Getting to grips with focuser backlash

[MG01]

I've been trying to get my head around focuser backlash, why it's a problem and what to do about it.

I'm still a little confused an so I decided to draw up my understanding of the problem and ask you all for your input.

As far as I understand, the problem is that (using the example above) movements OUT will always result in B having an undetermined exact position where after A has finished moving, B could move further (creep) over time up to the amount of the backlash (10 steps in this case).

Movements IN that are smaller than the backlash would also potentially be subject to similar creep. 

The amount of creep would be dependant upon numerous factors, e.g. the friction in the system, weight of imaging kit attached (wheels, cameras OAG etc) and the angle of scope...and as such is difficult to measure precisely.  But the maximum creep would be equal to the backlash.  

Obviously the solution would be to always make a final move inward that is greater than the backlash. 

But how do you accurately measure the backlash?

As far as I can tell from reading a number of other posts and videos it's not possible to get an exact number but you can get pretty close and then apply the good old engineering safety factor of x1.5 to give a number of steps to make your final inward move.

My focuser has a maximum extension of 42000 steps and I normally get focus around 26400.  So this is what I did:

First I established the amount of focuser travel per step

1. Go OUT max position at 42,000, then return IN to position 40,000. This should ensure gears engaged.
2. Measure focuser extension with digital calliper (A=105.66mm)
3. Move IN 30,000 steps to position 10,000.
4. Measure extension again (B=26.41mm)
5. We now know that 30,000 steps equals Measurement A - Measurement B. (105.66-26.41=79.25mm)
6. So focuser travel per step = 79.25mm/30,000 = 0.0026416667mm or 2.642um per step

Then to approximate backlash

1. Go OUT halfway to position 21,000, then return IN to position 10,000.  Gears should be engaged.
2. Measure focuser extension with digital callipers (C=26.42mm) notice slight difference caused by inaccuracies of measurements. NOTE: Calliper accuracy of +/-0.01mm equates to approx. 4 steps of focuser, so chasing perfection is going to be very difficult!!
3. Move OUT 1000 steps to position 11,000. We should now have moved 1000 steps minus the backlash.
4. Measure extension again (D=28.96mm)
5. Measurement D - Measurement C gives us the Measured travel for 1000 steps. (28.96-26.42=2.54mm)
6. Calculated 1000 steps should be 1000 x travel per step calculated above = 2.642mm
7. Backlash equals 2.642-2.54mm = 0.102mm which when divided by travel per step gives backlash of 38.6 or approximately 39 steps

SO if I set a final inward travel of 1.5x39 = 58.5, lets say 60 steps I should have eliminated backlash right?

Apologies if this ended up a bit of a brain dump...but hopefully that makes sense and if I'm way off the mark then someone will be able help set me back on course.

Cheers!

 

[Tomatobro]

I never worry about backlash. The image is either in focus or it isn't. I would never just go on the focuser reported position as an indicator of image quality.

[tomato]

In NINA and I guess other image capture software, the routine moves to the in focus position to take the final image and so will always need to reverse direction. I always visually check this image although it is rarely a problem.

Likewise I never rely on filter/temperature focus offsets, just like with guiding performance, the image is the acid test.

[MG01]

3 hours ago, tomato said:

In NINA and I guess other image capture software, the routine moves to the in focus position to take the final image and so will always need to reverse direction. I always visually check this image although it is rarely a problem.

Likewise I never rely on filter/temperature focus offsets, just like with guiding performance, the image is the acid test.

I've not tried NINA yet properly as I've been spending my time getting to grips with APT.  

I've generally used a bahtinov mask for focusing, but more recently have been using whole image HFR making use of ASTAP.  This led me down the path of auto-focus and then I started looking at the settings and the various options for backlash compensation and measurement and my head started to hurt so I had to try and work things out so I could understand it.

As far as I could tell APT uses a similar process to NINA and SGP for auto-focus in that it analyses the image and plots HFR or FWHM against focuser position to generate a v-curve to which an algorithm is applied to find a best fit and establish the focus point.

But so that these focuser positions are correct it requires some form of backlash compensation to be applied.

I've gone down the road of applying a final inward move rather than a specific backlash compensation because I'm not convinced that backlash remains the same at all positions due to minor manufacturing differences in gear train components and shifts in system weight etc.

I agree tha filter and temp offsets also feel unreliable and would much rather just set to auto focus after a filterchange and on temperature change....to what degree of temperature change I'm not yet experienced to know what works best as I'm still dabbling.

I'm assuming you couldn't find any glaring errors in my assumptions regarding what backlash is and how backlash affects things?  I got myself so tangled in understanded the autofocus routines that I started to question if I'd even got these basics correct!!  At one point I was even questioning the significance of backlash when we're aiming for a critical focus zone that's 50+ steps wide....but I guess every little helps

[MG01]

4 hours ago, Tomatobro said:

I never worry about backlash. The image is either in focus or it isn't. I would never just go on the focuser reported position as an indicator of image quality.

I don't think the focuser saying it is at a focus point suggests anything, but how do you assess image quality?  I would have thought HFR or FWHM analysis of an image is a pretty good indicator and certainly to my eye seems to be  repeatable and more precise than I've managed with just a bahtinov mask.

[900SL]

The way I do it is I blue tack a 4" piece of bamboo skewer to the fine focuser to act as a visual indicator after having moved the focuser out a short distance to clear backlash. I then reverse direction using the ASI air EAF control 10 steps at a time, until I see the pointer move. On my TS CF APO it's 40 steps, on my WO GT71 it's around 60

[MG01]

1 hour ago, 900SL said:

The way I do it is I blue tack a 4" piece of bamboo skewer to the fine focuser to act as a visual indicator after having moved the focuser out a short distance to clear backlash. I then reverse direction using the ASI air EAF control 10 steps at a time, until I see the pointer move. On my TS CF APO it's 40 steps, on my WO GT71 it's around 60

Love it....a low tech solution to calibrating our high tech gear!

I still think it's incredible how tiny the focuser steps are...we're super lucky to be using such precision equipment, even if it doesn't feel like it some nights!

[Starflyer]

I think you're over thinking the problem.  You don't need to compensate for the exact amount of backlash, any amount of compensation over is fine, any amount under will show as a dogleg in the first part of the U curve.

If you want to get it close to save a second or two on each focus run then decrease it until you start seeing a dogleg and then add a decent percentage to cover any variation.

Example of backlash, taken from the N.I.N.A page.

Edited February 19, 2023 by Starflyer

[MG01]

That makes sense starflyer.

[tomato]

A lot will depend on if you intend running unattended sessions. If you are there, you can always approach the focus point from the same direction and hence take care of any backlash that way, but I don't know if you can make that happen automatically from the focus routine, I will need to check if it is possible in NINA.

I use the HDR numbers rather than the visual image, with my ageing eyesight the focus can drift quite a bit before I see it show up in the sub image.

[MG01]

In APT there's a setting to make a final inward move. If the focuser is moved outward by you or the auto routine, it will always move out the additional number of steps specified before returning in by the same number of steps.

I'm assuming NINA has something similar. 

[MG01]

@Yoddha Hi Ivo, apologies for cross posting from the APT Forum, but I couldn't attach my diagram (in post #1) to the thread over there!

I'm still struggling to understand how backlash can be cumulative, which is why I tried to draw it out.  When using the APT backlash aid I'm getting vastly different results depending upon the number of iterations.  I would have thought that any inward move that is greater than the backlash will effectively cancel out the backlash of that move.  Obviously I'm missing something here as you and the team are the experts, but I'm really keen to understand better and hope you don't mind helping when you get the time....no rush though as I know you're always super busy on the next release! 

Many thanks! 

Michael

[Yoddha]

Hello Michael,

Somehow didn't get email for the mentioning  

Yesterday I had a talk in AstroWorldTV that covers this topic  Hope you will find it useful -