i am running Astroberry on a Pi4 with 2 GB but i have serious problems with the new polar alignment when i set the refresh method to "Plate Solve". I use a IMX533 Sensor with my 200 PDS, meaning i have a very narrow FOV of about 40' x 40'. Normally Plate Solver works very well, regardless which one i use. I have tried the internal solver, ASTAP, Local astrometry. Everything fine.
When i try polar alignment 3 images are taken and solved. So far so good.
But then comes the Refresh phase. I set a refresh rate (i think this is the exposure time, set identical to the exposure time set in the align tab) and click on "Refresh" while the widget on the right says "Plate Solve".
EKOS then captures the first image and a text "Image 1" appears between the 2 widgets. After 10 seconds i get the message "Refresh solver failed: 3,4s" in the log area. Regardless what i try: I have no success. I have setup a profile simulating my equipement with the simulator drivers to check what goes wrong.
I have looked in the source code on Github and found 2 strange things:
1. In polaralignmentassistent.cpp i found in Line 137 (PolarAlignmentAssistent::setEnabled) that the Polar Alignment Helper only can be used for FOV>0.5 degrees.
Looks like this value is not checked in the rest of the code.
Is this really a restriction? Why? Plate Solving with my small FOV works very well - why should i not be able to plate solve during the refresh phase of PA?
2. Another point in PolarAlignmentAssistent::startSolver, Line 152 sets the solverTimeout to 10 seconds.
When i use ASTAP as a solver it looks like ASTAP is started during the refresh phase but somehow filled with strange parameters. The call of the external solver does not appear in the "Solution Results" tab nor is the command with parameters logged. But ASTAP opens a small window in the upper left area and runs through different FOV. Looks like there are missing parameters and ASTAP does a complete blind solve.
If this is true, ASTAP needs more than 10 seconds for blind solving (which should be not necessary for just confirming the movement of the polar wedge in relation to the last image taken during the first phase).
For normal Plate Solving you can specify the Maximum Time in the Align Options under "Align Options Profile Editor". Default is set to 600s i think. Why is this value not used during the refresh rate of the PA routine?
And why is the command line used to start the external solver not shown in the log (if logging is set to verbose)? And why do the solver requests during refresh do not appear in the solution results? Looks like the refresh phase of the PA does not respect the solver settings at all? Maybe this leads to errors during solving? Instead it doubles the search radius and sets the timeout to 10s. And there is no display of the error that has caused the solver to abort - just a generic message when there is no success. Should be fixed
I tried rising the log level of the local astronometry but this produces a huge amount of log entries that seem to freeze kstars on the raspi when they are listed in the log area - found no way to ONLY redirect them to a file... I tried different settings for the logs and managed to see a logfile "externalExtracorSolver_?.log" for each of the 3 images taken in the first phase of the PA (and fits files for the images taken). But no log for the refresh phase (but fits files are created in /tmp). Only timeouts with ASTAP after 10s.... Or "Refresh solver failed: ?.?s" with other solvers.
Looks like there is some room for improvement in this area Any hints how i can get this running? Due to my very small FOV the normal "MOVE Star" method is difficult as even small polar alignment errors show a triangle much larger then the visible image. I have to do the polar alignment "old style" a couple of times until the error is small enough to be able to really move a star to the target. I have hoped that this is a lot easier with the new plate solve method - but could not get it working.
Is there anyone out there with a simular setup and a working PA with the new Plate Solve Refresh?
P.S.: I have installed all required and recommended index files and currently try to install the remaining ones - maybe it works with adding this huge files index-4200-* and index-5200-* - but this can take a while over the slow wifi of the raspi;-)
I have tested refresh rates of 1, 2, 3, 5, 10 15s (in simulator) and alway get the message "refresh solver failed". Without a meaningful error message.
I will try tonight under CS I hope but I fear this will not change with my real equipment.
I have tried up to 10s, but really when doing PAA the plate solve usually takes less than 2 seconds. With 'plate solve' for adjusting, it always times out. Is there any debug logging we could turn on to help diagnose?
10s or more lead to a solver timeout as 10s are set in the refresh code as max duration for the solver.
Less than 10s lead to a "failed" without meaningful error message and without entries in the solver table on the other tab.
Tested it today again with my real hardware: No success.
I have the same issue, where the plate solve takes longer then the refresh time and just fails.
Ideally refresh would just be a time interval between the plate solves. This way you can slow down the refresh to not be constantly plate solving and have time to make the adjustments, but also not cause any issues with plate solving if for some reason it does take longer than the refresh rate.
It is unusable to use with a raspberry. it could be very interesting because with a CEM70 you do not have a visive support like an EQ6 where you can use the optical polar scope to align near the pole.
I think that the procedure cold be modified in this way:
take a frame
do a platesolve
retake a frame
In this way the next frame is taken only after the previous solve.
Actually i switched back to the star move + calc error method, but I need to repeat the procedure many times at the first time
athe behaviour on ma windows pc is identical. No timeout but plate solver failing during refresh phase.
This is not a problem of the Raspi, but a matter of implementation. At least meaningful error messages should be forwarded from the plate solver and the solve attempts should be added to the solver list. Currently it is simply impossible to debug the process.
IMHO. The plate solve PA alignment would be better if it used the the set exposure time and the refresh button replaced with a time delay. That would give time to make an adjustment between exposures.
Also, a timer countdown or an array of LED indicators on screen that show how much time remains until the next exposure would help us keep our hands off during a capture and avoid spoiling an exposure.
INDI/KStars on Raspberry Pi 4, 4gb
Raspbian Buster with AstroPi3 script configuration
Skywatcher HEQ5 Pro Mount
Canon 600D Camera
Orion SSAG/ASI120mm @280mm Guide Scope
Waveshare Stepper Motor Board - DIY Focuser
Adafruit GPS Module
Generic Bluetooth Joystick.
Startech 7 port powered USB Hub.
I have the same problem. The first 3 images are resolved successfully, and all "refresh" images failed within a second. Please see attached video.
One thing I notice is that the RA value from the mount is largely different than the RA value from plate solving (of the 3rd image). I am wondering which one is sent to the plate solver for refreshed images, and if that is the cause of the problem.