Huge difference in RA guiding on different sides of flip

prastro

Hello,

There is a significant difference in guiding and tracking behaviour for my AM5 before and after the flip.

The test data from last week is given below. However, on going back through the logs, this has been the pattern since I got the AM5 in November. It became more noticeable after I started using an RC6 on it.

Some points.

The error is still within the 0.5-0.8 range quoted by ZWO.
This issue is only when imaging in the East at a pixel scale of 0.56, again less than the 1.0 or so recommended by ZWO for this mount.
Still, the mount has run for couple of hundred hours now, and the issue is always when pointing East (i.e., scope is West of Pier). It would be great to understand what is happening and be able to take whatever action is needed.
It works beautifully even at this focal length and pixel scale when pointing West. So obviously I would like to have the same performance when pointing East :-)

1. First, Summary of data (Order of settings / results is RA/Dec/Total)

Pointing East / West of Pier

Time: 1h17m, 1.5sec, Gain 65%, Hysteresis, Minmove 0.10/0.35, Aggression 80%/80%: 0.67/0.35/0.75

Time: 1h27m, 1.2sec, Gain 70%, Hysteresis, Minmove 0.10/0.35, Aggression 70%/80%: 0.64/0.29/0.70

Pointing West / East of Pier

Time: 1h21m, 1.2sec, Gain 70%, Hysteresis, Minmove 0.10/0.35, Aggression 70%/80%: 0.47/0.34/0.58

Time: 1h27m, 1.2sec, Gain 70%, Hysteresis, Minmove 0.10/0.35, Aggression 70%/80%: 0.50/0.37/0.62

A CN user OzAndrewJ was extremely helpful and analysed the logs, which can be seen at https://www.cloudyni...lub/?p=12505082

Basically, the RA when WoP is more rough and complex - no discernable pattern. After the flip its much more stable and like the ZWO chart.

Second, the behaviour chart for RA guiding RS for each sub, from the NINA metadata plugin

Third, the PEMPro FFT result

These are with guide moves added back, and set to "Favour Frequency" (though "Favour Amplitude" is similar)

When WoP, the dominant period is around 1100 sec but its 431 sec when EoP
The 1100sec period reaches a massive 38as when WoP but just 8as when EoP. However, the 431 sec amplitude is around 17as on both sides.

So something adds that huge 30as sec displacement with a period of 1100sec when the mount is pointed east.

Assuming it is a balance issue, I have so far tried

Rough balance in Dec with a dovetail counterweight
Reducing tripod height to stabilise the setup.
PPEC instead of hysterisis as the guide algorithm
Changing guide exposures between 1.0 and 1.5 sec, and aggression between 45-70%

No significant change.

Scope: RC6 + accessories, guiding with OAG, imaging at 1370mm FL, total weight around 8kg
Tripod: TC40, 5kg weight in weight bag.
Control: ASCOM/NINA/PHD2

Any suggestions / ideas are requested.

prastro

More details

1. Raw (guide moves added back and drift subtracted) RA curve for WoP,
Continuous

With cycles overlapping

Raw (guide moves added back and drift subtracted) RA curve for EoP,

Continuous

Cycles overlapping

EoP the RA cycle is more periodic, and different cycles more or less coincide with each other.

WoP , it much more irregular, and the different cycles all seem to have a different period (as also shown by the FFT).

w7ay

Jhaunton it may be the old advice to slightly unbalance in RA such that the gear train is always loaded against the turn of the drive applies

Yep, I had experimented with my RST-135 and can confirm that a perfect RA axis balance hurts guiding. Just like worm gears, you want the splines to be loaded, even though there are more teeth in contact than a worm.

However, the OP's measurements appears to last for multiple periods;and that should be enough time for the spline gears to fully mesh. The FFT also shows a predominant subharmonic, so there is something odd happening, beyond just bad gear meshing.

prastro

Did you window the data before doing the FFT? FFTs are estimates of the Fourier Transform only for periodic signals where the analysis frame has an integral number of periods. Without windowing, you defintely could see large sub-harmonics. Even a simple Hamming or Hann window could confirm if that large sub-harmonic spike is an FFT artifact or a mount artifact:

https://en.wikipedia.org/wiki/Window_function

Chen

Jhaunton

Hi,

The one thing you have not mentioned is use of the CW or not. You have 5Kg in the weight bag and I know you should not need it with an 8Kg payload BUT, if you want the best possible results, then it may be the old advice to slightly unbalance in RA such that the gear train is always loaded against the turn of the drive applies. My testing to date shows I get better performance with my payload on the WEST of the mount - I am in the SH so it matches your experience. Preload to be heavy WEST in SH, EAST in the NH. This is the opposite of worm based advice iirc - but as the HD rotates backwards (vs spline rotation) I guess it makes sense. Might be worth a try...

prastro

Jhaunton
I have been thinking about this actually.
No RA CW.
I did try a rough balance on Dec with a dovetail counterweight - dont think it really helped.

The issue here is I normally have unattended runs, so switching the weight at flip is not possible.

But still worth a try, maybe I can find some sweet spot where both sides are acceptable

prastro

w7ay
Confession - I did FFTs many many years back :-)

These graphs are straight from PEMPro software, which is designed to analyse the mount behaviour.
The output is almost identical when using the FFT analysis in PHD Log Viewer as well.
So both use the same approach - but I do not know if it is windowed as they have very little documentation.

Let me see if I can run it on PECPrep, which IIRC does have a window option. EDIT: it does, but analyses only residual PE, not Raw PE

Even without that, don't the Raw PE graphs shown in post 2 have completely different shapes for EoP and WoP? Something is definitely strange.

w7ay

prastro don't the Raw PE graphs shown in post 2 have completely different shapes for EoP and WoP?

Yeah, very strange for any gear to behave like that.

You do expect some subharmonics, since the flexible spline of the strain wave gear has some give -- i.e., one 430.82 second period to the next one do not have the same average value, but does not explain why it would be that different in two different rotational directions. (The variation from period to period is what makes perfect predictive PEC impossible; you would need to train the PEC over a sidereal day :-).

If you want to read about behavior of strain wave gears (albeit with the gears from the Harmonic Drive™ company, and from 1991), take a look at this paper:

https://ntrs.nasa.gov/api/citations/19910015306/downloads/19910015306.pdf

Some historical perspectives here:

https://www.esmats.eu/esmatspapers/pastpapers/pdfs/1999/ueura.pdf

Chen

prastro

ZWO team, any comments?

Thanks

prastro

To summarise what the RA performance is currently, pointing east (WoP)

Near celestial equator Dec+/- 10 : 0.65 - 0.70, with quite a few "excursions"

Midway Dec +40-50 : 0.5-0.6

High Dec +69: 0.35-0.40

RS pointing west is consistently 0.4-0.5.

Dec rms is very consistent at 0.3-0.4 and responds well to setting changes. I normally bump it up a bit to keep it closer to the RA RMS.

The Raw RA and frequency plots from PEMPro for the 3 different Dec positions are below.

The dominant period for all Dec positions is around 1100sec
As Dec increases, the Raw PE seems to take on some periodic shape. Dec10 is seemingly random, Dec69 more well defined.
As Dec increases, the amplitudedecreases, from 25as at Dec10 to 19as at Dec42 to 9as at Dec70. Quite expected, and also probably why it becomes easier to guide.

Maybe some mechanical engineers can chip in. Could it be a loading issue - at Dec close to 0, the scope body is more or less perpendicular to the RA axis, which closer to the pole, it is more parallel?

Dec10:

Dec42

Dec69

w7ay

prastro Could it be a loading issue - at Dec close to 0, the scope body is more or less perpendicular to the RA axis, which closer to the pole, it is more parallel?

(1) When you did these measurements, what was the guide camera's angle? Is it possible to redo the measurements so that the declination axis is almost parallel to the camera's x- or y- axis?

All these measurements are measured by PHD2 using pixel coordinates, and they have to be rotated into equatorial coordinates by a matrix that is obtained during guide calibration. If the guide camera angle is rotated with respect to the declination axis, the off-diagonal terms of the rotation matrix are non-zero, and their accuracy can mask what we are seeing. For example, conflating a declination axis problem into an RA axis problem.

(2) For what its worth, there is one major difference between pier-west tracking and pier-east tracking. When you are tracking a target that rises in the east, the RA motor (and gears) is "resisting" (actually braking) the weight of the instrument on the way up to the Meridian. After the target has flipped to the west of the Meridian, the RA motor is now "pulling" with all its might to track the instrumentation towards the west. It is kind of counter-intuitive, but you can visualize it by loooking at the mount.

This happens to all mounts from Fraunhofer's original clockwork (why they are called "German" mounts) to the strain wave geared mounts today.

(3) You might also try to balance the instrumentation on the declination axis as well as you can. With my mount (RST-135), I place the OTA and its declination dovetail plate on a flat table, with a pencil underneath the plate as a fulcrum, to get a position where the OTA "balances" best, and mark that location. When I mount the OTA on the declination on the dovetail clamp, I would try to line that mark near the center of the clamp. I don't try to balance it perfectly, since a little off balance is actually a good thing to have -- but too much imbalance, and it might become a problem.

Chen

ASIMount@ZWO

We noticed you are using RC6 with no focus reducer.
Do you ever try another OTA?
If the 1100-sec behavior remains, maybe we need further analysis with your mount.

prastro

ASIMount@ZWO
Thank you for your response.

I have used AM5 for RC6, Redcat51 and ED80.

Guiding performance on east has always been much worse for all 3. However, I have just analysed the period, which I had not done earlier. The 1100sec is only for the RC6 - for the others, it has been around 430 sec or so.

Caution - for some reason most of the imaging with those scopes was while pointing west, and the east pointing sessions were short.

I will confirm this by trying the Redcat51 tomorrow and having at least 1.5 hour guiding pointing East. If this continues to be 430 sec, then it would seem to be due to the RC6 load distribution.

Thanks for the question - had completely missed checking the harmonics for the other scopes.

I will get back in a day or so with the results.

To repeat what I had said earlier, I understand I am using it beyond the recommendations. But since it guides so well in one direction, it is natural to want it to guide equally well in the other direction as well :-)

prastro

ASIMount@ZWO

Hello, here are results with Redcat51, which is much shorter FL and much lighter.

WoP and EoP imaging times are 1-2 hours and roughly the same on each side.

Please note that I have tried to equalise guiding parameters as much as possible. So all cases are 1.2 sec exposure, hysterisis, Aggression 70%/80%. The only thing which cannot be truly equalised is min-move as, the guidescale for RC6 is 0.87" and for Redcat is 6.45". Minmove in RA is 0.07" and in Redcat is 0.32".

So I would rather focus on the difference between EoP and WoP guiding in RA for a given target rather than the actual numbers.

As you can see

The guiding on both sides of meridian for Redcat is almost the same for a given target (green). Whereas for RC6, the difference is much more (highlighted in red)
Even for Redcat, there is still the 1100 sec harmonic, but it is much smaller, and can be easily guided out. The main harmonic is 430 sec or so (green). This is similar to what can be seen even for the RC6 when EoP.
Only in RC6, when WoP, the primary period is 1100 sec (red).

So is the weight of the RC6, or some weight distribution is causing this behaviour?

If this is correct, then the good news is that it can possibly be corrected with a counterweight - I will test and see.

The bad news - does this mean the weight limit of 13kg without counterweight and "no balance" feature of AM5 is restricted only to smaller and lighter scopes?

Request views please.

prastro

w7ay

Chen, thank you for your suggestions.

w7ay (1) When you did these measurements, what was the guide camera's angle? Is it possible to redo the measurements so that the declination axis is almost parallel to the camera's x- or y- axis?

Yes, the guide camera RA angle from the PHD guidelog was -176 so it would seem that the difference is only about 4 degrees from x/y. This I can confirm as when the calibration happens, visually the E movement is along the horizontal axis more or less.

w7ay (3) You might also try to balance the instrumentation on the declination axis as well as you can.

I have done that already. After the first set of measurements, I checked and found the scope was very back heavy. I then added a dovetail conuterweight in front and used the same pencil approach to keep its CoG approximately over the dec axis. It did seem to improve a bit, but not enough (I feel) to distinguish from seeing differences.

However, it is not well balanced in the 3rd axis due to a filterwheel which is at an angle. There is an OAG on the opposite side, but the FW is obviously much heavier. i suppose I could try that. Though the Redcat 51 had the same filterwheel at a similar angle as well.

w7ay (2) For what its worth, there is one major difference between pier-west tracking and pier-east tracking.

As of now, this seems to be about the only possibility. As suggested in an earlier post in this thread, I will try with a counterweight and see if it reverses if I make it East heavy when pointing East. It may not be a solution that works for unattended imaging, but at least there will be an answer.

ASIMount@ZWO

prastro

You're welcome.
For improving guiding performance, you can refer to this review
https://astronomy-imaging-camera.com/news/systematically-improving-guiding-quality-and-zwo-am5-mount-review.html

Kevin_A

Every worm driven mount I have ever had has always performed better with scopes that are heavier and all my light scopes seems to guide roughly in comparison. On worm drives a slight east heavy and objective heavy setup seemed to work best with worm gears but with light scopes it did not matter much at all. I ran my AM5 with a very slight off balance too with a heavier scope and it ran well, but the very light scope ran just fair. Too much of a coincidence? Not sure! Just my 2c worth.

prastro

Kevin_A Both sides of meridian with same off-balance?

Kevin_A

prastro i noticed that the worm driven mounts were really susceptible to meridian change but the AM5 was not so much, but still worse overall guiding when using light scopes. I think that with lighter scope the the off balance has less effect. I will be trying my lighter scope soon with a larger dec unbalance and see if affects dec results.

Kevin_A

One thing I did to ensure my tripod was stable and had a better centre of gravity without added upper mass was this additional lower counterweight bar. No chance of tipovers and better rigidity.

ASIMount@ZWO

Hi prastro,
We are investigating all your results.
Could you please send a copy of your guiding log to us?
Thank you for all your experiments!

prastro The bad news - does this mean the weight limit of 13kg without counterweight and "no balance" feature of AM5 is restricted only to smaller and lighter scopes?

There are differences between strain wave gear mounts and worm gear ones.
The payload capacity numbers are determined when considering the length of the arm of force is 20cm( from RA axis to the mass center of the payload),i.e. AM5 can hardly drive(especially in slew mode) a Sharpstar SCA260 plus a DSLR mounted on the upper dovetail plate even with a standard counterweight. The total weight is luckily within the recommended numbers but the length of the arm of force is way beyond 20cm.
Therefore we recommend a shorter focal lenth.

Again, we are analyzing your results and will reply to you ASAP.

prastro

ASIMount@ZWO
Thanks for the reply.

Regarding the moment, my scope CoG is about 23cm or so from centre of RA Axis. So torque is 8x9.8x0.23 = 18NM, which is still much less than capacity of 13x9.8x0.2=25NM. I appreciate the arm makes a difference but then 23cm and 20 cm are not that different actually.

Once again, the guiding is very good even with RC6 on EoP. Question is how to get it equally good on WoP :-)

I look forward to your analysis. To which email should I send the logs?