Image misaligned after sequence recovery

Last night I was imaging the Deerlick group. At about 12:15 clouds apparently rolled in and the sequence went into recovery mode. At about 12:45, the sequence recovered, but the subs were significantly misaligned from those before. Luckily, the meridian flip occured about an hour later so plate solving re-aligned the image to where it was supposed to be. Otherwise I would have lost the whole rest of the night.

I’m not sure if maybe PHD locked onto a different guide star or something? But I guess I was under the impression that sequence recovery included a solve and center operation. Is that not correct? If not, it seems like it should be an option at the least.

Here’s a link to my Log file: Dropbox - File Deleted

And Here’s a link to the sequence file I was using: Dropbox - File Deleted

Your log appears to show that the target was successfully centred after recovery. What might have happened is that after centreing there were still a few clouds around, PhD lost the star but quickly detected noise as a star or perhaps even detected a different star in the search box, which moved the mount off centre until the meridian flip. In this case SGP recovery is not triggered. This has happened to me countless times and is perhaps the only disappointment I have with PhD, because no amount of adjustment of settings will stop PhD detecting noise as a star. Fortunately this only happens on nights when sky conditions include the passing clouds.

Maybe check your PhD guide log to see what happened during the time after recovery succeeded and before the meridian flip.

Does the original star need to be obstructed? PHD is using one star to guide and will be assuming that the star that is nearest to where the star was last seen will be the correct one. If there’s been a delay then there could be enough drift that a different star is selected.

I don’t think there is a way to prevent this with any guiding application that uses a single star, they all look the same.

What might work is a guiding application that uses multiple stars and does some sort of pattern matching using a number of stars to determine the guide position. Essentially some sort of plate solving. I don’t think this is done by any guiding application. If it was it could be very powerful, It could recover from large drifts and maybe even acquire the same guide position after a merdiian flip. There would be a limit, maybe 50% of the guider FOV.

1 Like

OK, I checked the PHD guide log after 12:47am. It looks perfectly normal.

In the SGP log, at the beginning, there is a line that says “Centering on Target” when the centering operation is in effect. This is a clear indication that centering occurred at that point in time.

However, after recovery at 12:47, there is just a status that says “Sequence Recovery was successful (CenteringAndGuiding) !”. That is a status message that occurs at the end of the recovery process. There is no status message that says the centering is occurring. Of course, I don’t know the code so maybe it did occur. But all other phases have a specific status message when it’s actually occurring such as “Running auto focus…” or “Settling Guider …”. That kind of status message shows up at the beginning of the session for centering, but not during the recovery.

In any case, it’s pretty clear from my images that after recovery occurred, the image was significantly shifted. If centering actually did occur, either it failed with no failure message, or something happened afterwards to move the mount. PHD shows no signs of that.


Interesting Dan.

Dan, can you post your PhD guide and debug logs?

Was the very first sub after recovery (frame 69) off-centre?

Were all subs after recovery (and before mflip) off-centre by the same amount, or was there further movement between subs?

Hello Chris,

What might work is a guiding application that uses multiple stars and does some sort of pattern matching using a number of stars to determine the guide position.

A user successfully modified PhD to guide on several stars, but the motivation was for better centroid estimation accuracy by averaging the seeing across those stars, given that atmospheric dispersion is to some extent spatially independent.

I agree that a form of simplified plate solving would help overcome unwanted and undetected shifts in the guider FOV. I would welcome such an addition to PhD.

SGP will recenter during recovery under certain conditions. Generally if it’s a lost guide start we do not recenter and just start imaging again…assuming that you should still be “pretty close” to on target. But we may need to recenter no matter what just to be sure.


Here’s the PHD Guide Log:

And here’s the Debug Log:

As I look through the Debug Log, I see there is oodles of activity during that 12:15 to 12:45 time period. I would have thought that SGP would pause guiding while in Recovery mode? Otherwise, as clouds come and go, PHD will keep losing stars, finding new (potentially different) ones, and possibly walking over time.

Yes the very first frame after recovery was off center, and the remaining frames showed no movement until the meridian flip.


Hi Jared:

As mentioned in my response to Ross, PHD is doing a ton of activity during the recovery phase. Is that normal or should SGP have paused guiding during recovery?

In general, I think it’s a good idea to add centering as a consistent part of the recovery process. But at the least you could leave it up to the user with a “Force center after recovery” checkbox in the Sequence Recovery section of the Sequence Options tab.


1 Like