DEC Arc Model drift question


ap@CaptivePhotons.com
 

On Mon, Jul 18, 2022 at 11:57 AM, Ray Gralak wrote:
Yes, three arcs separated by 1-degree would provide the highest accuracy. The more points, the better.

Which scope, image scale, and exposure duration? How much drift are you seeing?
I didn't want to hijack the prior thread, since this is a different mount, so am starting a new thread, apologies on any context confusion. 

This is all experimentation; I guide, so I am not concerned, but I am curious if I am doing something wrongly. 

The model was 37 points each side of the meridian, in 3 arcs of 2 degrees apart.  The model looked like this: 



This was made with a NP101is (4" refractor, 540mm, F5.4) on an AP1100AE, freshly polar aligned.

I don't know what to make of the huge numbers on the west side (that's pierside west, so pointing east).  The RMS error there was hugely greater for reasons unclear.   My test was done pierside east at hour angle 1, so high, and on the center arc.  I did two guiding assistant runs which gave similar results, except opposite axis.  The first had DEC drift about 1" over 6 minutes, and the second had RA drift about .75" over 8 minutes with DEC hanging in close to the exis (max peak -0.46, with most well under that). 

Here is that first one.  I think it was Roland who said it should be much better, so am curious if I am doing something wrongly. 



The refractor has a nice solid focuser and M68 imaging train with an OAG, so there's no guide scope flex involved. Not saying nothing could move, of course, but compared to my SCT it's a really stable setup.  The model was freshly made and extended to 30 degrees altitude on both sides and the target was at 70 degrees altitude (DEC 43d). 

I looked at a 1 degree x 3 arcs model I made a couple nights earlier, and it did not have the wacky numbers in the west side, but had similar drift, in that case about .75 in RA and 1.5" in DEC over 14 minutes.  



Again... I guide, it guides well.  This is not affecting my imaging. 

This is mostly curiosity, but you never know when digging into something will uncover something useful.  So I appreciate any insights you (or anyone) may have. 

Linwood


Roland Christen
 

It looks almost like your model was never turned on. Or perhaps your Pattersnips were misaligned.

Rolando


-----Original Message-----
From: ap@... <ap@...>
To: main@ap-gto.groups.io
Sent: Mon, Jul 18, 2022 12:01 pm
Subject: [ap-gto] DEC Arc Model drift question

On Mon, Jul 18, 2022 at 11:57 AM, Ray Gralak wrote:
Yes, three arcs separated by 1-degree would provide the highest accuracy. The more points, the better.

Which scope, image scale, and exposure duration? How much drift are you seeing?
I didn't want to hijack the prior thread, since this is a different mount, so am starting a new thread, apologies on any context confusion. 

This is all experimentation; I guide, so I am not concerned, but I am curious if I am doing something wrongly. 

The model was 37 points each side of the meridian, in 3 arcs of 2 degrees apart.  The model looked like this: 



This was made with a NP101is (4" refractor, 540mm, F5.4) on an AP1100AE, freshly polar aligned.

I don't know what to make of the huge numbers on the west side (that's pierside west, so pointing east).  The RMS error there was hugely greater for reasons unclear.   My test was done pierside east at hour angle 1, so high, and on the center arc.  I did two guiding assistant runs which gave similar results, except opposite axis.  The first had DEC drift about 1" over 6 minutes, and the second had RA drift about .75" over 8 minutes with DEC hanging in close to the exis (max peak -0.46, with most well under that). 

Here is that first one.  I think it was Roland who said it should be much better, so am curious if I am doing something wrongly. 



The refractor has a nice solid focuser and M68 imaging train with an OAG, so there's no guide scope flex involved. Not saying nothing could move, of course, but compared to my SCT it's a really stable setup.  The model was freshly made and extended to 30 degrees altitude on both sides and the target was at 70 degrees altitude (DEC 43d). 

I looked at a 1 degree x 3 arcs model I made a couple nights earlier, and it did not have the wacky numbers in the west side, but had similar drift, in that case about .75 in RA and 1.5" in DEC over 14 minutes.  



Again... I guide, it guides well.  This is not affecting my imaging. 

This is mostly curiosity, but you never know when digging into something will uncover something useful.  So I appreciate any insights you (or anyone) may have. 

Linwood

--
Roland Christen
Astro-Physics


ap@CaptivePhotons.com
 

On Mon, Jul 18, 2022 at 01:31 PM, Roland Christen wrote:
It looks almost like your model was never turned on.
I agree.  But when I look in the APCC log I see tracking corrections. 

0526243 2022-07-11 21:30:26.426:       Info,  DecArcTracking, Setting RA rate to -82.7901946840076 arc-sec/hour. External=0, RA Tweak=0
0526244 2022-07-11 21:30:26.426:       Info,  DecArcTracking, Setting Dec rate to -7.20680335389745 arc-sec/hour. External=0, Dec Tweak=0, Invert Dec East/West=False/False
 

On Mon, Jul 18, 2022 at 01:31 PM, Roland Christen wrote:
Or perhaps your Pattersnips were misaligned.
Are those Jabberwocky related?


Roland Christen
 


Or perhaps your Pattersnips were misaligned.
Are those Jabberwocky related?
Actually, please disregard. The Pattersnips is something i ate at our local Norwegian restaurant yesterday. I confused that with the Rohenfuegel adjuster nut on the Whensforth axis.

In any case, next time you're out, try turning off the model and do a 6 minute unguided tracking. You should then see if your actual drift rate is 8.279 arc seconds (which corresponds to 82.79/hour that your model predicts.) In other words, does the model agree with the actual drift that you experience with the model turned off? Was the model accumulated just before the imaging session, or was it taken on another day?

The rate of 82.79 / hour is quite large and suggest that you were imaging somewhere far from the meridian, probably 3 - 4 hours east of the meridian. Things change rapidly in that area of the sky, so your model needs to be denser there than overhead.

Rolando

-----Original Message-----
From: ap@... <ap@...>
To: main@ap-gto.groups.io
Sent: Mon, Jul 18, 2022 12:37 pm
Subject: Re: [ap-gto] DEC Arc Model drift question

On Mon, Jul 18, 2022 at 01:31 PM, Roland Christen wrote:
It looks almost like your model was never turned on.
I agree.  But when I look in the APCC log I see tracking corrections. 

0526243 2022-07-11 21:30:26.426:       Info,  DecArcTracking, Setting RA rate to -82.7901946840076 arc-sec/hour. External=0, RA Tweak=0
0526244 2022-07-11 21:30:26.426:       Info,  DecArcTracking, Setting Dec rate to -7.20680335389745 arc-sec/hour. External=0, Dec Tweak=0, Invert Dec East/West=False/False
 

On Mon, Jul 18, 2022 at 01:31 PM, Roland Christen wrote:
Or perhaps your Pattersnips were misaligned.
Are those Jabberwocky related?

--
Roland Christen
Astro-Physics


ap@CaptivePhotons.com
 

On Mon, Jul 18, 2022 at 01:55 PM, Roland Christen wrote:
Actually, please disregard. The Pattersnips is something i ate at our local Norwegian restaurant yesterday. I confused that with the Rohenfuegel adjuster nut on the Whensforth axis.
Better you than me, I have found it best not to consume things I cannot pronounce. It would make it difficult if I had to explain to the ER doctors.  :) 

 
In any case, next time you're out, try turning off the model and do a 6 minute unguided tracking. You should then see if your actual drift rate is 8.279 arc seconds (which corresponds to 82.79/hour that your model predicts.) In other words, does the model agree with the actual drift that you experience with the model turned off? Was the model accumulated just before the imaging session, or was it taken on another day?
 
The rate of 82.79 / hour is quite large and suggest that you were imaging somewhere far from the meridian, probably 3 - 4 hours east of the meridian. Things change rapidly in that area of the sky, so your model needs to be denser there than overhead.
I can do that.  The model was done literally minutes before I tried it.  My target did not rise for about 2 hours after dusk so I was experimenting.  For the guide assistant run I picked a spot on the Dec Arc that was about an hour off the meridian, in the case of the first it was 1 hour and 70 degrees altitude, so should be relatively unaffected by refraction.  Wind was calm, and seeing pretty good, guiding in that area was low 0.20's. This also put it well away from the roof of my house (guiding stinks near it, concrete tile roof in Florida).  I was trying to give it all the advantages I could.

Next time I'll do a GA run with and without the tracking corrections turned on, see how it looks.

Next time, based on the weather forecast, may involve an upgrade from J2000 to J2100. :( 

Linwood


Roland Christen
 

There are two possibilities that can produce errors. One is doing plate solving with outdated co-ordinates. The second is using a coordinate database that is only accurate to +- 2 arc seconds. That can produce small errors such as you experienced.

If you really were imaging within 1 hour of the meridian, and your mount was properly polar aligned, your RA drift rate should be closer to 10 - 15 arc sec per hour, not 82. Also, as you approach the pole, the RA drift rate becomes harder and harder to model because a 1 arc sec error in RA might be only a tiny fraction of a pixel (cos theta = 0 at 90). Therefore a 1.5 arc sec error as shown on your guider graph begins to have no effect above 70deg Dec on the roundness of the star in pixels.

Rolando

-----Original Message-----
From: ap@... <ap@...>
To: main@ap-gto.groups.io
Sent: Mon, Jul 18, 2022 1:04 pm
Subject: Re: [ap-gto] DEC Arc Model drift question

On Mon, Jul 18, 2022 at 01:55 PM, Roland Christen wrote:
Actually, please disregard. The Pattersnips is something i ate at our local Norwegian restaurant yesterday. I confused that with the Rohenfuegel adjuster nut on the Whensforth axis.
Better you than me, I have found it best not to consume things I cannot pronounce. It would make it difficult if I had to explain to the ER doctors.  :) 

 
In any case, next time you're out, try turning off the model and do a 6 minute unguided tracking. You should then see if your actual drift rate is 8.279 arc seconds (which corresponds to 82.79/hour that your model predicts.) In other words, does the model agree with the actual drift that you experience with the model turned off? Was the model accumulated just before the imaging session, or was it taken on another day?
 
The rate of 82.79 / hour is quite large and suggest that you were imaging somewhere far from the meridian, probably 3 - 4 hours east of the meridian. Things change rapidly in that area of the sky, so your model needs to be denser there than overhead.
I can do that.  The model was done literally minutes before I tried it.  My target did not rise for about 2 hours after dusk so I was experimenting.  For the guide assistant run I picked a spot on the Dec Arc that was about an hour off the meridian, in the case of the first it was 1 hour and 70 degrees altitude, so should be relatively unaffected by refraction.  Wind was calm, and seeing pretty good, guiding in that area was low 0.20's. This also put it well away from the roof of my house (guiding stinks near it, concrete tile roof in Florida).  I was trying to give it all the advantages I could.

Next time I'll do a GA run with and without the tracking corrections turned on, see how it looks.

Next time, based on the weather forecast, may involve an upgrade from J2000 to J2100. :( 

Linwood

--
Roland Christen
Astro-Physics


ap@CaptivePhotons.com
 

On Mon, Jul 18, 2022 at 02:35 PM, Roland Christen wrote:
There are two possibilities that can produce errors. One is doing plate solving with outdated co-ordinates. The second is using a coordinate database that is only accurate to +- 2 arc seconds. That can produce small errors such as you experienced.
How can one use outdated coordinates?   My lat/long were right, if that is what you mean, and I am using the database provided with ASTAP (H18).  Time was GPS sync'd to PC and thence to mount. I know Ray has often said different tools have different accuracy (though I do not think he ever determined, or at least said, which one(s) were right).   I no longer have TSX so cannot compare with it.

I don't see how but is there anything in using an app as a camera (specifically NINA)?  I think all it does is capture the image, ASTAP is separately used to solve. 

Linwood


Ray Gralak
 

Linwood,

For the guide assistant run I picked a spot on the Dec Arc that was about an hour off the meridian,
Is it possible the mount shifted, or settled into the ground after time? That could cause polar alignment to change and thus the dec drift near the meridian

As for plate solve accuracy, I used a much larger database with SkyX, so I considered it the reference. I'll have to find the spreadsheet to get the actual numbers.

-Ray


Roland Christen
 

I think you ignored what I said about RA operation above 70 degrees Dec. At 70 degrees the right ascension error of a plate solve is 3 times what it is at the celestial equator. At 80 Dec it is 6 times as high.

Rolando

-----Original Message-----
From: ap@... <ap@...>
To: main@ap-gto.groups.io
Sent: Mon, Jul 18, 2022 1:41 pm
Subject: Re: [ap-gto] DEC Arc Model drift question

On Mon, Jul 18, 2022 at 02:35 PM, Roland Christen wrote:
There are two possibilities that can produce errors. One is doing plate solving with outdated co-ordinates. The second is using a coordinate database that is only accurate to +- 2 arc seconds. That can produce small errors such as you experienced.
How can one use outdated coordinates?   My lat/long were right, if that is what you mean, and I am using the database provided with ASTAP (H18).  Time was GPS sync'd to PC and thence to mount. I know Ray has often said different tools have different accuracy (though I do not think he ever determined, or at least said, which one(s) were right).   I no longer have TSX so cannot compare with it.

I don't see how but is there anything in using an app as a camera (specifically NINA)?  I think all it does is capture the image, ASTAP is separately used to solve. 

Linwood

--
Roland Christen
Astro-Physics


 

It's possible to have an epoch mismatch


On Mon, Jul 18, 2022 at 11:42 AM ap@... <ap@...> wrote:
On Mon, Jul 18, 2022 at 02:35 PM, Roland Christen wrote:
There are two possibilities that can produce errors. One is doing plate solving with outdated co-ordinates. The second is using a coordinate database that is only accurate to +- 2 arc seconds. That can produce small errors such as you experienced.
How can one use outdated coordinates?   My lat/long were right, if that is what you mean, and I am using the database provided with ASTAP (H18).  Time was GPS sync'd to PC and thence to mount. I know Ray has often said different tools have different accuracy (though I do not think he ever determined, or at least said, which one(s) were right).   I no longer have TSX so cannot compare with it.

I don't see how but is there anything in using an app as a camera (specifically NINA)?  I think all it does is capture the image, ASTAP is separately used to solve. 

Linwood




ap@CaptivePhotons.com
 

On Mon, Jul 18, 2022 at 03:20 PM, Ray Gralak wrote:
Is it possible the mount shifted, or settled into the ground after time? That could cause polar alignment to change and thus the dec drift near the meridian
It's hard to say anything is impossible, but unlikely.  I have the Planet's spikes in self-centering holes on pavers with a gravel base, haven't moved in 18 months or so.  And this was a light setup.  But... hard to say anything is impossible.  Next time I can redo the polar alignment run after the test just to see if it appears to have settled. 

I guess I could do a verify run in APPM though not sure if that helps if the source of the plate solve is off, not sure if that just ensures it looks good (same error again, or would tell me it was bad. I guess it would depend on HOW the plate solve is off, if uniform or random.

I wish I still had my TSX subscription, I could try a separate run with it, but didn't keep it when the MyT went.  So I think it's ASTAP for right now.  If that's it, no real issue since I guide but am interested if I'm doing something else wrong. 

It is a time of night when a lot is changing in the atmosphere, this time of year it goes from overcast and stormy to clear (if it is going to clear at all) near dusk.  There is not a huge temperature change but it is probably most rapid in the hours I was testing.

Linwood


ap@CaptivePhotons.com
 

On Mon, Jul 18, 2022 at 03:27 PM, Roland Christen wrote:
I think you ignored what I said about RA operation above 70 degrees Dec. At 70 degrees the right ascension error of a plate solve is 3 times what it is at the celestial equator. At 80 Dec it is 6 times as high.
Sorry, I thought I had commented I was at 43 degrees DEC for one target and 31 degrees for the other.   My altitude was 70 degrees if I recall. 

Linwood


ap@CaptivePhotons.com
 

On Mon, Jul 18, 2022 at 03:29 PM, Brian Valente wrote:
It's possible to have an epoch mismatch
Is there a setting that I need to worry about for that, or is it all handled between APPM and ASTAP?  I do have the "Process J2000 to JNOW" set but the documentation implies that only affects the log file. 

I do run this at 1200x and a 1 second settle time, with a 2 second exposure.  It's a very light OTA and a very beefy mount, and I assumed if it wasn't settled it would streak stars and not solve (I had zero errors).  But is there a chance there is something not settled that gives bad coordinates? 

Linwood


 

Regarding epochs, leave everything at default values. The only time i see people having epoch-related issues is when they start fiddling with it. 


>>> But is there a chance there is something not settled that gives bad coordinates? 

It's possible, but I don't know if that explains your results

I found the best way to evaluate this is to enable 'save all images' for an APPM run, and then just blink through them when it's completed. I found that plates could be solved successfully with streaking stars, so i upped my settle time. It's worth seeing what that looks like for your run



On Mon, Jul 18, 2022 at 12:48 PM ap@... <ap@...> wrote:
On Mon, Jul 18, 2022 at 03:29 PM, Brian Valente wrote:
It's possible to have an epoch mismatch
Is there a setting that I need to worry about for that, or is it all handled between APPM and ASTAP?  I do have the "Process J2000 to JNOW" set but the documentation implies that only affects the log file. 

I do run this at 1200x and a 1 second settle time, with a 2 second exposure.  It's a very light OTA and a very beefy mount, and I assumed if it wasn't settled it would streak stars and not solve (I had zero errors).  But is there a chance there is something not settled that gives bad coordinates? 

Linwood




Eric Weiner
 

I found the best way to evaluate this is to enable 'save all images' for an APPM run, and then just blink through them when it's completed. I found that plates could be solved successfully with streaking stars, so i upped my settle time. It's worth seeing what that looks like for your run
+1

Eric


ap@CaptivePhotons.com
 

On Mon, Jul 18, 2022 at 03:55 PM, Brian Valente wrote:
I found the best way to evaluate this is to enable 'save all images' for an APPM run, and then just blink through them when it's completed. I found that plates could be solved successfully with streaking stars, so i upped my settle time. It's worth seeing what that looks like for your run
That's a good idea, I can also run through astrometry.net and compare, to Ray's point. 

Linwood


 

>>>  I can also run through astrometry.net and compare, to Ray's point. 

yes, that too. Although a simple visual inspection via blink would show if there was any settle time issues.  

On Mon, Jul 18, 2022 at 1:09 PM ap@... <ap@...> wrote:
On Mon, Jul 18, 2022 at 03:55 PM, Brian Valente wrote:
I found the best way to evaluate this is to enable 'save all images' for an APPM run, and then just blink through them when it's completed. I found that plates could be solved successfully with streaking stars, so i upped my settle time. It's worth seeing what that looks like for your run
That's a good idea, I can also run through astrometry.net and compare, to Ray's point. 

Linwood




ap@CaptivePhotons.com
 

I had an unexpected clear night, with clouds threatening, so decided to burn a few hours collecting data on this subject.

The short version of what I found was: I don't know what I found and think I have no useful data.  But this is a long ramble about what I did collect: 

About 2/3rds of the way through all this I decided to go back and redo the polar alignment (I was using NINA).  I started with 23" of total error, and built a model and did a bunch of testing and maybe 2 hours later reran the polar alignemnt and was off by about 2'.

I did numerous slews, and returns and reruns of the polar alignment and it moved around a fair amount, e.g. 

1'43" x -1'08"
57" x 43"
1'24" x 1.07"

Those are consecutive runs after slewing 10 degrees or so away and back.

I did some more tests and then ran PA again in NINA, adjusted to nearly zero, reran (minimal slews) and was off 7", then ran sharpcap and it said 47" off.  Reran NINA and it now said 2'13".

I just have no clue what is going on, but I suspect it invalidates anything I am doing with the models.  Either that or NINA's polar alignment routine is not working well, but it was consistent when i reran it more or less in place, without slewing around.

This is a 4" refractor, using an OAG, nice hefty M68 image train on a reasonably quality OTA (NP101is). It really doesn't seem like it should be moving all around.  The mount is on a Berlebach Planet with the spikes in holes (to self center) on pavers which in turn are in a bed of gravel, and have been undisturbed for over a year.  

Anyway, in the FWIW department, in terms of modeling - I did a 74 point DEC Arc, 3 arcs, 1 degree spacing.  I grabbed the second image it took and ran through astrometry.net and it agreed with astap: 

ASTAP = 237.89375035703895, 26.634782287087955
Astrometry=237.894, 26.634 


At worst that last DEC angle has a different rounding, but basically they agreed.  It's too tedious to run them all through astrometry to test, but this spot check said it was on point.

I then did a whole series of 400s guide assistant runs, first aimed east, then aimed west, both times about an hour away from the meridian.   Seeing was very good (for example now that I'm guiding I am getting 0.16" RMS error even though i just realized I forgot and left it on 1.5s exposures -- really good seeing (lousy transparency, but good seeing).  I did not refocus during any of those (so no shift from focusing, though it really does not shift significantly even if I did). 

Tests ((east and west are imaging direction not pierside): 

1) With model, east, error 2.1", -2.62"   (these are peak RA and DEC error, not peak-to-peak, just peak). 
2) Without model, east, error 1.77", -2.63"   (effectively no difference)
3) With model, PA adjustments off, east, error 0.80", -3.53" 
4) With model (repeat of 1), -1.10", -4.43"

Approaching meridian, checked PA, adjusted back (at least in theory it now matched where I started) 

5) With model, west, error -5.76", 1.04"
6) Without model, west, error -6.1", 2.07"

Did PA check again, clouds came in, moved back east: 

7) Without model, east, 2.15", 1.5"
8) With model, east, 2.13", 2.05"
9) With model, without PA terms, east, 1.74", 2.37"

The bottom line for me is I see so much variation that I do not think the differences in with/without the model, with/without the PA terms, or east/west make any difference.  Since I think the PA runs itself are likely to be consistent calculations, I think something else is moving around.  I am just at a loss as to what. 

Fortunately I am happy to guide.  Still at 0.16" RMS guiding, happy. 

But puzzled.  I will give it a more thoroughly look in the morning, but just not sure what could be moving around.  Especially in azimuth -- I might believe there's give in the ground for the heavy leg to settle a bit, but left/right? 

Anyway... Apparently I have nothing useful to contribute to the "what's wrong with the DEC Arc models". I'm on the "what part of my setup is put together with rubber instead of metal". 

Linwood

PS. If anyone wants to guess at the problem with the setup, I suspect photos will help: 

https://photos.smugmug.com/AstroEquipment/i-8CC6BMk/0/2f978863/X5/PXL_20220709_002945842.MP-X5.jpg
https://photos.smugmug.com/AstroEquipment/i-D7KwHGg/0/2800ffd1/X5/PXL_20220709_003003935-X5.jpg
https://photos.smugmug.com/AstroEquipment/i-VpN4DMn/0/6d841d0b/X5/PXL_20220709_003019197-X5.jpg
https://photos.smugmug.com/MyT/i-FgWvLKk/0/8ddc992f/X2/A9B00797_111541-X2.jpg




Ray Gralak
 

Hi Linwood,

This is a 4" refractor, using an OAG, nice hefty M68 image train on a reasonably quality OTA (NP101is). It really
doesn't seem like it should be moving all around. The mount is on a Berlebach Planet with the spikes in holes (to
self center) on pavers which in turn are in a bed of gravel, and have been undisturbed for over a year.
Yes, the mount could settle and you probably could not tell that it had. It doesn't take very much to throw off alignment by an arc-minute. Even cement piers can move when the ground swells or contracts from moisture or heat.

BTW, I saw on the news this evening that a van got swallowed in the street into a sinkhole in the Bronx, NY:

https://www.ny1.com/nyc/all-boroughs/news/2022/07/19/giant-sinkhole-swallows-up-van-in-the-bronx


-Ray


ap@CaptivePhotons.com
 

On Wed, Jul 20, 2022 at 12:30 AM, Ray Gralak wrote:
Yes, the mount could settle and you probably could not tell that it had. It doesn't take very much to throw off alignment by an arc-minute. Even cement piers can move when the ground swells or contracts from moisture or heat.
I'm in an area not prone to sinkholes, though there are many other hazards - usually involving naked crackers and meth, sometimes with alligators.   But so far they haven't intruded at night.   (Anyone not familiar just google "Florida Man Story"). 

So I spent some time staring at the setup this morning when I brought it in, and it seems to me there are three likely possibilities: 

1) The ground moves
2) The wooden tripod is moving
3) The only thing I left a bit loose is the four altitude locks, they are finger snug not tight

The first is possible but I really doubt it because the indicated errors on the polar alignment are not consistent over time, if it was shrinking or expanding as the night went on, I would think it would have a more consistent direction.  But possible, just don't know how to test it.  The ground around here is all the same (even under driveway pavers, still sand/soil). 

The second is more interesting - I love the wood's deadness to vibration, but I do go from 50% humidity to 95% each night, and it sits there as the humidity rises further. It seems possible, despite being well sealed, that wood is expanding, moving, stretching or something.  Not sure hot to test that either other than buying a different tripod/pier.  I am curious if others using a Berlebach Planet, especially in nightly setup situations, have noticed this when shooting unguided.

The latter -- well, not sure.  I leave them snug, then as I adjust I lower it and try to achieve alignment on an upward motion, but I do not crank down on the knobs, it (I thought) seems to hold OK since it has the adjustment built lifting it.  I guess it is POSSIBLE that I am introducing some overall slop, even side to side, by leaving them less than tight.  I can certainly change that next time out, but I am not hopeful - the fit between is very uniform, and there's a fair amount of weight (even with the light OTA) sitting on those bolts.  And I do have the azimuth very tight, working against each other to achieve PA. 

I'd welcome any other things to explore, but at the moment my best guess is the tripod. 

Though -- to a question I asked a while back -- this is a good example of why guiding is not ready to die.  This setup guides with excellent results, well under the C11's image scale even with a 3.7um sensor.  So I do not have a PROBLEM.  But I do have a mystery. 

Linwood