14 replies to this topic

[jpengstrom]

I was wondering if anyone has experience that could help me with a problem I'm having while autofocusing my 10" Quattro in NINA.  The basic problem is that HFR in NINA (using hocus focus for autofocus) doesn't reliably measure stars with donuts.  That's not surprising, but my focus range without producing donuts on my 10" newt is very narrow.  So my autofocus runs will typically have problems with HFR dropping dramatically as I go out of focus, especially on the inner focus side, because all the large stars become donuts and aren't recognized as being stars.  I've tried to limit my step size and the autofocus offset steps, but even then it's occasionally a problem.  And even when the limited step size and offset steps work, it produces a very shallow curve (goes from just above 3 to just below 2). Would increasing the exposure time (currently 5 seconds) up to 20 seconds help "fill in" the center of the donut and have the larger stars be detected as a real star?  Any other suggestions to deal with the problem?  I could always keep an eye on the HFR during an imaging sequence and manually focus with my bahtinov mask, but I'm addicted to sleep. 

[Forward Scatter]

One has several options to enhance NINA's AF options tab to detect and measure the HFR as the donuts roll in:

1. Increase exposure time (as you mentioned).

2. 2X2 binning sometimes helps

3. Increase gain 

 

That said, on my Newty, it's 6 seconds, 400 gain on my 533, and 2X2 binning. The largest HFR tends to be ~10, but typically the AF run HFRs are between ~2 and ~5, with a nice smile parabola.

[jpengstrom]

One has several options to enhance NINA's AF options tab to detect and measure the HFR as the donuts roll in:

1. Increase exposure time (as you mentioned).

2. 2X2 binning sometimes helps

3. Increase gain 

 

That said, on my Newty, it's 6 seconds, 400 gain on my 533, and 2X2 binning. The largest HFR tends to be ~10, but typically the AF run HFRs are between ~2 and ~5, with a nice smile parabola.

Thanks Jay.  I looked for a gain setting in the autofocus options and didn't see it.  It's quite possible I just missed it - I'll go back an check.

[Forward Scatter]

Thanks Jay.  I looked for a gain setting in the autofocus options and didn't see it.  It's quite possible I just missed it - I'll go back an check.

Go to the camera options and set the default gain there. 

[jpengstrom]

Go to the camera options and set the default gain there. 

Gotcha - set a default value which will be used for autofocus and then override that gain for the camera exposure commands in the sequence.  Makes sense.

[kathyastro]

I am having similar problems.  Following.

 

Frequently, I will get a nice, if shallow, parabola, but NINA fails to recognize it.  It tends to go off into la-la land going farther and farther out of focus.  And if it gives up, it doesn't adjust:it just does the same thing all over again.

 

There's a lot to like about NINA, but SGP had better autofocus.

 

The next clear night, I am going to have to spend it calibrating the autofocus instead of taking pictures. 

[fmendes]

Where are you connecting the EAF? I connect mine to the reduced knob. With that I can use ~150 EAF steps in NINA to get a nice graph. If you're seeing donuts during autofocus, you may be going too far.

[italic]

Some info I gleaned but haven't been able to test yet:

 

George (Hocus Focus creator) did a stream talking about how HF works and gave some good advice for autofocus starting points.

 

Start at focus and note the focuser step position. Step out until your HFR is 3-4x the in-focus HFR. Note the step difference. Divide this number by 4 and put it into step size and make the number of initial offset steps 4. This means autofocus should be roughly starting at your largest HFR (at 4x what you expect), move through focus, then end up at the other maximum inside focus.

 

I haven't gotten to test this with my own newt yet (yay cloudy SF summer skies!) but it's the next thing I'm testing. My computer is doing other things at the moment and is having trouble with video, but he starts talking about HF at 16 minutes and there's a section about basic AF settings. I could be mis-remembering the suggestions given.

[jpengstrom]

I am having similar problems.  Following.

 

Frequently, I will get a nice, if shallow, parabola, but NINA fails to recognize it.  It tends to go off into la-la land going farther and farther out of focus.  And if it gives up, it doesn't adjust:it just does the same thing all over again.

 

There's a lot to like about NINA, but SGP had better autofocus.

 

The next clear night, I am going to have to spend it calibrating the autofocus instead of taking pictures. 

Kathy, I'm right there with you.  Although I do have a window Sunday night where clouds are predicted to move it shortly after midnight which means I can setup then spend a couple hours dialing in autofocus.  I'll post back with what I did if I get consistent successful autofocus runs.

 

Where are you connecting the EAF? I connect mine to the reduced knob. With that I can use ~150 EAF steps in NINA to get a nice graph. If you're seeing donuts during autofocus, you may be going too far.

I have mine connected to the course focus knob.  The problem doesn't seem to be step size being too large, it's just that the curve is very shallow and if the HFR won't increase more than about 2 - 2 1/2 times the lowest value without donuts appearing and throwing off the star detection.  I did reduce the step size and the offset steps to try and keep that from happening, and it mostly works but occasionally gives me a problem.  And the curve is very shallow.  If I can increase gain, lengthen exposure, and try binning and that prevents the donuts from appearing until the HFR is larger then my autofocus runs will be consistent.  That's what I'm trying to achieve.

 

Some info I gleaned but haven't been able to test yet:

 

George (Hocus Focus creator) did a stream talking about how HF works and gave some good advice for autofocus starting points.

 

Start at focus and note the focuser step position. Step out until your HFR is 3-4x the in-focus HFR. Note the step difference. Divide this number by 4 and put it into step size and make the number of initial offset steps 4. This means autofocus should be roughly starting at your largest HFR (at 4x what you expect), move through focus, then end up at the other maximum inside focus.

 

I haven't gotten to test this with my own newt yet (yay cloudy SF summer skies!) but it's the next thing I'm testing. My computer is doing other things at the moment and is having trouble with video, but he starts talking about HF at 16 minutes and there's a section about basic AF settings. I could be mis-remembering the suggestions given.

Yeah, that's basically the procedure I follow to setup autofocus and it works great on my refractor.  The problem is with my newt that I can't get to 3 - 4x the in-focus HFR.  I get donuts in my larger stars before getting to 3x the in-focus HFR and then all h*ll breaks loose and the donut stars aren't detected as being stars so the HFR drops off significantly because the large stars are no longer being used.

[andysea]

The same happens with my epsilons which are essentially fast Newtonians. I can't use the 4x HFR rule.  I reduced the step size by trial and error, now I get a fairly shallow curve but it always nails focus. I use hyperbolic as a fitting method. 
Also, it goes without saying, but make sure that backlash compensation Is setup correctly.

[jpengstrom]

The same happens with my epsilons which are essentially fast Newtonians. I can't use the 4x HFR rule.  I reduced the step size by trial and error, now I get a fairly shallow curve but it always nails focus. I use hyperbolic as a fitting method. 
Also, it goes without saying, but make sure that backlash compensation Is setup correctly.

Thanks Andy.  Good point on the backlash compensation.  With respect to that, I've set it to eliminate any flat part of the curve and then bumped it up quite a bit from there.  As near as I can tell the only problem with backlash being high is it take the focuser a little bit more time to overshoot and come back.  You're not aware of any other disadvantage to a higher than necessary backlash compensation, are you?

[andysea]

No disadvantages!  You can set the overshoot as large as you like as long as it's more than the actual backlash

[Rasfahan]

The same happens with my epsilons which are essentially fast Newtonians. I can't use the 4x HFR rule.  I reduced the step size by trial and error, now I get a fairly shallow curve but it always nails focus. I use hyperbolic as a fitting method. 
Also, it goes without saying, but make sure that backlash compensation Is setup correctly.

I second that. When using the 4x rule with my Epsilon the steps are too coarse to calculate focus reliably with 7 positions. Before I got the Leo the focuser in-travel was too small to even get to that point. Voyager automatically detects the necessary focus range. But even there I had to manually tweak parameters that I don‘t need to touch for my other scopes.

[ngc2218]

I was wondering if anyone has experience that could help me with a problem I'm having while autofocusing my 10" Quattro in NINA.  The basic problem is that HFR in NINA (using hocus focus for autofocus) doesn't reliably measure stars with donuts.  That's not surprising, but my focus range without producing donuts on my 10" newt is very narrow.  So my autofocus runs will typically have problems with HFR dropping dramatically as I go out of focus, especially on the inner focus side, because all the large stars become donuts and aren't recognized as being stars.  I've tried to limit my step size and the autofocus offset steps, but even then it's occasionally a problem.  And even when the limited step size and offset steps work, it produces a very shallow curve (goes from just above 3 to just below 2). Would increasing the exposure time (currently 5 seconds) up to 20 seconds help "fill in" the center of the donut and have the larger stars be detected as a real star?  Any other suggestions to deal with the problem?  I could always keep an eye on the HFR during an imaging sequence and manually focus with my bahtinov mask, but I'm addicted to sleep. 

Could be that the step size is wrong.

If I remember correctly it goes something like:

step_size_for_V_curve = 2 * CFZ/ focuserStep

 

and :

CFZ = 0.00225 * Seeing_FWHM * sqrt(allowed_error) * A * fratio^2

allowed_error for focusing 1-5%

A = telescope area

 

Reference:

http://www.goldastro...dfocus/ncfz.php

 

I use this formula with a Optec Leo, and never had problems with V curves

Edited by ngc2218, 10 August 2024 - 07:02 AM.

[fmendes]

I have mine connected to the course focus knob.  The problem doesn't seem to be step size being too large, it's just that the curve is very shallow and if the HFR won't increase more than about 2 - 2 1/2 times the lowest value without donuts appearing and throwing off the star detection.  I did reduce the step size and the offset steps to try and keep that from happening, and it mostly works but occasionally gives me a problem.  And the curve is very shallow.  If I can increase gain, lengthen exposure, and try binning and that prevents the donuts from appearing until the HFR is larger then my autofocus runs will be consistent.  That's what I'm trying to achieve.

 

NINA shows the curve with the Y axis scaled. If you look at one of my recent graphs, it will appear flat on a bigger scale. See that HFR actually increased very little until a curve was obtained in that Y scale. Note that I have the focuser connected to the fine knob, so 50 steps are about 1/4 turn of the fine knob.

 

 

Anyway, assuming that you're using Hocus Focus as star detection, you can try changing star sensitivity in Options>Imaging>Image Options>Advanced Settings>Star Sensitivity.

 

Another thing, how is your unguided tracking? Poor tracking may prevent HFR to go below certain value. If you use a guide scope, try guiding while focusing.