Odd AOX Jumping Behavior

AOX guiding in MDL 7.1.4 (Aluma 4040/AFW10/SG4) via a Paramount MX+ shows the guide star jumping infrequently, but randomly (see attached). The jumping occurs over a range of cycle rates (I looked at everything from 1 to 10 Hz) and independent of max move or aggressiveness settings. These position jumps do not arise when guiding without the AOX (see attached). Frankly, this looks like it is introduced somehow in the processing.

Any ideas?

1. AOX Video

2. Mount tracking

 

Yes, my bad, it is an SC-4 and on a scope with a 2563mm focal length.

In the video, binning is 2 (there are AOX calibration difficulties with 1x1 that I haven't sorted out enough to ask about yet...). I see equivalent behavior with binning 3 and 4.

I also should have mentioned that I've been following the other thread where you made this configuration suggestion and it's already been applied. I've also run at various exposures to check the possibility that that algorithm might be biting on an image artifact for a single frame. The behavior doesn't change, but this would still be the most probable suspect. If you watch the video carefully, you'll see that the tilt change occurs at the same time as the jump (not after), and consistently similar angles which still argues for chasing an invisible image artifact, or manufacturing one.

Now that I think about it, my AOX is configured for simple dark calibration. Given that Sum and Clip software binning must show some benefit, is there a chance that dark calibration isn't being applied (or applied inconsistently) and the detection algorithm is picking up a hot pixel (or generating the equivalent of)? Admittedly, I haven't messed with the dark calibration. I will sniff around that to see if there are useful symptoms, as skies permit.

 

Experiments last night didn't reveal anything startling. Turning on or off autodark reduction did not have any obvious effect. 3x3 and 4x4 binning effected the visual amplitude of the jump (as one might expect if the jump is caused by an image artifact) but did not have a noticeable effect on frequency or direction. Nothing consistently varied over a range of target SNR's either. All dead ends, apparently.

 

This has been a long standing problem with the AOX/SC4. Check out my posts on the matter. Still waiting on DL to fix it.

 

Fair enough. Must be a tough problem -- not easily replicated in vitro. I just figure the more we can help narrow it down, the better.

Awaiting instructions for further experiments...

 

Have you watched the guide star also? Turn on ‘Watch Guide Star’ for a few minutes and you will see it jumping all over the place. Also, regarding calibration. It is still my experience that you can only calibrate the mount and AOX bin 1x1 but you have to guide bin 4x4. If you try to calibrate in any other bin mode it fails….with the calibration routine seemingly starting with some odd spot in the upper left of the calibration window.

 

I have seen the calibration behavior you describe, but I haven't been able to force it consistently yet (there are times that I can calibrate in all modes) so I haven't posted on that yet.

Yes, I've watched the guide star, even recorded it (see link in original post). I've also pretty much ruled out mount contribution (shake or move) by comparing AOX and non-AOX tracking on the same target (see original post).

When you break still-shots out of the recording some details as to the tracking behavior pop out. This particular track has a nice, high SNL 2X binning and .2 second cycle rate:

In the recording, tracking will idle along for a while at low tilt corrections.


Then, within one cycle, the "twitch" occurs. Note the trailing blur. Note the tilt change (in X only). That pretty much rules out an internal shift of the sub-frame as the culprit. The most likely explanation is that a tilt is occurring during the imaging (or the mount is moving, which is not on the suspect list, at this time). The most likely explanation is that the tilt has been instructed to chase an image artifact other than the target star.


In the next frame, the AOX seems to be trying to recenter the prior twitch with a tilt correction.


But then, another twitch to a similar location occurs. Trailing occurs. Note that in both just prior to the first twitch and in this frame, no image artifact is visible to chase. No tilt change in X, yet showing. Hmmm...


Next, a centering tilt correction is again applied. Big tilt change in X now. Some, but not much, in Y.


Lastly, we're back at center with small tilt corrections.




My recollection is that this twitch did not occur in the old SBIG AOX tracking code, so I'm guessing the problem may be in some difference between the old and new implementation of the detection algorithm. But that's for DL to figure out -- I'll be happy to continue to run and record experiments that might help out, though. It would be nice if there were acknowledgement that there is a problem, that it is on DL's radar screen, and that it will be worked on as time and priority permits... or not.

 

Oh I agree. I have used an SBIG STXL camera with their FW8G and AOX for years and years and it worked beautifully, so its not an AOX problem obviously, it is the SC4 and the software that supports it.
FYI - you could get a much higher guide rate if your ADU count and SNR is that high as shown in your screen shots and if you bin 4x4. I am surprised you managed to calibrate the mount and AOX at the other bin modes....never worked for me. Yet, still no sign of a fix from DL...I seriously wonder what the problem actually is and why this has gone on for so long, but I am delighted another person has posted identical results.