Hi folks,


I'm using a Primalucelab Esatto 3 focuser with varying auto focus results. Sometimes I have a perfect focus after 5 iterations, sometimes auto focusing fails after 25 iterations.
I've tried tons of different settings, but I keep getting mixed results.

Are there colleagues here who would like to share their experiences and optionally share/display their settings?


Thanks !

These are my settings that work 'reasonably' but with inconsistent results:









I'm beginning to suspect that the determination of the backlash is playing tricks on me.
Does anyone have a suggestion? Or are there other users with better results?

Hi Fitchie,

I don't have your focuser but here are some "general" suggestions on focus.

Just looking at the parameters you have selected I see you are using the Polynomial Algorithm. Is there a specific reason for selecting this? In general, Linear 1 Pass will deliver better results.

I see also that you have Sub Frame selected so you are using a single star for focus. From the CFZ tab you have a 107mm scope at F# 6.54. I would suggest using multiple stars for focus by selecting Full Field.

You have an initial step size of 3000 on the mechanics tab but CFZ tab is showing a CFZ of 8. So something probably isn't quite right there as these 2 numbers should be comparable. The CFZ will default all but 1 parameter from your Optical Train. So on the CFZ tab you have Step Size set to 5 microns which I think is the default so I'm guessing you haven't measured it? It would be worth setting this correctly as it will give you a clue as to what to set the Initial Step Size to.

As far as backlash is concerned you have driver backlash set to 5000. I can't say if that is right or wrong for your setup; you'll have to measure it yourself. Its worth having a read of the Backlash section in the manual if you aren't familiar with it.

I would recommend using the Focus Advisor (Advisor tab) and let that default most parameters for you. You will have to setup Step Size on CFZ tab and backlash yourself (or AF Overscan if you prefer).

Here is a link to the Focus section of the Kstars Handbook...
docs.kde.org/trunk5/en/kstars/kstars/tool-ekos.html#ekos-focus

If you are still having difficulties post a picture of the V-Curve along with the parameter settings as that is useful in seeing what's going on.

Good luck!

Hi John,


Thank you very much for your very detailed answer.

If the weather gods are in my favour I will follow your advice and do the necessary testing later. And if it doesn't work out tonight, it will be for tomorrow or the day after, I'll definitely keep you posted.

My previous words were not even cold when the clouds came up...

But in the meantime, based on your advice, I have checked my settings and put a number of things back to default.

The biggest mistake was the step size, which is apparently 0.04 micron, so I was wrong by a factor of 125. Oeps... Not that the hardware of the Essato actually achieves that precision, but that's another discussion.

And by adjusting the step size, the advisor values ​​also change to more logical values. So it's starting to make sense. Once the clouds will bother people on other places, I will test and send feedback.

The focus results are consistently better, that's the good news.
But they do include more iterations than before, and that has the disadvantage of consuming time.

As requested, here are some hyperbolas:

Hi Fitchie,

Thanks for sharing the screenshots. The purpose of the Focus Advisor is to get focus basically working which it is now. By playing around with the parameters you can probably improve things.

I usually look at the max / min HFR on the v-curve. In your case its something like 1.6 / 1.25 = 1.3 which is low. I'd try to increase that by increasing your step size. With a steeper curve the position of the minimum will be more accurate.

You seem to have plenty of stars which is good.

I see you are not specifically setting the Gain (using whatever Capture / Indi is using) which may be fine, but you could play around with Gain / Binning and Exposure to see if you can improve results. Obviously, lowering exposure will speed up the process. But you don't want to do that at the expense of quality of focus.

FocusAdvisor should be recommending 2s exposures, bin 1x1 and useWeights for your config. Maybe you've been playing around with the settings and these work better for you?

Good evening,


The max / min HFR on the v-curve is indeed low, but I'm not able to get it steeper. Play around with Gain / Binning / Exposure does not improve results. Increasing the step size gives steeper but unpredictable results.

Please see below screendumps:







Any suggestion/feedback is more than welcome.

What do you mean by "use weights for your config"? Weights in the sense of ballast as counterweights, or weights in the sense of weighted values?

Hi Fitchie,

You have the useWeights Focus option "off" in the first 2 screenshots and "on" in the third.

This is where the compromise comes in. You can increase the max/min range but at a cost. There's no right or wrong answer. Its upto you to set the parameters that give the best / most reliable answer for you. Its always good to experiment as it gives you a feel for how Focus works along with its limitations.

It did take some time and just like you say it's a bit of weighing and trying...

But finally I managed to systematically get this curve:



What strikes me most is that when I choose SEP / Linear 1 Pass / FWHM the focus results are consistently very neat. But when I choose SEP / Linear 1 Pass / HFR the focus results are inconsistent and much worse.

In the last case I see the stars getting more blurry but the HFR values ​​getting smaller. So I don't really understand that last one.

Looks like a good curve with FWHM - nice to see! Looks like the first and last points aren't so good (nut within the errors bars).

A way to compare curves is the R2 (0.96 in your screenshot). If all points lie on the curve it will be 1 so you ideally want a number as close to 1 as you can. So if you are comparing HFR with FWHM as a curve fitting measure then keep an eye on the R2.

Thanks for posting.

My attempts to get a steep curve are starting to pay off.

The 1st run looks good:



But on the 2nd run the curve is suddenly less flat again and measurements 1 - 2 - 3 are very strange:



What could be the cause of this?

The difference between run 1 and run 2 also worries me: 322 K steps at versus 308 K steps while the temperature remains very constant.