Some guidance with AF for NB Filters Please

I’m still struggling to get autofocus to work with NB filters. The L, R, G and B filters all produce a reasonable graph with a focuser setting of 38411. When I start from that value and try any of Ha, OIII or SII the graphs are like craggy mountains and the routine ends with “too many range shifts”. These are the settings that work OK for the LRGB filters:
Backlash = 1200 IN, Step size = 55, Data Points = 9, Min star size = 4
The exposure durations I have set are:
L = 6 sec, RGB = 10 sec, Ha = 15 sec, OIII = 15 sec and SII = 20 sec.
I’m using a Celestron 8" Edge HD with a Celestron motor focuser. I had assumed that as things worked for the LRGB then the only thing I would need to change would be the exposure durations for the NB, but that doesn’t seem to help.

The short answer here is that AF does not work well with NB filters. These filters are just too restrictive. What people do here is determine the difference in focus between LUM and each NB filter. Then direct auto focus to go through the routine with the LUM filter, then apply the filter offset.

More here:

After you have this data set, make sure you you tell AF to perform its duties with the LUM filter (in the AF settings dialog)

1 Like

Ken’s advice is, of course, right on. NB filters can be very hard to auto focus because they just are not gathering nearly as much light as LRGB do. You should set your exposure lengths for them even higher than you have. Maybe several times higher like 60 seconds each. Clearly you are not going to want to focus with them on a regular basis.

Spend at least one night just doing autofocus runs on all your filters. Then plot each filter versus temperature and find the least fit line. Choose a temperature midway between the warmest and coldest of the temperatures you had that night. Given the formula for each filter’s line, calculate what focuser position this one temperature would give you for each filter. Calculate the offsets from L. Now you always focus with the L filter, and SGP applies these offsets you have come up with as each different filter is loaded.

Why do you choose a temperature midway between the high and low temperatures of the night? Rarely are all the T vs Position lines for the various filters going to have the exact same slope. Picking a midpoint temperature at which to compare them minimizes this variable factor.

The more focus points for each filter that you can provide, the more accurate your offsets are going to be. If your focuser gives the same focus positions for given temperatures on different nights, you can lump several nights focus runs together, to improve your accuracy.


Thank you, that’s very helpful. I was wondering how to proceed, now I know :-):smiley:

I am relatively inexperienced, but I have been having good results autofocusing NB filters repeatedly during the night. My NB filters are ZWO and 7nm. At 530mm f5 with ZWO 1600mm camera with unity gain 131 and 9 sec exposures I have no problems getting autofocus to work and well. Always plenty of stars at least for the wide field nebula I’ve been imaging. I can imagine a narrower bandwidth filter like 3nm would present much more of a problem as well as higher focal ratios and longer focal lengths with narrower fields of view.

1 Like

I would echo Bob_P. I have been having excellent focusing results using ZWO NB filters (Ha, SII and OIII). I use a 826mm fl refractor and a 2350mm fl SC. If you want to focus using NB filters I would recommend using binning options and setting the gain for the binning option high eg:

Control Panel -> Camera -> Bin 2x2 = 200
Control Panel -> Focus -> Use Auto Focus -> Set Options -> Bin 2x2

This gives independent gain settings for imaging and focusing if you are imaging at Bin 1x1.

I also set up offsets for my filters but they are just used as a starting point for the AF routine so that my non parafocal filters do not throw the AF starting position too far from the optimal focus position.

It is also important to determine optimum step size and back lash settings for your system.

Since starting using the new curve fitting AF routine I can honestly say that I have not had a bad focus result in the 5 overnight sessions that I have been lucky enough to experience @ 9 focus runs per night.



1 Like

I too would echo Bob-P.
I have a Celestron EDGE 8" with 0.7 FR and a ZWO ASI1600MM with Celestron motorized focuser.

With 2x2 binning and 20 sec exposure, my Baader NB filters focus well.
They are almost parfocal but I record the positions when focused. They change a bit over the night (e.g. 5C to -5C right now), but generally the focusing change is minimal. I can then update the Filter settings in SGPro as required.

Dmert2016 - How do you measure the backlash of the focuser? I’ve determined the optimum step size for my system, but not sure how to measure the back lash on the focuser.

1 Like

Hi David,

My optimum back lash setting is enough so that back lash isn’t a problem for the AF routine but not so much that the focuser overshoots too far, wasting time. Basically I used trial and error.

As you probably know SGPro AF routine only requires back lash compensation parameter to be greater than the actual back lash in the AF system. First I made sure that the AF sequence was always running the focuser against gravity which requires a reversal in the stepper motor direction when switching between my refractor and the SCT. Luckily this is handled easily in the focuser driver I do not change my SGPro settings for this, I use Rigel n-step system. Once the correct focuser direction has been set I started off in focus and set an arbitrary back lash parameter. If the RHS of the “V” curve exhibited a shoulder or flattening I would double the back lash parameter and re-run AF (making sure that I again started off in focus). If the RHS still exhibited the shoulder I would again double the back lash and re-run AF. However if the “V” curve was good I would set the back lash parameter to halfway between the the current setting and the previous setting and re-run AF. If the shoulder or flattening came back I chose the previous back lash setting as my parameter, if the shoulder did not come back I just used the current back lash setting as my parameter. Having set the back lash parameter this way I monitored subsequent AF runs for signs of repeated shouldering If I see any then I just increase the back lash settings by 25%. Not very scientific really but I seem to have very consistent AF runs that don’t take forever to run with my NB filters.

I think there is a way to calculate the back lash settings from the slope of the “shoulder” but I never bothered to try and get so precise since SGPro doesn’t need it and I expect my relatively inexpensive focus systems will have variable back lash over time.

Two other points I am using 9 AF data points to generate my curves and I have smart focus switched off.



1 Like

By the way here are the plots of 10 AF runs from last night using my AF127. I am generating a 6 panel mosaic of the Heart Nebula. All these AF runs are using the ZWO Ha filter. As you can see all the curves are really good. My focuser on the AR127 seems to like a different position after the meridian flip but (AF run 7) the AF routine handles it well.




1 Like

Once again guys thanks for all the information, you are giving me lots to work on (whenever the clouds let me).
The step size and backlash settings I have all seem to work well for the LRGB filters, so I assume that those settings should be the same for the HOS (?).
Everything has been running so well with SGP that I haven’t upgraded recently, I’m on I suppose that I should come up to date.
Derek, can those graphs be generated using the latest SGP version?

Hi Robin,

These curves were generated by a program that was written by an SGPro user - Mikael
He very kindly shared his program and provided a download link in the forum thread called SGP Log Viewer (I think). It was posted a while back and I think the author was also involved in helping debug the new AF curve fitting routine. His latest version will also provide temperature coefficients for your filters if you have multiple runs over a range of temperatures using the same filter.

I am using Beta 313 of SGPro at the moment so I am a little behind on the betas. It is hard to change when things are running so well but I should update and help with Beta testing … better to find any bugs now than later.

Good luck with your NB focusing.



just adding to the chorus - with an 8300M at bin 2x2 and astrodon 5nm filters, using 20-30s focus exposures SGP’s autofocus works great. it is slow though. i would be better served by dialing in my filter offsets and using G or L instead but i am just far too lazy.

It’ some time since I raised this question but have been put of by having to dedicate a whole clear night to focusing runs. We get so few clear nights (only 1 or 2 mer month recently) that I’d rather try to wing the focusing and get on with imaging.
But sadly the results have been rubbish so I’m going to have to get to grips with it.
My desire is to get accurate offsets for the NB filters. Is it essential to have temperature data to achieve this?
My Celestron motor focuser doesn’t seem to have any capability to record temperature.
I’ve found a device called MBox that operates stand-alone and integrates with SGP via ASCOM. Does this seem suitable?

Also some guidance on how to find a least fit line and so calculate the offsets from L would be appreciated.

Please ignore my last question. My maths knowedge is pretty weak, its nearly 60 years since I did any at school. Anyway, I’ve been doing a bit of on-line research on linear regression where
y = bx+a and think I now understand and can see that temperature is essential.
So now I need a temperature sensor and a clear night where there is a temperature change. The last time I was out from 7.00pm to 1.00am the temperature (from a radio controlled clock) was a pretty constant 12 degrees C.
Would a couple of degrees temperature change be sufficient?

Actually there is a bit of confusion here. There are 2 properties related to focusing that it is very useful to determine.

  1. Temperature Coefficient - this is the amount of focus shift that occurs for each degree of temperature change. Example: 5, for each degree C that the temperature drops, focus will decrease by 5 steps.
  2. Focus Offset for each filter versus L - this is the most common focus offset that is used. Example: Ha is -37 steps relative to L.

These two are completely independent of each other. Temperature Coefficient is determined by your optics and focuser gearing. Focus Offset in determined by the optical properties of your filters. They have virtually nothing to do with each other.

In practice you can determine both of them with a single series of focus runs, as long as the temperature changes a fair amount from the beginning to the end of the focus run. Running all night long doing nothing but continuous focusing is the best way to do this. Then just use the great program by MichaelA to give you both of these values automatically.

However, if you are only looking to determine the Filter Offsets, you don’t need any temperature change whatsoever. In fact, having 0 temperature change over the course of your focus runs is ideal. The reason is you don’t have to worry about how your focus changed because the temperature changed. The only factor that is making your focus change is the change in filters.

“Actually there is a bit of confusion here.” That’s a very polite understatement. :grinning: I never realised there was so much to focusing. I thought that once I had the focus motor everything would be sorted.
I’m now determined to do everything thoroughly so I’ve ordered a temperature sensor and am praying for a clear night with a significant temperature change.
Thank you once again for your help, you’ve been very generous with your advice,