2.5.X - Bad column and Autofocus troubles



Sometime this summer, my SBIG ST-8300M developed a stuck pixel about 3 quarters up the frame that then produces a hot column to the bottom of the frame. This has been causing some grief when using Autofocus.

AF dark subtraction is enabled, but due to the random nature of the column’s intensity (since it isn’t a true bad column, but the result of hot pixel spilling down) it’s effectiveness is random.

Setting 10 pixel minimum star size helped, but when the column is continuously larger than 10 pix it picks up as a huge HFD reading (again, because the randomness of the columns intensity with dark subtraction, some times the column is broken up and no problem arises).

Is there a way for the HFD to check the geometry of the star and look for circular pattern instead of 1 pixel (or 2 pixel since 2x2 binning is used) wide by X pixels long linear pattern?


Here is some recent logs and AFpacks which may help.
SGP log from the night of 09/14/2016-09/15/2016:

SGP log from the night of 09/15/2016-09/16/2016:

PHD guide logs (did not include the huge phd debug logs):

SGP AF packs:


I guess I’m not really sure. With respect to the “hot column” nothing seems amiss with your AF runs. In all 4 packs, not a single frame in any of them tried to consider any part of the hot column as a star. For this, I indicated that the minimum star size at 1x1 was 15px and 2x2 was 7.5px.

The only things that seemingly need to change are:

  • Minimum star size to 15 px
  • Smart focus turned ON (if you would have continued going, I suspect SGPro would have found the curve)
  • Increase your step size to 9

Pack 12:

Pack 13:

Pack 14:

Pack 15:


Hi Ken,

I think your looking at the wrong AFPacks. Coincidentally named, as I’ve uploaded AFPacks_12 through 15, but my steps are in the 3000’s.

AFPack_14 in particular shows the behavior with the fourth frame. The behavior is random. I will upload more packs that show the issue.


BTW my setup is

GSO 8RC (Ritchie-Chretien, 95mm central obstruction)
Moonlite CS2 with highres stepper and mini v2 controller
Starlight Xpress SX-AO-LF active optics
SX Slim Off axis guider
SX Lodestar x2
SX USB Filter wheel
Celestron CGEM

So I have been keeping the Disable Smart Focus checked, and actually only tried it out for the first time on the 14th-15th. I worked once, and then it pushed it too far out of focus. For last night, I tried it again and kept it on all night, and it seemed to have worked quite well even with the many hot column erroneous data points.

I’ve uploaded the entire nights AFPacks (minus those which I canceled the autofocus routine) here (2016_09_17_AFPacks.zip):

And here is the nights logs:

Thanks for your help on this one!


No, these are yours, the step size is irrelevant when I am running it through our AF simulator…

My original recommendations still stand (I think).


Am still experiencing many HFR mis-reads cause of this hot column. Will post tonight’s data, and the graphs I am getting, cause they look nothing like what you’ve simulated there. Could the simulations differ from real world results?

My attention has also been brought to something that’s been happening on and off at least since July/August 2016. Every now and then I will look at the focus module and the minimum star size has changed from 10 pixels to 4 pixels!! Even though my equip profile says 10 and the sequence starts out with 10… And I continually set this back to 10!

It’s random. I haven’t seen it in a while, but tonight I am resuming my sequence from last night, and SGP keeps changing the “Minimum star Size at 1x1 (px)” back to 4 pixels… And it picks up the hot column very easily.

Just to confirm this, I checked the current running sgp log and sure enough, the routines were run with (from log) “Star detection using min star size of 2px…”. (FWIW the max was 40px)

What is changing my AF parameters on me?

Also, here is what I see my graphs from the night above as. Note Pack 14’s spike:

Pack 12

Pack 13

Pack 14

Pack 15


Here is a screenshot of a mis-read. I have AF dark subtraction, but the nature of the column is variable.

Just thinking out loud here, but a bad pixel/bad column map would solve problems like this entirely, no? Probably a pain to write something that in code though.


Looking at the images in your first post…
Pack 12 seems fine and shows that focus is near to the 3438 position. It doesn’t have the left hand side of the V but if 9 positions had been used instead of 7 it would have been fine.

Pack 13 seems similar except that there is even more missing from the left hand edge of the V. The last point, at 3246 looks as if it would have been close to focus had more points been used.

Pack 14 is a mess, no idea why but this seems to happen occasionally.

Pack 15 also only seems to be showing the right hand side of the V focus curve with a focus point at 2951. If more than 7 points had been selected this may have given a good focus.

So I would start by endorsing Ken’s recommendation that you increase the number of steps. He suggested 9, I think, to start with at least, 11 may be better.

Ken’s other suggestions may also help.

One other thing, the focus positions that seem to be found are very variable, the data shows 3438, 3246 and 2951. It seems strange that the focus position isn’t consistent from run to run. This will be read from your focuser, is there a reason it could be changing?



Your focus graphs might point at a Crayford working at maximum lifting capacity. I have had this – and can easily reproduce it – when the focuser partially slips. You can test it by trying a focus run with the focuser axis closer to horizontal.



The other thing to note is that the focusing is much better if the collimation is good. This was particularly noticeable with my 10" RC. I would get variable results depending on the positions of the stars that SGP selects (which changes with each exposure) when the collimation was not accurate.


Temperature shifts up are substantial so moving a couple hundred steps over a few degrees, seems normal.

Beginning of Pack 12 at 1:35 am
[9/16/2016 01:34:58] [DEBUG] [Sequence Thread] Set start frame temp to 9.32266666666667…

Beginning of Pack 15 at 1:35 am
[9/16/2016 03:28:26] [DEBUG] [Sequence Thread] Set start frame temp to 6.888…

Reason for posting the graphs was to illustrate the random nature and the effects of the Hot Column. Pack 12 normal and results were focused, Pack 13 normal and results were focused, Pack 14 - BOOM hot column, Pack 15 normal and results were focused.

Enabling Smart Focus does not help at all (my scope has an obstruction), especially when using 7 data points. If the AF doesnt get a line on the right, it backs out the focuser 3 spots further than the initial farthest position. When frames move so far out of focus, that stars are faint donuts, the chance of the HFR routine picking up on the hot column increases substantially.

I’ll try increasing the number of AF points to 9 but have not had success with that in the past as it causes the problem to get worse, for the same reason above that those frames that are so far out of focus the AF routine can’t pick up any of the faint donuts and just see’s the hot column. 7 seems to be the happy medium in terms of how far out of focus it goes.

Again, using darks do not help, because the hot column is actually a stuck pixel that bleeds downward, so the intensity varies.

When all is well and good the AF works great. 7 data points does the trick and my 75 step step size brings me in and out out the CFZ no problem. A defect map would be the perfect solution I think.

I should add that the problem is much worse when using Narroband filters.


Hi Horia, running a moonlite 2", with a 5.5 lb load. When scope is at zenith, there is no slip and has been adjust to lay on the safe side of things.

Great observation though, as I have had it slip on me in the past. The give away were the graphs moving hundred of steps inwards, with little to no temperature change. By the end of the night the focuser thought it was at 0 when it racked halfway out.

Anyways mechanically, everything is seemingly peachy. :slight_smile:


Hi Buzz,

I don’t thing collimation is a problem here. Stars are seemingly round across the image. I’m sure my collimation isn’t perfect, but I think it is collimated well enough to not cause problems with the AF routine. If the hot column is a problem during a night, SGP has no troubles getting it’s v-curve and in situations where it doesn’t get the v-curve, the final focus position is usually calculated right on the mark.

My step size is 75 and my FL is 1624mm with 1x1 pixel scale of .69 arcseconds. So if the AF routine is thrown off by slight variations in the frame, generally the CFZ size is large enough to accommodate where it my land.


One thing I noticed in all the AF graphs was that they are very asymmetrical. Of course the V curve will only be “perfectly” symmetrical if the camera is in focus to start with. SGP has an option to enable temperature compensation. When used, SGP will make a focus adjustment prior to starting the AF routine. If your TC is accurate, this adjustment might get your camera close to re-focused before the AF routine takes a crack at it. In my own AF graphs, I typically see 5 data points on the right and 4 on the left. My worst HFR is typically about 3x the best HFR.



I have an idea. You say you use dark frame subtraction. Can you make a dark frame that is from a longer period, so your light line becomes a dark line? You could almost manually edit it in.


Great idea!! I’ll give that a try.