Rick,

If you use GPG, then the RA section in the lower right isn't used for RA (except perhaps maximum pulse), but rather RA is controlled by the GPG settings in the Guider Settings menu. I guess that should be improved...
As you correctly point out, the DEC settings there are used.

Hy

Thanks for the clarification on that. Good to know.

Hello Hy & Rick,

Yes, that is indeed what I was looking for. However, being a first-timer with the internal guider, there was no "as it was" for me (i.e., I didn't know there is such a tab), and for me the tab name wasn't visible/readable when the default Drift Plot was active

So I gather I need to specify this in ms, and do a proper arcsec->ms conversion myself based on the calibration number. At least for DEC that should be constant, so I've now set it to 45ms, at 128.4ms/" that would be 0.35". Will see next time if that improves the scatter plot shape

Last nights guiding went absolutely stable BTW (6h, including meridian flip). Really impressive!

Just a quick update: Works like a charm with the changed settings:

That is AWESOME! What focal length guide scope are you using?

I started using the internal guider in early 2020, and I too couldn't find those settings, and were told of them, probably by Jo or Wouter.
If I had "guts" I'd just remove them from there, and move all those settings to the guider options menu
To this point I've been hesitant to make changes like that to the UI, as I know people have been using this software for a long time
(so I wind up doing what makes most sense to me for new features, but not disrupting old ones).

Perhaps it is time to rearrange the guider settings?
E.g. I could keep that settings tab there, but just have a note on it saying "moved to ...".

Hy

PS That guiding is incredible! I wish I could achieve that.

Thanks Jo

I use an OAG (ZWO), so it's the main scope = 910mm focal length, guide cam is a 290MMmini binned 2x2, i.e., 1.08"/px for guiding.

Yes, I'm extremely pleased by that performance! (But seeing is exceptional ATM, making things easier)

Hy,

I fully agree, ATM it's a bit chaotic having to adjust settings in several different places. I'd definitely welcome if you take the burden to clean this up a bit. I think at latest with the Analyze Tab you've proven your capabilities in UI design!
A good site helps a lot. But nothing worth without the software

Do it!

Hy,

I'll also endorse combining the guide controls into a single place (either on the guider tab or in Options -> Guider).

A couple of questions about these:

1) Proportional Gain and Integral Gain: What are the units? Is the left column RA and the right column DEC?
2) If I want a very light touch on RA, should I set one of these to about 1? Or about 10? Or about 0.001? I just don't have any sense of this.
3) What's the tradeoff between these two "Gain" settings and the Predictive and Control Gain in the Options -> Guider -> GPG RA Guider? Or are they unrelated?

Thanks!
Scott

TL;DR Control gain: It's unit-less. It's a multiplier on the error. You shouldn't use anything > 1. Probably should be near 0.5.

Detail The idea is that if there's say an error of 1 arc-second North in DEC, then in theory, the guider will pulse the DEC motor enough to move it 1 arc-second South when using a control gain of 1.0. If your gain had been 0.5, it would've pulsed enough for a movement of 1/2 arc-second South. Typically you don't want to try to fully correct these errors, (that is, you don't want to use a gain of near 1.0) because given the noise in the measurements, that tends to lead to overcorrection and bad guiding. So typical numbers are in the 0.5 ballpark, but of course you're free to experiment.

That said, I said "theoretically" for the following reason. The internal guider's DEC guider does not really know how much to pulse the motor to move the mount 1 arcsecond. It has a good guess, a hardwired 133 ms/a-s value (though that's user change-able). It does estimate the number of milliseconds of pulsing to move the mount 1 arcsecond in DEC in calibration, but doesn't use it. So, since in the end it's
correction_pulse_milliseconds = control_gain * error_in_arcseconds * milliseconds_of_pulse_per_arcsecond
and if the milliseconds_of_pulse_per_arcsecond number was too low, you could compensate by increasing control_gain. Doesn't matter which number you change.

I changed a lot of things in the Guider, but one thing I didn't change is the default control algorithm.
Honestly, the reason I didn't is that I didn't want to break something that
was working well (at least for me and others I talked to).

Now, if you are using GPG (for RA guiding), that does use the milliseconds_of_pulse_per_arcsecond
value that was estimated in calibration. So, the control gain for GPG is closer to the control gain I discussed above.

I don't have much experience with Integral Gain, but I assume it's like Control Gain. Integral gain won't be used with GPG. It isn't on the GPG settings page.

For a light-touch on GPG RA guiding, lower its control gain.

For fun, follow along with me on GPG code (remember, I wrote the interface (first 2 steps below, but the entire GPG algorithm is Edgar Klenske's code and Phd thesis, he gets the credit).

Look at:

1) We call gpg->computePulse( ERROR_IN_ARCSECONDS, ...) and then use the pulse value directly (except we make sure it's below the max pulse allowed).
invent.kde.org/education/kstars/-/blob/m...uide/gmath.cpp#L1024

2) ComputePulse calls the GPG library, then multiplies the result by cal.raPulseMillisecondsPerArcsecond() on line 226, and breaks out the direction and the magnitude of the pulse.
invent.kde.org/education/kstars/-/blob/m...alguide/gpg.cpp#L221

3) GPG code.
invent.kde.org/education/kstars/-/blob/m...cess_guider.cpp#L291
After doing all its fancy math and setting up its training and predictions...
line 344: Computes a standard proportional error (control_gain * input_error_in_arcseconds)
line 364: Adds in the prediction term multiplied by the prediction gain.
line 373: Possibly blends with "hysteresis_control" which is a backup control system, but that weight goes to 0 after 1 period or so, so most of the time there is no hysteresis control.

Bottom line, GPG adds the prediction of how much periodic error it expects (weighted by prediction_gain) with a standard proportional error (weighted by control_gain) and outputs that, and that is later multiplied by pulse_ms_per_arcsecond and sent to your mount. Both those controls should probably be somewhere near 0.5.

Hy

100% agree, do it! I've been using the internal guider for years and always wondered why the all-important settings were hidden like that.