Periodic Error HEQ5 Rowan Belt Mod - Can PHD2 Cope?

[Ken Richardson]

There appears to be some very knowledgeable people here on this forum. I've been trying to improve the guiding on my HEQ5. So far I have disassembled the mount and replaced all bearings, cleaned all the old sticky grease out of it and lightly greased with SuperLube. After reassembly I adjusted my worm backlash and have it setup about the best as it's going to get. DEC backlash is pretty good, I left a tiny bit of RA backlash and instead focused on keeping the worm gear nice and free turning. I also have the Rowan belt mod on my HEQ5.

The other night I was out and guided for several iterations of the worm gear. I loaded my PHD2 log file into PECPrep and learned a few things.

So that's the cycles retrieved from the PHD2 log file. Here is what FFT tells me.

So the low frequency component is the worm gear as expected for PE. The high frequency component is the belt meshing with the large pulley on the end of the worm gear. I highlighted the small belt pulley on the stepper motor as well but it's contribution is much smaller and likely insignificant.

My problem is the high frequency caused by the belt meshing. Does anyone have any experience with the Rowan belt mod on the HEQ5? Can this be tuned out with belt tension? Are there PHD2 RA algorithms that are well suited to guiding with this type of high frequency variation? That high frequency is a 13.6 second period.

I am hopeful that if I can reduce my PE my guiding will improve by allowing PHD2 to work less or use a PHD2 algorithm to cope with it better. I'm thinking if I apply a high frequency PEC curve to my mount (like below) it may not "sync" up exactly right with the belt mesh and I would be no better off (maybe worse).

[peter wolsley]

Ken,

Can you please post your corresponding guide and debug logs.  Your graphs are great but most of us here are more familiar with studying the PHD2 logs.

Peter

[peter wolsley]

Ken,

I am not a HEQ5 owner but a quick web search yielded this link 

https://stargazerslounge.com/topic/304113-heq5-pro-rowan-mod-problem/

This individual had the exact same 13.6 second oscillation...bottom line was belt tension.  The way I read it...it seems that the teeth on the belt will catch on the teeth of the motor gear if the belt tension is too low.

Good Luck

Peter

On Friday, December 7, 2018 at 6:17:18 PM UTC-5, Ken Richardson wrote:

[Ken Richardson]

Peter,

Thank you for the link! I failed to find that in my search.

I attached the logs. There are multiple guides and calibrations in that file. The PE data came from the last one which is quite long compared to the others. At the end I lost the guide star as a cloud passed by.

Any recommendation would be appreciated.

Ken

[bw_msgboard]

Hi Ken.  The 1 hour session suggests your PE problem is elsewhere:

 

 

 

 

This should be a lot easier to deal with using conventional PEC.

 

Hope this helps,

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.

[Ken Richardson]

Hi Bruce,

Yes, the largest component is the worm gear PE with a period of 638.2. The smaller spike on your chart at 13.6 is the high frequency signature I'm seeing on top of the large PE. For some reason it appears at a much lower amplitude in the software you used. What is that software by the way? In PECPrep the13.6 has a magnitude of 78.0 and the longer 638.2 period has a magnitude of 100.0. Very different than what your chart shows. Which is correct?

Ken

[bw_msgboard]

Andy and I use the PHD2 Log Viewer tool for this sort of thing.  The analysis is done on the guide star excursions seen *without* any of the PHD2 guide corrections applied.  This isn’t entirely accurate because it presumes the mount moved precisely as commanded.  But with RA, this is a pretty reasonable assumption and what we’re interested in is the behavior of the mount before guiding.  I don’t know what PecPrep is doing in this regard.  I think the view I showed you gives you a representation of what sort of work PHD2 needs to do to keep your mount on track.  So it makes sense to me that you’d try to deal with the easier problem first, that being the long-period component.  Once that’s moderated, you can decide whether the short-period component is actually a limiting condition for your imaging. 

 

This is just my opinion, of course, I normally choose to deal with the easier things first.

[Ken Richardson]

So went back and read the EQMOD PEC documentation. As it turns out I was doing it wrong. The documentation states when using a PHD2 log file for PE analysis you must not be guiding. Well I was so likely my result posted above is not valid. It's interesting that while guiding the 13.6 period was so prominent. Not sure what that means? PHD2 was correcting for that interval?

The documentation also states you can use the raw PEC data from EQMOD's AutoPEC that was recorded while guiding. So I did and this gives me a different result. It looks more like what Bruce posted above with only the worm interval to deal with. The 13.6 higher frequency component is virtually non existent.

http://eq-mod.sourceforge.net/docs/eqmod_vs-pec.pdf

[peter wolsley]

Ken,
There are great analysis tools in PHD2 Log Viewer.  The guiding can be analysed in two ways. 1)How tight is my guiding 2)How much is PHD2 having to work.  Bruce showed you the results from the "How much is PHD2 having to work" analysis which is also called "Analyze selected - raw RA"  The first graph clearly shows you that PHD2's RA workload is dominated by correcting for the 638 second PE.  The second graph is a frequency analysis of this workload.  The 638 second PE is roughly +/- 7 arc-seconds in amplitude.  This is a straight-forward problem that PE training can easily address...that's why Bruce has highlighted this issue.  The 13.6 second oscillation is also very significant.  PHD2 is moving your RA axis +/- 1 arc-second in an attempt to correct this.  PE training is not going to correct this issue...it's simply too fast.  That's also true of PHD2...this is too fast for PHD2.

PECPrep's display is similar to PHD2's "How tight is my guiding" analysis or "analyze selected frames".  This analysis only looks at the guidestar deviations which is why PECPrep states that you must not be guiding.  PECprep wants to see how big the errors become without anything else making corrections.  PECPreps charts look very similar to the guiding graphs that you normally see when using PHD2.  PECprep's FFT analysis is the analysis of the guidestar deviations which shows the 638 second oscillation being roughly the same size as the 13.6 second oscillations.  I believe PECprep's display simply scales the Y axis so that the strongest oscillation is always set to 100%.  This is why the 13.6 second oscillation is 78% when compared to the 638 second oscillation.

So you need to work with PECprep to address your 638 second PE issue.  You will also need to make physical adjustments to your RA mechanism...most likely belt tension...to address your 13.6 second oscillation.

Good Luck

Peter

[Ken Richardson]

Peter, Bruce,

You know what? I didn't know about the analysis functions in PHD Log Viewer. Very cool.

I'm not real clear on the difference between "Analyze Selected Frames" and the "Raw RA" charts. Let me think this through...

The Raw chart is showing the star position excluding any corrections so effectively if I disabled guide output this is what the guide star would do? I see a 7.4" magnitude of PE for the worm and 1.0" for the belt mesh. So that represents the mechanical performance of my mount I think? It assumes the mount did exactly what the guide pulse asked for. So if the pulse requests 0.5" movement it assumes the mount actually moved that much. Now I understand what Bruce was getting at.

The other chart shows the star position after the RA corrections so effectively what PHD did to reduce the PE? Here I see the worm is down to 1.1" magnitude and the belt mesh is still 0.9". So is it correct to say that PHD reduced the worm PE by 6.3" by issuing guiding corrections? The belt mesh is effectively unchanged so as you stated, PHD is unable to cope with this higher frequency. Which answers one of my primary questions.

I think I answered my own question but can you confirm that I have it right in my head now?

Thanks! That's a big help! Now to heat the garage up and go watch the belt interacting with the pulley with a microscope. :)

[Ken Richardson]

Here's where I think the 13.6s PE is coming from.

https://youtu.be/cEQntQ-ergE

Notice the tooth on the belt is contacting the corner of the tooth on the pulley and then sliding down the face. Not easy to see in the video but watch close.

I was unable to eliminate this going tighter and looser with the tension. Going to try tighter belt tension on the next clear night. Hard to know what the "right" tension is.

[john gibb]

Hello Ken.

I posted a similar issue "Help or advice on my setup and guiding session" in October.  Sad but I gave up trying to resolve this issue on my mount. I could never adjust the rowan belt to eliminate.

Let me know if you work it out.

John Gibb

[Ken Richardson]

John,

Will do. I sent an email to Rowan as well. I'm going to try and lube the belt next. See if it will slide down the tooth face smoother. The belt is polyurethane and Super Lube is a compatible lubricant. If it fails no big deal, belts are cheap. 👍

[peter wolsley]

Ken,

I think your interpretation is very good.  Hidden in the details is PHD2 measuring pixel values and converting these into arc-seconds and also how well the mount responds to the guide pulses.  There is no direct measurements of arc-seconds...they are all inferred.

I would suggest using your mount's slew commands at their slowest values for studying the belt tension.  I suspect there is a slew speed where you can quickly assess performance and avoid that "waiting for paint to dry" feeling.  

Peter

[Ken Richardson]

Update... Rowan Engineering is aware of the issue as it was reported by a few Customers now. They appreciated my data and have mailed me a "test" 9T pulley which they think will resolve the issue. Since my mount is tuned to the point that the 9T signature is very noticeable and repeatable I guess I've become a tester. ;) Also, they do have a tension spec on the belt. 2mm deflection with a 9oz force at the midpoint of the long side of the belt. It's undocumented because most people do not have a method of measuring a 9oz pull (or push) on the belt. Now that's Customer Service!

Ken

[john gibb]

Thanks Ken for the update.

I don't want your post to go down a rat hole but when you sent the information to Rowan, did they indicate if this is a problem, are they going to contact people who purchased these kits or do we need to contact them referencing your case?

Again great work and I hope the 9T pulley resolves the issue.

John

[Ken Richardson]

Folks,

I received the test 9T pulley from Rowan Engineering. It completely resolved the issue I was seeing. The magnitude of the 13.6s PE is now 0.1" which is effectively lost in the noise. The new pulley is slightly larger in diameter but still 9T. They must have widened the tooth spacing to better match the belt.

John, they did not mention about contacting existing customers. They did refer to the pulley as a "test pulley" so maybe they are not distributing them yet. I provided my feedback and data to them.

Ken

[peter wolsley]

Ken,

Sounds like you got a great Xmas present.  Congratulations!

Peter

[Brian Valente]

nice work and thanks for following up

Brian

--

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 

Brian Valente

portfolio brianvalentephotography.com 

[john gibb]

Great news Ken. Thanks for the update.

I did reach out to Dave from Rowan and he is also going to ship me a test 9T pulley to try. Excellent Support from Rowan....

If it worked for you then it should work for me.

Fingers crossed.

John