Dog Eared V Curves and auto focus missing the mark by one step move


#1

Hello,
I generally have reasonable focus although at times it could be better. In the process of trying a few things I noticed that when I get the message that the HFR of X may be higher than Y after an autofocus run, if I jog the focuser by the set step size in the autofocus dialogue box (mine is set at 75) the stars will be very tight and good focus is achieved.
While researching this last night it was suggested that it might pay to set the data points higher which I tried and then my V curves showed dog ears as the focuser moved further away from the CFZ, resulting in very poor focus.
Do I need to change the step size when I change the data points and does the star size need to be changed also?
Appreciate any suggestions, thanks.

Dog ears
AfGraph_DP-9_S-75_B-2x2_SS-6.0.jpg by Graham Meyer

Normal-ish
AfGraph_DP-7_S-75_B-2x2_SS-6.0.jpg by Graham Meyer


#2

Your photos don’t come up on my browser, but your problem should be easily solved from your description. The instructions for setting up auto focus are spot on and easily followed, but I find better results if I set the increase in FWHM at about twice the in focus result. If you get dog ears then the step size is too big. It sees things that aren’t there and incorrectly calculates a slightly better result.


#3

I did think about the step size being too big, although that is about right going by the SGP approved setup method. We haven’t had great weather so I did not want to play too much during an imaging session and my obs is also 60km from my house so if something goes wrong, it’s a 2-3hr return trip. Full moon is coming up so I will have more of a play then, and actually drive up.

Anyone else have any thoughts or tips?


#4

Ok, so I tried reducing my step count to various sizes under what I’m using now down to 35 and raised the min star to 8 and data points to 9 and had terrible results. I seem to get the best focus when I have a step size about 75-90 with 7 data points and 6-8 min star size. I’m sure I measured the step size at 110 originally and that was too big and gave fluffy results. Maybe my min star size is too small? I’m also waiting for some clear moonlit nights to set up temp compensation properly and perhaps filter off sets that may help skirt around these problems.

I can up load some of my AF run saves to my One Drive account if anyone wants to look. I’m getting reasonable focus, I just know it can do better as I have seen it.

Any further suggestions?


#5

Hello, anyone home?

Any help here? Any thoughts?


#6

OK - so step sizes and all that stuff are relative. What aperture (f/ratio) are you using and what is your micron step size? … also is this a refractor or reflector?


#7

Hi meygray3833

Following on from Buzz, it is not clear from what you have written whether you have tried the following (where relevant), so therefore these are suggestions for you:

  1. If you have a mono ccd are you taking your F&F subs binned - bin x 2, bin x 3? If not, experiment to see if it improves your focus.
  2. What length of F&F sub are you using - have you tried increasing?
  3. Have you tried a sub-frame so excluding a certain percentage of the whole frame?
  4. If a mono ccd or cmos, are you using filter offsets? If so, you can adjust the offset to get the required filter offset plus your one step size adjustment. Eg use Lum filter for NB and RGB filters, use R for Lum.
  5. Are you using a backlash setting? Have you experimented with increasing the size?
  6. Have you checked for any possible mechanical causes within the stepper motor/focuser interface?
  7. Have you tried switching off (or on) smart focus to see if it improves focus?

HTH.


#8

Thanks for the replies, yes, tried all of that.

I tried again last night with very small step sizes of 25, 9 data points and 10 for star size, 8 sec exposures binned 2x and had some of the worst curves I’ve ever seen although focus wasn’t too bad.

I also set to focus only using the Lum filter and tried to do filter offsets, even though they are supposed to be par focal. Now I need to wait until I have another clear, cold night to double check temp compensation and I can put it into full test. I tried focusing in different parts of the sky with reasonable success as well.

I find it frustrating that there needs to be this level of mucking around to get it to work when another stand alone focus package gives excellent results all the time. Might be a lesson in that.


#9

Sorry you are having this experience.

I use 4 different scopes, all refractors though (not sure what you use), two in my home observatory in the UK and two in a shared remote observatory in Spain, and establishing focus parameters and offsets was straightforward giving repeatable excellent results. So your experience certainly is not a universal one.

What scope, mount and image scale do you have?


#10

Hi Barry,

I am using a William Optics FLT132 with a Sharp Sky Pro focus motor on an AZ-EQ6. The camera is a QHY9M. I am using a WO Pflat IV version 2 as a flattener so image scale is 1.46 arc seconds per pixel.

I am going back to the observatory tonight to double check the focuser adjustments, stiffness or any sticking and I will physically measure how far my focuser moves in 1000 steps and calculate some more.

Cheers.


#11

The scope and flattener are very familiar, I use the same in my home observatory with a Lakeside focuser and QSI683. I have the 3.5” FT focuser and also use the 2” extender to the focuser tube to reduce the overall extension of the focuser tube when imaging at F7. I am confident therefore that you can get good focus: look at my Astrobin images for examples.

How far is the WO focuser extended when at focus? Is it rock solid and not sagging?


#12

I’m using the exact same scope and flattener with a Lakeside motor. I get 5.2u / step
My SGP AF settings are : 9 readings 25 steps apart, with 6-second Lum exposures on KAF8300 sensor


#13

Wow, did not realise it was such a popular setup.

I am on the right track. Step size is 7.14 microns, data points I have at 9 with 30 steps, bin 2 x for 8 sec, Lum filter only, with offsets for RGB. I will rcalculate offsets when I open the filter wheel and replace the narrow band filters. Focuser is easy to move, not binding or giving image shift. It is out less than 50 mm and I have checked and adjusted in SGP for backlash.

I have noticed that if I focus just on a star field it seems to work very well. If there are objects in the field it tends to go wonky. Some times with an object in the field it works well with the scope near the meridian but not at lower altitudes. I normally can,t get too low as the observatory walls prevent really low pointing.

I wonder if it is bad seeing that could be causing it?


#14

I would drop to 20 steps. The scope will not drift that much to warrant such a large overall spread.


#15

Ok, I’ll give it a go and see what happens.


#16

I haven’t had a good chance to do any long testing although I did get an hour in last night. The curves started off O.K., although more of a flat bottom U at 20 steps, 9 data points, 10 star size then the last two runs in different parts of the sky were in the shape of a J, from outside of focus to inside of focus being the short side, with no intersecting lines, only a yellow line on the outside of focus side. The strange thing was focus was still pretty good. I will go back to 30 steps and see what happens. Any other thoughts?
Cheers.


#17

It can be that the bottom of the V is flat, especially in poor seeing. When you get a J rather than a U, it means you were some way off when you started. In those cases SGP will highlight the lowest points in orange and make a pretty good estimation from those few points. The second run will center up on that and you will get a full V/U. If you don’t have a central obstruction and if you enable the smart focus option, an automatic second run will recenter on that first estimation to get you the V.


#18

Hi Buzz,

Thank you for replying. I can see what you mean now re the curves. I think I have it reasonably sorted now. The turning point really was physically measuring the distance of 1000 steps with a digital vernier rather than the setup method in SGP Help Files and advice from you guys and a friend of mine.

So thanks for the help, I really appreciate your time and effort.

Proof of focusing
M77

Kind regards,

Graham.


#19

nice image and target - I should try that some time. Should be possible at Lat 51.


www.mainsequencesoftware.com