Suggestions for Guiding Improvements

[Bill Richards]

I've been using PHD2 for about 1-1/2 years now and just love the tool.  I'm currently using a CEM40 mount and Esprit 100 telescope, autoguiding with an ASI290MM on on OAG.  My tracking accuracy is about 1" RMS on most nights, but I'm wondering if there are changes I could make to improve it.  I've seen some people with tracking plots that are super smooth and mostly within 0.5", whereas mine appear quite "noisy" by comparison.

In the longer runs in the attached log (log entries 4 and 5, each a little over 2 hours long), does anyone see anything I could do to improve guiding performance?

[wave...@talktalk.net]

Hi Bill.ri...,

I'm still learning here, but I suspect a little PEC training would improve your results. There's a noticable PE every 390 seconds or so which is presumably your mount's worm period.

Hope the experts don't shoot me down in flames!

Cheers,

Jack

[Bill Richards]

Thanks for the advice, Jack.  The CEM40 has pre-programmed PEC correction from the factory, and it can be redone manually by the user, but it will only record a single cycle.  From what I've read on various forums, that's better than nothing, but not ideal since the recorded correction curve will contain small anomalies unique to that particular cycle.

When you recommend "PEC training", are you referring to reprogramming the curve inside the mount or doing something with PHD2?

I've also read some conflicting advice about using the built-in PEC at the same time as PHD2's PPEC.  Should I be using just one or the other?

Bill

[Bruce Waddington]

The guiding here is actually pretty good, certainly good enough that you should be getting nice round stars.  I think you’ll need to assess your seeing conditions on nights like this, you might be guiding at the seeing limit.  Looking at your Dec guiding, I suspect that’s where you are.  Of course, a quick improvement would come from upgrading to the dev4 release and using multi-star guiding.   That will probably reduce the guiding RMS numbers but not necessarily your main images.   The RA and Dec guiding RMS numbers are pretty similar so you’re at a point where you need to be careful about further tuning.  If you improve one axis, you’ll probably need to improve the other axis by a similar amount or risk getting elongated stars.  Unless you can develop good evidence that your seeing conditions should support better guiding results, I would leave it alone other than moving to multi-star guiding.

 

Good luck,

Bruce

--
You received this message because you are subscribed to the Google Groups "Open PHD Guiding" group.
To unsubscribe from this group and stop receiving emails from it, send an email to open-phd-guidi...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/f38d7a7c-2dc1-47e9-b283-299192245247n%40googlegroups.com.

[wave...@talktalk.net]

Hello Bill,

Using my learner's eyes, I noticed some residual PE at 399 seconds and wondered whether it could be trained lower. From your comment on this, the manufacturer's PEC programming can only be replaced rather than Imodified. So I think I'd not risk compromising that. Here's what I'm referring to:

Maybe try the PPEC Algorithm set at 399 seconds? I've never tried that one myself, so perhaps Bruce will advise if that will achieve anything?

Cheers,

Jack

[Bill Richards]

Jack - How did you generate those graphs? 

Bruce - Thanks for the feedback and yes, I am getting nice round stars.  But eventually I want to migrate to longer focal lengths and would like to smooth out the guiding as much as possible so I'm trying to understand what's causing the variations that I'm seeing now.  What do you think about trying PPEC?

[Brian Valente]

Bill that's the PHD Log Viewer

https://openphdguiding.org/phd2-log-viewer/

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/180c9805-9990-45d7-b39b-b30f599a0f92n%40googlegroups.com.

--

Brian 

Brian Valente

portfolio brianvalentephotography.com 

[Bill Richards]

Yeah - I have that and use it all the time, but I've never seen the Frequency Analysis plots and can't figure out how to view them.

[Bill Richards]

Never mind - I found it.  Thanks for showing me that this tool was there to use.  It's very helpful.

[Brian Valente]

You first make sure you have the portions of the guidelog you want to view (for example, excluding dithering)

righ-click on the graph, it will give you options to view selected frames, view raw RA, etc. that will get you the window you are looking for

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/f7b62471-39b4-46cb-81a5-55a740f0ebaan%40googlegroups.com.

[Bryan]

Jack

My 2 cents worth...and worth every penny!

The graph suggests periodic error on the order of +/- 5-10 arc-seconds peak to peak.  That is typically the level that a mount's PEC should correct.  I think it will be difficult for PHD2 to manage that much PE.  That said, it only costs a little time to compare the default algorithm and PPEC algorithm.

Also,  PPEC algorithm is not a replacement for internal mount PEC.  That and PPEC work well together, but the mount PE should be set up first. If you think that the factory curve is good, then I would follow Bruce's recommendation.

Bryan

[Bill Richards]

I checked, and the PEC is "off" by default on the CEM40 and I don't remember enabling it.  So my first step would be to enable that on the mount and leave PHD2 alone.  I'll see how that works tonight.  Then maybe the next step will be to enable PPEC within PHD2.

[Brian Valente]

Hi Bill

>>> checked, and the PEC is "off" by default on the CEM40 and I don't remember enabling it.  So my first step would be to enable that on the mount and leave PHD2 alone

Mount-based PEC usually involves a training and correction curve implementation, and does require some knowledge of how it works in your mount. If you aren't familiar with it, i would not enable that until you are.

I apologize in advance if this is obvious to you, but a bad PEC is worse than no PEC at all

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/78f2f4e6-9666-409f-a0ca-2653e5031542n%40googlegroups.com.

[Bill Richards]

The CEM40 comes with the PEC curve already pre-programmed, as shown in the attachment.  Do I need to redo it?

[Brian Valente]

Hi guys

i wanted to add a bit on how to analyze the RA

Most people go to raw RA, which shows the mount performance

But that's half the story - the other half is the straight "analyze selected frames" which shows the residual error

For this mount's raw RA is looks like it's about 7.5" peak-to-peak. that's actually pretty reasonable.  

PHD does a great job on longer period errors like this. 

If you choose "analyze selected frames" from the same graph, you can see the residual error i.e., this is what PHD could not correct

Here we can see that the main period is still the biggest "leftover" part that it could not correct, but it is around 0.4"

If we could get it down to "everything else" that is around 0.15" error, so we're looking for maybe 0.25" of improvement in that main period error

Yes you can do Mount-based PEC to address that, and it will probably do a great job. But you really need to know what you're doing with mount PEC. "Bad PEC is worse than no PEC at all" (i say it all the time haha)

you should be able to make that tiny amount of improvement with some tweaks to the PHD guiding, most notably by switching to the PPEC algorithm for RA guiding. Someone suggested setting that up with a 399 second period, and i think that's a great first step. 

Your guiding is very good and bsed on your guidelog i don't think it's not going to take much to get further

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/b930c94b-cf5b-473b-9725-546593866b1dn%40googlegroups.com.

[Brian Valente]

If you haven't fiddled with it, sure turn it on. if it gets worse, turn it off!

i just posted another email on this a swell

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/f8b5512e-ca6e-429c-b3a0-870b0be7717fn%40googlegroups.com.

[Bill Richards]

Brian,

After doing some research, I am now not convinced that the PEC comes pre-programmed from the factory.  The user guide has come simple instructions for recording it, however.

"In order to use the PEC function, the Go2Nova® hand controller first needs to record the periodic error. The periodic error of the worm gear drive will be used to correct periodic error. 

To use the PEC function: 

1. Setup the mount with a telescope in autoguiding configuration by connecting a guiding camera via the mount’s Guide Port or using the ASCOM protocol; 

2. Select “MENU=>Settings => Set Guiding Rates”. Set a guiding speed from 0.10X to 0.90X. The default setting is 0.50X; 

3. Then press the BACK button and select “PEC Option” from the menu. Use the ▲ and ▼ scroll buttons to display the “Record PEC” option and press ENTER to start recording the periodic error. 

4. It takes the worm gear 400 seconds to make one complete revolution. After 400 seconds PEC will automatically stop recording. The PEC value will be permanently stored inside PEC chip on R.A. motor drive until a new data are recorded."

So if I decide to do this tonight, should I just set everything the way I normally do (polar align, calibrate near the meridian, start auto-guiding), and then execute this procedure?

I would love to get tracking as good as the image shown (also from a CEM40, but with an even heavier payload than mine).

[Brian Valente]

Bill that is great research i'm glad you checked into it

Personally I recommend just trying the PPEC algorithm and seeing how that goes

it's really easy to do, and you can turn it on/off. (just remember it needs at least one worm cycle to start really improving the predictive part)

the reason i mention this is because

A) it may be all you need
B) if it does show some improvement, you know that PEC helps, and you can pursue mount-based PEC as a next step. Conversely, if it doesn't show any improvement, PEC may not help

C) I said this before, doing good mount-based PEC can be tricky. Most first timers mess it up the first few or dozen times (and I speak in part from my own experience). so it may be better to dip a toe and see what happens than to work with the mount PEC

either way, this seems like the right place to be looking

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/3ddf95f3-4df5-4685-a6b6-9d0bc2f94c3cn%40googlegroups.com.

[Bill Richards]

Brian,

Thanks a ton for your advice.  I had already enabled PPEC in PHD2 for my imaging session tonight.  I was going to disable it and try the mount's PEC but based on your advice, I'll just leave PPEC enabled and see what happens.  

So all I need to do after calibration is let it run (guide) for at least 400 seconds before I start imaging?

[Brian Valente]

>>> So all I need to do after calibration is let it run (guide) for at least 400 seconds before I start imaging?

No need to wait. i would use it normally and just image the same way you do

it's not like your guiding isn't already really good

I was only mentioning this because a lot of people look at the first 30 seconds and say "it's not any better" without understanding how it actually works

just do normal and run a full night and then look at the logs in the morning

Also you should disable auto adjust period in PPEC

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/1fa4d04b-05f4-4f53-b607-4c85acae9fa0n%40googlegroups.com.

[Bill Richards]

> Also you should disable auto adjust period in PPEC

Good to know - I saw that feature and noted that it was enabled by default, so just left it.  I will be sure to disable it.

I'll see how it goes tonight and post the results sometime in the next few days.

[Brian Valente]

👍

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/55ed79f5-8a82-4c15-aa3c-e87695f61c88n%40googlegroups.com.

[Bill Richards]

I enabled PPEC in PHD2 last night and made sure that PEC was disabled in my mount.  But it didn't seem to change anything (see images below).

Also, something very strange happened about 12:30AM - the mount stopped tracking and I couldn't get it to resume.  I eventually had to shut everything down and start all over.  Does the guide log contain any details on what might have happened?

[Bill Richards]

Brian or Jack,

Any thoughts on this?  Is there some way to ascertain why enabling PPEC had no effect?

What should be my next step?

Bill

[Bill Richards]

Can anyone help me resolve this?  Enabling PPEC didn't seem to make any difference.  The guide log clearly shows that PHD2 is trying to do *something* with PPEC, but the resulting plots show no difference from the plots without PPEC.

[Brian Valente]

HI Bill

sorry i think i missed this one.

did you (or can you)upload the relevant guidelogs?

--
You received this message because you are subscribed to the Google Groups "Open PHD Guiding" group.
To unsubscribe from this group and stop receiving emails from it, send an email to open-phd-guidi...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/07ab519d-380c-4693-aa8a-079810ce640an%40googlegroups.com.

[Bill Richards]

Thanks for getting back to me, Brian.  The logs were uploaded in the message I posted on Saturday (the one with the embedded plots).

[Brian Valente]

ah okay, i think i found it (leo triplet.zip?)

what was the original guiding you were comparing to and what was that RMS? or maybe better if you can just reply with that specific guidelog

sorry - we go through a lot of data here, so i have to catch up with the past stuff ;)

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/fdef56ef-2661-466f-a598-563731baf197n%40googlegroups.com.

[Bill Richards]

The original guide log was posted in the original message of this thread, but I attached it here for convenience.

[Brian Valente]

got it 

let me look

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/6d9aafc8-31a4-49dc-a7d1-478f435cbf78n%40googlegroups.com.

[Brian Valente]

Hi Bill

they are pretty close for sure, but it's tricky to compare because you don't have exactly the same conditions and run times between nights

If you compare "raw RA" they are going to look very similar because that is the same mount with the corrections backed out. you need to look at the residual error to see how well it was guided out

i have a couple observations:

I think PPEC did do some good here. looking at the residual error, the primary period went from 0.42-3" to 0.35" and the secondary went from 0.23" to 0.15"

there's a period at 37.9 which went up with PPEC, but i think that could do with settings (see below)

your PPEC settings had default aggressiveness which imo is a bit low, but definitely lower than your aggressiveness for hysteresis (the RA guide algorithm for the 'before' run)

i suggest increasing these to control gain = .7 and prediction gain to .7

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/6d9aafc8-31a4-49dc-a7d1-478f435cbf78n%40googlegroups.com.

[Bill Richards]

Hi Brian,

Thanks again for taking the time to get back to me and analyze the data.  You are right - I looked at the corrected data from two previous sessions (2/6 and 3/5), and the corrected frequency domain charts show a peak at 390 ranging from 0.38" to 0.44" in magnitude.  But the charts from 4/2 are better - the 3 long runs had PEC peaks of 0.23" (run starting at 20:36), 0.36" (starting at 22:00), and 0.20" (starting at 23:22).  So the first and last runs were much improved but the middle run was about the same as before.  Do you have any idea why that 2nd run was so much worse than the ones before and after it?

Does the PPEC algorithm have to "re-learn" the adjustments every time you restart a guiding session or does it retain what it learned from prior sessions?

I will make the adjustments to the control and prediction gains that you recommended and give it another try next time we get some clear weather around here.  I've also downloaded the Dev4 version and enabled multi-star guiding. 

- Bill

[Brian Valente]

>>>Do you have any idea why that 2nd run was so much worse than the ones before and after it?

not really. there are a lot of variables there and you are talking a very small amount, so it could be anything

>>>Does the PPEC algorithm have to "re-learn" the adjustments every time you restart a guiding session or does it retain what it learned from prior sessions?

yes, you can control that

hovering over things brings great knowledge!

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/a19c80ab-d6cb-4248-9e6f-266534402ef3n%40googlegroups.com.

[Bill Richards]

OK, cool.  But I don't understand what that percentage does.  Why not retain 100% all the time?  Why is 40% the default and what advantage does that have over, say, 50% or 100%?

[Brian Valente]

That's a good question. i don't have a definitive answer

Maybe Bruce has some input here

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/23b113a5-eee9-4a47-aa55-9e8f03974df5n%40googlegroups.com.

[Duane Melvin]

Hi Bill.  What guide camera and guide scope are you using.

--
You received this message because you are subscribed to the Google Groups "Open PHD Guiding" group.
To unsubscribe from this group and stop receiving emails from it, send an email to open-phd-guidi...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/07ab519d-380c-4693-aa8a-079810ce640an%40googlegroups.com.

[Bill Richards]

Hi Duane,

ASI2600MC-Pro imaging camera and ASI290MM-Mini guide camera on an OAG.

[Bruce Waddington]

This is a complicated question but I’ll try to boil it down.  Nearly all mounts have a basic encoder mechanism of some kind that lets the mount firmware know the absolute rotational position of the worm gear.  It’s a zero-point or registration-point, whatever you want to call it.  Internally, we call it gear-angle.   This value establishes the zero-time point for the periodic error curve.  When the mount firmware supports a permanent periodic error curve, it uses the registration point as the “start” of the curve.  In other words, it always knows the phase of the curve.  PHD2 doesn’t have any way to obtain the registration point info because there’s no standard interface to get it.  So an important part of the training period for PPEC is to measure the periodic error and establish an internal registration point.  The phase is therefore deduced and must have some measurement uncertainty.  When the mount isn’t being guided or is slewed in RA, the original PPEC algorithm always forced another training period.  We later loosened that constraint by implementing this model retention value.  Now, the PPEC algorithm estimates how the phase of the curve needs to be shifted by looking at elapsed time and the size of any change in RA position.  Of course, this also contributes uncertainty to the process, and trying to do this for long time periods or large changes in pointing position will be very inaccurate.   But if the error is reasonably small, it will tend to be self-correcting, which is one of the reasons that the periodic length number can change.  The default value of 40% was based on our best estimates of how much of this adjustment could be tolerated.  I suppose you could experiment with it but I think you could be skating on thin ice unless you’re very careful and really know what you’re doing.   IMO, this parameter should have been placed in the ‘Expert’ group.

 

Hope this helps,

Bruce

 

From: open-phd...@googlegroups.com [mailto:open-phd...@googlegroups.com] On Behalf Of Brian Valente

Sent: Thursday, April 08, 2021 1:31 PM
To: Open PHD Guiding <open-phd...@googlegroups.com>

To view this discussion on the web visit https://groups.google.com/d/msgid/open-phd-guiding/CAJa45i646me349tKvcRz4Qd6jx4AbbpyJo8t4%2B0WbYQZh3B5Mg%40mail.gmail.com.