Hi.
Something that is constant about performance of the motor in SGP’s autofocus is that as SGP pulls the focuser in there is little change in the HFR when the HFR should actually be dipping as step gets closer to the middle of the curve.
Anyone else has similar experience with this focus motor and SGP’s auto focus?
Have you setup the auto focus parameters?
https://mainsequencesoftware.com/Content/SGPHelp/UnderstandingAutoFocus.html
Thanks,
Jared
Yes, I have.
Sounds like you need to increase your step size. Beyond that we’ll need a log with auto focus runs to be helpful.
Thank you,
Jared
following on from Jared’s post - for example I am using an f/8 10-inch RCT. The important parameter is the f/ratio and the movement between measurements
I’m taking 7 focus measurements with a stepsize of 35. Each step is 3.7 microns (so 0.13 mm spacings). This is as small as I would want to go. If am using a faster f/ratio, I can drop down a little.
Thanks for the input. I thought I was familiar with the manual and I had never read that auto focusing has to do with the f-number as you say. I am reading the manual again to see what I may have missed. So far I am not able to find a reference to F ration of f-number.
Yeah, I just finished reading up on autofocus and as I said I found no reference to the influence of OTA parameters there. And I have done exactly as the manual says.
We find the step size by pushing the focuser out until HFR is 3 to 5 times the HFR at the desirable focus position. I have been going out five time good HFR. We take the difference between that position (P2) and the initial “at focus” position (P1) and divide it by the number of steps on each side of the V that we want in order to get step size. Number of steps is number of data points minus one divided by two (for two sieds). So for the recommended 9-data point condition, which is what I use, number of steps is 4 (on each side of initial focus position).
Maybe what Jared is saying is not to increase the step size but to push the focuser out even farther than 5 times the in-focus HFR and then do the math which will result in larger step size.
I don’t see anything more magical about auto focusing parameters in SGP. It is very straightforward.
What is a mystery to me is the relationship between size of each step here and the size of each positional step in the Celestron focus motor. I think the answer is somewhere there. The step size should be also a function of the size of motor step that can in fact result in a meaningful change in focus that can be measured by SGP routine.
It is to do with Physics. SGP suggests you have a range of HFR values over the focus run. That range is a function of the physical travel and the f/number. That is why in FocusMax they asked for the f/number so that they could use it in their characterisation of a focuser.
@Farzad_k I don’t think Jared meant that you should go further than 5 times the best HFR. In fact, with the new quadratic routine it is now recommanded not to go beyond 3 times best HFR, it usually gives best result at about 3 times (see here for instance: SGP curve looks great but ends with "...too many range shifts..." - SGPro Support - Main Sequence Software).
To help you we would need to see some example of log files. There is useful information in those, such as your AF settings, including step size.
Mikael
Thanks Michael. I will update the manual. Adas soon as there is clear skies I will dedicate one session to automatic focusing and submit the log file.
Farzad
From my experience, the SGP autofocus algorithm has a much harder time with an SCT than a refractor. My Esprit100 refractor creates a nice smooth curve if set up with the right step size and number of iterations, and can cover the recommended 3x to 5x HFR range without a problem.
My Edge 8 SCT cannot. If I go much beyond 2.5x, the wings of the curve start to angle off badly, and asymetrically, making the curve fit fail. Seems like there’s a couple of issues.
First, because of the long focal length and narrow field of view, the number of stars in the field can be pretty limited, depending on what part of the sky you’re targeting. And when you move too far from focus, the number of stars detected for HFR calculations drops way down. Of course, you can increase your exposure time, binning, camera gain, etc. and that helps. But longer exposures also increase the auto-focus cycle time. For my SCT I’m now using 10 second exposures, and that seems barely enough.
Also, with an SCT, the out-of-focus stars take on a donut shape rather than a round ball. I think that might make it harder for the algorithm to detect stars or accurately measure the HFR.
In any case, setting up the auto-focus parameters to get reliable focusing is MUCH harder on an SCT, especially when you’re in a star-poor area of the sky.
-Dan
Thanks, Dan. The manual should be talking about the differences between a refractor and an SCT (and other scopes).
With the MoonLite focuser on my refractor I used to get v-shaped curves, and I have not used that scope in a while.
Farzad
Have you inspected the gearing of the focus motor? I opened mine up and initially discovered a little play of the gear that is attached to the servo motor. I opened up that housing and by the time I got to the motor, I was able to spin the gear freely. Whatever that was locking the gear to motor broke off by then.
I don’t know if this was a progressive issue with my focusing problem in SGP from when I put the motor in use a year ago or a combination of things. I have not heard back from Celestron yet, but I am hoping to get a new motor since I have a 2-year warranty.
My focus motor is a PrimaluceLabs Sesto Senso. I don’t suspect there’s anything wrong with it because I just took it off my refractor and put it on my SCT. It was working perfectly on the refractor. It seems to move through the focus range correctly. There is some backlash (play) in the focus system, but I am using the backlash compensation setting in SGP so when moving the mirror backwards, it goes past the target position and then approaches it from the other direction. That way the final mirror motion is always upwards, against gravity, to avoid backlash issues.
Good for you. I agree that every inward movement should be made with plan to finish it in an outward correction. We are suspecting something may be wrong with the ASCOM driver for the Celestron focus motor.
I use SGP with a 10-inch RCT. When it is not collimated, the focus curve is not symmetrical and SGP is simply reporting what it measures. If you go too far out - you get ‘dog ears’. Poor collimation causes star shapes to become irregular, very quickly and confuse the focus algorithm. When you are a long way out, you start to get donuts. It sometimes happens though that there are irregularities in intensity in the donuts, caused by seeing and SGP will believe these hot spots within the donut are a star and similarly get confused.
What actual step size are you using with your 9 data points? I have not used my C925 (Which has a Celestron focus motor on it) for some months but I was using 9 autofocus data points and a step size of 20. If you are getting more or less a flat line then you are probably using too small a step size.
It does not need too much by way of over thinking or you can just go down a bottomless rabbit hole. I did the maths to arrive at I think a step size of 25 but that made it a bit unreliable sometimes as you ended up with obvious doughnuts at the ends of the run that could fool the HFR measurement and I experimented my way down to 20 step size as a reasonable compromise betwen a good shaped curve and reliably getting the HFR measurement at the ends of the curve (Such as if it was a bit out of focus to begin with)
One thing that does require thinking about with the SCT focuser and Celestron motor is direction of movement. I selected “Reverse direction” so that a commanded “IN” movement was what Celestron consider an “Out” movement (Mirror moving away from the rear of the tube) so that “IN” focus matched what you would expect to see from a normal focuser. Then I selected backlash compensation of 800 steps in the “IN” direction.
You can leave the normal direction and select “OUT” for the backlash compensation (You really do need the compensation) but the focus run will take longer as more of the focuser moves will have the compensation applied.
I am not a fan of reversing the direction because it makes things even more difficult to think about. I do know that it is best for SGP to pull the mirror IN and then step it out because then it can deal with the backlash on the first step out.
I have used various step sizes. I don’t remember which and when.
It looks like someone else in this discussion is having great success with just basic standard parameters.
I am getting a new focus motor and hopefully some of the issues will then go away.
Farzad
Any new focus motor will need to be set up so that the first move is down the tube with a backlash move back up so you will probably have to reverse the direction of the new one anyway. However you set it up, the first move in a focus run needs to be down the tube with a backlash move back up to position and the final move to the calculated position needs to be the same, otherwise 8 out of ten focuser moves in a focus run will have backlash compensation applied, and it needs to be about half a turn of the focus knob. Depending on how fast the focus motor moves it could add a significant amount of time to each focus run, and if the temperature is changing, an SCT needs focusing often!
I am not sure if the option of first move direction is available with the Celestron focus motor. I will look into it once they manage to send me a replacement.