Hi,
I'm not having much success using the Ekos autoguide module. I've watched the videos and read the webpages and it seems pretty straightforward, but I'm obviously doing something wrong. Here is my procedure:

1. Start Ekos/INDI.
2. I'm using the Telescope Simulator since my mount is dumb (Losmandy G-11 with 492 digital drive). I setup the aperature and focal length for the imaging and guide scope (guide scope is 50mm/~250mm and imaging scope is 90mm/600mm).
3. I select the CCD tab, capture and image and perform the calibration. If I use the default calibration values, I need to select 2x guide rate (which seems extreme)on the 492 drive for the calibration to succeed. If I use a more sane guide rate of 0.5X, I need to bump the max pulse duration up to 3000 for calibration to succeed.
4. Once calibration has succeeded, I capture an image and start autoguiding. Typically, it immediately starts applying wild corrections (which aren't needed) and things get worse and worse until the guide star is lost (presumably because it has moved too far due to the "corrections" that were applied).

This sounds to me like an inappropriate gain value, but I've experimented with different values with little success.

My guide camera is a Meade DSI Pro, my guide port is a homemade ST-4 module based on the ArduinoAstroControl hardware using a hacked up gpusb driver. I have verified that the ST-4 port does what it should be doing so I'm pretty confident that's not the problem.

I noticed the guide module webpage indicates there should be an autoguide log (indilib.org/about/ekos/guide-module.html) that seems like it might be useful in debugging my problem...but the file doesn't exist.

Any suggestions??

Thanks,
Ben

First of all, are you using KStars-bleeding? On what distro/desktop environment? Run the following to get the location of data files:
Have you tried guiding with lin_guider/PHD? If you're building KStars from source, you can enable extensive debug output, just uncomment line 28 of gmath.cpp and recompile again.

I had much better luck last night. I found that making the guide star box smaller seemed to help and I was able to guide for several hours before the guidestar got lost in the trees. I also used a guide rate of 0.5x. The guide graph looked great on one object, but quite bad on another...though maybe that's an issue with my mount or balance.

I'm using kstars 2.2.0 on Linux Mint 17. I'll try kstars-bleeding to see if things improve.

Thanks!
-Ben

Well, nothing changed with respect to guiding in kstars-bleeding, and you wouldn't be able to get kstars-bleeding under Mint 17 AFAIK.

@bgilsrud

This is a while back, but ...

Were you able to narrow down what causes Ekos built-in guiding to over correct?
I am just starting with autoguiding, using OnStep and sending the pulses to the mount directly (no ST4). The correction over compensate, and I do not know how to reduce them.

Try lowering the proportional gain.

Attached are some guiding logs.

What should I lower the Proportional to? 1.11? 1.0?

Should I lower the minimum pulse as well?

So the recommended proportional gain for a 0.5x guide rate is 133x which is the default. What's the guide rate for your mount?

I can indeed see that the mount reacts strongly to the guiding pulses, but I think a big factor here is the time between exposures. 7 seconds is too long when guiding. Ideally, it should be between 1-4 seconds (counting download time as well).

The minimum pulse should be increased, not decreased. So if the guider decided it needs to pulse for 150ms, but the minimum pulse is set to 200ms, then the 150ms pulse is completely discarded. Think of it as a "dead-zone", any pulses below the minimum and above the maximum are ignored

It is set to 0.5X.


That is odd.
I never set it that high. I think it was 4 seconds.
Maybe the download doubles it.

I will lower it to 2 seconds then next time I am out.


So it is better to set the pulse range to very low and very high (say 20 and 5000)?
That way, no discarded instances?

Interesting stuff here. I wasn't aware of these issues (pulse timing and wait times) that might have influence on guiding. I've been using the default times in the internal guider except for bining 2x2 (which seems to solve issues right-away).

Does the "EQ Mount" driver know what should be the best timings for a Skywatcher EQ6-Pro, or should I be testing some other values.

How to find the "guide rate for your mount"?

I'd say so, yes. I do run my mount with 20/1500. 5s pulses are a bit over the top IMO but it won't hurt either.