PHD2 Guiding Assistant recommended min-move settings

[Davor Dvorzak]

Hi everyone

 

I'm a novice in the hobby and currently trying to learn the ropes.

I have a question, well actually several :-), but I can start with this one.

It concerns the min-move values suggested by the guiding assistant.

As I understand the min-move values are suggested as to be just larger than the High-frequency Star Motion, logic of it being that the high-frequency motion is caused by seeing and therefore can't be corrected.

This coincides with what I saw on various forums from people posting print screens of their guiding assistant windows.

However in my case although the High-frequency motions is 0.09 px for RA and 0.04 px for DEC the guiding assistant suggest to set the min-move at 0.24 and 0.30.

Wouldn't this be under-correcting the mount?

I'm guiding on a ZWO178MM connected to an 80/400 achromatic refractor on a CEM60 mount.

 

I appreciate any advice you can give me.

 

Regards

 

Davor  

 

[Bruce Waddington]

Hi Davor,  good question.  But before I answer it, I want to encourage you to send us at least a guide log when you want to know something about a guiding session.  Screen snapshots don’t provide anywhere near the full picture – for example, we don’t know what your guiding image scale is.  Anyway, on to answering your question.

 

The min-move recommendation algorithm has changed since the 2.6.5 release because we found they were consistently too small for most mounts.  As the image scale became finer (focal length got longer), the recommendations didn’t react enough.  Having a too-small min-move value is very common for many imagers so they don’t see it as a “problem”.  But for Dec especially, a too-small value (chasing the seeing) usually causes a high rate of direction reversals.  And direction reversals in Dec are not welcome because they often trigger behavior like backlash delays and sometimes stiction.   Especially on lower quality mounts, the overall Dec guiding degrades badly when direction reversals are common.  As you know, the Dec motor isn’t running at all unless it’s being told to make a corrective move.  In a perfect world, the Dec axis would almost  never need guiding,  but of course we don’t have that luxury.  But with good polar alignment and sound mechanics, Dec guide pulses should be rare, probably occurring in the range of 10-20% of the guide exposures.  When you look at a guiding graph and the Dec axis is getting corrections at a higher rate than that, you probably have the min-move set too low.

 

In the original GA algorithm, we used a high-pass filter to try to sample the seeing fluctuations.  But this proved to not work very well with typical guide cam exposures of 1+ seconds.  So we changed our approach and now measure the history of the guide star movement in Dec, correct it for drift created by polar mis-alignment, then compute the standard deviation of the results.  Assuming the displacements are caused by seeing and are random, we can then compute min-move values that should result in Dec guiding activity of 10-20%.   They will probably seem larger than people expect but that’s probably a good thing, especially for longer focal-length guiding.  Remember, there is an effective floor on the min-move values because we assume the centroid calculation (the position of the star) is accurate to about 0.1px in most cases, so there’s no point in trying to react to star movement smaller than that.  The RA min-move is harder to nail down because the RA motor is running continuously and there are lots of mechanical effects embedded in the data, so we currently apply a reduction term to the min-move value computed for Dec. 

 

With the new approach, there is a potential for getting bogus recommendations if you’re not paying attention to what’s happening during the GA run.  If the guiding hasn’t stabilized or there are large displacements caused by wind gusts or other things that jostle the scope, the measurements aren’t really valid.  One case I’ve seen is when someone ran the backlash test, then immediately repeated the GA run before the mount had recovered from the huge excursions caused by the previous backlash measurement.  In that case, the min-move recommendations were bogus.

 

The min-move values are probably the most important tuning parameters for most people although they often ignore them.  So you can adjust them as you see fit and get your own sense of what works best for your setup.  And of course they are affected by seeing conditions so they often need to be adjusted for that.  I normally run the GA at the start of each imaging session to get a good starting point.

 

Sorry for the long-winded response, but the question is very topical because of the changes that show up in the 2.6.6 release.  And don’t hesitate to ask questions, we’re happy to try to help.

 

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.
For more options, visit https://groups.google.com/d/optout.

[Brian Valente]

Super helpful and insightful Bruce. Thanks

Brian

--

Brian 

Brian Valente

portfolio brianvalentephotography.com 

[Davor Dvorzak]

Dear Bruce

thank you for a very detailed answer. I certainly have some food for thought and further analysis. 

Here is the guide log from the session. As you can see I was fiddling quite a bit with the settings and experimenting also with the guide rate. That is why there are two calibrations in the session. The guiding was generally OK with total RMS around 0,65", before I started the assistant. 

I would be grateful for any advice or input you might have.

https://openphdguiding.org/logs/dl/PHD2_logs_nrld.zip

Best regards 

Davor

[Andy Galasso]

Hi Davor,

We can see from the guide log just what Bruce was describing.  Your Dec min-motion value of 0.20 pixels was a little too aggressive (too low) causing somewhat frequent unnecessary dec axis correction direction reversals:

Your mount is very well-behaved, so the the consequences of the back-and-forth dec corrections were quite minor, only a few very small over-shoots here and there.   Other mounts would have a lot more trouble than yours did.

Later when you ran the guiding assistant it gave you a recommendation of 0.30 px for Dec min-move.  That would have smoothed out the dec guiding.  You can tweak the min-move settings a bit as Bruce describes.

Your RA guiding was also quite good, mostly seeing-limited, but there's room for a little improvement there too.  A couple things to try:

   a)  lower the exposure duration a little bit, to about 1.5 to 2 seconds and restore the RA aggressiveness to the default value of 70%. This will allow the guiding to react more quickly to the RA error than you were getting with 3.0s exposures.

   b) you may want to try the Predictive PEC guide algorithm option on RA, I think it would work well with your mount.  Give it a try, and if it looks good you can try increasing the predictive weight to around 80% or so.

Andy

[Bruce Waddington]

Hi Davor.  Congrats, this is really good guiding, you’re already in good shape.  Since you seem to be interested in the details and the parameters, I recommend reading this:

 

https://openphdguiding.org/tutorial-analyzing-phd2-guiding-results/

 

I think a lot of the experimentation you were doing with the low-level parameters was either futile or counter-productive.  Your mount is actually performing very well, so you aren’t faced with most of the problems that plague other beginning imagers.  I can see from the guide log that the Dec activity level was right about where you want it.  Most of the direction reversals were self-correcting, which means they were seeing induced.  I have a couple of suggestions to try after reverting the guiding parameters (excluding min-moves) to their default values.  First, try running the PPEC algorithm in RA – figure out the native worm period and enter that value as the ‘period length’ parameter.  Then let it run for a good long while and see how the RA guiding works.  Second, you can try using hysteresis on the Dec axis because your mount’s Dec backlash is so low.  This should improve the responsiveness when there is a reversal that warrants a guiding correction.    But as you do these things, keep an eye on the RMS values of the two axes and don’t let one get a lot better than the other.  If that happens, you’ll start to get elongated stars and that will just defeat the purpose.  Details depend on your imaging setup, but differences of more than, say, 50% over the life of a main camera exposure may lead to visible elongation.

 

Have fun,

[Davor Dvorzak]

Dear Bruce and Andy

Thank you both for the great advice. I really appreciate it. I can’t wait for some clear skies to tryout the settings you suggested.

Best regards

Davor

[Davor Dvorzak]

Dear Andy and Bruce

 

In the recent weeks I continued to play with PHD2 settings and basically I found that the single most influential factor on my guiding performance is the quality of the selected star, meaning the SNR and the HFD.

It differs from a total RMS of around 0,55" with good SNR and relatively low HFD (tapered spike in the Star profile display) to anywhere up to 1.2" if the SNR is low or the peak is clipped in the Star Profile, all other things being the same.

All the other factors (MinMo, Aggression even the choice of Algorithm) change the total RMS maybe 0,1" to 0,2" up or down.

 

1) Having said that, my first question is do you have any tips on how to maximize the SNR and get low HFD (better focus, tighter stars)?

Currently my guide scale is 1,24"/px. I'm guiding on a 80/400 achromatic refractor and the guide camera is ZWO 178 MM. 

I should also mention that my light pollution is pretty bad, sky brightens ranges from 18,30 up to 19,30 magSQM/arcsec2 (on good nights). 

 

I have been experimenting with binning the camera.

2) Would binning the camera yield better SNR at the expense of some resolution?

3) If I bin at 2x2 would an image scale of 2,48"/px still be OK with regard to resolution?

4) Should I apply the noise reduction to boost the SNR?

 

Also I have some difficulty focusing.

The focuser on the Omegon AC 80/400 is very basic but I try to focus using the FWHM tool in SharpCap and bring down the FWHM values as low as it goes. However I still get a suggestion from the "guiding assistant" to try to improve my focus.

 

5) Do you think a semi APO filter from Baader or a similar product would be useful to get the stars a bit tighter?

It would probably come at a cost of some signal coming in but maybe it would be worth the effort.  

6) Do you think an upgrade to an APO as a guide scope would give better results regarding SNR and focus?

 

Finally I have a question regarding PEC training the mount. I have given it a go and the first results do show some promise.

I have first recorded the PE curve of the mount using PemPro. (I run it for 5 cycles, one 30 min session) and created a PEC curve.

Now I know that some people had trouble running PHD2 together with the CEM60 internal PEC, myself include.

I found that when you run the CEM 60 internal PEC the mount just becomes unresponsive to RA guide commands sent by PHD2.

My workaround idea was to record the PEC curve in PemPro and connect PemPro to the mount using ASCOM POTH.

Then playback the PEC from PemPro and also connect the mount to PHD2 using POTH so not to get conflicted commands from PHD2 and PemPro. (I wanted to avoid the mount getting instructions from PemPro via ASCOM POTH and from PHD2 via Ioptron ASCOM driver at the same time).

I have done this last night and running the PHD2 with the PemPro playback I got some really nice regular periodic shift in my RA graph. (See the screen shot) 

I know it might be due to a million other factors but it seemed to me that the errors occurring in the RA were more regular and smoother. I intend to look into it a bit further.

Also I run PemPro playback with the hysteresis algorithm in PHD2 and it would probably work even better with the Predictive PEC algorithm.

Probably the best way to test the effect of PemPro PEC would be to disable "mount guide output" in PHD2 and watch the RA graph with and without the PemPro playback.

 

7) Any way my question is if you have some suggestion towards how to get a better PE curve in PemPro.

 

Please find enclosed the guide logs from last night as well as a screen shot of the guiding with PemPro Playback as well as the PemPro .ppc and .txt files.

 

Any advice you might have regarding these issues are mostly appreciated.

Also sorry for the lengthy email and the many questions but what I found about this hobby is that answering one question just leads to two or three new ones. J

 

Best regards 

 

Davor  

 

[Steve Winston]

I've found that local seeing has a huge impact on HFD, assuming your focus is good.  PHD2 though is very good at determining the centroid of the guide star, so it shouldn't really matter too much.

re: image scale, given you are already at 1.24", I would not bin.

[bw_msgboard]

Hi Davor.  I think you are probably chasing smoke here. J  Higher SNR on the guide star does help with the centroid calculation but that calculation rarely is the limiting factor for guiding.  What you’re discussing is probably the side effects of seeing conditions.  In poorer seeing, star profiles become distended and more variable and may even develop a “flat top” even when the brightness is nowhere near saturation.  If you’re working in an urban or suburban area, your seeing conditions can be considerably worse than whatever is forecast by sites like ClearSky Chart, and seeing-induced variations in guiding RMS of 0.5 to 1.2 arc-sec are not unexpected.  If your mount is capable of guiding with 0.5 arc-sec RMS accuracy, it is going to need good seeing conditions to do it consistently.  

 

There’s a short intro to seeing and its effect on guiding in the appendix of this document: https://openphdguiding.org/tutorial-analyzing-phd2-guiding-results/

 

This is a big topic and there are many other references on the web depending on how much math and detail you want to get into.  But you’ll need to remember that the formal models for seeing – the ones used by professionals – are focused on high altitude atmospheric phenomena.  This is the major reason people locate observatories at high elevations.  In an urban/suburban area, there is much more contribution from surrounding structures – buildings and trees that break up laminar air flow, flat surfaces that radiate heat long past sunset, etc.  

 

You can certainly experiment with binning but I wouldn’t expect it to help you much – neither would it hurt you.  

 

I don’t have any insight into what PemPro is doing or how you’re trying to use it.  I know that with some mounts, the “playback” is ignored by the mount firmware once one worm period has elapsed.  I think you’d be better off trying to figure out how to program the mount with a good PE curve and not try to trick it with playback sessions. Using the PHD2 PPEC algorithm would also be a logical thing to try.

 

Good luck,

Bruce

 

---