ZWO ASI120MC bad pixels

[Datalord]

So, decided to go into the deep end and bought myself a guidescope and camera. Fun never ends.

Anyways, I start up PHD2 to play with it and build the dark library. I start it with the cap on and in a relatively dark room (23 C) and to my horror I see this:

Really? On a 1s exposure? Is this to be expected?

[knobby]

Pretty naf ! But darks should sort it if you're just using it to guide.

What gain / gamma was that at ?

[vlaiv]

Yes, PHD2 will try to stretch image automatically to find faint guide stars (it can actually find stars without stretching image so much - it is there as a visual aid so you can see faint stars and select one you like) - this means that if you took a normal exposure of 1s with your cap on you would get black image - this image is only stretched to extreme because there are no guide stars to be found.

[Datalord]

5 hours ago, knobby said:

Pretty naf ! But darks should sort it if you're just using it to guide.

What gain / gamma was that at ?

It was at default gamma, didn't touch it IIRC.

 

4 hours ago, vlaiv said:

Yes, PHD2 will try to stretch image automatically to find faint guide stars (it can actually find stars without stretching image so much - it is there as a visual aid so you can see faint stars and select one you like) - this means that if you took a normal exposure of 1s with your cap on you would get black image - this image is only stretched to extreme because there are no guide stars to be found.

Ok, fair enough. But how am I going to pick a real star amongst all this madness? I tried building a dark library and enabling it, but then PHD turned into something like white noise.

[vlaiv]

4 hours ago, Datalord said:

Ok, fair enough. But how am I going to pick a real star amongst all this madness? I tried building a dark library and enabling it, but then PHD turned into something like white noise.

Random noise is perfect since you don't have any signal.

You will be able to pick out star among all of that random madness just fine . That auto stretching just means that phd2 for display chooses range of detected signal on picture to go from dark to bright to emphasize details in the image. In this case since there is no real signal - it emphasizes only noise so you get what you call white noise picture (like old TV sets when there is no channel tuned in). When you start doing actual exposures stars will be bright and all this noise (still there) will just be shades of dark gray - almost black - it will look like really dark background. If you think about signal level as percentage (ranging from 0 to 100 percent) - this what you are seeing now is bottom 2-3% but shaded so that 0% is black and 3% is white. When you start doing real guiding - bottom 2-3% percent will still look like this - but there will be stars that are 20%, stars that are 50% and those that are over 100% - saturated ones. Don't pick saturated ones for guiding. But this time PHD2 will set white to be lets say 60% - and you will see nice whitish stars and this 2-3% noise will be almost the same brightness - very dark uniform looking background.

[newbie alert]

Maybe too much gain ..

[Datalord]

2 hours ago, vlaiv said:

Random noise is perfect since you don't have any signal.

You will be able to pick out star among all of that random madness just fine . That auto stretching just means that phd2 for display chooses range of detected signal on picture to go from dark to bright to emphasize details in the image. In this case since there is no real signal - it emphasizes only noise so you get what you call white noise picture (like old TV sets when there is no channel tuned in). When you start doing actual exposures stars will be bright and all this noise (still there) will just be shades of dark gray - almost black - it will look like really dark background. If you think about signal level as percentage (ranging from 0 to 100 percent) - this what you are seeing now is bottom 2-3% but shaded so that 0% is black and 3% is white. When you start doing real guiding - bottom 2-3% percent will still look like this - but there will be stars that are 20%, stars that are 50% and those that are over 100% - saturated ones. Don't pick saturated ones for guiding. But this time PHD2 will set white to be lets say 60% - and you will see nice whitish stars and this 2-3% noise will be almost the same brightness - very dark uniform looking background.

Makes total sense. The white noise started after I made the master dark, which as you say, will cancel out the bad pixel and the whole thing go tits up. Thank you!

[Jannis]

The amount of hot pixels are normal for that sensor and temp btw. Cool it to 0c and you have only like 1/3 of it. Cool to -25 and you have only a few bright ones on a high gain 5 min exposure. And dont worry about phd, it only show that noise if its too dark (clouds, or out of focus, or naturally cap left on) or too bright.

[Datalord]

I'm at wits end. Went out to try it out tonight, had it pointed to Arcturus and still a ton of noise. How do I verify this thing isn't completely broken?

[Jannis]

like i said its normal for that sensor, you just need to take dark frames to calibrate it...

i have two cameras with that sensor myself, so i know how insane the noise is at high temp (as in more then 0c).
if you want a confirmation, start a sequense indoors with 60 sec at max gain, and while capyuring with the cap on, put the camera in the freezer.  first exposure will be almost white in just noise at room temp, give it 10 min and youll see half the noise. 30 min later ypull barely have any noise.  dont worry about the camera, its designed for -20 to -30c, but disconnect it before you take it out of the freezer and allow to dry properly before next use.

after this test youll also understand why people buy cooled cameras...

 

[Datalord]

but but, my 6D doesn't remotely behave like this.

So, what do I do? Put the camera in the freezer before I go out to shoot?

[Jannis]

your 6d raw sensor data also have these, but a dslr is programmed to cover up these things (also, canons long exposure noise reduction is simply calibrating the image with a dark frame).

putting it in the freezer before imaging wont help, you will have to deal with the noise by taking dark calibration frames. the darks frames will remove all those white dots, so trust us and dont worry about them.

heres an example of a single frame captured with the same sensor as you have. one have no calibration frames, the other have. also a dingle of the dark frames used for calibration.you can see the huge difference it makes on the background noise. then you stack enough of them and youll end up with a clear and mostly noise free picture. any hot pixels (those white dots you have) will also be removed completely.

[Jannis]

I realized my example didn't show the hot pixels very well due to the sensor being too hot and exposure too long. Just thought I'd give you another more similar example to yours...

When you have captured your images, say you've taken 30x 60 sec exposures at gain 30, at 10c, then to take matching dark frames, simply put the "lens cap" on and keep capturing 60 sec exposures at same gain and temp. It's recommended to take at least 10-20.
When you subtract the image with the lens cap on from the image with the lens cap off you're basically done removing the hot pixels and some other noise.  You won't have to worry much about how to do this because a stacking program like deep sky stacker will handle it all automatically for you - you just need to capture it.
What you've taken is in fact simply a single 1 sec dark frame at 23c.

Here's a single exposure with the lens cap off, and on, where I've subtracted the image with the lens cap on. As you see, you end up with very usable data after this process.

For more basic information about calibration frames used, you DSS have described it here: http://deepskystacker.free.fr/english/faq.htm#lightdarkflatoffset

[Datalord]

I think something went wrong in the communication here. I'm not looking to use the ZWO ASI120 as an imager, I have my Canon 6D for that. I literally just want to be able to get one star identified in PHD2 so I can start guiding, but everything is absolutely horrible.

I just came home from a trip and I'm going to see if I can get any image in broad daylight out of this camera, otherwise I'm going to return it as faulty.

[Datalord]

ok, so I put on the lens that comes with the camera. With a bit of tinkering it gave me the image below, which is quite ok. Making out the strings in the ricepaper lamp.

So, I'm thinking I have an issue with focus on the finder scope, but I can't find a way to set any focus on that scope. If I just use my eye to look into the scope, the focal point is about 20cm back from the end of the scope. Halp?

This is the scope: https://www.widescreen-centre.co.uk/astronomy/guidescopes/william-optics-50mm-guidescope

[Datalord]

The plot thickens. I simply didn't have proper focus on the camera. It's quite tricky to obtain on this scope, but I managed to get a nice shot of the tree in the garden.

[Jannis]

3 hours ago, Datalord said:

I think something went wrong in the communication here.

Whops - i think that "something" would be me, i was thinking about imaging, not guiding. I should have read the post more carefully. Sorry

PHD auto-stretches the image, so that might be why. I haven't seen the same image as you got though, i usually just see noise - but if sensor temp increased after taking darks - it's likely PHD will detect the new hot pixels, think they're stars, and stretch according to those. Your last two images look quite good at least, so camera appears fine.

I'm not sure how to focus your guide scope, but maybe it's possible to twist the front (like on many other finder scopes)? On my finder scope i found it easiest just to move the camera in/out and lock in position once quite close. Getting perfect focus like this is nearly impossible, but luckily for guiding that isn't needed.
I focus mine with a different program then PHD. Firecapture or sharpcap for example. Any program that allow manual exposure and live view without stretching will make the job significantly faster and easier.

[vlaiv]

Yes, focusing guider can be quite problematic. There is just a tiny focus range on F/4 guide scope where stars will show up on screen - usually as donuts if not properly focused - we are talking millimeter range.

Larger out/in focus than that and stars will be too spread out to register (same amount of light from a star spreads over large surface so it becomes too dim for camera to pick up).

I second Jannis suggestion to use either FireCapture or SharpCap to do initial focus - check where focus position approximately is. Just turn your guider to very bright star, set long exposure - like second or two and increase gain to max.

This way even if star is really out of focus you will see large donut on screen (probably fairly dim). Then you start focusing and observe diameter of donut. If it is getting smaller you are focusing in right direction - if it is getting bigger - switch direction (in/out).

If you by any chance reach end of focus before donut turns into single spot, then you can't reach focus in such setup. If you need more out focus that is Ok, just add some sort of extension to nose piece of camera to make further away from guide scope. If you don't have

enough of in focus - try removing nose piece and use T2 thread if available on guide scope - camera should have it.

[Datalord]

For good measure I thought I'd show off the final result of all this. Yesterday evening I managed to get a peek through some clouds and LP and got everything working. The focus was fiddly, but I got it done on Vega. Then I went to M13 to test it all out.

10 minutes exposure and not even the slightest star trail. I'm quite psyched. Except, obviously, I need to move to a dark sky place, preferably on the mountains in the Canary Islands.

[Jannis]

Glad you got the guiding working. 10 min is not bad at all!

A darker location would help, but i think you can still capture a lot from that location. Both clusters, galaxies and brighter nebula. Your image contains very little red color data, but assuming you're shooting in raw the color should be corrected quite easily.
Hope you don't mind, but i did some small adjustments to your image to better show what you captured. The color should be possible to fix from the raw files, and more details is should be possible by gathering more data even from that location.

 

 

[Datalord]

Thanks!

Yeah,  I have to get some proper imaging time on these bright clusters when the clouds give me the opportunity. The blue comes from the LP clip filter I use. Probably why the red is almost gone, but I can't avoid that in my location.