Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Larger pixels and better RMS?

Equipment Imaging Mount
  • Please log in to reply
10 replies to this topic

#1 gfstallin

gfstallin

    Apollo

  • *****
  • topic starter
  • Posts: 1,467
  • Joined: 08 Aug 2015
  • Loc: Rosebud Observatory - Cheverly, Maryland USA

Posted 10 October 2024 - 10:17 AM

Four nights ago mid-session, the camera I had been using for a guiding, an ASI183MM, died (sort of). I suspect this is an issue with ZWO drivers and PHD2 (cords and Pegasus Astro hub all working fine). This has been happening recently, but I didn't feel like chasing the issue down since I had another camera I could use. I replaced the ASI183MM with a Player One Apollo-M (IMX174) I use for solar imaging, created a new profile in PHD2 and recalibrated. 

 

I noticed immediately that my guide graph looked a lot "better." The jumps in error I'd see were much more contained and my total RMS for the session was about 30% lower than it had been 30 minutes prior. Over the past several nights, this has held up. Whereas before it was standard to see average RMS over a session between 0.70" to 0.80", I have been between 0.40" and 0.60". Seeing hasn't been particularly better (and noticeably worse one night) and I've been imaging the same two objects as I had been before switching out cameras and imaging them at the same times each night. The stars in the PHD2 fov are crisper, cleaner and rounder with the Player One and guide scope combination.

 

Could the significantly larger pixels in the IMX174-based camera lead to a lower report RMS? Note that I am not arguing that my guiding is "better." I just know that the jumps are I see are much smaller in both the graphs and in the PHD2 logs I've viewed. 

 

Background: 

I'm somewhat luxuriously using a 90mm triplet (600mm fl) as a my guide scope for the time being. Considering I'm guiding a 115mm triplet reduced to 523mm fl, it is probably overkill, but I have the mount capacity for it (Losmandy G11). Everything is attached using ADM and Losmandy clamps and dovetails. The cameras weigh about the same (quite light) and my focuser on the guide scope is not racked out at all. 

 

The reason I'm even bothering to look this gift horse in the mouth is because I will be upping my game over the next several weeks with a mount axis upgrade (Losmandy G11T Titan axis upgrade) and attempting to image at 1960mm focal length. Any info I can get on what could cause such a decrease in RMS values (whether real or just as reported by PHD2) would be helpful. 

 

Regards, 

 

George


  • Spaceman 56 likes this

#2 bbasiaga

bbasiaga

    Skylab

  • *****
  • Posts: 4,404
  • Joined: 10 May 2006

Posted 10 October 2024 - 10:33 AM

If I understand right - the centroid calculation in guiding can get to around 1/3 of a pixel accuracy.   So your pixel scale defines the maximum accuracy that can be achieved.  Therefore it tracks (!) that your RMS looks better at the increased  guider image scale. 

 

Brian


  • BlueMoon likes this

#3 Oort Cloud

Oort Cloud

    Fly Me to the Moon

  • -----
  • Posts: 6,864
  • Joined: 19 Nov 2020
  • Loc: New Jersey, USA

Posted 10 October 2024 - 10:56 AM

If I understand right - the centroid calculation in guiding can get to around 1/3 of a pixel accuracy. So your pixel scale defines the maximum accuracy that can be achieved. Therefore it tracks (!) that your RMS looks better at the increased guider image scale.

Brian

Brian,

I think you might have this backwards; OP increased their pixel size, thus the guide scale is coarser than it was. That should bring less accuracy, not more. My guess would be that they aren't sampling the seeing as well with the larger pixels, and thus, not chasing those erroneous "changes" to the star position anymore.

Also, I've seen it stated that PHD2 can track a centroid to 1/10th of a pixel, but my own use has suggested it may be even more accurate than that.

Edited by Oort Cloud, 10 October 2024 - 10:57 AM.


#4 bbasiaga

bbasiaga

    Skylab

  • *****
  • Posts: 4,404
  • Joined: 10 May 2006

Posted 10 October 2024 - 11:27 AM

Brian,

I think you might have this backwards; OP increased their pixel size, thus the guide scale is coarser than it was. That should bring less accuracy, not more. My guess would be that they aren't sampling the seeing as well with the larger pixels, and thus, not chasing those erroneous "changes" to the star position anymore.

Also, I've seen it stated that PHD2 can track a centroid to 1/10th of a pixel, but my own use has suggested it may be even more accurate than that.

I think we are saying the same thing.  If his original image scale was say 6 arc-s/px, and the new one 12 arc-s/px, then 1/3 of each of those (the detection/accurracy level of the guide algorithm) is 2 and 4 respectively.  For a change to be detected, it would need to be a larger deviation in the second scenario, so the RMS looks lower as it is detecting less changes.    

Similarly, you can sometimes get lower RMS by extending your guide exposure which averages the changes more.  

 

Both can lower reported RMS, but that does not mean the image in the main camera is better - in fact it can actually be worse despite the lower reported RMS in both cases.  

 

Brian


  • gfstallin and Oort Cloud like this

#5 bbasiaga

bbasiaga

    Skylab

  • *****
  • Posts: 4,404
  • Joined: 10 May 2006

Posted 10 October 2024 - 11:28 AM

The 1/3 vs 1/10 is sometimes debated.   But the math above still tracks either way.  The 'rule of thumb' is that a guide scale to image scale of 3-5 is as far as you want to push your guide set up supports the 1/3, but could perhaps be conservative. 

 

Brian


  • Oort Cloud likes this

#6 idclimber

idclimber

    Fly Me to the Moon

  • *****
  • Posts: 7,340
  • Joined: 08 Apr 2016
  • Loc: McCall Idaho

Posted 10 October 2024 - 11:40 AM

PHD2 needs both the pixel size and your focal length to calculate rms in arc seconds. It will attempt to get the pixel size from the camera driver, but if that info is not there you need to enter it manually. When you changed cameras did you make sure the pixel size was adjusted?

 

It is also a good idea to double check the focal length you entered. I have forgotten when changing scopes and also wondered why the numbers looked wrong. 

 

The primary reason PHD2 will oscillate back and forth is seeing related. If the system is constantly making 0.25" corrections those tiny corrections can self amplify to larger movements. The primary solutions to this is longer guide intervals, multi star guiding and most importantly letting PHD2 analyze and adjust the settings with the guiding assistant. If my system was not operating as well as I thought it should, I would immediately try redoing those setup and calibration tasks.  

 

There are a number of very experienced imagers here who know how to look at a PHD2 log and help analyze problems. I am not one of them. Post a log to a shared drive and one of those folks will take a look. 

 

You mention using a 90mm refractor as a guide scope. I think this may be part of your problem, even if this is only a remote possibility. Any flexure or movement between these two scopes will confound guiding and more importantly tracking. You may think there is no way this is possible given your gear. I can assure you it can and it does. Any software that does sky modeling measures this and accommodates for it to do unguided imaging. 

 

For your 500mm focal length imaging a 50 to 60mm guide scope is plenty for 0.5" guiding. If you stay at this focal length it might be worth trying. 

 

You do however mention you want to go to two meter focal length. IMHO the best solution at the longer focal length is an off axis guider. This presents some upgrade obstacles as you need that and a slimmer version of the IMX174. The full sized planetary versions like you have typically do not fit. 

 

I moved to an OAG on my camera setup that can be moved to my SCT or either refractor. I typically do not even need to adjust the focus. I just pull up the appropriate profile in PHD2 and calibrate the guider and start my imaging session. 


  • bobzeq25 likes this

#7 Jared

Jared

    Cosmos

  • *****
  • Moderators
  • Posts: 8,249
  • Joined: 11 Oct 2005
  • Loc: Piedmont, California, U.S.

Posted 10 October 2024 - 12:16 PM

Assuming that PHD2 is calculating everything correctly with the new camera--that you got the arc seconds per pixel working properly--then I suspect the improvement comes from the better SNR you are getting with the larger pixels assuming you left the exposure length the same. The interesting question is not whether your guide numbers look any better, but whether your actual exposures are better. Unfortunately, that can be hard to tell since seeing varies from night to night. What sort of SNR is PHD2 reporting from your exposures? And what sort of values did it report with the old camera?


  • gfstallin likes this

#8 gfstallin

gfstallin

    Apollo

  • *****
  • topic starter
  • Posts: 1,467
  • Joined: 08 Aug 2015
  • Loc: Rosebud Observatory - Cheverly, Maryland USA

Posted 10 October 2024 - 03:09 PM

PHD2 needs both the pixel size and your focal length to calculate rms in arc seconds. It will attempt to get the pixel size from the camera driver, but if that info is not there you need to enter it manually. When you changed cameras did you make sure the pixel size was adjusted?

 

It is also a good idea to double check the focal length you entered. I have forgotten when changing scopes and also wondered why the numbers looked wrong. 

 

This was a good catch and the first thing I thought of. I made sure to check this before I started guiding and checked it again once those results started looking suspicious. The guide scope had not changed, but the pixel size certainly had. 

 

George



#9 gfstallin

gfstallin

    Apollo

  • *****
  • topic starter
  • Posts: 1,467
  • Joined: 08 Aug 2015
  • Loc: Rosebud Observatory - Cheverly, Maryland USA

Posted 10 October 2024 - 04:17 PM

Assuming that PHD2 is calculating everything correctly with the new camera--that you got the arc seconds per pixel working properly--then I suspect the improvement comes from the better SNR you are getting with the larger pixels assuming you left the exposure length the same. The interesting question is not whether your guide numbers look any better, but whether your actual exposures are better. Unfortunately, that can be hard to tell since seeing varies from night to night. What sort of SNR is PHD2 reporting from your exposures? And what sort of values did it report with the old camera?

Jared, 

 

I won't say you nailed it because I have no idea, but this thought occurred to me on my lunch break (when the serious problems of life are mulled over). A couple other things actually changed as well, which my newbie level of experience had not thought to mention. I'd like to thank others here as well for making me think of them. I increased my guiding exposure time by 50% to 1.5 seconds. Those crisp, clear guide stars that had been so absent from the 183 with its tiny pixels were the result of both larger pixels and increased exposure time. Perhaps less noisy stars for guiding + less chasing of the seeing with longer exposures could explain this? 

 

I can say definitely that my images aren't any better. This isn't entirely surprising since my image scale (I'm reduced to f/4.5, 1.48"/px) is pretty forgiving of tracking errors with this specific mount. I typically check each 5-minute exposure and I have not noticed an obvious relationship between spikes in guide/tracking error and average FWHM values in my images. This does not mean that these values are unaffected by those spikes, but any such relationship isn't obvious to my inexperienced eyes and relative ignorance when it comes to DSO imaging (I'm a planetary guy). Similarly, I've seen head-scratching variation in FWHM values during periods of smooth guiding, so I am guessing seeing is probably the elephant in the room.

 

Planetary imaging really gets you focused on the concept of moments of good seeing. At 5m to 7m of focal length, seeing is absolutely everything. Tracking? I've seen spectacular planetary images from hand-tracked Dobs. When I first started DSO imaging last winter, I thought seeing wouldn't be a factor. "I don't know what DSO folks complain about. 500mm is cake. Oh, so the moon is out. Wah wah wah." Sigh...a little ignorance goes a long way and a little learning doesn't go nearly far enough, but I'm learning. Folks in this thread alone just doubled the sum of my knowledge of DSO imaging, improving my stupidity to signal ratio considerably. grin.gif

 

George



#10 acrh2

acrh2

    Gemini

  • *****
  • Posts: 3,186
  • Joined: 16 Mar 2021

Posted 10 October 2024 - 06:14 PM

If I understand right - the centroid calculation in guiding can get to around 1/3 of a pixel accuracy.   So your pixel scale defines the maximum accuracy that can be achieved.  Therefore it tracks (!) that your RMS looks better at the increased  guider image scale. 

 

Brian

1) You are way off. Modern centroid algorithms have precession of 1/50th of a pixel. 

2) Your logic doesn't track - from your own premise, the larger the pixels, the higher the pixel scale, the less the precision. 

 

 

What is likely happening here is the OP gets better SNR with larger pixels, so his guiding is better because of that. 



#11 bbasiaga

bbasiaga

    Skylab

  • *****
  • Posts: 4,404
  • Joined: 10 May 2006

Posted 10 October 2024 - 07:09 PM

1) You are way off. Modern centroid algorithms have precession of 1/50th of a pixel. 

2) Your logic doesn't track - from your own premise, the larger the pixels, the higher the pixel scale, the less the precision. 

 

 

What is likely happening here is the OP gets better SNR with larger pixels, so his guiding is better because of that. 

 

Yes, the larger the pixels (or equally the shorter the focal length of your guide scope)/ the larger the image scale in arc-s/px,  the lower them maximum performance of your guide corrections will be.    This is why you need a longer guide scope as your main scope gets bigger.     Whatever the centroid accurccay is wether 1/10th or  1/50th of a pixel, that is still less precise when the pixel covers a larger area of the sky than it is when that pixel covers a smaller area of the sky.  

 

Brian




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics





Also tagged with one or more of these keywords: Equipment, Imaging, Mount



Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics