Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Has NINA removed backlash measuring?

  • Please log in to reply
17 replies to this topic

#1 unimatrix0

unimatrix0

    Viking 1

  • -----
  • topic starter
  • Posts: 806
  • Joined: 03 Jan 2021

Posted 29 March 2021 - 11:26 PM

I'm watching videos on how to get the backlash out of my autofocus, but I don't have that button? ("Measure Focuser Backlash") 
Was this removed or my NINA removed this or something weird going on with my version (1.10HF2) ? 

Or is it for specific focusers only?  The button supposed to be under the "Start Autofocus" as I'm watching the videos. I tried scrolling and stretching that window and there are no other buttons. 

Attached Thumbnails

  • backlash.JPG

Edited by unimatrix0, 29 March 2021 - 11:29 PM.


#2 jonnybravo0311

jonnybravo0311

    Vanguard

  • *****
  • Posts: 2,424
  • Joined: 05 Nov 2020
  • Loc: NJ, US

Posted 29 March 2021 - 11:49 PM

You imaging in orbit? That temperature of -127 sounds awfully cold :p

 

On topic - I don't have any "measure backlash" button, either. I'm on 1.11 nightly 065.


  • limeyx and unimatrix0 like this

#3 dghent

dghent

    Apollo

  • *****
  • Vendors
  • Posts: 1,200
  • Joined: 10 Jun 2007
  • Loc: Maryland

Posted 30 March 2021 - 12:00 AM

Backlash measurement was removed because its results were not reliably consistent. The better solution is to use the Overshoot backlash compensation method.


  • dswtan, limeyx and jonnybravo0311 like this

#4 unimatrix0

unimatrix0

    Viking 1

  • -----
  • topic starter
  • Posts: 806
  • Joined: 03 Jan 2021

Posted 30 March 2021 - 12:09 AM

You imaging in orbit? That temperature of -127 sounds awfully cold tongue2.gif

 

On topic - I don't have any "measure backlash" button, either. I'm on 1.11 nightly 065.

 

I don't have a thermostat so it's either Martian nights or Mercury afternoons! 
 

Ok, so I added 30 after I saw those 3 measurements not showing a change, 10 steps each. Looks ok to me now, although I could probably do a better job. 


  • jonnybravo0311 likes this

#5 Oyaji

Oyaji

    Ranger 4

  • *****
  • Posts: 347
  • Joined: 05 Sep 2018
  • Loc: Central Illinois, USA

Posted 30 March 2021 - 12:31 AM

Funny you should post this today.  I have been working on N.I.N.A. autofocus issues over the last couple of days and nights.  I have been using the autofocus for close to a year, but it never seemed to work right, and I finally decided to spend some time understanding why.  DUH.  

 

Yes, N.I.N.A.  has removed the backlash measurement tool.  The overshoot method works like a champ, as I discovered last night (3 scopes with motorized focusers). 

 

This topic is explained in a number of places on the net, but I found these to be most useful.  

 

https://www.youtube....h?v=XsjT-We9Wn0

https://www.youtube....?v=zfBFIXdujww 

https://nighttime-im...ashmeasurement/

https://nighttime-im...nced/autofocus/


Edited by Oyaji, 30 March 2021 - 12:38 AM.

  • mikefulb and limeyx like this

#6 limeyx

limeyx

    Surveyor 1

  • -----
  • Posts: 1,597
  • Joined: 23 May 2020
  • Loc: Seattle, WA

Posted 30 March 2021 - 01:07 AM

Overshoot is amazing, so simple & easy !



#7 Der_Pit

Der_Pit

    Vanguard

  • *****
  • Posts: 2,332
  • Joined: 07 Jul 2018
  • Loc: La Palma

Posted 30 March 2021 - 10:29 AM

Hmm, you can use the overshoot method to measure BL.  Do an AF run, then change the direction so that 'Focus inward' actually moves outwards.  The difference between the two is your BL.

 

I'm using both, a measured BL and overshoot focussing.  That way having BL spot-on isn't crucial, but you get better response if you do manual focus operations...



#8 Whichwayisnorth

Whichwayisnorth

    Gemini

  • *****
  • Posts: 3,345
  • Joined: 04 Jul 2011
  • Loc: Southern California

Posted 30 March 2021 - 12:09 PM

You can manually do it with a focus mask.

  Focus in one direction until it is perfect and take note of position. Then focus the opposite direction in small increments until the focus point moves.  Note that number.  That is the backlash in that direction.  Now overshoot the original perfect position by at least double the measured backlash and then focus the opposite way until the focus is perfect, note that position, then reverse focus my tiny amounts again until it moves.  You have your IN/OUT backlash numbers.

 

Some auto focus programs have a way to set both in and out backlash compensation.  I believe NINA is one of those.   Also a note about programs like Sharpcap that have a bahtinov focus mask analyzer built in.  That can be helpful in this process.



#9 Jinux

Jinux

    Messenger

  • *****
  • Posts: 480
  • Joined: 25 Jun 2007
  • Loc: Bay Area, CA

Posted 30 March 2021 - 12:41 PM

Overshoot is THE answer. 

I tried many other ways to measure absolute value of BL but none worked reliably but overshoot works.

Jared explained this to me on my CN thread below.

 

https://www.cloudyni...acklash-amount/

 

The thing is, NINA doesn't tell the user what is the HFR measurement result after final focus position measurement compared with estimated HFR, unless it's far worse than starting point and coming back to original, which actually never worked for me.

So, even after all these, it always left small anxiety whether BL is really cleared and the focus point is achieved.

My method of confirming this is using "Image Annotation" in "Imaging Option" menu. Enable this and start Autofocus while watching 1:1 image.

NINA displays HFR numbers on detected stars while Autofocus. When it's finishing, it moves to target point and measure again, so you can actually see the HFR values NINA is getting every and final step. 

Also, I takes several pictures with the same exposure as focus run and double check whether targeted HFR is really achieved. 

Of course, this is tested only during BL setting and when I'm present in front of the rigs. Once sequence starts, everything is automated and you're at the mercy of NINA algorithm and seeing.

 

One note. Overshoot and Absolute menu change is only permitted while focuser is "NOT" connected. Had to dig in NINA manual to figure this out.

 

Hope this helps.

 

-Jinux


  • limeyx and unimatrix0 like this

#10 unimatrix0

unimatrix0

    Viking 1

  • -----
  • topic starter
  • Posts: 806
  • Joined: 03 Jan 2021

Posted 01 April 2021 - 11:11 AM

Overshoot is THE answer. 

I tried many other ways to measure absolute value of BL but none worked reliably but overshoot works.

Jared explained this to me on my CN thread below.

 

https://www.cloudyni...acklash-amount/

 

The thing is, NINA doesn't tell the user what is the HFR measurement result after final focus position measurement compared with estimated HFR, unless it's far worse than starting point and coming back to original, which actually never worked for me.

So, even after all these, it always left small anxiety whether BL is really cleared and the focus point is achieved.

My method of confirming this is using "Image Annotation" in "Imaging Option" menu. Enable this and start Autofocus while watching 1:1 image.

NINA displays HFR numbers on detected stars while Autofocus. When it's finishing, it moves to target point and measure again, so you can actually see the HFR values NINA is getting every and final step. 

Also, I takes several pictures with the same exposure as focus run and double check whether targeted HFR is really achieved. 

Of course, this is tested only during BL setting and when I'm present in front of the rigs. Once sequence starts, everything is automated and you're at the mercy of NINA algorithm and seeing.

 

One note. Overshoot and Absolute menu change is only permitted while focuser is "NOT" connected. Had to dig in NINA manual to figure this out.

 

Hope this helps.

 

-Jinux

Great! I have read the thread, I sort of understand what's happening. Although - excuse my wording- not sure why my parabolic measuring graph (first post) has "t1ts" lol.gif  . If it scensored, I meant to say ****.  How did NINA find a closer/ better focused -  value 3 steps over in both directions-  from the best focus point (at the middle)  - if I'm reading my own graph correctly? 
Just a hard guess, that's caused by the backlash itself? 
This is what I was expecting  ,  Bad.->...tick ..tick ...tick -> Better-> tick...tick..tick -> Best <- tick..tick..tick.<- better..tick tick tick <- Bad
Instead I got:   Bad.....tick.tick..tick ---> good->..tick tick t -> Bad-> ..tick tick ->Bad .....-> Best<-...tick tick<- bad<-..tick tick <- Bad<-..tick tick <- Good<- tick tick <- Bad
 

Though I did add 30 value in both directions, the focus seems ok, but I still think it's not the best, but I will try the overshoot option also! 


Edited by unimatrix0, 01 April 2021 - 11:21 AM.


#11 Jinux

Jinux

    Messenger

  • *****
  • Posts: 480
  • Joined: 25 Jun 2007
  • Loc: Bay Area, CA

Posted 01 April 2021 - 01:00 PM

It's probably your variance of HFR due to seeing and measurement error is bigger than variance of HFR due to focus change, IMHO.

What I mean is, your "Y" scale change over focus run is merely 3.2 ~ 3.6 while the range of error that NINA displays with vertical bar is almost as big as your whole HFR range during focus run. So you're actually watching the star shape change by seeing on top of small focus change.

 

What I have found during my investigation was that NINA does NOT use the same star set for HFR measurement for each frame. You can limit this by changing number of brightest star options in NINA but it still doesn't reliably select the same stars for analysis. It selects "some stars" by algorithm and white dots are just average number and it shows upper and lower limit by error bar. 

 

What I would do (I was on the same boat) is, to increase "Auto Focus Step Size". This looks counter intuitive as it comes like the finer focuser moves, the better NINA finds the best focus. But it doesn't seem to. What we want to do is make 'sufficient' move on focus knob to find 'focus curve trend'. NINA will find the 'minimum spot' of the curve and will move focus to there, regardless of whether there was the measurement on the spot or not during focus run.

 

My focus run changes HFR by +/- 2~3 typically in the middle of imaging, for example. And I feel comfortable for NINA to choose the minimum value from the curve with sufficiently big autofocus step size. My worry was how NINA can move focuser back to the minimum point by overcoming BL. And Overshoot method does that better than Absolute way. 

 

I still have some suspicion on it as when NINA searches minimum in right (focus out) direction, it keep using Overshoot to overcome BL. But it usually ends up sub-optimal than finding minimum in left (focus in) direction. I suspected my overshoot is too small but even with big number it didn't change. I need to solve this puzzle sometime later.

 

-Jinux

Attached Thumbnails

  • post-355785-0-92150800-1617078336.jpg


#12 limeyx

limeyx

    Surveyor 1

  • -----
  • Posts: 1,597
  • Joined: 23 May 2020
  • Loc: Seattle, WA

Posted 02 April 2021 - 10:56 AM

Overshoot is THE answer. 

I tried many other ways to measure absolute value of BL but none worked reliably but overshoot works.

Jared explained this to me on my CN thread below.

 

https://www.cloudyni...acklash-amount/

 

The thing is, NINA doesn't tell the user what is the HFR measurement result after final focus position measurement compared with estimated HFR, unless it's far worse than starting point and coming back to original, which actually never worked for me.

So, even after all these, it always left small anxiety whether BL is really cleared and the focus point is achieved.

My method of confirming this is using "Image Annotation" in "Imaging Option" menu. Enable this and start Autofocus while watching 1:1 image.

NINA displays HFR numbers on detected stars while Autofocus. When it's finishing, it moves to target point and measure again, so you can actually see the HFR values NINA is getting every and final step. 

Also, I takes several pictures with the same exposure as focus run and double check whether targeted HFR is really achieved. 

Of course, this is tested only during BL setting and when I'm present in front of the rigs. Once sequence starts, everything is automated and you're at the mercy of NINA algorithm and seeing.

 

One note. Overshoot and Absolute menu change is only permitted while focuser is "NOT" connected. Had to dig in NINA manual to figure this out.

 

Hope this helps.

 

-Jinux

Autofocus does save a .json file each time it runs. The NINA discord server has a "autofocus" channel where you can upload your files and a bot will analyze them and give you advice for improvement (humans too sometimes)

 

Can be very helpful

 



#13 limeyx

limeyx

    Surveyor 1

  • -----
  • Posts: 1,597
  • Joined: 23 May 2020
  • Loc: Seattle, WA

Posted 02 April 2021 - 11:00 AM

It's probably your variance of HFR due to seeing and measurement error is bigger than variance of HFR due to focus change, IMHO.

What I mean is, your "Y" scale change over focus run is merely 3.2 ~ 3.6 while the range of error that NINA displays with vertical bar is almost as big as your whole HFR range during focus run. So you're actually watching the star shape change by seeing on top of small focus change.

 

What I have found during my investigation was that NINA does NOT use the same star set for HFR measurement for each frame. You can limit this by changing number of brightest star options in NINA but it still doesn't reliably select the same stars for analysis. It selects "some stars" by algorithm and white dots are just average number and it shows upper and lower limit by error bar. 

 

What I would do (I was on the same boat) is, to increase "Auto Focus Step Size". This looks counter intuitive as it comes like the finer focuser moves, the better NINA finds the best focus. But it doesn't seem to. What we want to do is make 'sufficient' move on focus knob to find 'focus curve trend'. NINA will find the 'minimum spot' of the curve and will move focus to there, regardless of whether there was the measurement on the spot or not during focus run.

 

My focus run changes HFR by +/- 2~3 typically in the middle of imaging, for example. And I feel comfortable for NINA to choose the minimum value from the curve with sufficiently big autofocus step size. My worry was how NINA can move focuser back to the minimum point by overcoming BL. And Overshoot method does that better than Absolute way. 

 

I still have some suspicion on it as when NINA searches minimum in right (focus out) direction, it keep using Overshoot to overcome BL. But it usually ends up sub-optimal than finding minimum in left (focus in) direction. I suspected my overshoot is too small but even with big number it didn't change. I need to solve this puzzle sometime later.

 

-Jinux

I'm curious about this "My focus run changes HFR by +/- 2~3 typically in the middle of imaging, for example"

 

Do you mean 0.2 / 0.3 HFR or really 2-3 steps ? 2-3HFR seems like a huge change.

 

It's pretty cold consistently here right now which I think is keeping my focus fairly stable

 

I focus every 30 mins or 2 degrees change (it almost always goes the 30 mins) and the focus position moves maybe 1-3 steps between focus runs which tells me I might want to do it every 45-60 mins in winter at least and get more imaging time back.

 

There is almost no HFR difference.

 

I do get cases where my actual HFR seems a bit too much higher than the calculated minimum one. Playing with step size made the NINA warning go away but I am still not sure if I can do something to improve my actual HFR and get it closer to the theoretical



#14 Jinux

Jinux

    Messenger

  • *****
  • Posts: 480
  • Joined: 25 Jun 2007
  • Loc: Bay Area, CA

Posted 02 April 2021 - 12:33 PM

I'm curious about this "My focus run changes HFR by +/- 2~3 typically in the middle of imaging, for example"

 

Do you mean 0.2 / 0.3 HFR or really 2-3 steps ? 2-3HFR seems like a huge change.

I "literally" mean HFR by 2~3. Specific range would be HFR 6 ~ 4 over ~10 focus steps. My focuser setting has 10 count per step but this doesn't mean anything as every electric focuser has different angle per each count and every telescope has different focus amount change per focus knob angle change.

In my case, this is specific to C11/0.63 reducer with Celestron Motor Focuser. 

 

My point being, finding best focus by going over every and very fine step is NOT the way NINA is after, IMHO, because due to the measurement error range and seeing variance per measurement, i.e. time varying.

What I'm after is to find the focus trend by stepping over around best focus and NINA algorithm seems designed for it. (I didn't design it so it's my best guess.) There are three types of focus trend you can use. Parabolic, Hyperbolic and Linear (trend line, a.k.a. linear regression). Critical focus point can't pop up somewhere in the middle of nowhere. It has specific trend per optical design. So, say if that is Hyperbolic, if you find nice hyperbolic fitting points over 10 measurements, then focus will be on the minimum point of the hyperbolic curve. And you don't have to find this point by going over every point around but by just curve fitting.

 

Going too small steps in narrow range may (or does, in my case) cause undesirable fittings due to bumps in the middle (like your case) caused by temporary seeing turbulence and/or measurement error and/or tracking error (during focus run, guiding is disabled and stars can be elongate) and/or whatever vibration/winds etc.

 

If you really don't like the large steps, then you can try increase "Auto Focus Initial Offset Steps". This will make NINA Step Size * Initial Offset Steps away from current position and search inwards. So, it will search wide range and make curve fitting better while you can keep fine step search at the cost of search time.

 

But just remember, NINA does NOT select minimum value it searched and measured as focus point but select the minimum point on curve fitting. So, there is really no point of going fine resolution around optimal focus point.

 

-Jinux



#15 limeyx

limeyx

    Surveyor 1

  • -----
  • Posts: 1,597
  • Joined: 23 May 2020
  • Loc: Seattle, WA

Posted 02 April 2021 - 12:48 PM

I "literally" mean HFR by 2~3. Specific range would be HFR 6 ~ 4 over ~10 focus steps. My focuser setting has 10 count per step but this doesn't mean anything as every electric focuser has different angle per each count and every telescope has different focus amount change per focus knob angle change.

In my case, this is specific to C11/0.63 reducer with Celestron Motor Focuser. 

 

My point being, finding best focus by going over every and very fine step is NOT the way NINA is after, IMHO, because due to the measurement error range and seeing variance per measurement, i.e. time varying.

What I'm after is to find the focus trend by stepping over around best focus and NINA algorithm seems designed for it. (I didn't design it so it's my best guess.) There are three types of focus trend you can use. Parabolic, Hyperbolic and Linear (trend line, a.k.a. linear regression). Critical focus point can't pop up somewhere in the middle of nowhere. It has specific trend per optical design. So, say if that is Hyperbolic, if you find nice hyperbolic fitting points over 10 measurements, then focus will be on the minimum point of the hyperbolic curve. And you don't have to find this point by going over every point around but by just curve fitting.

 

Going too small steps in narrow range may (or does, in my case) cause undesirable fittings due to bumps in the middle (like your case) caused by temporary seeing turbulence and/or measurement error and/or tracking error (during focus run, guiding is disabled and stars can be elongate) and/or whatever vibration/winds etc.

 

If you really don't like the large steps, then you can try increase "Auto Focus Initial Offset Steps". This will make NINA Step Size * Initial Offset Steps away from current position and search inwards. So, it will search wide range and make curve fitting better while you can keep fine step search at the cost of search time.

 

But just remember, NINA does NOT select minimum value it searched and measured as focus point but select the minimum point on curve fitting. So, there is really no point of going fine resolution around optimal focus point.

 

-Jinux

Wow OK. Seems you must need to focus very often then ? Is it due to temp drop or how your scope works ? My HFR generally sits around 2.9-3.2, but I dont see a huge change in it after focusing

 

The variance I think is just the slop in the Sky Guider mechanics, not the temp changing - just based on it only moving a couple of steps. I don't know how much distance 2 steps is but I would guess a fraction of a mm (I have the ZWO EAF)



#16 Lead_Weight

Lead_Weight

    Vanguard

  • *****
  • Posts: 2,276
  • Joined: 21 Jul 2016
  • Loc: Houston

Posted 02 April 2021 - 01:48 PM

One other setting in NINA lets you average out the seeing fluctuations by taking multiple HFR readings with multiple shots at each focus point. There's a setting for how many shots to take at each focus point. I usually have it set to three. It takes three shots, and averages the HFR measurement, then uses that average as the HFR reading for that focus point. This helps with big up/down jumps like you're seeing in the V curve.


  • limeyx likes this

#17 jonnybravo0311

jonnybravo0311

    Vanguard

  • *****
  • Posts: 2,424
  • Joined: 05 Nov 2020
  • Loc: NJ, US

Posted 02 April 2021 - 02:56 PM

And to add to what was just written, you can also set a ROI in the center of the frame... so if your back focus distance isn't quite right, or you've got some coma at the edges, you can use the AF Inner Crop Ratio. Set it to something like 0.8 to make a rectangle 80% the size of your full image (same center point as the full image). Values are from 0 to 1.


  • limeyx likes this

#18 starbuckin

starbuckin

    Ranger 4

  • *****
  • Posts: 391
  • Joined: 03 Mar 2019
  • Loc: Massachusetts

Posted 23 April 2021 - 10:11 PM

In my focuser config, the Backlash Compensation method is greyed out. Is the overshoot method not part of the released version




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics