ASI2600MM users?

[rob pfile]

i put this on cloudy nights, but maybe there are some people locally that are using the ASI2600MM and can help me out.

after passing the 30-day return window, i found that my new ZWO ASI2600MM has random horizontal banding in darks and bias frames. ZWO wasn’t much help and never gave me any feedback but eventually agreed to bring the camera in for service; i sent the camera back to agena. now agena is telling me this kind of banding is normal for this camera, which i can’t really believe.

if anyone on the list is using an ASI2600MM pro would you mind sharing some darks and/or bias frames with me so i can see if my camera behaves the same as others?

youtube really mangled the quality of this video, but this is what my bin2, gain 0, offset 20 180s darks look like:

https://youtu.be/RQTQCk_57eE

only the keyframes look good and all the interpolated frames are bad but you might be able to get an idea of what i’m talking about in that video.

thanks,

rob

[Francesco Meschia]

Not to that extreme, but I have similar bands in my ASI294MM. They’re actually random, so they average out with integration. I reckon it’s their form of read noise.
There seems to be, though, some dependency from the amount USB throttling I set. Is it the same for you?
Francesco

> --
> You received this message because you are subscribed to the Google Groups "SJAA AstroImaging" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sjaa-astroimag...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/sjaa-astroimaging/ED3D7CA6-9865-4547-A1FA-363FB70EC2B2%40gmail.com.

[Karen Bieber]

Hi Rob,
Are those your dark frames? I have a ASI2600MC-Pro. Mine is color not mono, but I always get banding in my darks. It’s never been a problem during integration.
Karen

> On Jun 8, 2022, at 5:53 PM, rob pfile <rpf...@gmail.com> wrote:
>

[rob pfile]

On Jun 8, 2022, at 6:11 PM, Francesco Meschia <francesc...@gmail.com> wrote:

Not to that extreme, but I have similar bands in my ASI294MM. They’re actually random, so they average out with integration. I reckon it’s their form of read noise.
There seems to be, though, some dependency from the amount USB throttling I set. Is it the same for you?
Francesco

i have not messed around with the USB throttling; i just left it set to whatever default was set in NINA. which way does the correlation go with the USB throttling? i’m not even sure what that knob does, are higher values higher speeds or lower speeds? this is not something that i can test now as the camera is back at agena (who are now telling me this is totally normal and they might not send it back to ZWO for repair.)

On Jun 8, 2022, at 7:00 PM, Karen Bieber <kaleb...@gmail.com> wrote:

Hi Rob,
Are those your dark frames? I have a ASI2600MC-Pro. Mine is color not mono, but I always get banding in my darks. It’s never been a problem during integration.
Karen

yes, those are 180s darks. it’s true that with enough darks everything should smooth out. the only reason i started looking at this is because i noticed a diagonal moire-like pattern in the background of my integrations as i was processing them. then when i checked the darks/bias, i saw this pronounced banding, which i’ve never seen on any CCD sensor, anyway.

of course if this banding is in the darks, then it is in the lights too. my assumption is that the noise was being enhanced by registration similar to what you get in a narrowband image with very little signal in the background.

perhaps this kind of banding is a normal phenomenon but the magnitude in my camera is worse. i guess i need to try quantifying this problem. actually - i just made a 32 frame master dark and i can still see banding in it - both horizontal and vertical. so in fact the banding may not actually be as random as i thought it was.

thanks,

rob

On Jun 8, 2022, at 5:53 PM, rob pfile <rpf...@gmail.com> wrote:

i put this on cloudy nights, but maybe there are some people locally that are using the ASI2600MM and can help me out.

after passing the 30-day return window, i found that my new ZWO ASI2600MM has random horizontal banding in darks and bias frames. ZWO wasn’t much help and never gave me any feedback but eventually agreed to bring the camera in for service; i sent the camera back to agena. now agena is telling me this kind of banding is normal for this camera, which i can’t really believe.

if anyone on the list is using an ASI2600MM pro would you mind sharing some darks and/or bias frames with me so i can see if my camera behaves the same as others? 

youtube really mangled the quality of this video, but this is what my bin2, gain 0, offset 20 180s darks look like:

https://youtu.be/RQTQCk_57eE

only the keyframes look good and all the interpolated frames are bad but you might be able to get an idea of what i’m talking about in that video.

thanks,

rob

-- 
You received this message because you are subscribed to the Google Groups "SJAA AstroImaging" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sjaa-astroimag...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sjaa-astroimaging/ED3D7CA6-9865-4547-A1FA-363FB70EC2B2%40gmail.com.

-- 
You received this message because you are subscribed to the Google Groups "SJAA AstroImaging" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sjaa-astroimag...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sjaa-astroimaging/6B1B52EE-DCA8-42B9-8393-76AECDF9058A%40gmail.com.

[Francesco Meschia]

On Jun 8, 2022, at 9:21 PM, rob pfile <rpf...@gmail.com> wrote:

i have not messed around with the USB throttling; i just left it set to whatever default was set in NINA. which way does the correlation go with the USB throttling? i’m not even sure what that knob does, are higher values higher speeds or lower speeds? this is not something that i can test now as the camera is back at agena (who are now telling me this is totally normal and they might not send it back to ZWO for repair.)

The meaning of USB throttling (or “limit” as NINA calls it) is kinda dark… here’s what one of the NINA developers wrote:

The USB Limit setting is a general, vendor-unspecific knob for camera parameters that camera vendors commonly implement for their cameras. The exact meaning of the setting and of its values will differ from vendor to vendor, and even from model to model.

Here’s a stack of 24x180s darks at gain zero with my camera:

The 294 has amp glow, which can make detecting faint patterns in the master dark more difficult. To make it more visible, I did a new auto-STF on a preview that does not include the starburst pattern:

Now the pattern can be seen. It’s quite faint and it’s swamped by the noise of any light frame, but you may want to notice that there are people advocating for calibration with just bias, and even synthetic bias, for the ASI2600 – since it doesn’t have any amp glow. Here’s an interesting article: https://siril.org/tutorials/synthetic-biases/

Francesco

[Francesco Meschia]

By the way, you wrote:

perhaps this kind of banding is a normal phenomenon but the magnitude in my camera is worse.

what magnitude are we talking about? If it’s compatible with the read noise stated by the manufacturer, you’re probably fine.

Francesco

On Jun 8, 2022, at 9:42 PM, Francesco Meschia <francesc...@gmail.com> wrote:

On Jun 8, 2022, at 9:21 PM, rob pfile <rpf...@gmail.com> wrote:

i have not messed around with the USB throttling; i just left it set to whatever default was set in NINA. which way does the correlation go with the USB throttling? i’m not even sure what that knob does, are higher values higher speeds or lower speeds? this is not something that i can test now as the camera is back at agena (who are now telling me this is totally normal and they might not send it back to ZWO for repair.)

The meaning of USB throttling (or “limit” as NINA calls it) is kinda dark… here’s what one of the NINA developers wrote:

The USB Limit setting is a general, vendor-unspecific knob for camera parameters that camera vendors commonly implement for their cameras. The exact meaning of the setting and of its values will differ from vendor to vendor, and even from model to model.

Here’s a stack of 24x180s darks at gain zero with my camera:

<PastedGraphic-1.png>

The 294 has amp glow, which can make detecting faint patterns in the master dark more difficult. To make it more visible, I did a new auto-STF on a preview that does not include the starburst pattern:

<PastedGraphic-2.png>

[Tarun Kottary]

For 1600mm pro, higher usb limits of  80-90 was recommended . I use 80 with my 6200mm/2600mc pro 's.  Do you see the banding in lights too ?

TK

--
You received this message because you are subscribed to the Google Groups "SJAA AstroImaging" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sjaa-astroimag...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sjaa-astroimaging/5501BBC3-EC53-4A26-BF11-1AF67619648A%40gmail.com.

[jyp...@gmail.com]

Hi, rob,

I can see the banding on the dark frames and integrated master dark with my AIS2600MM. 

I used Sequence Generator Pro to collect the dark frames using the default USB throttle setup.

See attached screenshot. I saved two dark frames for each exposure time. I can upload them for your reference.

Cheers!

Jian Yuan Peng

[rob pfile]

yes if you could please upload those someplace so i can compare…

For 1600mm pro, higher usb limits of  80-90 was recommended . I use 80 with my 6200mm/2600mc pro 's.  Do you see the banding in lights too ?

TK

i can’t really see the banding in the lights, but the reason i started investigating this in the first place is that there was a very clear moire noise pattern in my integrated lights, down near the noise floor. the background of my integrations was not smooth at all.

there could be another explanation for this, which is that my dither is not large enough for the FPN of this sensor. i just checked and on average i had 16 sensor pixels of dither at bin2. maybe this is just not enough.

On Jun 8, 2022, at 9:45 PM, Francesco Meschia <francesc...@gmail.com> wrote:

By the way, you wrote:

perhaps this kind of banding is a normal phenomenon but the magnitude in my camera is worse.

what magnitude are we talking about? If it’s compatible with the read noise stated by the manufacturer, you’re probably fine.

Francesco

i don’t know yet, i haven’t done the analysis. basically once i saw the random banding in the darks i ran it up the flagpole at ZWO and basically never got a response. i am surprised at how “bandy” the 30x180s master dark is though. at least one person on CN is using 1000 bias frames and 100 dark frames with this sensor. it’s possible that between the smallish dither and “low” number of dark/bias subs i’m seeing problems. someone sent me a master dark from a QHY camera with the same sensor and it has vertical banding (which kind of looks like typical CCD bias banding) but very little horizontal banding.

is the IMX455 (the full frame sensor) the same generation as the IMX571 (this APS camera?). i have some darks from an ASI6200 and at least on visual inspection do not suffer nearly as much banding. also someone on CN gave me a dark from their QHY (also IMX571) and if i normalize the two images and apply the same STF the banding in mine is way worse. it is not a perfect comparison as the exposure lengths, temperature and offset are different, but the gain and bin is the same. i can’t take any test images right now since agena has my camera.

rob

[Glenn Newell]

Is it more in the bias frames than the dark?

If so, my understanding is for CMOS you don't need to do BIAS, as that signal is included in the darks, and just do darks and dark flats.

That should avoid any nonsense from 0 length shutters. Then for flats, avoid 1.0 seconds. as that is the point where the cameras/software switches from to/form hardware/software based timing.

Lastly, yes to USB throttling, but I usually turn it up some from the minimum default. For the ASI1600 I was running 80-90. For the ASI294, 55.

I have A USB 3 cable going directly from the imaging camera to a USB 3 port on the computer, no hub. Everything else on the rig goes through a hub, on the OTA.

[Francesco Meschia]

On Jun 9, 2022, at 8:56 AM, rob pfile <rpf...@gmail.com> wrote:

i don’t know yet, i haven’t done the analysis. basically once i saw the random banding in the darks i ran it up the flagpole at ZWO and basically never got a response. i am surprised at how “bandy” the 30x180s master dark is though. at least one person on CN is using 1000 bias frames and 100 dark frames with this sensor. it’s possible that between the smallish dither and “low” number of dark/bias subs i’m seeing problems. someone sent me a master dark from a QHY camera with the same sensor and it has vertical banding (which kind of looks like typical CCD bias banding) but very little horizontal banding.

Yeah… keep in mind that the 2600 is known for very smooth large-scale darks, so that PI’s auto STF can and will stretch in an extreme way, making small banding effects very visible.

If you want to share two bias frames, I’ll be curious to measure the read noise to see if it’s consistent with what ZWO says.

Francesco

[rob pfile]

the banding appears both in bias frames and dark frames. it’s really pronounced in the bias frames but that could be because there is very little signal in the bias frames thus the banding really stands out on STF.

when i first set this camera up and tried to take a dark, there were all kinds of weird artifacts in the dark. eventually i realized that there were copious light leaks from the OAG and filter wheel. even after taping everything up there were still ‘optical’ artifacts in the darks. since i could more quickly take bias frames, i initially just used 50-100 bias frames to calibrate everything. somewhere along the line i decided i should just use darks, but the light leaks made taking them kind of difficult. so i ended up with 20-30 darks.

i didn’t know about the 1.0 second thing. i definitely have sky flats that are shorter than that. i can’t get my arduino-based panel controller to work reliably with NINA, but if i could i’d have much more control over flat exposures.

as for USB cables, etc., i definitely did try all kinds of stuff, including:

new USB cables (not the flat ones that come with the camera)
ferrite chokes on the USB cables
different 12V power supplies including the ZWO-branded power supply
ferrite chokes on the 12V power supply cables
disconnect filter wheel
disconnect guide camera
a completely different PC (a MacBook) with a completely different USB cable (USB-C -> USB3)

nothing changed…

i can’t actually go direct to the PC because the Intel Stick PC i’m using has a single USB3 port, so i’m stuck with a hub. one thing i didn’t try was a powered USB3 hub. however the fact that the problem persisted with the MacBook makes me think it doesn’t have much to do with the USB topology or the cables on the imaging rig.

anyway it seems that random banding is a fact of life with this sensor, however, it does seem that my particular camera has it a little worse than others. the question is if this is out of spec.

rob

> --
> You received this message because you are subscribed to the Google Groups "SJAA AstroImaging" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sjaa-astroimag...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/sjaa-astroimaging/CACdhie3wbPtdREhF_EofNnm0NFF9EoB7ng-d07MW0PpBHn%2BHdw%40mail.gmail.com.

[Francesco Meschia]

On Jun 9, 2022, at 9:52 AM, rob pfile <rpf...@gmail.com> wrote:

the question is if this is out of spec.

Let’s quantify the randomness in electrons and find out! :)

Francesco

[rob pfile]

yes i believe this. however the comparison with the QHY is pretty interesting. it would be better though if i could get some exactly matching dark subs from another ASI2600MM to see just how much mine deviates, if at all.

assuming the right procedure for finding the read noise in a frame is to take sdev(frame1-frame2)/sqrt(2), then convert the floating point representation in PI to DN and then convert to electrons using the gain graph given by ZWO, i get:

gain0: 3.84e- (ZWO gives 3.25e- for gain0 read noise)
gain100: 0.97e- (ZWO gives ~1.5e- for gain100 read noise)

so it seems i’m mostly within spec. it could be the higher offset in the QHY image that is masking the banding visually.

there are of course a lot of clamped-to-0 pixels in the subtraction; do you need to add a pedestal or rescale to get better statistics? i think you should be able to carry out this procedure by subtracting a bias sub from a master bias; i wonder if that might give a cleaner result.

rob

[Arvind K]

Sorry you're going through this & zwo isn't of much help.

Unable to help answer directly to your question because my 2600 is on a remote location and I'm unable to compare the level of noise.

I did take darks when I first got the camera and noticed what you're describing ... but it was comparable across 2600mm, and 6200mm (I have both). I didn't think too much of the banding when I saw it, because, like you say it was random.

Secondly.. I don't actually use bias/darks with these cameras. The actual output once you add a reasonable number of subs (I don't go below 10 each filter) is outstanding even without any darks/bias calibration frames. A recent example: https://www.astrobin.com/03f0vo/

Not sure if you already clarified this: are you powering the camera with a good power source? is it a shared power source with something that has non-uniform power draw?

--

You received this message because you are subscribed to the Google Groups "SJAA AstroImaging" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sjaa-astroimag...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sjaa-astroimaging/38F8DF1B-3DA4-4EB7-9325-438A8C9EE8BB%40gmail.com.

[rob pfile]

i can certainly try without any calibration but it does seem that the sensor has a bit of fixed pattern noise so bias or darks should help with that. it all probably comes down to how much dithering is being done and i’m starting to suspect my 16 pixel dither is not enough.

as for the power supply, yeah, i did eliminate that as well by switching over to ZWO’s power brick. same look to the frames though i didn’t do the noise analysis that francesco just asked for on those frames vs. the ones take with NINA + a shared 12V power supply. although i think the camera electronics might be powered off of the USB’s 5V and only the cooler runs off the 12V. of course that doesn’t preclude noise from seeping over from the 12V rail to the 5V rail and any rails derived from the 5V. it really depends on how carefully they designed the power supplies in the camera.

rob

> To view this discussion on the web visit https://groups.google.com/d/msgid/sjaa-astroimaging/CA%2BrYWU%2BKiTxoe9C6rTUY%2BRJiDoTxDnkrAc37uKPuO4m7_T_qug%40mail.gmail.com.

[Francesco Meschia]

On Jun 9, 2022, at 10:42 AM, rob pfile <rpf...@gmail.com> wrote:

assuming the right procedure for finding the read noise in a frame is to take sdev(frame1-frame2)/sqrt(2), then convert the floating point representation in PI to DN and then convert to electrons using the gain graph given by ZWO, i get:

gain0: 3.84e- (ZWO gives 3.25e- for gain0 read noise)
gain100: 0.97e- (ZWO gives ~1.5e- for gain100 read noise)

so it seems i’m mostly within spec. it could be the higher offset in the QHY image that is masking the banding visually.

Thanks for doing this. The frame1-frame2 thing is actually used to remove the influence of fixed pattern noise: given the analysis you want to do, you may want to compare this with the result of the stdev of a single frame. Also keep in mind that standard deviation is very sensitive to outliers, I use a Python script that fits a Gaussian curve to the data for better estimation, or you can use the normalized MAD in PixInsight, which is more robust than stdev.

Francesco

[rob pfile]

i can post the bias frames so you can use your script to double check my work. i guess i could run a gentle CC on the single bias frame to try to get a better sdev.

rob

[jyp...@gmail.com]

Hi rob,

I uploaded one bias, and one dark frame here.  https://drive.google.com/drive/folders/116XttE50_6Wv6zvzvEY2Rp-p5R22vgR8?usp=sharing

Cheers,

Jian Yuan Peng

[rob pfile]

thanks for the frames, they do look similar to mine.

francesco, here are 2 zip files with 2 bias frames, gain 100 and gain 0, if you want to try your script on these images.

https://drive.google.com/file/d/1CyN1qgjarQzPHeHxKEOZ4ORzl3qtwQEL/view?usp=sharing
https://drive.google.com/file/d/1EKECUM6kFnERALhhyl844N9o7Dpaf3ip/view?usp=sharing

thanks,

rob

> rob

>
>
>
> --
> You received this message because you are subscribed to the Google Groups "SJAA AstroImaging" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sjaa-astroimag...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/sjaa-astroimaging/3b93eed9-86aa-494f-910b-85697e066ccan%40googlegroups.com.

[Francesco Meschia]

Hi Rob,

Here’s the (quite perfect!) Gaussian fit on the difference of your two gain-0 subs:

sigma/sqrt(2) = 3.267 ± 0.006 ADU

This corresponds to 2.509 e-… lower than what ASI says.

Here’s for gain 100:

sigma/sqrt(2) = 3.85 ± 0.01 ADU

which corresponds to 0.94 e-, again lower than what ASI suggests.

Not sure if I’m doing something wrong, but it seems a keeper to me.

Francesco

To view this discussion on the web visit https://groups.google.com/d/msgid/sjaa-astroimaging/458C2581-5788-4134-8268-33E853A6D799%40gmail.com.

[rob pfile]

thanks, that is similar to what i got. 

rob

On Jun 10, 2022, at 9:39 AM, Francesco Meschia <francesc...@gmail.com> wrote:

Hi Rob,

Here’s the (quite perfect!) Gaussian fit on the difference of your two gain-0 subs:

sigma/sqrt(2) = 3.267 ± 0.006 ADU

This corresponds to 2.509 e-… lower than what ASI says.

Here’s for gain 100:

To view this discussion on the web visit https://groups.google.com/d/msgid/sjaa-astroimaging/4E4E434C-0FBD-4E9D-84D0-605847EB4198%40gmail.com.

[Kai Yang]

You guys make me have a second thoughts on 2600 and ASI. Ha. 

To view this discussion on the web visit https://groups.google.com/d/msgid/sjaa-astroimaging/5F195B49-B737-49EA-9D78-3A9128253948%40gmail.com.