4.1 bugs

[Borodog]

Hmm. I never noticed the lunar tracking within SharpCap. The mount is set to sidereal, and when i want to track the Moon I set it to lunar there as well.

Mostly clear night so I will download the latest version and give it a shot.

[Borodog]

First, on the feature tracking scatter plot, RA +ve is shown in black, which makes it impossible to see with the dark theme.

Second, I realized you have not posted an updated beta, but I tried the last one again anyway, with the SharpCap tracking switched to sidereal instead of lunar. I'm not sure it made a difference. The difference between lunar and sidereal tracking is about 3s per arc-second, so about 20" per minute. That would take something like 13 minutes to drift the diameter of Jupiter right now, so I don't think it's too important, but it probably wasn't help.

In any event, polar alignment once again excellent. I always made sure the DEC gear was engaged north before attempting calibration. Minimum movement was always 250 pixels. I tried various combinations of settings, basically at random. After many failures I finally got a successful calibration, the first in 4.1 and the first in weeks (months?), with the following settings:

Movement Speed: 4x
Initial Step Size: 0.5 s
Required Movement: 250 pixels

Do you have a script that can pull out the calibration data from the guide log to recreate the scatter plots in post? I don't have the time to do it manually. There are a bunch of failed calibrations in the log using various settings before I finally got a successful calibration. As expected, once it finally succeeded, guiding worked just fine.

GuidingLog_2023-01-03T20_24_38-19380.log

(182.24 KiB) Downloaded 25 times

[admin]

Hi,

the tracking image in SharpCap should match the setting on the ASCOM driver - SharpCap polls the rate regularly and updates the image as appropriate. Anyway, as you say, the difference isn't significant.

I am hoping to get a new update out in a few days time - missed out on this Monday what with New Year etc

I will dig through the log to see what I can make of it. I don't have a tool written to draw the graphs, but could make one if it looks useful.

cheers,

Robin

[Borodog]

A better option moving forward would just be to log a csv file that could easily be plotted.

[admin]

Hi,

that was exactly what I decided - I put the effort into getting the logging done in a useful way in future versions rather than picking apart the old log files. In fact, if you download and install yesterday's beta build, you will find that it puts something like this into the guiding log after each axis calibration

Code: Select all

Data for YPositive - mount movement (X=RA/Az,Y=Dec/Alt) then image movement (X,Y) then magnitude of image movement
0, 0, 0, 0, 0
0, 20, 0.023571976, 0.353724, 0.35450855
0, 40, -0.050104026, -0.13766795, 0.14650214
0, 60, -1.827749, -17.61481, 17.709383
0, 80, -3.6565957, -35.475647, 35.663597
0, 95, -5.1035013, -48.390415, 48.65879
0, 106.25, -5.978201, -58.143322, 58.44985

You can copy that straight into Excel and then plot charts as required. It will also log this data for all axes that have had calibration attempted if the calibration fails.

cheers,

Robin

[Borodog]

Robin,

Thanks for changing the color of RA +ve on the scatter plot to yellow.

The tolerances for Feature Tracking calibration are simply too tight for my rig. Here's 5 attempted calibrations in a row using the settings that finally succeeded the other night. The first two succeeded, the next three failed. There was no change in settings or any change to anything on the mount between the successes and the failures. The calibration results are not significantly different between the successes and the failures. When calibration finally (randomly) succeeds, tracking works just fine. The tolerances simply don't need to be this tight; it does not really make any difference if the rates are significantly different in opposite directions. It doesn't really matter if any given correction is somewhat high or somewhat low. In fact I don't think we even need to have a single rate for both directions; it's simple enough to have one rate for positive corrections and another for negative.

success1.jpg (376.89 KiB) Viewed 413 times

success2.jpg (390.2 KiB) Viewed 413 times

failure1.jpg (388.78 KiB) Viewed 413 times

failure2.jpg (388.95 KiB) Viewed 413 times

failure3.jpg (388.93 KiB) Viewed 413 times

[Borodog]

Also, the Focus Assistant is completely broken in the latest beta, at least for the Fourier Detail and Edge Detection methods. SharpCap crashed multiple times in a row trying to either adjust the size of the focus box or even simply to open the Focus Assistant. Multiple online crash reports were submitted.

[admin]

Hi,

all the plots look a nice shape, so I suspect it is the scaling in the two opposite directions that is causing the failure. Can you find the numbers that it is complaining about for me, so I get a good feeling for how far I need to push the limits (I don't want to remove it completely as - for instance - a 10 to 1 disparity probably means that something is badly wrong).

I added an option to those focus tools to do the 2x2 binning and colour channel selection - I suspect that is not behaving, which is odd, as I had a good play with it before releasing Let me check the bug reports.

cheers,

Robin

[Borodog]

Success 1:

2023-01-06 19:34:37,258 [Image Process Thread] INFO tracking - Dec +ve gives scale of 166.94135742187518 in direction -39.87298480174262
2023-01-06 19:34:55,957 [Image Process Thread] INFO tracking - Dec -ve gives scale of 152.74303787686387 in direction 136.64606611475185
2023-01-06 19:35:15,407 [Image Process Thread] INFO tracking - RA +ve gives scale of 101.2735053917117 in direction 48.599502375268344
2023-01-06 19:35:29,952 [Image Process Thread] INFO tracking - RA -ve gives scale of 163.95888061523456 in direction -130.85912437173926
2023-01-06 19:35:29,952 [Image Process Thread] INFO tracking - Moving calibration on to Finished

Success 2:

2023-01-06 19:39:16,340 [Image Process Thread] INFO tracking - Dec +ve gives scale of 167.51716657366077 in direction -39.35929791064429
2023-01-06 19:39:34,691 [Image Process Thread] INFO tracking - Dec -ve gives scale of 133.8047072311926 in direction 136.33616726010507
2023-01-06 19:39:55,701 [Image Process Thread] INFO tracking - RA +ve gives scale of 110.70078967866439 in direction 50.24010094286501
2023-01-06 19:40:08,736 [Image Process Thread] INFO tracking - RA -ve gives scale of 183.46889038085942 in direction -130.78220358034298
2023-01-06 19:40:08,737 [Image Process Thread] INFO tracking - Moving calibration on to Finished


All 3 failures were on RA:

Failure 1:

2023-01-06 19:51:43,588 [Image Process Thread] INFO tracking - Dec +ve gives scale of 149.58744419642875 in direction -37.53245663331744
2023-01-06 19:52:03,261 [Image Process Thread] INFO tracking - Dec -ve gives scale of 146.63808387292949 in direction 136.9864364828652
2023-01-06 19:52:23,936 [Image Process Thread] INFO tracking - RA +ve gives scale of 103.97332800002326 in direction 48.83471209742154
2023-01-06 19:52:37,237 [Image Process Thread] INFO tracking - RA -ve gives scale of 195.01723632812508 in direction -130.3261352705321
2023-01-06 19:52:37,237 [Image Process Thread] WARN tracking - Calibration Failed : Measured X scalings of 103.97332800002326 and 195.01723632812508 differ too much

Failure 2:

2023-01-06 19:54:09,256 [Image Process Thread] INFO tracking - Dec +ve gives scale of 153.5838963099889 in direction -39.70350505788541
2023-01-06 19:54:27,737 [Image Process Thread] INFO tracking - Dec -ve gives scale of 133.9115930942477 in direction 135.15846853488688
2023-01-06 19:54:48,311 [Image Process Thread] INFO tracking - RA +ve gives scale of 106.19717915852863 in direction 48.9067677186538
2023-01-06 19:55:02,777 [Image Process Thread] INFO tracking - RA -ve gives scale of 178.24557495117205 in direction -130.94082480425675
2023-01-06 19:55:02,777 [Image Process Thread] WARN tracking - Calibration Failed : Measured X scalings of 106.19717915852863 and 178.24557495117205 differ too much

Failure 3:

2023-01-06 19:58:28,162 [Image Process Thread] INFO tracking - Dec +ve gives scale of 152.7907121930805 in direction -42.791636512920284
2023-01-06 19:58:47,648 [Image Process Thread] INFO tracking - Dec -ve gives scale of 140.47616200889445 in direction 138.81806648309148
2023-01-06 19:59:08,700 [Image Process Thread] INFO tracking - RA +ve gives scale of 104.32918621244883 in direction 49.47018811930663
2023-01-06 19:59:21,915 [Image Process Thread] INFO tracking - RA -ve gives scale of 193.2697570800782 in direction -130.71309616526497
2023-01-06 19:59:21,915 [Image Process Thread] WARN tracking - Calibration Failed : Measured X scalings of 104.32918621244883 and 193.2697570800782 differ too much

[admin]

Excellent, thank you.

All of your figures seem to differ by a bit less than a factor of 2. The current threshold is roughly that the larger figure should be no more than 166.66% of the lower, which just about includes the two that works but rules out the other calibrations. I am going to change this to allow a difference of a factor of two, but not much beyond that.

Interesting that the direction plots for all of those calibrations had none of the 'drift in a direction at right angles) that we saw in some of the earlier runs.

I found out what was wrong with the focus code - with the 2x2 binning and colour channel selection, the code now needs special handling for selection areas that have an odd height or an odd width (odd as in not divisible by 2). I had written this code correctly in one place, but forgotten to make the necessary adjustments somewhere else. That meant it worked fine if the width/height of the selection was even, but not if they were odd.

cheers,

Robin