Autoguider sampling versus Imaging sampling

113 views
Skip to first unread message

comp...@gmail.com

unread,
Jul 15, 2022, 2:11:41 PM7/15/22
to Open PHD Guiding
Hello,

I suppose this question has been asked many times on this forum, but I haven't found the answer or its demonstration.
So I'm sorry to re-submit it.

I simply saw on a French amateur website the following assertion:
With an autoguider software such as PHD2guider, the autoguider sampling should not exceed 4x that of the imager.

Thank you in advance for your confirmation or demonstration,

bw_m...@earthlink.net

unread,
Jul 15, 2022, 5:20:10 PM7/15/22
to open-phd...@googlegroups.com

I think they’re probably talking about the image scale – the resolution of the guide camera/guide scope in units of arc-sec/sec.  The advice isn’t wrong but neither is it a requirement.

 

Bruce

--
You received this message because you are subscribed to the Google Groups "Open PHD Guiding" group.
To unsubscribe from this group and stop receiving emails from it, send an email to open-phd-guidi...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/dd17bb66-c735-477a-ba93-2eb43470eaf0n%40googlegroups.com.

Brian Valente

unread,
Jul 15, 2022, 6:46:54 PM7/15/22
to open-phd...@googlegroups.com
my personal rule of thumb is no more than 3x image scale between guider and imaging, but as Bruce pointed out there is no hard and fast rule

Brian


--

Bryan

unread,
Jul 15, 2022, 8:05:47 PM7/15/22
to Open PHD Guiding
There is an old thread on the relationship between the respective images scales of guide camera and imaging camera.  A rule of thumb, not a hard requirement.


Bryan

comp...@gmail.com

unread,
Jul 16, 2022, 3:34:42 AM7/16/22
to Open PHD Guiding
Thanks a lot for your useful comments,

And what is a "hard requirement" between the imager and the autoguiding system in term of optical performance ?

I assume that there is still some limitations with the PSF sampling between the two systems,
since the angular resolution can be very different ?

Let's assume the following case:
The main imaging system has a focal length of 1600 mm with an F/D 6.3. I use an ASI224MC camera with a pixel size of 3.75 microns.
On the autoguider I would use an ASI178MM camera with a 2.4 micron pixel size. With the previous rule, I see that I need a higher focal length of 252 mm. If we assume a factor of 3 between the two systems, we find that a self-guiding system must be of a focal length greater than 336 mm.
In this calculation, I do not take into account the turbulence or the seeing or the real spatial frequency response of the system + site (generally lower).

Sorry to ask these rather theoretical questions.
Or can we decrease in focal length ? Indeed, this would allow to have a much lighter optical system.
 
There have probably been studies on this subject ?

I am a researcher in optics/photonics, so the discussion can be of a high level.
No problem with understanding, but it's for my amateur astronomy activities.

Thanks for your advices,

Bryan

unread,
Jul 16, 2022, 9:12:06 AM7/16/22
to Open PHD Guiding
I suspect that there is NO hard requirement.  However, others more knowledgeable may have better insight.

The un-quantifiable variability from the factors that you mention are the reason. 

In the end, the quality of your stars is 'the standard.'

Bryan

bw_m...@earthlink.net

unread,
Jul 16, 2022, 2:53:37 PM7/16/22
to open-phd...@googlegroups.com

This is unfortunately a squishy discussion because there are so many free variables.  But I can explain the basic idea behind the desire for a rule-of-thumb.  From a guiding perspective, the goal is for the guided imaging system to be performance-limited by either mount performance or seeing, two things the guiding software can’t do much about.  We don’t want performance to be limited by the guiding software.  Assuming users haven’t chosen ridiculous guiding parameters and have gotten acceptable calibrations, the guiding should produce results that are limited by the accuracy of the centroid values of the guide stars.  The centroid accuracy is proportional to SNR which depends on a bunch of things we have no control over – site conditions, sky conditions, accuracy of guide camera focus, etc.  The empirical evidence across thousands of amateur setups and locations indicates that centroid accuracy is typically 0.1 px and as low as 0.05 px when multiple guide stars are available.  If we take the value of 0.1 px and apply the image scale of the guiding system, we get a performance “floor” for guiding, and we want that floor to be well below the tracking errors that arise from mount mechanics and seeing.  Then we have to consider what level of performance is needed by the main-scope imaging system based on its image scale.  Now let’s take a ridiculous example, but one we’ve actually seen.  Suppose the primary system image scale is 1 arc-sec/px and the guider is working with an image scale of 10 arc-sec/px.  That means the likely centroid uncertainty of 0.1 px will translate to a guiding uncertainty of 1 arc-sec and thus to an expected performance floor of 1 px on the main system.  It’s unlikely this would be even close to acceptable.  It’s often easier to work in the reverse direction by knowing the image scale and the optical performance of your main system and deciding how much centroid uncertainty you can tolerate.  You might want to be more conservative and use an expected uncertainty of 0.2px on the guiding system.  You can see why there’s no “hard limit” here.  The goal for having even the rough rule-of-thumb is to produce a result where the user would be highly unlikely to be limited by centroid accuracy, and the range of 3x to 4x does that.

 

You can see that this relationship between the two image scales would become increasingly important as the image scale of the main system increases (focal length increases or pixel size decreases).  As it turns out though, the topic becomes moot at these finer image scales because differential flexure between the two scopes becomes the dominant source of error – not the centroid calculations.  Once you drop below about 1 arc-sec/px on the main system, you are likely to be in a situation where you need to use an off-axis-guider.

 

Hope this makes some sense,

Bruce

 

 

From: open-phd...@googlegroups.com <open-phd...@googlegroups.com> On Behalf Of comp...@gmail.com
Sent: Saturday, July 16, 2022 12:35 AM
To: Open PHD Guiding <open-phd...@googlegroups.com>

--

You received this message because you are subscribed to the Google Groups "Open PHD Guiding" group.
To unsubscribe from this group and stop receiving emails from it, send an email to open-phd-guidi...@googlegroups.com.

Chris. S

unread,
Jul 16, 2022, 4:41:51 PM7/16/22
to open-phd...@googlegroups.com

To Bruce,

As a waiter in a well known British Comedy once used to say.... Que???????

So many Questions, but first I must get what is for me a reasonable result with my guiding, then maybe I will ask some questions

This is a good place to learn, Thank you..

Chris S.

 

 

From: bw_m...@earthlink.net
Sent: 16 July 2022 19:53
To: open-phd...@googlegroups.com
Subject: RE: [open-phd-guiding] Re: Autoguider sampling versus Imaging sampling

 

This is unfortunately a squishy discussion because there are so many free variables.  But I can explain the basic idea behind the desire for a rule-of-thumb.  From a guiding perspective, the goal is for the guided imaging system to be performance-limited by either mount performance or seeing, two things the guiding software can’t do much about.  We don’t want performance to be limited by the guiding software.  Assuming users haven’t chosen ridiculous guiding parameters and have gotten acceptable calibrations, the guiding should produce results that are limited by the accuracy of the centroid values of the guide stars.  The centroid accuracy is proportional to SNR which depends on a bunch of things we have no control over – site conditions, sky conditions, accuracy of guide camera focus, etc  The empirical evidence across thousands of amateur setups and locations indicates that centroid accuracy is typically 0.1 px and as low as 0.05 px when multiple guide stars are available.  If we take the value of 0.1 px and apply the image scale of the guiding system, we get a performance “floor” for guiding, and we want that floor to be well below the tracking errors that arise from mount mechanics and seeing.  Then we have to consider what level of performance is needed by the main-scope imaging system based on its image scale.  Now let’s take a ridiculous example, but one we’ve actually seen.  Suppose the primary system image scale is 1 arc-sec/px and the guider is working with an image scale of 10 arc-sec/px.  That means the likely centroid uncertainty of 0.1 px will translate to a guiding uncertainty of 1 arc-sec and thus to an expected performance floor of 1 px on the main system.  It’s unlikely this would be even close to acceptable.  It’s often easier to work in the reverse direction by knowing the image scale and the optical performance of your main system and deciding how much centroid uncertainty you can tolerate.  You might want to be more conservative and use an expected uncertainty of 0.2px on the guiding system.  You can see why there’s no “hard limit” here.  The goal for having even the rough rule-of-thumb is to produce a result where the user would be highly unlikely to be limited by centroid accuracy, and the range of 3x to 4x does that.




Avast logo

This email has been checked for viruses by Avast antivirus software.
www.avast.com


comp...@gmail.com

unread,
Jul 17, 2022, 2:33:14 PM7/17/22
to Open PHD Guiding
Dear Bruce,
Your answer is very interesting, and I will study it closely.
I will come back to you if I have any further questions.

I agree that there are many variables in the "system", and that the mechanical or atmospheric aspects should not be neglected.
However, it would be interesting to do a benchmark study according to the mount, the optical instruments, the observation conditions to see any statistical rules is observed.
Best,

bw_m...@earthlink.net

unread,
Jul 17, 2022, 2:47:40 PM7/17/22
to open-phd...@googlegroups.com

If I’m interpreting your initial comments correctly and you’re really trying to image at 1600mm focal length with a color camera that has 3.75 micron pixels, I don’t think you’ll be able to use a separate guide scope in any case.  Since it’s a color camera, you can’t bin it which means your main system image scale will be less than 0.5 arc-sec/px.  That puts you in a region where you will need an off-axis-guider.  Otherwise, even with perfect guiding, you will have problems with differential flexure. 

Reply all
Reply to author
Forward
0 new messages