To guide or not to guide?

I have the 183 too (same sensor)... small pixels, 15k full well... I know well why you avoid long exposures. That being said, I'm on the side of reducing files because this camera will gladly use up a google-server worth of space when you try to collect enough light at lower exposures... the lesser of the evils for me has been to expose long for the target if needed, run a different set of rgb for stars or fix in post, and always guide... but I don't have the pixel scale you're talking about and there's no upside to not guiding for me so I can't speak to that part... only to suggest that if it's exposure motivated, I have examples of running longer exps with the 183 and it's not a terrible option.

Rick Veregin:
I image without guiding on a CEM60EC with a theoretical resolution of 0.5"  on my C925 (though seeing is the limiting factor, rarely better than average for me). I get an rms typically < 0.3" for 1 minute exposures, 0.5" if I went to 2 minute exposures. I use SharpCap to do a very good polar align. I do benefit from the fact that my mount is the ec version, which means it has encoders to keep its tracking steady. When I turn off encoders, I get something that is 2 or so times worse in tracking. And I do note that if it is windy tracking is more challenging. 

The other thing is periodic error. There is one article here on Cloudy Nights about the CEM26: https://www.cloudynights.com/topic/759169-cem26-first-cloudy-night/page-3   The poster says he gets 8" of periodic error over 600 seconds. The maximum shift will be a little more than 8"/300 or about 0.03 "/s. for 180 s that gives you 5.4", which is an rms of about 2.7". So that sounds to be probably ok. But you would do better to implement PEC if you can, it sounds as if it is available for your mount. Or limit your exposures to 1 or 2 minutes. Remember it all adds up tracking errors, PEC, wind and polar alignment errors, so you can't let PEC be right at the limit of your resolution. 

And I do do what you suggested, I let it drift (with my mount it varies through the night, back and forth, or some drift in one direction and then the other, due to polar alignment, where you are in the sky (atmospheric refraction), etc. So I do have to occassionally recenter. Personally I only let it drift maybe 30 px in any direction, so that I don't loose FOV at the edges. But this is also enough to give me a good dither, which is important to limit camera noises.

I do suggest you do some test exposures. What I do is just take short exposures every 10 s for say 5 minutes. Then I use DSS to stack the images. It provides the offsets in pixels so you can see how much drift you are getting. So measure the shift for 1 minutes of the stack, 2 minutes, 3 minutes, etc. That way you can see how much you can tolerate by carefully examining the stacked images, as well as looking at the shift in the offsets (rms error is close to 1/2 of the shift in offsets. Your optimal exposure is something less than the image that shows no tracking defects, non-round stars etc, so you have some leaway. Note you will want to try this in different parts of the sky, it will affect your tracking. You will find with experience in polar aligning, making sure the mount is on a stable base, etc, that you will get better at unguided imaging than what you can do first starting out. I do like it because it is one less thing to worry about, not having to guide. And knowing it would be hard for me to do better guiding anyway. I live in a high Bortle 8 area, so even NB filters do not need more than 1 minute exposures generally. So I have been very happy with this approach.
CS
Rick

Markus Selbach:
Clear nights are so rare. It‘s a pain to throw away 50% of your subs because of errors that sum up to become visible. If you can guide I would recommend to do it.

Yes, using an optolong filter here too and agree that will definitely help you. Your F2 lens is a game changer. 10 hrs of clear skies is a dream-come-true in NE where I am, wishing you lots of luck!

Rick Veregin:
I image without guiding on a CEM60EC with a theoretical resolution of 0.5"  on my C925 (though seeing is the limiting factor, rarely better than average for me). I get an rms typically < 0.3" for 1 minute exposures, 0.5" if I went to 2 minute exposures. I use SharpCap to do a very good polar align. I do benefit from the fact that my mount is the ec version, which means it has encoders to keep its tracking steady. When I turn off encoders, I get something that is 2 or so times worse in tracking. And I do note that if it is windy tracking is more challenging. 

The other thing is periodic error. There is one article here on Cloudy Nights about the CEM26: https://www.cloudynights.com/topic/759169-cem26-first-cloudy-night/page-3   The poster says he gets 8" of periodic error over 600 seconds. The maximum shift will be a little more than 8"/300 or about 0.03 "/s. for 180 s that gives you 5.4", which is an rms of about 2.7". So that sounds to be probably ok. But you would do better to implement PEC if you can, it sounds as if it is available for your mount. Or limit your exposures to 1 or 2 minutes. Remember it all adds up tracking errors, PEC, wind and polar alignment errors, so you can't let PEC be right at the limit of your resolution. 

And I do do what you suggested, I let it drift (with my mount it varies through the night, back and forth, or some drift in one direction and then the other, due to polar alignment, where you are in the sky (atmospheric refraction), etc. So I do have to occassionally recenter. Personally I only let it drift maybe 30 px in any direction, so that I don't loose FOV at the edges. But this is also enough to give me a good dither, which is important to limit camera noises.

I do suggest you do some test exposures. What I do is just take short exposures every 10 s for say 5 minutes. Then I use DSS to stack the images. It provides the offsets in pixels so you can see how much drift you are getting. So measure the shift for 1 minutes of the stack, 2 minutes, 3 minutes, etc. That way you can see how much you can tolerate by carefully examining the stacked images, as well as looking at the shift in the offsets (rms error is close to 1/2 of the shift in offsets. Your optimal exposure is something less than the image that shows no tracking defects, non-round stars etc, so you have some leaway. Note you will want to try this in different parts of the sky, it will affect your tracking. You will find with experience in polar aligning, making sure the mount is on a stable base, etc, that you will get better at unguided imaging than what you can do first starting out. I do like it because it is one less thing to worry about, not having to guide. And knowing it would be hard for me to do better guiding anyway. I live in a high Bortle 8 area, so even NB filters do not need more than 1 minute exposures generally. So I have been very happy with this approach.
CS
Rick

Thanks for the clear explanation Rick, much appreciated. I hadn't looked at PPEC yet, as I went almost straight to NINA (after struggling to get Stellarmate to work). I definitely need test exposures. But I see a small challenge with DSS as I do post on a Mac, and DSS is Windows only. I could try DSS on my NINA system, a Mele 3, but that might blow it up, as it is not the fastest processor around. Do you have alternatives for the kind of analysis you described?