Frustrating night with autofocus

[AlistairW]

Hello,

Never had a problem with this before, but spend a few hours trying to get this to work last night in SGP. It's not the kit, definitely something I am not doing right.

Lakeside focuser and using a Baader 8nm Ha filter in a wheel.

I was pointing at M45 (Pleadies) and ran the normal autofocus routine, but I was not seeing my usual V curve (I think SGP has a different algorithm to FocusMax. SGP uses a large selection of stars to analyse an average position(?)). So I started playing with the number of steps and the step size. I think my question is around how large the 'range' should be. By this, I mean, should the routine only be concerned about small movements about a known approximate focus point ?

I started with a large range and then you could obviously see the dip towards focus. But if I limited the focuser range then I got more variance and indeed some near 'straightish' curves.

So yes I can get a V curve but only if the scale is 'very course' and the optimum focus will be not analysed (?). So , to reiterate, is autofocus about 'small' movements around an approximate known focus point ?

Also does it matter what I am pointing at to do this. I think with SGP, the more stars in the image the better, and I was using 60 second exposures for the Ha filter. I guess it is possible to choose a part of the sky that has fewer stars available and this would not help the algorithm - (I did do this later in the night - i.e. just pointing to a random part of the sky).

Thanks

Alistair

 

[AlistairW]

Thinking about it - maybe a simpler question would be:

If the bottom of the V curve is say 1 - then what should the top of the curve be ? ( I have read 3 to 4 times the bottom ) - so that would be a three or four then.

[Barry-Wilson]

Hi Alistair

If you are using your GT81 and 383L combination, I would start with a step size of 35, 9 AF points, bin 2, 4 for pixel size, 6 sec exposure and use Lum filter. Set backlash to something like 45 In.

You should aim for the top of the V-curve to read about 3 to 4 times the HFR value at the low point.

I would do this for LRGB filters and then input the numbers into the Filter set up. Repeat when your OTA has stabilised and temp is steady using the Ha filter (and your other NB filters) with 35sec exposure bin 2. Record the focus point and then set up the offset in the Filter set up. The help file gives a good explanation of what to do.

HTH

Barry

[ChrisLX200]

The 'seeing' last night was poor - around here at least. So instead of having a minimum HFR of 0.6-0.7 it was 1.2, and that gives a flat bottom to the 'V' curve. With M45 you add to the difficulties because the main stars don't focus down to a small point due to the nebulosity.

ChrisH

[AlistairW]

Thanks Barry and Chris.

Barry I tried you values for Ha and indeed managed to get a V curve, but only just before the clouds rolled in. So that is good progress.

I did it for HA as the only filters I have are Baader Narrowband and a Clear and a Moon\Skyglow. This may sound a silly question but I know the Narrowband filters are roughly parfocal, but would the clear filter be parfocal with the Narrowband (I am sure the answer is no).

Chris, what do you suggest as a target, I did think that that the Pleiades was a bad choice. I tried M32 (Spiral Cluster) today, and initially had all the same problems, until in SGP I cropped the area I was using. I think SGP can select stars that are already fairly small and this somehow disturbs the result.

Thanks

 

 

 

 

[Barry-Wilson]

Despite claims of being parfocal, in practice the filters rarely are, and an accurate focus is everything.

Focusing with a NB filter is quite a slow and painful process.  Try using your clear filter for a reliable V-curve and then setup offsets.

Glad to read that you managed a V-curve with the suggested settings.  You may yet need to adjust slightly for your setup but at least you're on your way now.

[ChrisLX200]

11 minutes ago, AlistairW said:

Thanks Barry and Chris.

Barry I tried you values for Ha and indeed managed to get a V curve, but only just before the clouds rolled in. So that is good progress.

I did it for HA as the only filters I have are Baader Narrowband and a Clear and a Moon\Skyglow. This may sound a silly question but I know the Narrowband filters are roughly parfocal, but would the clear filter be parfocal with the Narrowband (I am sure the answer is no).

Chris, what do you suggest as a target, I did think that that the Pleiades was a bad choice. I tried M32 (Spiral Cluster) today, and initially had all the same problems, until in SGP I cropped the area I was using. I think SGP can select stars that are already fairly small and this somehow disturbs the result.

Thanks

 

As Barry said, you really need to focus using each filter for best results, it may take longer (I use 10sec exposures 2x2 @ f/5.2) but worth the wait. For M45 I just slew North a little way, focus, then slew back. Sometimes when approaching focus stars that were intitially rejected as being too big now get included as you approach critical focus, that can bias the result just when you don't want it! I don't worry too much about getting a perfect V so long as the result is repeatable.

[AlistairW]

Thanks for the advice all - much appreciated ?