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

CEM60 or CEM60-EC

  • Please log in to reply
758 replies to this topic

#51 John Miele

John Miele

    Gemini

  • *****
  • Posts: 3452
  • Joined: 29 May 2005
  • Loc: North Alabama

Posted 28 June 2019 - 12:26 PM

So last night was quite a desaster frown.gif

I'll still have to go through the logs, which is somewhat a mess as PHD2 kept crashing a lot.

 

But the essence is that if I run unguided there is a drift of ~1.2 arcsec per minute in RA.  I have no idea where that is coming from.  ISTR someone had mentioned a similar issue in the firmware update thread - will have to look for that.

The drift-corrected errors in that unguided phase seem reasonable - something like an arcsec PtP, so around 0.3 RMS.  But as soon as I start guiding it begins to oscillate, amplitude increases to 2.5-3 arcsec PtP, RMS 1 arcsec.  While this might sound not too bad, it is (persistently over many hours) a factor 4 larger than the DEC RMS.  The spot diagram doesn't even look like a football - it's a cigar.

 

I had experimented throughout the night with various guide algorithms, aggressiveness settings, exposure times.  Some settings seem somewhat better than others, but never did I get substantially below 1" RMS.

 

One of the unguided periods suggested that there was no 5.5s peak at all - only a 2.8s one.  That one is also visible in several other power spectra, together with the 5.5 one.  I wonder if that is the base and 5.5 is a harmonic?  I had tried 2s exposure and adjusted the 'Time Lapse' to get a 2.8s update frequency.  My impression was that was one of the better settings.

 

Maybe someone has additional input/suggestions - would be highly welcome.  I'll post more details with logs later.  Right now I start wondering whether I should better have gone for the non-EC one frown.gif

Hi Der_Pit!

 

Sounds like you are experiencing exactly what I have been gong through with my 6 month old CEM60EC...

 

1. Guiding in DEC is excellent

 

2. Unguided tracking is excellent with low RMS (0.3 arc-sec) but there is a slow steady drift. I just discovered that recently.

 

3. RA guiding goes crazy with completely random large amplitude excursions not related to seeing that will not damp out. I have tried long guide durations high min motions and low aggression and everything in between and nothing will produce consistently good RA guiding.

 

You and I are not the only ones to see this. There are other posters saying they essentially gave up RA guiding their CEM60EC mounts and guide only in DEC or just run unguided. The mount can do very well unguided, but then dithering becomes an issue. I did not buy the mount to only use for unguided imaging. I venture we both bought our mounts expecting them to effectively guide in RA.

 

And the strangest thing is that the very first time I used the mount 5 months ago, it guided great in both RA and DEC the first night! It has never again guided well in RA since that first night. This strongly leads me to believe the solution will have to be software/firmware related.

 

cs...John


Edited by John Miele, 28 June 2019 - 12:30 PM.


#52 Der_Pit

Der_Pit

    Viking 1

  • -----
  • Posts: 860
  • Joined: 07 Jul 2018
  • Loc: La Palma

Posted 28 June 2019 - 12:46 PM

Hi John!
 

Sounds like you are experiencing exactly what I have been gong through with my 6 month old CEM60EC...


I was indeed thinking of you, as in the beginning I had claimed mine is working fine - which most likely only was because of the high declination object(s) I was observing frown.gif
 

1. Guiding in DEC is excellent
 
2. Unguided tracking is excellent with low RMS (0.3 arc-sec) but there is a slow steady drift. I just discovered that recently.


So you also see that? Have you measured it? Is it in the same ballpark as what I see (1.3"/min)? Would be a strong hint to some real firmware issue...
 

3. RA guiding goes crazy with completely random large amplitude excursions not related to seeing that will not damp out. I have tried long guide durations high min motions and low aggression and everything in between and nothing will produce consistently good RA guiding.

I wonder if you already tried tackling the drift by changing tracking rate.  
 

You and I are not the only ones to see this. There are other posters saying they essentially gave up RA guiding their CEM60EC mounts and guide only in DEC or just run unguided. The mount can do very well unguided, but then dithering becomes an issue. I did not buy the mount to only use for unguided imaging. I venture we both bought our mounts expecting them to effectively guide in RA.


I bought it mostly because I couldn't get the DEC guiding of my GP-DX to work.  RA was fine, but without DEC dithering....
 

And the strangest thing is that the very first time I used the mount 5 months ago, it guided great in both RA and DEC the first night! It has never again guided well in RA since that first night. This strongly leads me to believe the solution will have to be software/firmware related.

Same here, my first run was great....

One thing I'm wondering about is the influence of the meshing.  That is somewhat a variable, how much you close/open the knob, and likely the same setting does different things at different temperatures.



#53 OzAndrewJ

OzAndrewJ

    Mercury-Atlas

  • *****
  • Posts: 2887
  • Joined: 30 Nov 2010

Posted 28 June 2019 - 05:30 PM

Gday DerPit

 

Ah, so you say the drift is a known old thing?

It was the biggest error seen in the early tests done by 3 independent users.

No real idea of how the later units work as there isnt much recently posted data of long duration unguided logs.

( Drift tests can be done at a normal framerate so dont require special setup )

 

So do you have a clue were the drift comes from?

Nope. It could be the feedback loop or it could be an eccentric encoder.

I dont know what data you can read from an IOPtron, but with the Syntas, we can run the motors manually and read back the raw encoder and microstep counts. By comparing the 2 we can see what the real tracking rate is based on delta microsteps over time. The Synta encoder is far cruder than the IOptrons but it is useful for doing rate tests.
 

You mention the calibration - I assume that is the encoder calibration after FW update?

I believe thats what it is called. One of the three users ( Mr Whang) reported using it when it was first released and it appeared to help with drift but didnt fix his SDE. That was a year or more ago, so no idea whats happened since.

 

 

For the CEM60EC, you can indeed disable guide signals, but they are on by default.
I started wondering if it might improve things to use the ST4 port instead.

I wasnt talking about "you" disabling the signals, i was thinking that if the mount itself uses an averaging function on the encoder, it might disable the encoder, do the guide, recalc the average, restart the encoder.

If the second guide arrives before it does the recalc and restarts, then the encoder is effectively disabled.

As to the ST4 port, worth a try. I know in my AZEQ5 that with PPEC ON, if i pulseguide, i have to disable PPEC, guide, enable PPEC for it to work properly, but if ST4 is used, the motor cards deal with it internally.

Maybe yours works the same way with the encoder ???

 

Andrew Johansen Melbourne Australia



#54 OzAndrewJ

OzAndrewJ

    Mercury-Atlas

  • *****
  • Posts: 2887
  • Joined: 30 Nov 2010

Posted 28 June 2019 - 09:26 PM

Gday Der Pit

 

It's enough if the correction arrives at the mount half a phase too late - that way it will double/amplify any inherent periodic error.

Just thinking on this a bit more.

After doing a normal calibrate, can you set the guide app to have a very small min move AND also a very small max move

In this scenario, the guider will never be able to send a large correction, but it will send something that the mount has to deal with and process.

If the RA still goes "relatively" crazy, then its not due to the guide pulse being out of phase

as the pulse would be too small to actually show any effect.???

 

Andrew Johansen Melbourne Australia


Edited by OzAndrewJ, 28 June 2019 - 09:27 PM.


#55 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 8837
  • Joined: 12 Oct 2007

Posted 29 June 2019 - 07:52 AM

With regard to RA drift - if it seems like the RA rate is wrong, one way to remove possible effects of refraction or polar alignment would be to leave the mount tracking and simply time how long it takes to turn 90 degrees in RA with continuous tracking.  Ideally you would want to have raw axis angle values for the ra axis, rather than hour angle, so it doesn't include a sky alignment model and it is just the physical angle of the axis.

 

Unless the tracking is at King rate or something, the rate you calculate should match sidereal rotation, one 360 degrees in 23 hours 56m 4s.  Measuring over 6 hours and noting the exact time it rotates through 90 degrees should give an accurate value for the true rate.

 

Frank



#56 Der_Pit

Der_Pit

    Viking 1

  • -----
  • Posts: 860
  • Joined: 07 Jul 2018
  • Loc: La Palma

Posted 29 June 2019 - 08:46 AM

After doing a normal calibrate, can you set the guide app to have a very small min move AND also a very small max move


The small minmove is what am usually doing when running fast cadence. Also limiting max I've never tried. But it would make dithering very cumbersome....
 

In this scenario, the guider will never be able to send a large correction, but it will send something that the mount has to deal with and process.
If the RA still goes "relatively" crazy, then its not due to the guide pulse being out of phase
as the pulse would be too small to actually show any effect.??? 

 My impression so far is that any size of correction triggers a counter-correction by the mounts internal control loop.  The example was rather more general, to explain why processing delays can strongly affect the efficiency of a control loop.
 

With regard to RA drift - if it seems like the RA rate is wrong, one way to remove possible effects of refraction or polar alignment would be to leave the mount tracking and simply time how long it takes to turn 90 degrees in RA with continuous tracking.  Ideally you would want to have raw axis angle values for the ra axis, rather than hour angle, so it doesn't include a sky alignment model and it is just the physical angle of the axis.

 Oh, interesting approach. I'm afraid though it would be limited by the accuracy that I could determine the 90 degrees.  And I shouldn't neccessarily trust the mount coordinates - it could be a self-consistent error, isn't it?

 

Unless the tracking is at King rate or something, the rate you calculate should match sidereal rotation, one 360 degrees in 23 hours 56m 4s.  Measuring over 6 hours and noting the exact time it rotates through 90 degrees should give an accurate value for the true rate.

I had a more successful run yesterday, manually tweaking the tracking rate.  Going to 15.043 (instead of the nominal 15.041) I could get good tracking without RA drifts.

Maybe King rate would also have worked (it also accelerates while rising, is it?  I had first got that wrong....).  But then again, it was like that more or less up to zenit.  I'll have to read up about King....

 

Still, with the changed rate I could get 0.6/0.3 arcsec rms guiding.  As long as phd2 didn't try to guide :(

I was running with fast cadence (0.5s, subframe readout) and had set minmov to 0.8 (somewhat more than an arcsec) as that was the max amplitude the SDE oscillations reach unguided.  One can easily see that guiding is substantially better as long as no guide pulses occur in RA.  As soon as they do, strong(er) ascillations start, and they really give the impression the mount wants to pull back to the position it was before.  Somewhat like already having predictive positions where to be, and just going there no matter from where :o

 

For those interested, last nights log is here (too large to attach):

ftp://pit.changeip.net/PHD2_GuideLog_2019-06-28_220327.txt.gz

2nd set shows the play with guide rate and the effect on the drift.  Zooming in on the long sets shows the described reaction to guide pulses...



#57 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 8837
  • Joined: 12 Oct 2007

Posted 29 June 2019 - 08:52 AM


Oh, interesting approach. I'm afraid though it would be limited by the accuracy that I could determine the 90 degrees. And I shouldn't neccessarily trust the mount coordinates - it could be a self-consistent error, isn't it?

I wasn't clear how far off the RA appeared to be. Refraction and polar alignment will make the RA rate look wrong if it is based on drift of a star.

As long as the drift is small it is no surprise and is expected to be there due to the reasons stated. And correcting it in RA is easier than in Dec. with guiding.

But, yes, I think measuring the true rate should be very accurate by timing rotation over 6 hours. It won't be affected by polar alignment or PE or anything - as long as you can read off the raw RA axis angle.

But it may not be needed if the RA rate error appears to be small. It could just be refraction or polar alignment issues.

Frank

Edited by freestar8n, 29 June 2019 - 09:33 AM.


#58 555aaa

555aaa

    Vendor (Xerxes Scientific)

  • *****
  • Vendors
  • Posts: 1442
  • Joined: 09 Aug 2016
  • Loc: Lynnwood, WA, USA

Posted 29 June 2019 - 11:23 AM

A drift error of one arc second per minute is about a one in one thousand rate error. It's possible that this is from temperature variations plus manufacturing variations in the oscillator that drives this axis. It's also likely from pa misalignment. If you are on the meridian and dec= 0 and you have no north south drift then ra drift is only from altitude adjustment error. I think that if it's too fast then it can only be oscillator error because there is no alignment error that make the mount run faster at that point. There's nothing wrong with running at an adjusted ra rate and it would be a nice feature for PHD if it would use the ASCOM variable tracking rates for mounts that support it when it sees drift.

#59 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 8837
  • Joined: 12 Oct 2007

Posted 29 June 2019 - 11:54 AM

If it really is 1" per minute that is pretty negligible.  If you correct every 5s that is about an error of 0.08" per correction.  In addition, since each correction will lead or lag by the same amount - the impact on the image will be even less (about half the actual error).

 

And if you time it over 6 hours of rotation, 90 degrees would happen early or late by about 3.5 minutes - so it would be easy to measure.

 

But a rate error that small could be dependent on temperature - and is to be expected.

 

As for correcting it while guiding - I doubt it would show at all either way if it really is that small.  Especially if you correct every 1s as I do - in which case the RA error introduced is only 0.017" and in the noise of the centroid error.

 

Frank



#60 Der_Pit

Der_Pit

    Viking 1

  • -----
  • Posts: 860
  • Joined: 07 Jul 2018
  • Loc: La Palma

Posted 29 June 2019 - 12:14 PM

I wasn't clear how far off the RA appeared to be. Refraction and polar alignment will make the RA rate look wrong if it is based on drift of a star.

As long as the drift is small it is no surprise and is expected to be there due to the reasons stated. And correcting it in RA is easier than in Dec. with guiding.

At least if the mount co-operates with guiding that is. Which it isn't frown.gif

 

Difficult to say if it is only refraction.  I've never seen it before when using my old mount.  Then again that one had a (substantial) periodic error that might have buried the drift.  But for sure the drift had been one major ingredient of the bad guiding performance the day before...



#61 Der_Pit

Der_Pit

    Viking 1

  • -----
  • Posts: 860
  • Joined: 07 Jul 2018
  • Loc: La Palma

Posted 29 June 2019 - 12:45 PM



As for correcting it while guiding - I doubt it would show at all either way if it really is that small.  Especially if you correct every 1s as I do - in which case the RA error introduced is only 0.017" and in the noise of the centroid error.

This is for sure true for standard conditions.  But look at this:

 

guiding.jpg

 

Look at the period starting 03:58.  This is (IMHO) quite nice guiding.  No correction pulses.  As soon as there are corrections, amplitudes increase and an oscillation starts.  Eventually it reaches the 'dead zone' again where no guide pulses are sent, and guiding is much better again.  This pattern is easily seen repetitive in the guide log.  

There is something fishy in the way the EC mount handles guide pulses.  Yes, one can find better parameters that help gilding the excursions.  But you'd rather want to remove the cause, not the symptom...



#62 John Miele

John Miele

    Gemini

  • *****
  • Posts: 3452
  • Joined: 29 May 2005
  • Loc: North Alabama

Posted 29 June 2019 - 01:31 PM

Here are two examples of my RA drift. I disabled the RA guide corrections and just let the mount run for a few minutes. The drift is small but there. The first plot is at Sidereal rate and the second plot is with King rate. Notice how small the RA error would be if you corrected that drift. The encoders do a good job at that.

Attached Thumbnails

  • drift siderial.jpg
  • drift.jpg

Edited by John Miele, 29 June 2019 - 01:32 PM.


#63 OzAndrewJ

OzAndrewJ

    Mercury-Atlas

  • *****
  • Posts: 2887
  • Joined: 30 Nov 2010

Posted 29 June 2019 - 04:39 PM

Gday DerPit

 

The small minmove is what am usually doing when running fast cadence. Also limiting max I've never tried. But it would make dithering very cumbersome....

I wasnt suggesting it as a permanent setting, just a quick test

ie, if the act of sending a guide triggers the problem, if you had a min move of say 0.1 and a max move of say 0.2

then it will constantly chase the seeing but the pulses sent would probably be to small to be "seen" in the data

If you stil get 1 or 2arcsec oscillations from a max move of 0.2, then its not the guide itself that made it move that far.

Andrew Johansen Melbourne Australia


  • Der_Pit likes this

#64 OzAndrewJ

OzAndrewJ

    Mercury-Atlas

  • *****
  • Posts: 2887
  • Joined: 30 Nov 2010

Posted 29 June 2019 - 08:42 PM

Gday Der Pit

But it would make dithering very cumbersome....

And i see you like your dithering lol.gif

Just looked through all yr data as raw output ( hence can see the amount of dithering )

The very last run in the log appears to highlight the problem well

Have attached an expanded view of it from the PHDlogviewer

and also a precis of the log file showing only the RA guides.

In all cases, the first RA guide after a settled period starts the oscillation

( I just expanded the graph in PHD viewer and read the frame no as i moved the mouse to see what frame the oscillation started at )

Virtually all guides were listed as being 40-50ms ( so if i have my calcs right, that means 0.25arcsec ), so no way it could create the large excursion by itself.

All very intriguing,

Andrew Johansen Melbourne Australia

Attached File  GuideTest_2019-06-28_220327Oscillate.txt   1.58KB   6 downloads

Oscillations.jpg

 

 

 

 

 

 

 

 



#65 RossW

RossW

    Vostok 1

  • -----
  • Posts: 112
  • Joined: 15 Jun 2018
  • Loc: Lake Biwa, Japan

Posted 29 June 2019 - 10:59 PM

 

Look at the period starting 03:58.  This is (IMHO) quite nice guiding.  No correction pulses.  As soon as there are corrections, amplitudes increase and an oscillation starts.  Eventually it reaches the 'dead zone' again where no guide pulses are sent, and guiding is much better again.  This pattern is easily seen repetitive in the guide log.  

There is something fishy in the way the EC mount handles guide pulses. 

 

Hello Der_Pit. It has been reported in numerous threads here on CN and iOptron forums that the EC mounts (60s and 120s at least) do not like being hit by rapid-fire guide pulses. And nor should they need to be hit with rapid-fire guide pulses if the mount is working correctly, IMHO. You have virtually no periodic error to correct, your PA and RA tracking rate errors appear to be very small, the CEM60 has next to no Dec backlash (when adjusted correctly), and I assume you don't want to chase the seeing, so I can't understand why you wish to use 0.5 second guide pulses and zero min-move. Perhaps you can explain your motivation to do so? Apologies if I have missed something obvious. If for some reason you need to guide rapidly (AO guiding?) I think the EC mounts would be a very poor choice.

 

I may be wrong, but I think the poor RA guide performance you are experiencing is due to your choice of using a very fast rate of guiding. Looking at your guide logs you're certainly chasing the seeing at 0.5s exposures, and even a 2-sec guide rate appears to be too fast for EC mounts. You also have your min-move set at 0, so your mount will definitely be interrupted by a guide pulse every 0.5s. Would you mind retesting your mount using, say, 5 to 10 second exposures and a reasonable "seeing-based" min-move setting, and uploading the PhD guide log?

 

By the way, your SDE measures to 0.2" RMS in your latest 2019-06-13 guide log, so it looks like you have a very good mount in that regard:

 

Der_Pit.png

 

Cheers,

 

Ross


Edited by RossW, 30 June 2019 - 04:26 AM.


#66 RossW

RossW

    Vostok 1

  • -----
  • Posts: 112
  • Joined: 15 Jun 2018
  • Loc: Lake Biwa, Japan

Posted 29 June 2019 - 11:08 PM

Here is the FFT measured from log section #3 (the previous graph was from section #1).

0.1" RMS SDE is a fantastic result:

 

Der_Pit2.png



#67 OzAndrewJ

OzAndrewJ

    Mercury-Atlas

  • *****
  • Posts: 2887
  • Joined: 30 Nov 2010

Posted 29 June 2019 - 11:43 PM

Gday Ross

The high framerate was so any residual SDE can be detected, but as part of testing the users noted that the mount would start to oscillate badly.

Have attached a small section of run 8 where i analysed only the oscillation section, and according to the viewer, it gets up to 2.5arcsec pk-pk over a minute or more, vs the averaged 0.2 sec for a full data set.

The low min/max move test was to see if the act of sending a tiny guide would cause major oscillation, to exclude the theory that a slight delay in the guiding may be reinforcing/harmonising with the encoder feedback loop.

In all cases, guides of only 40ms appear almost immediately as 2-3arcsec swings or more.

The gold standard test here would be to use PHD to log the data at a high framerate with guiding disabled, but then use a third party app in parallel to manually send short pulseguides at know time spacings. That way you dont hit the mount with a high "guide rate" but can use the high rate to see what effect widely spaced guides have, and what is the lowest guide period you can go to before it oscillates..

Andrew Johansen Melbourne Australia

Guide Effects.jpg

 



#68 RossW

RossW

    Vostok 1

  • -----
  • Posts: 112
  • Joined: 15 Jun 2018
  • Loc: Lake Biwa, Japan

Posted 30 June 2019 - 12:37 AM

Hello Andrew,

 

Yes I understand you requested guide logs with fast cadence and low min-move. But several comments by Der_Pit in this thread suggest that running fast cadence and low min-move is Der_Pit's usual guiding style, rather than the norm of 5 to 10 second cadence that most CEM60EC/120EC users are guiding with. And the negative guiding performance (RA instability) presented in this thread seems to have occurred mostly (only?) when guiding with fast cadence. I see you've analysed another log but I can't seem to find it in this thread. Anyway, I may have misunderstood something so in that case I hope Der_Pit will set me straight grin.gif

 

I'd be very interested to see what Der_Pit's guiding is like using say 5 sec exposures and 1 arc-sec min-move.

 

I agree that the iOptron EC mounts do become unstable when guided at fast cadence, and I understand you are trying to determine why that is, but I do not see this as a fault of the mount given there should not be any reason to guide a mount with encoders at such a fast rate anyway. There may be certain circumstances where you'd want to guide at, say, 0.5s cadence but unfortunately the EC mounts may not be suitable for that unfortunately. 

 

Isn't it the case that encoder mounts of other manufacturers are usually guided at a slow cadence and almost never at 0.5 to 1 sec?

 

The gold standard test here would be to use PHD to log the data at a high framerate with guiding disabled, but then use a third party app in parallel to manually send short pulseguides at know time spacings. That way you dont hit the mount with a high "guide rate" but can use the high rate to see what effect widely spaced guides have, and what is the lowest guide period you can go to before it oscillates..

 

Yes indeed that would be a thorough test to determine the upper limit on exposure rate for the EC mounts, and I believe it is possible to run two instances of PhD2 at the same time, so it should be easily accomplished. In fact if my EC2 mount had returned from iOptron repair I would try this very test, but alas, no mount yet bawling.gif

 

 

 

Cheers,

 

Ross



#69 OzAndrewJ

OzAndrewJ

    Mercury-Atlas

  • *****
  • Posts: 2887
  • Joined: 30 Nov 2010

Posted 30 June 2019 - 01:34 AM

Gday Ross

Anyway, I may have misunderstood something

Nope, as i dont think anyone really understands these beasties yet :-)

Its currently leaning towards an exercise to see what guide rate is best for the mounts.

IIRC, one earlier poster didnt want to go to too long guide periods due to the effects of RA drift, which has also been quite variable between mounts..

I'd be very interested to see what Der_Pit's guiding is like using say 5 sec exposures and 1 arc-sec min-move

I would need to look, but IIRC there are some logs with around a 5 second rate.

Have attached 2 plots from a very short section at the end of run2 in DerPits last log

The frame rate dropped right back to about 6 seconds and the tracking "appeared" to flatten right out, but 6 secs is so close to the SDE freq that you dont know if this is real or not, and no guides got sent.

Once it drifted far enough, some guides ( little orange pillars starting at about 2694s ) cut in at a 1.2s freq and it started to oscillate.

The frame rate then stayed at 1.2s but not a lot of guides got sent

So what i find interesting here is that the oscillation continued for 30 to 60 secs after the last guide pulses. No idea if this is normal or an artifact of the guide settling down.

 

Andrew Johansen Melbourne Australia

SettlingWide.jpg SettlingClose.jpg


Edited by OzAndrewJ, 30 June 2019 - 01:38 AM.


#70 Der_Pit

Der_Pit

    Viking 1

  • -----
  • Posts: 860
  • Joined: 07 Jul 2018
  • Loc: La Palma

Posted 30 June 2019 - 04:33 AM

OK, I'll try to get some background together wink.gif

 

As Andrew mentioned, one main goal at the moment is to be able to see what the mount really does.  The SDE frequency is known and confirmed, so to measure it (and do so accurately) you do need fast cadence.  Unfortunately, that also can/does lead to fast guide pulses.  The ZFilter algorithm of PHD2 should be perfect for this, but it seems not to work properly, or I'm using it wrong.  The main reason for zero minmov had been a discussion with member Corsica about 'optimal guiding'.  His arguments are reasonable from control loop theory, so I wanted to give that a try.  It indeed seems to play not too well with the CEM60EC tongue2.gif

 

Yes, with 0.5s exposure / 1s cadence one starts chasing the seeing.  That however is around 0.6 arcsec, sometimes even better, so I'm not too concerned there.  

 

But one thing has been quite consistent so far.  As also Andrew mentions, there seems to be an immediate reaction of the mount already to the very first guide pulse, trying to put it back to where it would have been without the guide pulse.  Maybe iOptron uses something similar to a PPEC algorithm (trained with the encoder calibration) with a very high predictive weight?  That's one of the things I (and obviously also Andrew) want to find out.

 

Now, for the long exposures:  Of course with 5 (or 10) seconds exposure, you would just average over the whole SDE oscilation, no matter how strong it is.  So of course the reported guide errors are much lower then - the question is, what happens under the hood?  ISTR some claims of elliptic stars despite very good guiding numbers.  Verifying that indeed needs a two-fold setup, one cam doing the guiding (at low cadence), a second one recording at high cadence without actually guiding.  I'll try that - but not during New Moon grin.gif

 

As for more numbers on long(er) exposure guiding:  Here you go.  Last night I had dedicated to that mode, the log is attached.  As expected, numbers are much better, over some short periods even insanely good (like below 0.2" RMS).  And still it's best when I can leave the mount alone, not sending guide pulses (in RA).  Two things play foul there: 

  • The drift.  I could see (IMHO) yesterday that guiding quality worsened during the night, and improved again after I changed the tracking rate (was getting closer to horizon).  Someone mentioned it would be nice if PHD2 could monitor for such drifts and optionally adjust custom tracking rates.  I absolutely agree, that would be cool cool.gif
  • Sudden excursions.  You'll easily see them in the log:  Nice guiding, and suddenly a jump of 1.5 or 2 arcsec.  I have so far not the foggiest idea what that is.  Wind was low last night, and quite laminar.  Seeing, as mentioned, was below 0.6" all night.  So also rather not. Dirt in the gears?

So yes, the performance is for sure good enough for my setup (1.65" pixels main cam).  Even the less good guiding of the night before didn't get visible in the data (at my level of pickiness).  But there is clearly room for improvements.  A mount that works great under perfect conditions is nice, but reality usually isn't perfect. So I would appreciate if the mount would react a bit more graceful on external guide pulses.

One thing I noticed:  In the release notes of the latest CEM60EC firmware (20190424) is mentioned that it needs at least ASCOM 5.7.0 to work properly.  Does anyone know more about that?  What is done different there?  I don't have Windows, so I don't use ASCOM at all...

Attached Files



#71 Der_Pit

Der_Pit

    Viking 1

  • -----
  • Posts: 860
  • Joined: 07 Jul 2018
  • Loc: La Palma

Posted 30 June 2019 - 05:28 AM

Some more comments/remarks:
 

And i see you like your dithering lol.gif

Well, that's a tongue2.gif .  My ASI1600 demands some relatively aggressive dithering to rein the fixed pattern noise...

 

In all cases, the first RA guide after a settled period starts the oscillation
....
Virtually all guides were listed as being 40-50ms ( so if i have my calcs right, that means 0.25arcsec ), so no way it could create the large excursion by itself.
All very intriguing, 

 

 

 Yes, that is exactly what I notice when watching it live.  Already the very first pulse creates an IMHO way too strong reaction.  The size of the step seems right, minmov is 0.9px=1.15", with an agg of 20% that is 0.23"

 

Would you mind retesting your mount using, say, 5 to 10 second exposures and a reasonable "seeing-based" min-move setting, and uploading the PhD guide log?

Did so, see the other post.  But 5s really is an upper limit there, at least for me with short exposures (30-60s) and lots of dithering (every 6 or 10).  With 10s exposures, dither settling would easily exceed a minute (was some 30s with the 5-6s cadence)

 

But several comments by Der_Pit in this thread suggest that running fast cadence and low min-move is Der_Pit's usual guiding style, rather than the norm of 5 to 10 second cadence that most CEM60EC/120EC users are guiding with.

Well, it used to be typical (though with higher min-move) for my old mount that had a periodic error with some steep slope.  To catch that I had to be fast.  But that mount just did exactly what I told it, as it had no 'own brain' grin.gif The CEM is new - I don't have a usual guiding style there yet, I just want to learn how it's behaving...



#72 OzAndrewJ

OzAndrewJ

    Mercury-Atlas

  • *****
  • Posts: 2887
  • Joined: 30 Nov 2010

Posted 30 June 2019 - 05:55 AM

Gday Der Pit

 

Sudden excursions.  You'll easily see them in the log:  Nice guiding, and suddenly a jump of 1.5 or 2 arcsec.  I have so far not the foggiest idea what that is.  Wind was low last night, and quite laminar.  Seeing, as mentioned, was below 0.6" all night.  So also rather not. Dirt in the gears?

It could also be grot on the encoder????

Whilst the synta encoders dont actually control tracking, there are many examples of grot on the disk

causing quantum adjusts of the "perceived" position.

This doesnt affect the tracking as in the syntas the encoder doesnt actually "adjust" the motor "rates".

With the IOptrons, grot on the encoder will forcibly adjust the real position, so needs to be added to the possible list of culprits.

 

Andrew Johansen Melbourne Australia



#73 RossW

RossW

    Vostok 1

  • -----
  • Posts: 112
  • Joined: 15 Jun 2018
  • Loc: Lake Biwa, Japan

Posted 30 June 2019 - 07:03 AM

I would need to look, but IIRC there are some logs with around a 5 second rate.

 

Here is one section @ 5s exposures. Looks like RA guiding was turned off during this period, or min-move was set very high. You can see the RA gradually drifting east. Is this the RA tracking rate error that was discussed? If you look at the Dec guiding you'll see that all guide pules are in the same direction, so that would suggest Dec is also drifting, and therefore either the mount's polar alignment is out, or perhaps the guide camera's calibration in PhD isn't correct, i.e., a pure movement in RA is being seen as movements in both RA and Dec. So based on this little snippet of info at least, it's not possible to determine whether the RA has tracking error or not. It would be good to see 60 minutes of unguided data (unguided in RA and Dec using guiding assistant). If Dec stays on the x-axis but RA drifts off (at a rate greater than what atmospheric dispersion would cause) then that would suggest RA tracking error, correct?

 

Der_Pit_5000ms.png

 

The frame rate dropped right back to about 6 seconds and the tracking "appeared" to flatten right out

 

IMHO the RA graph is flattening out because of the long 5 sec exposures, not necessarily because the RA is performing any better or worse. At 5 sec exposures you're greatly filtering (averaging) star movement, and the flattening is exactly what we want to achieve (to not follow the seeing). Of course, the imaging camera still sees the full "unfiltered" star movement though.  

 

but 6 secs is so close to the SDE freq that you dont know if this is real or not, and no guides got sent.

 

I checked about a dozen places in the four main guide sections (#2, 4, 5, 8) and measured "long term" SDE values from 0.05" to 0.2" RMS. Nothing greater than 0.2" RMS, even when using 0.5 sec rapid-fire exposures. So perhaps SDE is not a major problem? I'd be over the moon if my mount had such low SDE estimates. Not as good as Wade's AP mount though  smile.gif  Having said that, I was able to measure SDE on some very short sections of the log where instability appears to occur, and it is indeed higher than 0.2" RMS, so perhaps the SDE sometimes momentarily appears after a guide pulse is sent?

 

Once it drifted far enough, some guides ( little orange pillars starting at about 2694s ) cut in at a 1.2s freq and it started to oscillate.

 

That would be this section. Yes indeed, I can see a little instability immediately after an RA guide pulse, and a breakout to an oscillation at the very end of the graph.

Der_Pit_500ms.png

 

 

So what i find interesting here is that the oscillation continued for 30 to 60 secs after the last guide pulses. No idea if this is normal or an artifact of the guide settling down.

 

We are back to 500ms sampling here (actually 1.2s including download and processing time), so we should expect to see a much rougher graph in comparison to the highly-smoothed 5s guiding. The two guide pulses at around 23:30:40 seem to cause some very short-term instability, but after that and until around 23:04:40 when the next RA guide pulse is sent I'm not sure whether what we see is continued instability due to the encoder or just seeing. However that "oscillation" at the end of the graph reminds me of my mount's SDE, and if I try to measure it the FFT does show a largish (0.8") peak at 5.5s suggesting SDE, but the data here so sparse we can't really rely on it.

 

Det_Pit I can easily understand your frustration with guiding that produces an RA RMS error that is two to three times as large as Dec, as your logs clearly show. It seems the RA encoder on these mounts really hates being interrupted by guide pulses. The effect of even a single guide pulse seems to cause instability for at least several seconds, and if another guide pulse is sent within that time the RA never settles down. I guess that is why most users are settling on 4 to 10s cadence, and that appears to be sufficient if your PA error is very low. Here's another section from your recent log that shows the effect of guide pulses very well. Left half of graph shows far larger RA excursions when guide pulses are being sent, while the right half with no guide pulses is much smoother (just seeing).

 

 

Der_Pit_3.png

 

 

 



#74 John Miele

John Miele

    Gemini

  • *****
  • Posts: 3452
  • Joined: 29 May 2005
  • Loc: North Alabama

Posted 30 June 2019 - 07:04 AM

Gday Der Pit

It could also be grot on the encoder????

Whilst the synta encoders dont actually control tracking, there are many examples of grot on the disk

causing quantum adjusts of the "perceived" position.

This doesnt affect the tracking as in the syntas the encoder doesnt actually "adjust" the motor "rates".

With the IOptrons, grot on the encoder will forcibly adjust the real position, so needs to be added to the possible list of culprits.

 

Andrew Johansen Melbourne Australia

If there was "grot" on the encoder or dirt in the gears, then wouldn't unguided tracking suffer as well? Both my mount and Der_Pit's mount do very well unguided achieving low RMS and no large excursions. All the troubles begin only when they are guided. The only way to avoid the problem while guiding is to increase the min motion so much that no guide corrections are sent. At which point you are essentially "not guiding"...John



#75 Der_Pit

Der_Pit

    Viking 1

  • -----
  • Posts: 860
  • Joined: 07 Jul 2018
  • Loc: La Palma

Posted 30 June 2019 - 07:24 AM



If there was "grot" on the encoder or dirt in the gears, then wouldn't unguided tracking suffer as well? Both my mount and Der_Pit's mount do very well unguided achieving low RMS and no large excursions. All the troubles begin only when they are guided. The only way to avoid the problem while guiding is to increase the min motion so much that no guide corrections are sent. At which point you are essentially "not guiding"...John

TBH, I've (so far) never run the mount unguided for really long periods (more than a few minutes).  It could be coincidence.  But in many (most?) cases the issue starts with a large excursion before there is any correction signal, so it's hard to understand how it should be caused by the guiding....


  • RossW likes this


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






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics