars. So coming to you question: it depends on the star region, guide scope, exposure time and seeing. But keep away from fully saturated stars.

3. no Bias with CMOS cameras. Take darks, Flats and Flats-Darks. Bias is contraproductive and leads to very weird effects. Reason is, that CMOS chips are not working strictly linear at their shortest exposures possible. And when using Darks and Flat-Darks the Bias noise is also contained there. CMOS requires very strict matching between light (also Flat) and darks exposure times. There is only very little tolerance. Take care to switch off dark optimization in PI, when using.

What I also have noticed: the 6200 has so little noise, that Darks sometimes make it worse and I use only Flats and Flat Darks.

cs
Rüdiger

Since i use bias with my ASI2600MM and i used bias with ASI2600MC one year ago to calibrate flats without any problem. I know the ASI294 is not linear in short exposures but i've no same news about ASI2600 and his sensor.