Auto focus not focusing


#1

New on here so firstly a big hello.
I have been using Sgp for a fair few months now and am pretty pleased with it, especially the platesolve, the big main reason for buying after the trial, everything works well apart from the auto focus which I have spent an age over the last few months on and off trying to get to work as it should. I have a qsi583wsg mounted on a wogt81 and a lakeside autofocuser, I have measured that my focuser travel is 4418 steps over 20mm which if my maths are correct equates to .0045microns, using the Wilmslow page my cfz is .081microns which I believe means a step size of 18 steps, if I’m wrong someone please tell me.
I have tried anywhere between 5 steps and 80 steps, using a multitude of different size backlash too but I end up with a real rubbish graph, that doesn’t bother me but the fact that I can obtain better focus with the bhatinov mask than the autofocus routine does. most of the time it leaves me out of focus and I end up going back to re-adjust with the mask again.
Anyone got any ideas why? Hoping it may be something simple that I’ve possibly overlooked or perhaps someone has had the same issues. So much imaging time lost due to this it’s unreal and am beginning to loose patience with it.

Thanks in advance,
Tony.


#2

Tony,
Can you post an image of or describe what your AF graphs look like? If the answer is “all over the place” can you specify what your high and low HFR is in these graphs?

A couple of things:

  1. If you have backlash you need to get this worked out first. Attempting to run AF with backlash that is not compensated will get you nowhere fast. Your backlash will likely be due to gravity since you’re using a refractor. This means you’ll need to compensate in the “IN” direction.
  2. We generally don’t recommend the math based solution for figuring out step size these days. It’s generally better to attempt to get your high HFR to be around 3x your low HFR in the AF graph. If your graph seems too flat then increase the step size. If it’s too steep then decrease it.
  3. What filter and exposure length are you attempting to use? I would recommend around 5 seconds for L and 10 seconds for RGB. And I would using the option to use a Lum filter when focusing with narrowband and using the “Auto Adjust Focus per Filter” option. But I would save that for later once you have these issues worked out.

Thanks,
Jared


#3

Tony,

I was having a big problem with autofocus on my Astro-Physics 155 refractor using USB-focus. I measured the backlash with a dial gauge, set it in SGP and still problems. I traced the problem to the pinyon block of the focuser.

On my setup, the motor attaches to a bracket which is bolted to the pinyon block. If the block is too tight, the pinyon wouldn’t turn, too loose and the block would wobble when the motor moved the focus. Also, when loose, the block would fall back due to the weight of the camera and filter wheel. As a result, my focus plots were more zig-zig and never really giving me the second half of the V curve.

I managed to correct the problem with a small piece of aluminum channel and four nylon bolts. I drilled holes in the channel to match the mounting holes of the focuser mounting bracket on the pinyon block such that the channel was perpendicular to the pinyon block. I drilled and tapped four holes to at the corners of the channel for the nylon bolts. Using the bolts, I tightened them up against the body of the focuser so the pinyon block would not wobble. You can use a plate instead of the channel, but the channel is stiffer.

Now my autofocus plots are textbook perfect V curves. Not bad for about $1 to correct.

So, check to see that your pinyon block is not wobbling or jerking when the motor turns. If that is the case (even small amounts are trouble), then the above may fix your problem. Then the only backlash comp used is to correct for backlash due to the gears in your focus motor which are pretty repeatable.

Frank Z…


#4

Thanks Jared,
Pics from Wednesday evening, can’t remember what step sizes are on which pic but a mixture of 18, 27 and 36 steps.
That evening was the first time using 18 steps, I have had some help getting everything in sgp set up and previously had had the step size set from between 20 and 80 and I have even gone as low as 5 myself. We would get a half decent curve but always usually end up with the first couple graph plots going upwards and I have been told this is probably due to backlash, whether that’s the case I don’t know, and as said in my first post have tried a number of different sizes in the ‘in’ direction but it has made no difference. The filter used for focusing is Lum at 7 secs and I have already done the focusing for my nb filters with the Bahtinov mask and the offsets are already entered.
I have clear forecast this evening and would really like to get this sorted, not much hair left to pull out.

Frank, thanks for that, I will check that out this afternoon.

Tony


#5

Tony,
Based on your graphs I would definitely check the mechanical connections first. I had issues like this and then discovered that the coupler holding the motor onto the focuser pinion shaft was loose.

Next, I would fairly dramatically increase your step size. Focus your LUM using the bahtinov mask and put your step size at 100 or even 150 (just as an experiment). The goal is to get your first HFR to be about 3-4x higher than when in focus. To me it looks like the focuser isn’t moving enough (assuming that the focuser mechanics are tight).

You should definitely be able to get much better results than you are currently able to get.

A couple other questions, are you using an ascom driver for the focus motor? And are you using the stock WO focuser or is this a moonlite or feathertouch?


#6

Hi Joel, thanks. I have had better graphs than that but have no pics to show, always the first couple of plots went upwards. will try that stepsize later this evening hopefully but am now about to check the focuser/ pinion block.
And yes, Ascom and stock r&p focuser.


#7

ok quick update, the tiny grubscrew that locks onto the focus motor spindle had worked loose somehow. wont know for sure till later now.


#8

That’s exactly what has happened to me in the past, so now on the rare time that I get a weird graph I first check that pinion set screw. Hopefully that was the problem!


#9

all is good now. that tiny grubscrew on the focus motor spindle had loosened itself. tightened up and the graph was much better, needed a few adjustments and a bit of backlash added and all spot on, step size of 25 with a inward backlash comp of 25.


#10

it does get better, can’t believe that that small grubscrew has given me so much grief, this was from last night so all working exactly as it should be.


#11

I’m glad you figured out the problem. I remember when I first discovered a loose screw on my focuser too and how annoying it was to lose a few nights of imaging because of it. But I bet you’ll always check that screw first thing if you ever get a wacky graph again!


#12

Nice! I would say that you can probably reduce your step size some…but if you get graphs that look like that I think you might want to leave it as is!

Jared


#13

Nice Curve!! glad u get it Working;)


www.mainsequencesoftware.com