CEM60EC Guide settings references

[cotak]

So a while back I posted this in another thread:

This is just me writing up what I have found with the CEM60EC so that others can reference it later on.

I find that hysteresis works the best it seems. The other more advanced algorithm tend to be attempts at stopping overreaction to seeing or transient events, which might be good if you are using a fast guide exposure and have a normal mount. I find with the encoders those ends up reacting later than necessary. I guess that make sense if you consider that my copy has p-p PE of about 1" (note this isn't normal PE but a measure of the transient spikes of an encoder mount) and RMS is about 0.2-0.3". There is just so little to correct for that if  you have good PA, sufficiently long exposures to avoid chasing seeing, all you are left with is PA drift or imaging chain shifts to compensate for.

So what I find is 80-95% aggression, 0 hysteresis and a minmo just slightly above the remnant seeing effects works best. PA becomes quite important with encoder mounts, with it off you'll find the mount tends to end up on one side of the guide graph while Phd keep trying to bring it back to alignment. I find it is much more sensitive than  regular mounts.

Originally I though the new Z algorithm would be ideal as it only correct for long term events. I didn't find it work too well maybe due to no control on aggression. However, it might be that I haven't gotten used to it yet.

 

 

[Starflyer]

Thanks for this, following as this will probably be my next mount.

[ChrisWhite]

Very interesting. Thanks for the writeup.  I've got a CEM60 NON-EC, and also a CEM60-EC on the way.  I have found that for the standard CEM60 the PPEC algorithm works very well.  I generally use 1.5 second exposures and adjust min/mo based on seeing.  I'm curious to see if I can get PPEC to play well with the encoder, but if not I'll use your hystersis logic as a starting point. 

[cotak]

3 hours ago, ChrisWhite said:

Very interesting. Thanks for the writeup.  I've got a CEM60 NON-EC, and also a CEM60-EC on the way.  I have found that for the standard CEM60 the PPEC algorithm works very well.  I generally use 1.5 second exposures and adjust min/mo based on seeing.  I'm curious to see if I can get PPEC to play well with the encoder, but if not I'll use your hystersis logic as a starting point. 

Hi Chris

This is what the frequency plot of my copy on a good night looks like:

Not much PE to manage with PPEC. I would worry that PPEC would end up chasing ghosts but it might be worth a try.

My guiding last night was pretty bad when I tried hysteresis,  Z and lowpass 2. I suspect that a few things (To be confirmed) to be important to this mount to work well. While hysteresis worked well previously last night it seems all the algorithm struggled with what I think was a combination of winds and bad seeing (it was still gusting to 20-30 kph when I started and all night the winds never dropped much below 20 kph). 

Next time my attempt will involve potentially improving my alignment in hope of making the guiding task as easy as possible. And I'll see what happens with a truly long exposure like 10s.

 

Finally, all the noise online about SDE and I wasn't seeing much of anything in my guide trace. So to recap supposedly the period of this "fast" oscillation in the CEM60EC is 5-6 seconds or a 54 fundamental. Often touted as caused by cheap encoder in iOptron mount, but don't believe everything you read online. Does it exist in a gen2 CEM60EC? Yes, but...

 

It's less than 0.3 p-p in amplitude. I am not sure why I got two peaks, it could be due to the short data set. However, it is within the range of measured SDE in the CEM120EC. Personally, I don't see any impacts at 0.54" imaging scale with a 1400mm focal length. So for all intent and purpose it's a non-issue. 

What is SDE? Well it's the error that results from interpolation between encoder ticks, as it is very expensive and technically challenging to have sufficient native encoder resolution to avoid this. Even the often worshiped absolute encoders use interpolation to achieve sufficient resolution. And it's not like interpolation is a super wild guess, the interpolation essentially process the analog signal from the reader to get additional information not available if you treat the signal as digital pulses. The problem occurs if you have alignment errors between the encoder and the reader head , and not as far as facts are concerned (as often stated without proof online) due to the lower cost of incremental encoders (plane wave used incremental as well). As absolute encoders are perfectly capable of having SDE as well. Possible solution to mitigate alignment error (which are in many ways unavoidable) are dual read heads (I know even AP only use one) or software solution to record and compensate for the error. There are some users who have measured AP encoder mounts and say no SDE, which sound good, until you see the noise floor is above the 0.3" range in that measurement and so if they do have similar amplitude SDE (and it would be a non-issue due to the low amplitude) you wouldn't see it with a very high noise floor. At the end of the day if your fast errors are all of 0.3" every 5 seconds or so and you are seeing limited you will be hard pressed to find it showing up in your images. 

Edited June 4, 2019 by cotak

[Jkulin]

I was using these settings with my CEM60EC. even when the seeing when she still created a fantastic guide graph: -

My mount really was brilliant, I've not managed as good with the 120EC yet, but what surprised me with the 60 was that even when the seeing went kaput it still managed to keep on track.

[Freddie]

The graph looks good when displayed in pixels but wouldn’t look anything like as good if displayed in arc sec which is better for comparisons.

[cotak]

Right.

I haven't seen as high as 0.7 on RA yet. Mostly mine comes in max at 0.55 on RA at this point. The problem is when the guiding appears to over correct. I am trying to see if longer guide exposures will help reduce this issue. I also am seeing when the scope's pointing generally straight up some issues with larger deviations when scope's on east side of pier. When the it's pointing lower in DEC it seems to have no issue, so it could be just my mirror but it does this for a a long time with some period of quiet so I am not entirely sure what the cause it. The other suspect is off balance RA which I'll have to do some experiments to chase down. 

[ChrisWhite]

@Jkulin:  If you switch to arc seconds on your graph it won't look quite as good.  .7" total RMS is not bad guiding by any means, but I would expect to see better if things were running optimally.  With my CEM60 NON-EC I generally guide below .6" total RMS and often around .5" or below with an 8in Newt OAG at 920mm FL.    That said, these numbers are not everything. If your stars are tight and round, then it doesnt matter.

Has anyone given PPEC a go?  It takes a couple of worm cycles to refine, but with the Non-EC it is brilliant.  For the first couple of cycles it behaves like a hystersis algorithm, but then things settle down.   I think you would want a pretty high min/mo and long-ish exposures like 3-5 seconds.  I hope to test this out Thursday when my mount arrives...

[souls33k3r]

9 hours ago, ChrisWhite said:

@Jkulin:  If you switch to arc seconds on your graph it won't look quite as good.  .7" total RMS is not bad guiding by any means, but I would expect to see better if things were running optimally.  With my CEM60 NON-EC I generally guide below .6" total RMS and often around .5" or below with an 8in Newt OAG at 920mm FL.    That said, these numbers are not everything. If your stars are tight and round, then it doesnt matter.

Has anyone given PPEC a go?  It takes a couple of worm cycles to refine, but with the Non-EC it is brilliant.  For the first couple of cycles it behaves like a hystersis algorithm, but then things settle down.   I think you would want a pretty high min/mo and long-ish exposures like 3-5 seconds.  I hope to test this out Thursday when my mount arrives...

Welcome Chris. I've been following your CN progress with great interest and glad to have you here. 

I'm very much on the fence between the EC or non-EC so your input will be super valuable

[cotak]

My skies are hopefully clear Thursday night, I can give it a 15 minutes run to see.

The posts by Gaston on CN is interesting. I think it is over simplified but there are interesting points to consider. I'll see if a zero or low min mo, long exposure and low aggression is better.

[souls33k3r]

13 minutes ago, cotak said:

My skies are hopefully clear Thursday night, I can give it a 15 minutes run to see.

The posts by Gaston on CN is interesting. I think it is over simplified but there are interesting points to consider. I'll see if a zero or low min mo, long exposure and low aggression is better.

Can you kindly send a link to that post? Ta

[Rainer]

32 minutes ago, cotak said:

My skies are hopefully clear Thursday night, I can give it a 15 minutes run to see.

The posts by Gaston on CN is interesting. I think it is over simplified but there are interesting points to consider. I'll see if a zero or low min mo, long exposure and low aggression is better.

Hi,

I have been guiding my CEM 120EC2 in the last weeks with 100% aggressivity, 0.005 minimum move and 15 seconds guide camera exposure time but with SkyGuide. Much better then before as when I still used PHD2. My RMS values are around 0.25" - 0.30" arc seconds with focal length of 2424 mm guiding using the OAG of my QSI 540wsg cmeras.

Why 0.005 sec minimum move because that is the time for one step and under that I do not think the stepper motor will react. Maybe it is even just my imagination.

Condition for guiding like this are mounts with really 0 zero backlash and that is what the iOptron mounts with their worm / worm gear assembly have.

Rainer

[Rainer]

25 minutes ago, souls33k3r said:

Can you kindly send a link to that post? Ta

https://www.cloudynights.com/topic/663250-cem60ec-some-questions/

[Freddie]

35 minutes ago, Rainer said:

Hi,

I have been guiding my CEM 120EC2 in the last weeks with 100% aggressivity, 0.005 minimum move and 15 seconds guide camera exposure time but with SkyGuide. Much better then before as when I still used PHD2. My RMS values are around 0.25" - 0.30" arc seconds with focal length of 2424 mm guiding using the OAG of my QSI 540wsg cmeras.

Why 0.005 sec minimum move because that is the time for one step and under that I do not think the stepper motor will react. Maybe it is even just my imagination.

Condition for guiding like this are mounts with really 0 zero backlash and that is what the iOptron mounts with their worm / worm gear assembly have.

Rainer

I think you are a bit confused there as min move is a value of distance from the zero position of the guide star in pixels not a value of time.

[Rainer]

1 hour ago, Freddie said:

I think you are a bit confused there as min move is a value of distance from the zero position of the guide star in pixels not a value of time.

Hi Freddie,

No, I am not confused. PHD2 yes specifies it in arc seconds but SkyGuide of Innovations Foresight https://www.innovationsforesight.com/ specifies it in seconds of time. They calculate how much seconds the mount has to move after having calculated the distance the star has moved so in both cases it is the same. 

See ? I am not confused as it is just a matter of how they did program it.

There are more guide programs and not only PHD2    😉

regards Rainer

[Freddie]

Apologies, as the OP is using PHD I assumed guiding references would be made to that but I do now see that you say you use SkyGuide.

[Rainer]

18 minutes ago, Freddie said:

Apologies, as the OP is using PHD I assumed guiding references would be made to that but I do now see that you say you use SkyGuide.

Hi Freddie,

You do not need to apologize. I am the one bringing chaos into this. 0.005 seconds move would be more or less a move of 15" x 0.005 = 0.075" 👍 this is at a guide speed of 1X and OK, at a guide speed 0.5X then it would be 7.5" x 0.005 = 0.0375" and so on ...

Edited June 5, 2019 by Rainer

[Jkulin]

20 hours ago, ChrisWhite said:

@Jkulin:  If you switch to arc seconds on your graph it won't look quite as good.  .7" total RMS is not bad guiding by any means, but I would expect to see better if things were running optimally.  With my CEM60 NON-EC I generally guide below .6" total RMS and often around .5" or below with an 8in Newt OAG at 920mm FL.    That said, these numbers are not everything. If your stars are tight and round, then it doesnt matter.

Has anyone given PPEC a go?  It takes a couple of worm cycles to refine, but with the Non-EC it is brilliant.  For the first couple of cycles it behaves like a hystersis algorithm, but then things settle down.   I think you would want a pretty high min/mo and long-ish exposures like 3-5 seconds.  I hope to test this out Thursday when my mount arrives...

Hi Chris, I just grabbed what images I could find since I swapped out my 60EC for the 120EC and forgot that was in px’s.

I have some images somewhere in arc secs and I think on exceptional nights it was half that.

With my 120EC I’m generally Guiding at 5-7secs, I’m Running my aggression on the RA between 30 and 50 and on the DEC between 50 and 100, really what ever the guide assistant suggests, I did find that increasing the Hys to 30 helped as well.

in my opinion the 60EC guided better that the current firmware of the 120EC but I knew that when I bought it, with a new firmware due soon then I am ever hopeful of even better.

[ChrisWhite]

14 hours ago, souls33k3r said:

Welcome Chris. I've been following your CN progress with great interest and glad to have you here. 

I'm very much on the fence between the EC or non-EC so your input will be super valuable 

Thanks for the welcome.  Tomorrow night will be a good test.  Seeing what cotak and Rainer have posted looks in line with my thoughts.  I'm surprised by 15 second exposure, but it makes sense perhaps.  I think you want a long enough exposure to eliminate chasing seeing and periodic error, as the EC mount *should* have  negligible PE.  Really, all you want to do is correct for drift from PA error or atmospheric refraction over time.  It seems that people are having problems with guiding and encoder when guide corrections are too frequent.  It might be due to how the guide corrections are issued in the firmware.  I have tried running a PEC model with the CEM60 and it didn't work.  I didn't try long guide exposures though.  I wonder if that would make a difference.  

[Rainer]

15 hours ago, ChrisWhite said:

Thanks for the welcome.  Tomorrow night will be a good test.  Seeing what cotak and Rainer have posted looks in line with my thoughts.  I'm surprised by 15 second exposure, but it makes sense perhaps.  I think you want a long enough exposure to eliminate chasing seeing and periodic error, as the EC mount *should* have  negligible PE.  Really, all you want to do is correct for drift from PA error or atmospheric refraction over time.  It seems that people are having problems with guiding and encoder when guide corrections are too frequent.  It might be due to how the guide corrections are issued in the firmware.  I have tried running a PEC model with the CEM60 and it didn't work.  I didn't try long guide exposures though.  I wonder if that would make a difference.  

Hi Chris,

At first when I got my two CEM 120EC 2 I was guiding at 5 seconds. After changing from PHD2 to SKyGuide and having many many e-mail exchanges with Dr. Gaston from IF I experimented with longer and longer exposure times and yes the result were getting better. Now I am testing the predictive guiding of Sky Guide and to my surprise even better the guiding.

The fact is the encoders do a good job killing the Periodic error and so the thing we have to guide out is our bad Polar Alignment as well as sudden atmospheric refraction changes. You can have a thin stripe of the Jet Stream pushing your star to some other position an then suddenly it disappears. Be aware that the EC2 mounts do use temperature and barometric pressure for compensating the atmospheric refraction when moving from the Zenith to the horizon or viceversa. This was confirmed to me by iOptron.

The fact is a lot of people out there still do not understand what encoders are for and so they recommend not to get them and I dare to say if you have them but do not use them there is no harm. The latest firmware v190318 is very stable and good. So far in the last 14 months since I have my mounts iOptron has done a lot of improvements based on many of my findings over the moths.

I would say the CEM 120EC2 as well as the other smaller siblings are ripe mounts.

Quote

I'm surprised by 15 second exposure, ...

I fully understand that but it is proven that EC mounts can go with such long exposure times. We guys with more then 15 or 20 years experience in this are also now confronted with new technology but we need to shake off what we got glued onto us and we need to think now out of the box ... Invest some time in learning this new technology and learn from people who are already doing it ...

Rainer

Edited June 6, 2019 by Rainer

[cotak]

So last night was a bit of a cluster (expletive). I did not end up trying out PPEC but was trying different settings in hysteresis.

I am suspecting there's something wrong with my edgehd 8. So here's what it does:

Target is high in the skies so scope is almost vertical (FYI was specifically chasing down this vertical issue as saw it with other targets), when on west of pier mount does Ok and guides well with some random movements here and there which I think are mirror movements but overall controllable. However, no matter how good the graph is there's elongation in the stars. On the east side the graph has random 2" spikes and again the stars are elongated slightly. Now the star gets better later on when the scope's tracked so the OTA is less vertical, this I also saw in a previous session. So that was a hint there.

Previously while talking to iOptron about this I got a beta firmware (now the official April 2019 one) and wanted to test. During that run I didn't see spikes in the guide graph but I didn't take any images. So yesterday night when I saw the issues again with east of pier I manually slewed to a lower DEC and there it was again, no spikes. Took a quick 3 minute sub and stars looks pretty good. Switched to sharpcap and took a look at what the main camera was seeing live (at around 5fps as I was on USB2) and other than the star moving around due to seeing there was no hints of RA movements or DEC movements.

Can it still be the mount or encoders? Well actually unlikely because this elongation even with great guiding was also seen before on my ieq45 pro when the scope was near vertical, but back then I just attributed it to 

So it seems I have elongation in stars from something to do with the optics when vertical and seemingly also some sort of mirror shifting, but before everyone scream mirror flop it doesn't look like other examples of mirror flop I have seen and refocusing correctly doesn't solve it.

Overall not the most enjoyable of night and now I wonder if I should consider replacing the edge with something more rigid. 

Tonight's another clear one so I'll see what else I can dig up. Maybe I'll try for a lower target and see how that goes for one night of guiding to try and confirm the above.

 

 

[cotak]

These setting seems to work well:

Both axis using hysteresis. Seems reasonable level of aggression and 4 or more second exposure results is very smooth trace. East of pier when pointed south is a pile of doodoo still, wondering if that's cause I am looking over a line of roofs and seeing thermals. As:

 

See the big house there right behind with the grey roof, that's directly south. And south is also downtown Toronto.

[cotak]

Ok so it looks like the most likely candidate for my south pointing issues is.... Balance.

Turns out my OTA being heavier on one side, while balanced in DEC, causes problems for the RA when it's pointed south, resulting in a west heavy situation. Also my dec was less balanced than I had thought as the weight bias means it felt more balanced then in reality.

So next step is to make an offset counter weight for the OTA and see if that fixes the issues I saw with guiding.

 

[cotak]

Update. Was chasing ghosts a bit. And I think it did harm as stars were slightly every so elongated. 

It seems there's something weird about how PHD handles frames downloaded at a slower cadence than your guide exposure settings (say due to a USB2 guide cam). Let me demonstrate with some PHD2 results...

Oh no! SAW TOOTH (bad subframe setting resulting in almost 6 second cadence vs 500ms exposures due to every other frame being corrupted + some extra delays):

 

But wait... using correct subframe size not to cause weird frames to be dropped every other frame (700ms cadence on 500ms exposure):

 

Back on CN there was a gent who struggled with his CEM60EC seeing saw tooth patterns unless he used very specific exposures and settings. I wonder if his issue was similar to mine.

Now with correct subframe I can get 2 second guide exposures with 75 aggression and 0.6 min-mo (works out to about 0.5" minmo). However, 4 second 75 agg and 0.6 min-mo still seems to get better guide results.

One important thing is that cadence thing resulted in higher measurement of the so called SDE, but once the cadence was faster that measurement dropped it actually disappeared completely in PHDlog viewer. So that's interesting. I wonder if the old SDE complaints were in some parts due to bad PHD2 vs cadence behavior?

Edited June 22, 2019 by cotak