Interesting problem Dec axis with iOptron CEM 70 EC2

[Sanjeev Joshi]

This is not a PHD2 issue but i am asking here in the hopes that somebody else might have brought it up and a workaround or cause may be known.  I have also reached out to iOptron to get their take on it.

I have a CEM 70 EC2 that guides perfectly well with PHD2, using 4s exposures, typical rms is 0.3 - 0.5 arc seconds with maybe 1 - 1.5 arc seconds peak to peak.

Recently I discovered that if something - could be wind or even a manual guide I initiated from phd2 causes a weird reaction from Dec.   

1.  Initiate a manual guide north pulse of 1000 ms, one click.   Roughly 8 arc seconds movement north.    PHD2 issues corrections, mount responds, Dec goes back to zero.

2.  Dec then stays at zero for 20+ seconds, looking very controlled, and then suddenly takes off accelerating in the opposite direction.   PHD2 issues corrections again, mount responds, Dec comes back to zero and this time it stays at zero.

I have confirmed that PHD2 is not issuing any additional corrections in that 20+ second lull that happens.   So it almost seems like a interaction between PHD2 and the Dec controller.. some serious gap in timing maybe? but it does not explain why the rebound then completes cleanly.   I tried lowering Dec aggression and pulse a lot.. you can see the increased smaller oscillation on the right and the rebound still happens.

Any thoughts appreciated!

Sanjeev

[Sanjeev Joshi]

Also to note, it does not matter what algorithm is used. the issue is not the level of corrections issued.   Just curious if people have seen this issue. TYIA!

[Brian Valente]

Sanjeev

Please upload your logs via the built-in log uploader in PHD:

https://openphdguiding.org/getting-help/

On Tue, Aug 11, 2026 at 5:35 AM Sanjeev Joshi <sanjee...@gmail.com> wrote:

Also to note, it does not matter what algorithm is used. the issue is not the level of corrections issued.   Just curious if people have seen this issue. TYIA!

--
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 visit https://groups.google.com/d/msgid/open-phd-guiding/2ed2b390-9843-4179-9a8a-41d731d22fa5n%40googlegroups.com.

--

Brian 

Brian Valente

astronomy portfolio https://www.astrobin.com/users/bvalente/ 

[Sanjeev Joshi]

Done!   Ty.  I have uploaded for 2 nights.    https://openphdguiding.org/logs/dl/PHD2_logs_ZuNg.zip

Aug 10 (to early am today) - I did a lot of experimentation, including turning off Dec guiding to see how it behaves in drift, how it responds to manual guide with dec guiding vs dec not guiding etc.. the most useful part is towards the end of this log.   I also uploaded the log from Aug 7 which shows the calibration steps.

One thing to note -- there are 2 separate issues I am investigating.

1.  For a mount and rig that has performed really well last several months remote and prior to that at home, seeing some very large excursions in Dec between segments of very typical guiding.  Prime suspects are the aluminum dew shield (if came loose during filter change) or guide cam / guide scope, if something is a bit loose enough to be trigged by roughly 8 to 10 mph wind / gust.  Something has made the rig unususally sensitive to wind.   Rig is 3d balanced and confirmed based on witness marks, no concerns there.  Backlash is practically zero.

2.  The second issue is what Dec is doing in response to the large perturbations.   The ying, then 20+ seconds, and then an independent accelerated yang in the opposite direction.    This caught my eye since I am able to reproduce it just with a manual guide in calm conditions, by issuing a 1000 ms signal to North.   Cabling has been changed, currently utilizing through mount previously tested - but behavior existed even prior.

thanks again,

Sanjeev

[Brian Valente]

Sanjeev

Based on your guidelogs I think there are a couple things going on

First, if this is mount has encoders in both axes, it looks like your dec encoder is not enabled or not working correctly. The kinds of excursions shown are typical of a mount with a lot of dec backlash that isn't managed well (more on this later), but for an encoder mount, you probably expect (rightfully so) you don't have to manage that. It seems particularly prone at higher altitudes. First thing I would do is test for the Dec encoder working properly, is there any troubleshooting or positive confirmation you can get from ioptron. I suspect if you showed them this result, they would agree.

Second, for the Dec performance data, your current settings (such as backlash disabled, low aggression) are working against you. Your dec settings are better (but not great) for an encoder mount, but that's not what your performance shows. You didn't perform a dec backlash analysis but i'm guessing it would report a high backlash value. When your mount changes direction, there are a series of increasingly higher guide pulses trying to change direction. Eventually it does reverse, but it overshoots the zero point. If the encoder was working correctly, this would not be an issue, so back to the first point. 

As an aside, for your mount i suggest you use lowpass2 for both ra and dec (this was the recommendation given to you from your GA run as well). The PPEC settings you have are not really doing much for you, lowpass2 would be a better option. resist switch on an encoder axis is not the best and can cause instability/oscillations. lowpass2 better here as well. 

Brian

To view this discussion visit https://groups.google.com/d/msgid/open-phd-guiding/5dac9ce2-a0c1-4e75-aa89-43128bb6583dn%40googlegroups.com.

[Sanjeev Joshi]

Ah yes, TY very much for the detailed analysis.   The selection of algorithms last few times was really just random.   For months I have been using PPEC and resist with gentle settings with no issues.  Knowing Dec was somehow not interacting correctly with PHD2, I decided to try a couple different algorithms with nothing to lose.

I am curious about the backlash and I will test it again tonight with guiding assistant and backlash measurement on.   But in the past, it has always taken max 2 to 3 nudges to center the star back in calibration and the recommended backlash correction in guiding assistant has been zero, and lowpass 2 recommended.  I typically run for more than the minimum 2 mins, I will run tonight with several minutes to confirm.

What is happening at higher altitudes as well (i tested a few different altitudes) is that the RASA 8 + dew shield is much more susceptible to wind initiated excursions.. why i dont know.  It has really done well in prior months.   But i tested both sides of the pier early this am and angled it so it was at a 65 deg - 70 deg altitude with the full length exposed to the prevailing wind towards the end.   I ran out of night time (it was several mins past dawn) to complete that test but will do tonight.

In addition to the excellent points you raised, the sense I got was .. Dec does not independently mis behave when guiding is off, and stays pretty true to a natural drift.  Even when I push it manually, it does not jump back or react in an unexpected way.  It stays where the manual push left.  But the moment I have guiding on + a large deflection, Dec goes ying yang.     So I suspect something with the encoder or firmware.   Hoping its firmware fixable (even turning Dec encoder off for example) so I can manage it remotely.   My other CEM 70 AW at home does quite well without the relative encoders in it.

TY very much!

Sanjeev

You received this message because you are subscribed to a topic in the Google Groups "Open PHD Guiding" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/open-phd-guiding/537rQCJeMTY/unsubscribe.
To unsubscribe from this group and all its topics, send an email to open-phd-guidi...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/open-phd-guiding/CAJa45i4n5hKh74TFxFfT4ZtQ5AEwhLfH2aB-EYbr_WaCP%3DjRrA%40mail.gmail.com.

[Brian Valente]

>>>Dec goes ying yang.     So I suspect something with the encoder or firmware.

Just to clarify, that "ying yang" is the backlash not being managed, and then overshooting. It's not an encoder or firmware issue (other than the encoder not working or disabled)

>>>the RASA 8 + dew shield is much more susceptible to wind initiated excursions.

I don't think this has anything to do with wind, they all point back to the same issue that when your dec reverses direction, it has a huge backlash to takeup, and based on your current (presumably encoder optimized) settings, it's overshooting.

To view this discussion visit https://groups.google.com/d/msgid/open-phd-guiding/CA%2BCsSt6HRdgSTR6J%2BRvUuCb5zN8Xk%3DPw4chTihC5r69JJL1azA%40mail.gmail.com.

[Sanjeev Joshi]

Ty, will confirm backlash tonight.

Sanjeev

Sent from my iPhone

On Aug 11, 2026, at 1:07 PM, Brian Valente <bval...@gmail.com> wrote:



To view this discussion visit https://groups.google.com/d/msgid/open-phd-guiding/CAJa45i7hVs3MwEo17AvfQiux3E5b4xMe5kNxaF0j-vknk7hWeA%40mail.gmail.com.

[Sanjeev Joshi]

That was a very insightful and perceptive call Brian!   I think the texas heat has rendered the Dec with significant backlash.   It was zero for sure about 3 months ago but that was just before the summer started.   I am curious to see if the reverse acceleration after 20 seconds also stops now.   Luckily this is a very quick fix and honestly, something I think will be needed twice a year at this location.   Peak of summer and peak of winter.   It is interesting the problem started on Aug 5 or so, somewhat abruptly.  But its signature is clear.

Will get this fixed and rerun the tests to confirm the whiplash is gone.

Thanks again,

Sanjeev

[Brian Valente]

>>>I think the texas heat has rendered the Dec with significant backlash

This mount has a dec axis encoder, correct? If so, there should be no backlash at all. Did you ask Ioptron about this?

To view this discussion visit https://groups.google.com/d/msgid/open-phd-guiding/9c6bac2e-8b94-4d1a-8138-a912f7ee1704n%40googlegroups.com.

[Sanjeev Joshi]

Waiting to hear back, but essentially there is a mechanical tolerance that needs to be respected before the relative encoder can do its job.  I know these mounts inside / out, have taken them apart and put them back together optimized.  The remote observatory team (at Starfront) is nice enough to do this fix for me based on instructions, so I will report back once thats done.  It takes about 10 mins to do.

 

1.  There is a belt for the Dec motor.   Within a margin, yes it will work fine.  But if the belt got really loose, the relative encoder has no chance

2.  Once the belt is within tolerance, the next place to check is the worm gear.   This controls the level of engagement the worm has with the Dec lock mechanism.   The 20 second delay and acceleration on the other side could potentially be due to excessive play that has developed from my original settings in more temperate temperatures.   As the Dec loads up, some of the backlash is likely taken up by the excessive play (which is spring loaded), and then it grabs.    On the other side, it can release and still have a lot of slack

I know these mounts well and its a fundamental trade off in the design and set up.  If really optimized for temp, they will likely need a tweak on the worm twice a year or so, this is guaranteed.  Those that are not tweaking have found a lucky mechanical setting, this cannot be set precisely, just feel.   I dont know yet if the belt needs adjustment, it might not.   So in my case, once this is set up for the summer, we should ideally leave a little (not a lot) of slack in the worm, so that in the middle of the winter there is not a reverse problem of stiction.  If no slack is left in the summer, then say in January, the worm will start freezing up against the radial gear (not temperature wise, but movement between gear teeth) and it will cause stiction.   The compromise is to leave it just enough slack in the summer so that the backlash is small and acceptable (say just a few hundred ms) mechanically, and it does not bind in the winter.   Within this range, the relative encoders will absolutely do their jobs.

If you noticed in my logs, without perturbations, the Dec encoders were helping maintain a smooth drift based on the roughly 2 arc min PA error on one specific night and orientation.   As I am sure you know, the relative encoder keep position, but do not correct it.   Absolute encoders like in the mounts that are roughly 5K more (mine was about 7K and change) do use advanced pointing models to minimize that drift as well.   

Sanjeev

 

To view this discussion visit https://groups.google.com/d/msgid/open-phd-guiding/CAJa45i4UteWuVhtUbuR3H%2B2GZzXMr93tOvA8GS-ztHeqjhG-ew%40mail.gmail.com.

[Brian Valente]

This is an aside from PHD, but your comment

>>>If you noticed in my logs, without perturbations, the Dec encoders were helping maintain a smooth drift based on the roughly 2 arc min PA error on one specific night and orientation.

i don't know what this means. If the Dec motor is not running, how is it that a motor encoder does anything for accuracy?  The ioptron marketing suggests it addresses periodic error in both axes, which again doesn't really mean anything if the motor isn't running. 

>>>Absolute encoders like in the mounts that are roughly 5K more (mine was about 7K and change) do use advanced pointing models to minimize that drift as well.

yes i'm familiar with these kinds of mounts ;) but even without sky modeling/tracking corrections ('advanced pointing models'), absolute encoders also eliminate backlash. 

To view this discussion visit https://groups.google.com/d/msgid/open-phd-guiding/CA%2BCsSt59sd3sC_NwtrOMnZ%3D34ng6v_s2TA2Xt7kGQT3AY7810w%40mail.gmail.com.

[Sanjeev Joshi]

Ah, i meant that PHD2 Dec guiding is off but mount dec motor is still working.  If I switch off guiding and I dont get a big spike, Dec motors and relative encoders keep it reasonably tight and narrow.

To view this discussion visit https://groups.google.com/d/msgid/open-phd-guiding/CAJa45i54YVYDvVWP7K%3Dh%2Be7n79rr-VfJkOnbpfxUP56CN%2B1vgg%40mail.gmail.com.

[Brian Valente]

>>>Ah, i meant that PHD2 Dec guiding is off but mount dec motor is still working.  If I switch off guiding and I dont get a big spike, Dec motors and relative encoders keep it reasonably tight and narrow.

The Dec motor doesn't track on GEM mounts (unless you have dual axis sky modeling), Dec motor only responds to corrections. There is no Dec tracking so the motor doesn't run at all unless commanded to (vs. RA which tracks in sidereal). One of the reasons backlash is an issue in Dec and not in RA.

That is, unless there's a missing detail that makes your situation unique. 

To view this discussion visit https://groups.google.com/d/msgid/open-phd-guiding/CA%2BCsSt6dwNm3ctOLb9y9SPA%3DR1aZZyW47QDQoFQpqH9dc6443g%40mail.gmail.com.

[Sanjeev Joshi]

Yes my CEM 70 CEM 70 EC2 mount has relative encoders on both axes.  Not absolute sky positions but it still earns its pay!   Once the mount is tracking (it does need to be tracking), the encoders on both axis are measuring how much each axis is moving (rate) in relation to what is needed and issuing corrections.  Where I ran into the issue is that the backlash is so much that the corrections are likely out of sync significantly with phd2.

Lets walk through this:

1.  Assume no polar alignment error (dont we all wish).    RA motor will be actively moving at siderial rate.   Dec motor will be in a "commanded hold" position with servo still running (you can hear it hum)

2.  Now lets introduce the PA error.   Dec is still in a "commanded hold" position with servo running, humming, and without any perturbation, the scope will drift at a rate implied by the PA error.  The relative encoder has no specific geometry to align with, so it is oblivious to the PA error

3.  Now lets introduce a small gust that perturbs the Dec axis.    Dec relative encoder sees the rotation, and tells the motor actively to either speed up or slow down or even momenatarily reverse.  It is actively adjusting but only in relation to the accumulation of the rate, not the absolute position (which it has no idea of)

Physical backlash is the enemy of such a mount.  The Dec encoder can never get it right and it will always be too late depending on level of backlash and how big the perturbation is.

 Hope this makes it clear.

Sanjeev

To view this discussion visit https://groups.google.com/d/msgid/open-phd-guiding/CAJa45i7aANqPeYhcHxr74DwTK_MKWgt%3DSuN76wVxpsiWOzVtBg%40mail.gmail.com.

[Franklin Shaffer]

On 8/12/2026 10:37 AM, Sanjeev Joshi wrote:
>   If no slack is left in the summer, then say in January, the worm
> will start freezing up against the radial gear (not temperature wise,
> but movement between gear teeth) and it will cause stiction.   The
> compromise is to leave it just enough slack in the summer so that the
> backlash is small and acceptable (say just a few hundred ms)
> mechanically, and it does not bind in the winter.   Within this range,
> the relative encoders will absolutely do their jobs.
>

Don't know what types of metals you've dealt before with but metals
shrink in cold so tolerances get looser not tighter to cause stiction.
It's the opposite in warmer weather. The metal expands. The term is the
"Coefficient of thermal expansion"

Greases used on the worms and on metal surfaces can get stiffer in cold
but most mount designers take this into account and use greases with
better cold weather performance.

[Sanjeev Joshi]

TY for input.  Are you basing this on actual notes from the CEM 70, because i own two and have taken them through both winter and summer.   I agree with your generalized statements about grease and metals.  The bigger driver is actually the rubber belt.   The coefficient of thermal expansion (or contraction) is very different for the rubber belt compared with the metals, so the slack in summer and tightness in winter is driven much more by the rubber belt behavior than the metals contracting or expanding.

Sanjeev

--
You received this message because you are subscribed to a topic in the Google Groups "Open PHD Guiding" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/open-phd-guiding/537rQCJeMTY/unsubscribe.
To unsubscribe from this group and all its topics, send an email to open-phd-guidi...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/open-phd-guiding/71cfdf94-5106-4ee7-8c2f-3569b83e70fb%40gmail.com.

[Sanjeev Joshi]

Co-efficients of thermal expansion for your reference :)

Typical values (linear CTE, ppm/°C):

• Steel: ~11–13
• Brass: ~18–19
• Aluminum: ~22–24
• Rubber (general elastomers, EPDM, neoprene, natural rubber): ~150–250

So rubber is roughly 10X metals.  When rubber shrinks, even if the metals are shrinking, it will cause binding.   When rubber expands in the summer, even if the metals expand, the net effect is still expansion and therefore more slack.

Hope this helps clarify.

Sanjeev

[Sanjeev Joshi]

Wanted to provide a quick update.  The Dec backlash was mechanically reduced from 4700 ms to 2200 ms.   So likely one more iteration to get it down to very low (we are talking turning a screw maybe 1/16th of a turn).

But mount is already happy.  For now just using resist switch and the backlash value, mount has been humming for over an hour in not ideal conditions.. including mild gusts that were causing 20 arc second jumps, and a PA that predictably got a bit worse with the worm engagement (this was expected).  Now with some additional mechanical work and PA tuning still left, it is performing to the level that I am used to under breezy conditions.   I will redo calibration and guiding assistant and likely go back to low pass 2 once the backlash has been fully fixed mechanically.   But at this point the mount is operating as usual.

Ty for all the help,

Sanjeev