Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Variable rate guiding vs Pulse guiding

Beginner DIY Dob EAA Equipment Mount Software Video Astronomy
  • Please log in to reply
7 replies to this topic

#1 boscoburrowes

boscoburrowes

    Lift Off

  • -----
  • topic starter
  • Posts: 2
  • Joined: 14 Mar 2021
  • Loc: North Carolina

Posted 25 September 2022 - 04:48 PM

Hello all,

 

I have been in the hobby for two years now. I have a 10" truss dobsonian that I built an VNS style equatorial platform for that is powered by one of the common equatorial motors. My experience with this setup has been great and I have even been able to take decent pictures of galaxies and small planetary nebula.

 

I recently purchased a dedicated astrophotography camera and started learning about the EAA tools and autoguiding. I have searched through the forums and it seems that the standard is to use a pulse guiding method where a software monitors a guide star and sends pulses to the mount to either speed up or slow down as the guide star drifts. This seems to be an extension of the hand controllers people use to move East or West (I'm only considering RA axis for now) to keep objects in the FOV. My understanding is normally the mount has three speeds (West, Sidereal, East) and when there is too much drift, the pulse guiding software will send a direction and a duration that it expects will bring the object back into the FOV.

 

I'm curious what the benefits of the pulse guiding method are over a variable rate method. Do variable rate guiding softwares exist? ASCOM drivers seems to support it with the TrackingRate and RightAscensionRate offset properites. I am writing my own ASCOM drivers to control the motor on my platform and I was wondering how a variable rate method with a PID loop would compare to traditional pulse guiding. I was thinking take the error measurement from the star drift at a periodic rate (1 second or less) and apply to a PID loop that sends a control output to the motor in the form of a PWM duty cycle percentage. There would be no more pulses, just a control output percent that is sent to the mount periodically.



#2 duck

duck

    Viking 1

  • -----
  • Posts: 906
  • Joined: 11 Jun 2020
  • Loc: madera ca

Posted 25 September 2022 - 07:42 PM

No doubt gross polar misalignment could be guided out by the I in the PID controller.  But the errors I've seen when I used to manually guide wouldn't seem to me to be better corrected with anything besides just the P at discrete intervals.  The D is going to be noisy as heck.



#3 555aaa

555aaa

    Vendor (Xerxes Scientific)

  • *****
  • Vendors
  • Posts: 2,512
  • Joined: 09 Aug 2016
  • Loc: Ellensburg, WA, USA

Posted 25 September 2022 - 07:44 PM

Variable rate is better in an encoder mount. I’ve asked some kind soul to add this to PhD but so far so takers. To be clear, the measured parameter is star position(s) from a camera. The drift error is measured and you use that to change the RA and Dec rates in ASCOM but not the tracking rate property which is an enumerated constant (think solar, lunar, king, etc)

Edited by 555aaa, 25 September 2022 - 07:48 PM.


#4 cuzimthedad

cuzimthedad

    Just Be Cuz

  • *****
  • Moderators
  • Posts: 10,668
  • Joined: 09 Apr 2006
  • Loc: Nampa, ID

Posted 26 September 2022 - 10:40 AM

Welcome to Cloudy Nights!



#5 boscoburrowes

boscoburrowes

    Lift Off

  • -----
  • topic starter
  • Posts: 2
  • Joined: 14 Mar 2021
  • Loc: North Carolina

Posted 26 September 2022 - 06:53 PM

No doubt gross polar misalignment could be guided out by the I in the PID controller.  But the errors I've seen when I used to manually guide wouldn't seem to me to be better corrected with anything besides just the P at discrete intervals.  The D is going to be noisy as heck.

I agree, the camera feed is way too noisy for derivative control.

 

Variable rate is better in an encoder mount. I’ve asked some kind soul to add this to PhD but so far so takers. To be clear, the measured parameter is star position(s) from a camera. The drift error is measured and you use that to change the RA and Dec rates in ASCOM but not the tracking rate property which is an enumerated constant (think solar, lunar, king, etc)

Yes, the measured parameter would be star position(s). Each scan cycle, calculate the error from one frame to the next. I understand now the TrackingRate property is the rate selected from those available in the TrackingRates collection (solar, lunar, king). These would be set to the expected duty cycle for those rates. The calculated error would then be used in the PI loop to find the control output for the RightAscensionRate which is an offset from the currently selected TrackingRate. I was hoping this already existed somewhere. I may take a look at the PHD github and see if I can experiment with it.

 

If we are only considering guiding and no go to capability, would an encoder even be needed? The closed loop feedback would be from the reported error from the star position. The motor doesn't necessarily need to know where it is, just to speed up and slow down based on the star position.

 

Welcome to Cloudy Nights!

Thanks! I've been lurking for about three years. Glad to finally start contributing. 



#6 duck

duck

    Viking 1

  • -----
  • Posts: 906
  • Joined: 11 Jun 2020
  • Loc: madera ca

Posted 26 September 2022 - 08:05 PM

With a camera in the loop, you should be able to turn on a control loop from zero set rates and let the I get the rate, the P and D would drive out errors around the set rates, in theory, even without any mount encoders.  But seeing will create problems that can be mitigated using the mount encoders.  Use the encoders for the P and D feedbacks, or use complementary filters to reduce the noise in the image for the P and D feedbacks.  Kind of speculating here, I'm only familiar with airplane controls. 


Edited by duck, 26 September 2022 - 08:05 PM.


#7 555aaa

555aaa

    Vendor (Xerxes Scientific)

  • *****
  • Vendors
  • Posts: 2,512
  • Joined: 09 Aug 2016
  • Loc: Ellensburg, WA, USA

Posted 26 September 2022 - 11:41 PM

OK so I am talking about high precision encoders which are more accurate than star measurements (at least in the short term). If we are talking about lower cost mounts then we still need a way to have a motor than doesn't blow up. So we have two commonly used kinds of motors, stepper motors and servomotors. Stepper motors run open loop but are "all cogging" meaning they run at a known rate based on the step input rate. Servomotors have to have a shaft encoder to control the speed (or position) and the controller inner loop is motor current. So in either case we have to have an inner speed loop which runs the motor at a known rate. We know the speed of the drive motor. We don't know the speed of the output of all the gearing and we don't know the polar alignment errors, flex, etc. Using star images works for all of those and then the choice of pulse vs variable rate comes down to issues like what is better for hysteresis / backlash and you want fairly fast corrections because the motor bandwidth is high (the motor is running hundreds to thousands of times faster than the telescope even when guiding). With a precision encoder on the shaft, there is already an outer control loop which is very accurate, and the bandwidth of that system is far lower, maybe 1Hz or so, and so you want slow inputs. Pulses are high bandwidth inputs.


  • hamishbarker likes this

#8 luxo II

luxo II

    Fly Me to the Moon

  • *****
  • Posts: 6,765
  • Joined: 13 Jan 2017
  • Loc: Sydney, Australia

Posted 28 September 2022 - 05:57 AM

There are other control algorithms besides PID and some can work significantly better in some circumstances, if you know the theory behind them.

 

For a start "floating integral" is in effect what "pulse guiding" tries to emulate. Then there are other analog control strategies like dead-beat, and feed-forward (which is what PEC does, using the periodic  errors measured in the drive train and applying a correction before the scope has moved off target). Dead-beat can be made continuously self-tuning, as well, which would probably obviate a need for PEC.

 

But these are all copies of analog strategies from the pre-digital era.

 

If you have a digital controller running with a fixed periodic cycle there is another - a difference equation - where the output at time T is a weighted sum of the previous P inputs, and a weighted sum of the previous Q outputs. "input" in this case is the error between scope position and the target position, in RA or dec, while "output" is the tracking rate (RA) or north/south in dec. You can add terms for backlash as well.

 

There is no analog equivalent to this controller, and it can perform surprisingly well. When properly tuned it resembles deadbeat. 

 

Mathematically, the output O at time T equals a weighted sum of the current and previous P inputs I(T) ... I(T-P) where a0 to aP are the weights (ie fixed constants), plus a weighted sum of the previous Q outputs O(T) ... O(T-Q) where b1 to bQ are the weights (ie fixed constants). 

 

When the time constant of the system is known (for a telescope with an autoguider, this is the time between updates from the guider) this controller can run with a cycle time equal to the system time constant, and P and Q are typically 2 to 4.

 

Tuning is done by setting all the values of a and b to zero initially, except a0 which is set to some value between 0...1, and is equivalent to the proportional control term in a PID controller; increase this till the system becomes unstable (oscillates) then reduce to 0.7 the value at which oscillation starts. Then do the same with the other constants. Often some of the weights can be left at zero.

Attached Thumbnails

  • Screen Shot 2022-09-28 at 8.47.10 pm.png

Edited by luxo II, 28 September 2022 - 05:59 AM.



CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics





Also tagged with one or more of these keywords: Beginner, DIY, Dob, EAA, Equipment, Mount, Software, Video Astronomy



Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics