SGPro Focus settings with Lakeside Focuser

[Uplooker]

Hi Everyone,

I would like some advice, if possible,  regarding the focus settings I should be using with SGPro and a Lakeside Focuser.

I am struggling to achieve good focus and therefore unable to get auto focusing setup.

The issue I am having is trying to understand what Step Size should I choose within Lakeside Utilities and then what Fine and Course settings should I choose within SGPro

Currently I have chosen a Step Size of 20 a Fine Focus of 5 and Course Focus of 20. This results in me having a Maximum Increment of 1255. I am very close to achieving good focus with a focus position of 463. I did manage to get the start of a "V" curve.

It is not made any easier by the limited time  and the long intervals I seem to get to be able to sort it.

It may well be that I just need to persevere, I get that, but I have been trying for months now

I know that I have not given a lot of information to go on. Please let me know anything you need clarification on.

Thank you in advance for any help given.

[kirkster501]

I feel your pain, so few clear nights to get stuff up and running.

Nearly every issue can be explained with the step size and backlash.  There is a routine you run through to determine this step size explained in the software help system.  The long and short is that when the focuser goes to its maximum "out" position, you should be between 3 to 5 times the "in focus" position (that you determined with a Bahtinov mask). Similarly, when the focuser is at the maximum "IN" position it should also be 3-5 tines the "in-focus" position. Also, set the backlash to a 2-3 times the step size in the IN direction.  Many times, people do not set the step size high enough and so not enough adequately spaced samples to get an accurate line intersection of the two arms of the "V" to determine the true focus position at that temperature.

Once this all clicks, it all works like magic and is a sight to behold it working out the perfect focus - almost magical....

[symmetal]

The 'Fine' and 'Coarse' settings are only used when you are focusing 'manually' with SGPro and move the focuser in or out by those amounts. They have no affect on autofocusing. You just need to increase your step size to 25 or 30 or maybe even more and see what happens and you'll get the characteristic V shape appearing. The default 'Auto Focus Data Points' at 9 is normally optimum. It doesn't matter that the 'step size' is significantly larger than your 'Fine' focus setting as when a decent V curve is plotted it will interpolate the results to determine the optimum focuser value. This most likely will be a value between your sampled step size values. As kirkster says choose the IN direction for your backlash setting, as then the focuser is always pulling the focuser up against gravity to achieve focus position so maintains gear meshing. As long as the backlash value is greater than your actual backlash is all that matters. If the backlash setting is say 30 and it needs to focus out by say 20 it will focus out by 50 and then in by 30. I have mine set to 40 which is probably more than it needs to be but it doesn't matter. 

Lakeside utilities is used to obtain an optimum 'Fine' focus setting where a 'Fine' movement is a 'just noticeable difference' in focus. For autofocusing you want the step size to be a significant difference in focus.

Hope this helps 

Alan

[Uplooker]

Thank you both for the replies.

Alan - I did not realise that the Fine and Course setting had no impact on Auto focusing.

I was using manual focusing so that I could get to a place that was close to focused as I thought you had to be quite close to focused so that Autofocus would do it thing.

I will certainly be having another few tries at getting this sorted, just not sure when as the forecast is dire for the rest of this week.

Thank you again

[symmetal]

Ian, You're right that you need to be roughly near focus before autofocus can work. I have the 'Course' setting at 100 to quickly nudge it near focus before running autofocus. Having set your backlash direction to 'IN' when you run autofocus it will focus out by (0.5 x AutoFocus Data Points x Step Size) + backlash value, then focus in by 'backlash value' to eliminate backlash. It then takes a series of images focusing in by 'step size' each time and expects the focus to initially get better and then get worse as it passes though correct focus generating the V curve.

If it initially gets worse because your starting focus position was a long way 'in' from correct focus, it will do a major focus 'out' jump before starting again. If your 'step size' is too small and it's approaching focus, the variations in 'seeing' can make the focus appear to get slightly worse for a sample or two and it decides to do a major focus out jump and things go rapidly downhill from there. Starting out with a 'step size' which is too big is much better than one which is too small as it's more obvious what's happening.

The 9 steps of 'Autofocus Data Points' will be automatically increased if autofocus thinks it has just gone through focus but hasn't done enough samples to generate the second half of the V curve.

Alan

[gnomus]

Yes.  The values shown on the interface are not what is used for autofocus.  The first thing to do is to ensure that you are using backlash compensation and set this at a healthy value.  I learned this the hard way!

The SGP manual is very good on how to set step size.  There are two methods described.  I have tried both.  The best method, in my opinion, is the easiest.  Get good focus using a Bahtinov.  Get SGP to calculate the HFR and note the stepper position.  (Let us say it is 1.00.)  Now rack out the focuser in stages and keep checking HFR until it becomes 3-5x the value of the in-focus HFR (3-5 in our example).  Note the stepper position again.  You want the step size to be such that it will take you out to the 3-5x HFR at the start of the routine.  I use a 9 point model, so ideally you want the in-focus to be position to be around step 5.  So, if your in-focus stepper position was - say - 1000, and your 3-5x position was 1200, then set your step size to be 200 divided by 4 or 5 (the outward number of steps) - 45 would be a good starting value.  You can always fine tune as you get experience.  

Another great tip that I got from Barry Wilson just a couple of nights ago - your focuser will tend to move inwards as the night wears on due to it getting colder.  On some scopes, it may shift so much that when you start up the next night, the autofocus racking out the 4 steps will actually put the HFR near the bottom of the V curve.  SGP can take care of this itself, but it greatly prolongs the autofocus routine.  So why not start the night by moving the focuser outwards a coarse number of steps?  It doesn't need to be precise, but it will depend very much on your scope and focuser.  So my WO Star 71 and Lakeside needs to be moved out 50 steps, whereas my Esprit 120, Feathertouch combo requires 200.

 

[Uplooker]

Hi,

I did manage to have another go last night, all to no avail. I tried a number of different step sizes from 20 to 50. I was already using backlash compensation but increased this from 15 to 50.

When I use a Bhatinov mask I am not getting the stereotypical cross with the line through it. The image is just not distinct, and no matter how small an adjustment I make the horizontal line would move well away from the crux. Not sure what is causing this. I am familiar with using a bhatinov mask for focus.

I believe I am able to get "close" to focus, enough to be able to Platesolve, after which I centre a particular star. I then try and run Autofocusing. It does do its thing and occasionally I would get an HFD reading of 1. It would then continue way past the point where I believe “good” is. It would then step out a mile and start going again. It seemed to step the focuser out so far that it would appear impossible for it to get a star anywhere close to focused. It continues until it displays the message “Autofocus failed”

Because when you start Autofocus at or near to focus, the autofocus routine steps out and then starts to step in as it tries to construct the V curve I placed the focuser at a point inside of focus so that when it stepped out as part of the autofocus routine it would be starting just outside of focus. My thinking was that it would step to or close the actual focused point. This didn’t work either.

I will keep trying. A bit tired today as I was out until 0115hrs – on a school night!!

[gnomus]

My backlash compensation is set at 100 (you could even try 200) - it does not need to be precise.

I find the main reason that I get indistinct Bahtinov patterns is if the star is not bright enough - or if I am using a NB filter.  Slew to a bright star for this purpose.  You won't need to do this once autofocus is all set up - you are only doing this to get the correct settings.  And use a Lum filter.  If you still have problems with the Bahtinov, just focus until it looks good enough on screen.  It needs to be quite far out of focus for the routine to fail.

The focus routine is supposed to run from well out of focus one way, through focus and then well out of focus the other way.  That is how it estimates the slope lines.  If you start from near enough good focus, then you should get a 'V' shaped, or U-shaped curve.  Do you get this?  If so what is the HFR being reported on the graph at the extreme end of the curve?  It should be around the 4-5 mark. 

2 hours ago, Uplooker said:

....Because when you start Autofocus at or near to focus, the autofocus routine steps out and then starts to step in as it tries to construct the V curve I placed the focuser at a point inside of focus so that when it stepped out as part of the autofocus routine it would be starting just outside of focus. My thinking was that it would step to or close the actual focused point. This didn’t work either....

I am not sure what this means.  I would strongly suggest following the routine suggested in the manual, which is, roughly speaking, what I said earlier - you want the step size to be such that it will take you out to where the first reading is 3-5x the HFR reported when you achieved good focus with the Bahtinov - NB this is likely to be around '3' to '5' on the Y-Axis of the graph.  It will then come through good focus (probaby a little under 1.0), and then it will continue racking in until it gets to around the 3-5x point again.  SGP then estimates where the point of best focus is, returns the focuser to that point and takes a final frame so that it can confirm good focus and give a final notification of the HFR achieved at that point.  

[symmetal]

Ian, How many stars are shown labelled when it displays the image on screen after each autofocus step. It needs a fair number to get an average HFR reading it can work with. I usually get over 100 stars on each frame when it's approaching focus and around 30 or so when it's furthest out of focus. Is your focus exposure setting set too low that not enough stars are being detected. On the autofocus 'Settings'  panel at the top of the 'Options' pane is the exposure times for filters or OSC. I bin at 2x2 for autofocus and use 3s for Luminance, 5s for R & G and 8s for B. This is for my FLT-98 at f5 and Atik One. For a slower scope or less sensitive camera these will need to be increased of course.

It's also important not to overexpose too many stars as this can lead to an increase in actual HFR reading when approaching focus as the stars start bloating. This tends to show as the V curve becoming flat at the bottom (more of a U shape) as star images near focus give similar HFR values. I have sometimes dropped the L exposure to 2s for a field of bright stars when I get a U shape graph at 3s but my default settings work for most of the time.

When your autofocus seems to be doing something giving HFR values does the graph plot have any indication of any side of the V shape. If it can't find a discernible slope autofocus may do a large jump to start again or just give up with an autofocus error.

Alan

[Uplooker]

Thank you guys for the support. You have given me plenty to go on. I will make this a short reply as I plan on trying again tonight. 

Two points thought:

gnomes, what step size are you using when you have the backlash set to 100?

Alan,

I have increased the exposure to 10s, yet it seems to only take a 1 or 2 sec exposure during Autofocus.

I will update later.

Thank you again for the assistance.

For information I using an APM 107/700 and QHY9 mono with TS 2 flattener and using a lum filter.

Wish me luck, I will need it as I was up late last night. Two late nights during a work week are going to kill me?

[gnomus]

I think my step size is around 50, but step size and backlash are completely independent.  

[symmetal]

There's a bit of a typo but did you say it only takes a 1-2s exposure even though you set it to 10. Are you filling in the right exposure settings panel. The exposure setting at top right of the autofocus options panel is for OSC. To set the exposures for filters you have to click on the 'For Filters' button and fill in the 'AF Expose' boxes on the 'Filter settings' panel which are in seconds. The 'Focus Pt' values are only used if you have the 'Auto adjust focus per filter' option selected on the Focus tab in the 'Control Panel'. I used to have this set to avoid refocusing on a filter change but I found it wasn't reliable, so I always refocus on a filter change. If the 'AF Expose' values haven't been filled in it might use the OSC value but I'm not sure.

Hope you have more success tonight.

Alan

[Uplooker]

2 minutes ago, symmetal said:

There's a bit of a typo but did you say it only takes a 1-2s exposure even though you set it to 10. Are you filling in the right exposure settings panel. The exposure setting at top right of the autofocus options panel is for OSC. To set the exposures for filters you have to click on the 'For Filters' button and fill in the 'AF Expose' boxes on the 'Filter settings' panel which are in seconds. The 'Focus Pt' values are only used if you have the 'Auto adjust focus per filter' option selected on the Focus tab in the 'Control Panel'. I used to have this set to avoid refocusing on a filter change but I found it wasn't reliable, so I always refocus on a filter change. If the 'AF Expose' values haven't been filled in it might use the OSC value but I'm not sure.

Hope you have more success tonight.

Alan

Hi Alan,

I think you may have hit on something there regarding setting the exposure in the correct boxes for filters.

Unfortunately, even though it is a beautifully clear night here I did get out again as friends came round and it was too late by the time they left, on top of the fact that I am just too tired from last night.

 

[Uplooker]

Alan,

Your post sparked a number of thoughts and it resulted in me realising that I had omitted to add a load of information regarding, but not limited to scope settling, filter exposure times andI now have a better understanding of the relationships between step size and focusing.

Steve (gnomes), you have also been invaluable in helping me sort things out.

It was clear again tonight, unfortunately I encountered any number if problems that ranged  from no platesolving not working, focuser continually disconnecting and a multitude of other problems- stemming from software and user ineptitude.

I have not given up yet and will persevere.

Thank you once again Steve and Alan, I would not have got this far without your assistance and advice. Not there yet, but an awful lot closer ?

[symmetal]

No problem Ian, happy to help. 

Here's a couple of screen shots showing what my autofocus curve looks like when it finished. Note the number of stars detected in the image. Sometimes you get a definite V but often times it has a flattish bottom. The slopes of the sides point to correct focus anyway.

As  a test I disabled the Backlash compensation and I got this which I suspected would happen. Between step 1 and 2 of the curve (top right) the 20 focuser steps were just taking up the backlash, hence the flat curve segment. So if the first line segment is always flatter than the second your backlash setting is too low.

Glad you're not giving up and will keep trying. 

Alan

[Starflyer]

Hi Alan,

If it's working for you then feel free to ignore me ?

But if you're getting a very flat bottom you may want to try increasing your step size slightly. I read somewhere that it's best to aim for a flat bottom, but with just two points, this gives you more data points on the slopes and gives a more accurate result.

 

Cheers

Ian

[symmetal]

Ian, you're right. I usually use a step size of 25 now. This was an old sequence I had loaded to get a few more subs and didn't notice until later that it was using my old step size of 20. 

Alan

[KyleStoke]

Hi Ian may sound silly but is the focuser moving during the steps, I had a similar issue when  mine was slipping.

[Uplooker]

Hi Kyle,

Yes the focused is definitely moving. Not much chance of slipping as it is a rack & pinion focused.

I think the main issue is operator error, as has been helpfully, but indirectly pointed out by Steve and Alan ?

I will provided an update soon, once I get the opportunity to get out again.

[Uplooker]

Hi All,

I am still failing miserably at getting Autofocusing with SGpro to work. I do not think that it is the settings or software at fault but rather it's something in the imaging train / scope.

When I adjust the focus I can get more stars to become visible as I get closer to proper focus. I continue until the stars are close to the smallest dots of light. The stars on their periphery are nice and round.

I then centre a bright star and put on a bahtinov mask. The image I see on the screen is barely recognisable as a cross. I then proceed to adjust. The image starts to improve, still not a distinct cross by any means. As I continue the image then begins to look more like an image of Mickey Mouse's head.

It may be that I have the spacing of the TS 2 field flattened incorrect but I would not expect it to have this effect on the image.

Any suggestions gratefully recieved.

[sloz1664]

I have 3 scopes with different cameras all with auto focus and using SGPro without issues. Yes there were teething problems which were step size and backlash oriented but the one biggest issue that took a time to sort out was attempting to auto focus on a set up which had field curvature> SGPro threw all sorts of Wobblies until I realised. Adjusted the spacing for a flat field and SGpro autofocused perfectly.

Steve

[symmetal]

Sorry to hear you're still having problems Ian. From your latest description it would appear to be related to field curvature as sloz says and maybe combined with image tilt as well  As SGPro goes through its autofocus steps different areas of your image come into focus so the HFR readings are all over the place.

I used CCD Inspector to sort out a slight tilt error in my setup and get best flattener spacing. You can use it free for a trial period. I found it so useful I bought the registered version although it's not cheap. The 3D display of the sensor 'flatness' I find the most imformative.

If you want to post or pm me a sample image, near the best focus you can manage, preferably fits, though a high quality jpg would work I can give you a display of your image flatness which might help.

Alan

[Uplooker]

Hi Steve and Alan,

Thank you both for your replies and suggestions.

I thought I had meticulously set the FF spacing correctly, have used various spacers and a vernier. Possibly not.

Tilt is also something that could be at fault.

In the past I have looked at CCD Inspector but found it hard to understand and interpret.

@symmetal Alan, that is a very generous offer to analyse one of my images. I would like to take you up on that.

Just not sure when I will be able to get something to you as I have not kept any images. I will endeavour to capture something this coming weekend. Between now and then I will have another look at the FF spacing.

thank you once again.

[RayD]

The important thing about using CCD inspector to analyse your images it to take the images correctly in the first place.  On the CCD forums the authors say it should be an even star field, taken with a sub in the region of 10 to 20 seconds (depending how long to get a well exposed field without stars blown out) and with tracking on but no guiding.

It is also best to have at least 10 images, but 20 seems for me to provide the most accurate results which you really can use with confidence for analysis.

Hope you get it sorted as I'm sure we all know how frustrated you must be.

[symmetal]

It's been recommended to do exposures for CCD Inspector pointing at the zenith to eliminate slop in your imaging train causing tilt though when I tried it there, there were insufficient stars for a usable result. I found pointing at an area of the Milky Way with no large bright nebulae or prominent galaxy clusters in view gave the best result. Also a high elevation is preferred, greater than 60 degrees or so to minimize seeing errors. Repeating after a meridian flip gives an indication as to whether there is slop in your imaging train.

As RayD says a number of images of each view are required to get a reliable result (individual images, not stacked).  Four or five will give a good indication of your problems, ten or more for fine tuning but I don't think you're at the fine tuning stage at the moment.

Alan