Autoguiding / Imaging resolutions

[OzDave]

Hi,

I am thinking about setting up an Autoguider and have been reading various things online about the resolution of the guider system and the relationship of that to the Imaging system, and in fact, whether of not that matters much.

So I started by calculating the resolution of what I planned to do. My imaging setup uses a scope at 1500mm FL and my camera has 5.2um pixels. This gives me a resolution of about 0.7 arcsecs/pixel.

I have been thinking about using an old finderscope as a guide scope with an Imaging Source webcam. The finder has a FL of about 190mm and the webcam has 4.65um pixels. This gives the guider a resolution of about 5 arcsecs/pixel.

Now I read that ideally the ratio of resolutions should be something like 1:5 (guider:imager), which presumably means that the guider needs to be somewhat higher resolution that the imager in order to detect star movements and correct them before the lower resolution imaging system 'sees' them.

However, I'm not really sure if that is correct as I've never seen anything mathematical about that ratio. It just seems to be a rule of thumb.

I am wondering if my proposed setup would actually work?

With modern guiding software able to work at sub-pixel accuracy, say at 1/10 of a pixel, it means my guider effectively has a resolution of 0.5 arcsecs/pixel, which is slightly better than the Imaging setup, but not by much.

I know 'seeing' is a major factor is all of this, but that aside, does anyone think that this setup could actually work ok? Or does my guiding system not really have enough resolution to detect star movement and make corrections quickly enough to avoid blurring the Imaging system view?

Opinions?

Thanks,

David

[roundycat]

you are right about comparing the two via image scale rather than focal length. I don't know your software capability but Maxim calculates centroids to three decimal places so 1/100th pixel is pretty much guaranteed. In general a ratio of ten between your guidescope and imaging scope is ok. So, if you image at 1500 and guide at 150 (assuming the same pixel size) it will all work. I often use 2500 for imaging and 450 for guiding with the guider binned x2 and it works just fine.

Dennis

[FraserClarke]

I would think your setup would be fine. I use a ratio of ~8:1 between guider and imager (guider is 8 times shorter FL/larger pixel scale), and don't believe this is introducing any adverse effects into the images. In practice you'll probably be limited by other factors such a sampling rate, backlash, flexure, etc...

I wouldn't trust the 1/100th pixel maxim numbers; just because it reports three decimal places doesn't mean it is as accurate as that! The test would be to take a series of exposures with many stars in, and see how repeatable the reported *relative* position of the stars is. Depending on the brightness of the stars, it's probably good to 1/10th of a pixel though...

[OzDave]

Thanks very much for the replies!

So it sounds like it may be ok based on your experiences, but I'm still a bit sceptical due to the low resolution of the guider.

Reading on the web, it seems that a typical star image in average seeing might occupy about 1 arcsec and may vary between 0.5 - 1.5 arcsec.

If the guider is seeing a star, lets assume that the image is hitting a particular pixel on the guide cam. There are 5 arcsecs on that pixel, so I guess the star has a bit of room to move about on that pixel. When the star does move, its not really possible to locate it accurately within that 5 arcsec pixel, only that the star is somewhere on it because there is no other information to form a more accurate centroid calculation.

On the other hand, the imager only has 0.7 arcsecs on its pixels. So more than likely, the star will span at least 2 pixels. As the star moves about within the guider pixel and we're unable to determine by how much, the imager will be seeing the star move about across the imager equivalent of 5 arcsecs, which would be about 7 pixels on the imager, and therefore look like blurring.

I guess I'm wondering how the guider centroid calculation can be done as accurately as say 1/10th of a pixel, when the star image is less than the "size" of a pixel. If it were bigger than a pixel and spanned several pixels I could see how it would be possible, but not in this case.

Does anyone know how centroid calculations are done to sub-pixel accuracy?

Regards,

David

[FraserClarke]

1 arcsec is a nominal number for good (i.e. top of a mountain) sites. Most sea-level sites are going to be ~2--3 arcseconds. You also need to consider the spatial resolution of your guider; it is likely to be diffraction limited rather than seeing limited.

You make a good point about sampling, and it's certainly a consideration. The 'size' of the image is usually given as the "Full Width Half Maximum" -- there is still significant flux outside of this diameter. As long as you're not undersampling by a large factor, you should be OK. 5" for a typical 3" FWHM should be OK (though you're right that your centering accuracy won't be as good as if you had 1" pixels for a 3" FWHM).

The centroid is usually defined as the (weighted) mean of the flux across the star profile. You do this in two directions to the X and Y centre.

Are you in the UK or Australia?? I'm not sure what the seeing statistics are like in Oz, but it's not famed for brilliant conditions; even at the 'good' site median seeing is 1-1.5".

[roundycat]

I have never seen a guide star that occupied less than about 9 or 10 pixels.

Dennis

[adamsp123]

David, to be honest I think you are over complicating the situation with too much analysis, been there done it believe me, proof of the pud so to speak is in the eating so here is a link to an image I done using a cheapo finder with QHY5 on a 120ED, 8 min subs and the stars are pretty darn small and round, job done IMO.

http://stargazerslounge.com/imaging-deep-sky/101357-m81-82-120ed.html

Whatever you go for good luck as there are many other things to muck up guiding.

Pete

[OzDave]

Well thanks for the advice all. I think I'll just try to hook everything together and buy the "glue" parts I'm missing and see how it goes.

Cheers,

David