Hi

I’ve just got a Celestron focus motor for my C6 and have got it basically setup and running however using the ekos autofocus seems very hit and miss.

I’ve tried the defaults, switching to polynomial, linear and linear 1, I’ve altered the initial step from 1 to 200 and a number in between but I don’t seem to get a V shape. In fact it seems to wander off detecting what I would guess is noise after the stars have gone out of focus and the noise seems to have quite a low hfr….

It doesn’t seem to recognise when it’s gone out of focus even when stars are big doughnuts or it cannot even identify stars anymore.

I’ve managed to manually focus it and then used autofocus with a small (5 or 10) step size to refine it but I’m not convinced it’s better than me doing it manually. I’m not even getting a hfr lower than what I used to manually twiddling the knob myself though I suppose I’m at least warm if rather frustrated.


Has anyone got a sample set of settings for a similar setup that could give me a starting point please? Also is there some video somewhere of it actually being used so I can get a feel as to what to expect?

Regards

Nigel

Hi Nigel,

A good place to start is the Focus section of the Kstars Manual...
docs.kde.org/trunk5/en/kstars/kstars/tool-ekos.html#ekos-focus

Within that search for the section "How to Setup for an Autofocus Run". It takes you through setting up the Linear 1 Pass algorithm.

You could practice with the focuser simulator in Kstars. Setup an optical train containing CCD simulator, tetescope simulator and focuser simulator. This would allow you to practice setting up a focuser in daylight so you become familiar with the process. The settings won't match your C6 but you will gain experience.

Sounds like the trickiest thing is getting the initial step size "right". Needs to make a noticeable change to the HFR but if you go too far with an SCT then the doughnuts will confuse the star detection, so its a balance.

I don't have a Celestron focuser so I can't tell you the exact numbers to set (others on the forum may be able to do so) but if you post some photos of the V-curve I can probably work out the settings.

Hi

I had read the manual and a number of forum posts unfortunately while I’ve (think I have ) got a reasonable idea of how it should work I’ve been only vaguely successful.

If I make the initial step size in the hundreds then it seems to head off past the doughnut stage hunting for the lowest hfr. From one post I read it seemed that I should try a step size around the size of my scope critical focus zone which is around 120 micrometers giving me a step size around 160, tried that but not successful. Low values around 10 look like they might work, though sometimes these turn into Ws rather than Vs and then it can end up hunting past that point. This is why I was hoping someone had details of a vaguely similar setup.

One thing that might (probably) be confusing the issue is that it’s been quite windy while I’ve been trying it out. My rms has been around 3” with quite large excursions. This would make the hfr vary all on its own. I’m using the limits settings in capture which means the focus keeps getting tried. I would think it would make sense for the refocus to be linked with the guiding so it doesn’t attempt a refocus based on one dodgy image or when the wind is blowing…. Do you use the limit settings to refocus?

Thanks

Nigel

Just did some googling on your setup. The CFZ (depending on which algorithm you take of course) is about 90 microns. I saw on a cloudy nights post that 1 step = 0.75 microns. If that's correct then an initial step size of 120 steps. When you say that wasn't successful when happens?

Setting a step size too small will result in a poor V curve which will mean the curve fitting will be unreliable.

A way to get the number of steps to use is to start at focus (point A) then move either in or out to the point that you get doughnuts and the HFR on the fits viewer stops increasing. This is the point you don't want autofocus to reach as it won't work so go back a bit (point .

So A - B is the workable range where: step size * number steps = (A - . This should let you choose "number of steps".

If wind is causing noticeable problems then you're going to struggle. I personally wouldn't bother imaging in those conditions unless you can shield your equipment from the wind. As far as I remember the checks are simple checks on the last image for hfr, temp change, time, etc. I'll check the code though.

Oh, nice of the website to insert an emoji!
(Point = (Point B )

Hi John

thanks for that. You’ve confirmed what I thought about the cfz just that I forgot to mention the reducer which probably explains the slight difference.

I will try working out the A to B distance etc and try using that. I had sort of done something similar using finger in the air maths. Also I could do with working out the back lash so a bit of fiddling coming up.

Yes must admit trying things out when the wind blows with my little mount with the C6 and a dew shield on it is probably asking for problems. Add the occasional cloud going across and it’s a wonder I could get anything

I’ve now had a play with the simulators.

Some of my problems look likely to be related to trying to use a small initial step at least the simulator was showing a similar up down zig zagging rather than a curve. So next time bigger steps.

The out step setting explained why it makes so many steps AFTER it has passed the minimum, I thought it was only for an initial move out of focus.

I can also see how some of the limits and back lash settings impact the moves as well.

Now all I’ve got to do is wait for a quiet night and try and do the calculations for my setup, though I suspect avoiding the doughnuts might be a bit hard. Slightly worryingly was that the simulator runs were a bit hit and miss though became more stable at the larger step sizes assuming it didn’t hit the limit or go too far out of focus.