Autofocus problem

Sequence Generator
#1

I am trying tonight to get first light with my new Edge HD 9.25. and am having a problem with running autofocus. When I tell it to run, the autofocus screen pops up, and it says it is moving the focuser to the next position, but nothing happens. Any help before dawn much appreciated! :smile:
Dean

#2

We will need more information:

#3

Using SG Pro version 2.4.0.27723 with Win 7. Here is the log file https://www.dropbox.com/s/v2pqbi1dsggm6fb/sg_logfile_20150407194405.txt?dl=0

Dean

#4

Thanks for the logs. I’m not sure if I can be of any immediate help. I notice that this occurs every time your run AF:

[4/7/2015 11:53:00 PM] [DEBUG] [Camera Thread] Moving focuser to next position (32440)...
[4/7/2015 11:53:00 PM] [DEBUG] [Camera Thread] Focuser moving to 32440
[4/7/2015 11:53:00 PM] [DEBUG] [Camera Thread] Focuser move call complete

The part about complete doesn’t mean that the focuser is done moving, just that request to move is complete (focuser has accepted it). After connecting the focuser, can you perform a manual move (through SGPro) with the focuser module (in or out)?

#5

Hi Ken,

No it won’t move manually in SG Pro either. It is a Micro Touch, and moves okay if I use the control box. It worked great overall last season with SG Pro…

Dean

#6

Can any other software application control it?

#7

SG Pro is the only software I have running with the focuser. I’ll install and try Focus Max.

#8

Hi Ken,
I can move the focuser using Focusmax.

update: Progress! I have no idea why, but once I moved the focuser in Focusmax, it now goes through the autofocus routine in SG Pro. So this problem appears to have solved itself, and now I just need to find the proper step size.

Dean

#9

Ok… weird. Step size shouldn’t take too long to find. Just find your HFR while in focus and aim for around 9 steps where the outside HFR is 2.5-3x your in-focus HFR. Since you have a central obstruction, you might be better off with 2-2.5x (depends where donut formation starts). You should also probably disable smart focus.

#10

Hi Ken,
I measured the focuser travel with callipers, and in about 29 mm of travel, it moved 5000 steps on the Micro Touch. That is .0058 mm per step. Critical focus zone of the f10 Edge HD 9.25 with 4.3 micron camera pixels is 232 microns. If I have done this right, that works out to 2.32mm divided by .0058mm equals a step size of 400. Seems huge? Step size for my Micro Touch on my ED100 is only 20, and it works flawlessly. I didn’t have any luck with that 400 step size setting tonight. Very inconsistent results, no matter what step size I tried, but I had a very light haze of cloud moving through that I think was the problem. I’ll try again ASAP, and will shoot for the 2 - 2.5 time the in focus HFR, whatever step size that takes. And I have disabled smart focus. What would be a ballpark figure for HFR for this telescope?

Dean

#11

Hi Dean,

A little off on your math. 232 microns is .232 mm so your CFZ is around 40 steps.

I haven’t used my SCT in a while (hope to put it on the mount this weekend) but I had to use 4x4 binning to get good autofocus runs due to the donuts. With my current setup with an AP155 and 4.5 microns pixels, I get a minimum HFR of about 0.9 pixels on a focus run with 2x2 binning (4 seconds exposures with L filter). My step size is set to 65 with with 9 steps which gives me HFRs around 2.4 to 2.8 at the limits when I’m centered. I get text book V curves which is mainly helped by getting all the slop out of the focuser that backlash couldn’t compensate for properly, such as pinion block wobble.

I haven’t worried much about the CFZ as I get very repeatable results with autofocus with three consecutive runs (used when setting up filter offsets) give me results within a few steps of each other.

Frank Z…

1 Like
#12

Hi Frank,

Ooops! Well 40 steps sounds a little more reasonable. I am using a Canon 60Da, so I don’t have the option to bin. Last night was the first time with an SCT, so I used a Bahtinov mask to achieve initial focus, then did an autofocus routine. Results were all over the place, even with a step size under 100 and as high as 800. I thought it was the thin haze layer moving in and out that was causing problems, but perhaps not. I had not thought of the donut problem once the autofocus routine takes it way out of focus. Do you know of a solution to this if one is using a DSLR and can’t bin? And how exactly does binning help this problem? And thanks for those HFR figures. That will give me an idea of where I should be.

#13

Well I am having no better luck tonight with auto focus under nice clear skies. Win 7, SG Pro 2.4.0.2773. Step size was calculated with callipers measuring the focuser travel. CFZ is 232 microns, step size 40. Edge HD 9.25 at f10. Using 10 sec exposure, at 800 or 1600 ISO. I have tried varying the number of stars sampled from 100 to 15 with no luck. Focus runs just make no sense at all - very erratic with no real pattern to the madness. :slight_smile: I have tried step sizes from 20 to 100, and all are no good at all. I have tried number of data points from 7 to 11. More than about a step size of 50 starts giving big donuts, so I doubt I want to go any bigger than that. Any help much appreciated.

Dean

#14

Logs from a representative run please

#15

Here ya go Ken. https://www.dropbox.com/sh/w0wotovs80k2lku/AABBPW5Ch23tSHs8xe-y09u1a?dl=0
Dean

#16

Dean,

Thanks, but that log doesn’t contain any attempt to auto focus.

#17

Hi Ken,

Ooops Am I sending you the wrong thing by sending autofocus packs? Here are the logs from tonight. https://www.dropbox.com/sh/wuvqxh0gfzn2h41/AAAy75ANnOXkEBnHeO5hr6kwa?dl=0

I also notice that even though I can see dozens of stars in the focus images, SG only seems to plot the HFR for a few of them, and not necessarily the bright ones. And sometimes only one out of the whole frame of dozens is plotted. Then next frame, it grabs totally different ones to sample, again leaving most of them untouched. It seems really random.

Dean

#18

My apologies. I am on dropbox a lot and when I clicked your link, it showed me a set of logs from another issue I was looking at… Not sure how.

But to answer your question, AFPacks are great, but will almost always require a log for correlation.

I’ll take a look through.

#19

@headdown

I took a look through that data you sent and I have determimed that there are a few issues here:

  • Whatever target this is, it is extremely data poor in terms of stars
  • This means that you need to keep your AF sample size down low to avoid forcing SGPro to lock onto hot pixels (there are 15-20 viable stars here)
  • Now the bad news… SGPro appears to be miscalculating centroids due to the bayer matrix. SGPro currently expects, in terms of intensity, to find the brightest points at center or near center of a star. The bayer matrix is unfortunately providing all sorts of black and gray pixels near center. Are you using a filter with your camera? Like an Ha filter or other?

Here is one of the stars I am talking about. You can pretty clearly see the matrix here:

I am wondering for OSC camera if we are better getting rid of the bayer matrix prior to running AF. The problem here is that it introduces bayer patterns and more complexity into AF, and I don’t even know if it would work…

Right now I don’t think your current target is a good candidate for AF.

#20

Hi Ken,
I have tried reducing the AF sample size to 15 with no luck. And it does no better in more star rich areas of the sky either. So it must be the Bayer matrix that is the problem. I am using no filters. Bad news indeed. I just bought the edge and Feather Touch hoping to use it with SG. Any chance of you doing some software magic to remove the Bayer matrix? The thought of having to focus by hand with the Edge is very discouraging…

Dean