Unexpected systematic drift while guiding
David Wilden
Hi,
Even though I've polar aligned to about 1 arc min overall error (using Sharpcap's tool) and stars are not elongated over 2 to 4 min exposures, I'm still seeing a sizable systematic drift superimposed on the random dithering applied by PHD2.
I’ve done some analysis of the PHD2 log files to try to reconcile the image shift between subs as measured in Deep Sky Stacker with the dither movements recorded in the PHD2 log files but cant work out where the discrepancy is coming from, so any help understanding this would be greatly appreciated.
My imaging system is a skywatcher 72ED refractor (FL 420 mm) with sv605CC camera, (pixel size 3.76 um) with an Orion Guide scope (FL 62 mm) with sv305 camera (pixel size 2.9um). This is mounted on skywatcher AZ-GTI mount on an equatorial wedge. I believe the system is well balanced on both axes.
I've attached two images shwoing graphical plots of the data I've analysed for two runs which show the problem. One is a run where I dithered after each exposure, and the other with dithering every third exposure.
Each plot shows:
In blue, the (X.XXX, Y.YYY) dither values from the lines in the PHD2 log which say “ INFO: DITHER by -X.XXX, Y.YYY, new lock pos = AAA.AAA, BBB.BBB
In green, the lock position values (AAA.AA, BBB.BBB) values from the same log lines.
In red, the image offsets calculated by Deep Sky Stacker when it aligns the subs during stacking.
Note: In order to be able to compare the three data sets on the same axes, I have subtracted fixed offsets from each data set to make the first data point in each series equal to (0,0) in eacgh case.
Also I have scaled the PHD2 data sets by multiplying them by 2 to convert from guide scope pixels to imaging scope pixels. (Pixel ration = 420 x 2.9/162/3.76 = 1.9996)
Data set 1 (2025-10-11)
Each sub was 4 mins with dithering applied after each sub with range (+/- 10 guide pixels = +/- 20 imaging pixels). In total there are 17 subs and 17 dithers.
The PHD2 log analyser calculates that the polar alignment error was only 1.6 arc mins for this run and that the rms error reported by PHD2 over the 17 exposures was 0.60 guide pixels (1.2 imaging pixels).
The blue dots (dither positions) in the figure show the expected range of dithering values, with pseudo random dithers within a +/-20 imaging pixel range.
The green line (PHD2 lock positions) and the red line (DSS offsets) both show a large systematic drift during the run which is is not what I expect given that the dither steps are roughly random in direction and magnitude.
Overall the drift over a period of 17 subs was more than 90 image pixels which equates to an average of more than 5 pixels per sub. Yet the star images in the subs show no significant visible elongation, which means that there cannot be a significant drift while each 4 min exposures was being made.
Data set 2 (2025-09-25)
Each subs was 2 min exposure and dithering of +/-5 guide pixels (+/-10 image pixels) was applied after every third sub. In total there are 60 subs and 20 dithers.
The blue dots show the expected range of random dither values.
However the green and red lines again show a large systematic drifty during the run.
The DSS data shows another feature which demonstrates that the systematic shift only occurs when dithering occurs. The DSS data points appear as closely spaced triplets. Within each group of three no dithers have been applied and the points are close together suggest very small drift. Then there is a large jump in position between each triplet when the next dither is applied. The mean shift between exposures within each triplet is only 1.0 imaging pixels, which confirms that no significant drift is occurring while the exposures are being made.
Conclusion
The captured PHD2 and DSS data for both runs show very large (up to 10 pixels or more) steps in position when each dither is applied and this seems to result in a systematic (i.e non-random) drift of up to 90 pixels over a 1 hour time period. This does not seem to be consistent with the fact that captured dither values do appear to be roughly random in magnitude and direction.
The data and the quality of the star images and the measured polar alignment accuracy all indicate that any drifts in star position during the actual 2 to 4 min exposure times+ were negligible ( less than 1 pixel). This means that polar alignment accuracy and periodic errors in the mount cannot be the cause of the systematic jumps seen after each dither.
I am note sure how the new lock positions are meant to be calculated when each dither is applied (there is no explanation or even a definition of what lock position means in the PHD2 manual) . I expected that each new lock position would just be the old position plus the new dither but that is not what the log files show.
I have noticed that PHD2 is reporting settling errors after many of the dithers were applied. Is it possible this is causing the drift? If so any suggestions on what setting to change would be much appreciated.
