Peter,

Phew! You have done some pretty interesting work here researching all the physical dimensions and tolerances, plus you obviously have an engineering or tech background. As far as HGM being able to precisely carve out a bearing seat to a certain ID and then use the precise insertion pressure to collapse the bearing ID to more accurately match the shaft OD perhaps is not even possible with their equipment and manufacturing techniques, of which we are not privy. Machining aluminum is not that difficult, but precisely honing down stainless to a tighter tolerance may be more a challenge. Perhaps you are treading more into hyper-tuning and customization than general manufacturing techniques. Higher precision equates to more expensive machinery and more elaborate QC analyses.? But if you do finally arrive at a user-practicable solution, please share.

Also I am not sure how you are replacing the Oldham coupler with a rigid Allen screw attached one, as shown. Unless the motor shaft and worm are really tightly axially centered and parallel, the set screws would probably loosen over time, or start grinding a depression into the shafts. Or the motor shaft or worm shaft would move around instead. This connection must need a Oldham coupler, flex shaft or universal joint of some type to allow for the differences.?

Best Regards,

John

David,

Without even looking at your logs, it appears you are on the right path. For a scope of this focal length, this guiding should be more than good. Now you may be in the way of just tweaking PHD2 settings for best performance. You probably have the PHD2 troubleshooting guide already.? As far as Dec going off sometimes you might be able to attribute to atmospheric turbulence, increasing exposure to 1.5 or 2 may help. My guiding error can range from 1.0 at 30 degrees altitude to 0.5 or less at the zenith on the same night. My Dec balance is always camera heavy. If you are too well balanced you could be bouncing back and forth between gear faces and having some Dec oscillations. The majority of your scatter plots are nicely centered, and your stars are pretty round as far as I can see. You seem to be set for doing some nice imaging going forward. Image quality is more satisfying to me than chasing numerical perfections :).

Good luck!??

Sorry, forgot how great 1990s technology Email lists are so here is a link to the full size photos.

Alan,

Since I have had the G11G2 mount, I have experienced two types of runaways. The first was from a GoTo command being sent to the mount and the firmware not being able to decide (in RA) what side of the sky to go to, east or west. If the wrong side was interpreted by the G2 unit, the mount would go on a wild slew which would not stop at the safety limits. The Beta ware was the fix for this, and that problem has not repeated in almost three years now.

A few months ago an intermittent problem started where I would get a high rev runway slew in Dec right after start up and if I touched any of the Dec buttons, on the HC or the virtual ones. When I mean high rev, I mean very loud and grindy, and perhaps 3 times normal Slew speed. Folks on the forum, especially Michael and Brendan, surmised this was due to a bad optical encoder, after I did other steps of elimination like swapping motors and cables, and checking the Dec and RA cable sockets on the Gemini PC board. Then 2 weeks ago I replaced the Gemini 2 PC board with a new one from HGM, the only way to replace the encoders without a micro-soldering job with the right chips. So far I think this has fixed the high rev slew problem, but I have been socked in under bad skies and have not been able to give it a full shake down.

But since you are not reporting high rev, and the bad slew is in RA, it still sounds like original firmware issue or bad commands from one of your apps. Removing all software and reinstalling is one option. Trying another PC is another. Attached is the copy of the Beta ware I used if you want to give that another flash.?

Best regards,

John

Okay so I went out Saturday night with halfway decent seeing, no wind, transparency not great.
Mount was adjusted according to videos.
Was tweaking the little screw next to the spring screw, ended up about half to three quarters a turn extra, meshing the worm tighter.
Ran guiding assist in PHD2 then did a quick four worm PEMPro curved program into the mount.

WOW! I think I have this thing dialed in pretty good now. Except for the random DEC excursions.
Attached are some PHD2 screenshots showing the guiding and the DEC excursions.
Also log files from PEMPro and PHD2.
I couldn't believe when I actually had sub .3 guiding for a little while.

I don't know if it is a software issue are not, but when I would have the guiding go bad, I would stop PHD2 guiding, quickly restart,
in the guiding would resume perfectly again. Kind of strange.

Guiding was pretty great all night except when I hit the deck excursions, which you can see jacked up the PHD2 to the .5 - .6 range.
Long PHD2 runs were 19, 20, 23, 25, 26. You can see it was guiding great and then it hit the DEC excursions.
Please excuse the choppy PHD2 log, I was working on camera tilt first.? It's nice to be able to do actual imaging.

Attached are also a quick picture of M13. 46 minutes of 30 second subs.
Iris Nebula total of two hours and 10 minutes mostly 30 subs.
Then I decided to experiment and did a quick 10 minute single sub of the Iris Nebula.
Once I get rid of the DEC excursions it will be nice to be able to do 5 minutes subs.
I was actually also able to incorporate dithering after building confidence in seeing the guiding.
Previously dithering, the mount would never settle down after a dither.

So with the logs posted, I would be interesting to hear back whether it could possibly be the worm too tight,
dirt somewhere or some other recognizable problem. In the PHD2 logs, I can't seem to find any repeatable time
for the excursions but some of them do look like they have the same pattern i.e. shape.

Here is my equipment profile:

Orion EON 115mm Triplet Refractor 805 focal length F/7 -? Feather Touch Focuser
Losmandy G11G
QHY247c? Pegasus Focus Cube V2
QHY-Medium OAG? ZWO asi174mm
GPS, Ioptron Ipolar
2X 12v Deep Cycle Batteries? Pure Sine Wave Inverter
Lenovo Mini PC M710Q 16GB RAM 256GB SSD W10P with a router connecting Via Ethernet to laptop in toasty warm car
SharpCap Pro, NINA, Pixinsight, PHD2, DSS, Photoshop, PEMPro V3.

Thank you in advance for all the help.

Allen,

I wounder?if the difference in TVC values for ?2s ?backlash is due to your step size being 0.56 arc-sec, while mine is 0.17 arc-sec?

Just a thought,

Peter

?Allen which mount? What was the measured PHD2 backlash or what did you estimate it to be with TVC=0?

Let's hope not else this really becomes an intractable problem for people. If your coupler is exposed please try my method looking at the daylight between the key and base plate. I did try your laser pointer method but this is better at seeing the tiniest of movement. You were my inspiration for doing this. ?I also found setting the worm mesh to be a +/- 0.001" affair. I see no way a spring set the right gap. Absolutely no way! Also, good/perfect balance is key. Best done with the worm removed in my experience.

I should point out to folks you must be in guiding rate.

Peter

I determined my TVC values via the laser pointer method as 2 seconds = TVC 10.? So maybe my TVC points are worth 10x yours.

On Sun, Jul 18, 2021 at 01:24 PM, Brian Valente wrote:

but here we are talking about TVC and guiding which are two software systems for backlash compensation. They will fight with each other if not handled properly. This is the recommendation of Losmandy and openPhD.?

?

Brian

Brian,

They should only fight if you are in the unstable zone per my sketch.

Peter

Brian,

As a representative of the manufacturer I appreciate that you responded to my two posts, but I would ask that you please read them carefully as you are in part providing your customers with I feel incorrect/imprecise information. ?The backlash was measured via the Allen key method outline previously and did not involve PHD2.

If gravity plays a role then it in likely due to the needle bearing problem I outlined in earlier postings. I.e. Shifting of the Dec Axis due to a slop in the needle bearings sure to the weight and position of the scope. The shaft is out of spec for these bearings. These bearings where designed to be pressed into place. When done correctly the bearing ID is reduced slightly. The bearings are a light press (Scott himself has said as much). I pressed them in with my fingers! The shaft needs to be exactly 1.25". Mine is 0.004 to 0.006 under size this makes a big difference.?

Did you read read my explanation for how PHD2 works? Yes it does issue a single large pulse when reversing compensation, but it takes that amount of time to take effect. It does not matter if it dynamically adjusts through the night or session to session. It's irrelevant. If the backlash reversal pulse is longer than the guide imaging rate you have a big problem. This will undoubtably lead to instability within PHD2.

Lastly, you absolutely do not want to use a conservative value for TVC, you want to use the correct value. What do I mean by that? For those who use auto focus there is a curve where the minimum point on the parabola is the point of focus. This is true of TVC. Too small a value and backlash remains. Too large and backlash increases but in a negative way, which no doubt leads to the instability you noted. I think it's your job to study this in a precise meaningfull way. Replace my Allen key with a timing circuit. This way you can precisely measure TVC as a function of backlash. I see no reason why thus is not a fixed function. I take your point about being conservative as you do not want to be on the wrong side of the parabola. But setting an arbitrary, often too small a value, makes no sense in my opinion.

Look the backlash problem in my setup was due the excessive pressure put on the thrust bearings. This resulted in too much force needed to move the ring gear. This in turn forced considerable pressure to build in the gearbox and coupler resulting an uncorrectable large backlash. This was resolved with new thrust bearings and determining the correct value for TVC. Until I built my open worm assembly, where I had access to the various elements, I struggled to understand what was really going on, and certainly did not appreciate how important TVC was to eliminating backlash.

Peter

--

I think as much as possible, considerations of DEC backlash on the mount should be de-coupled from PHD2's backlash compensation which is a feature of a particular piece of software.? DEC backlash, whether with the SLW or not, can be objectively measured using whatever method (but apparently not using PHD2 for some reason from what I've read).? Reducing DEC backlash by mechanical means is obviously the best route, followed by software compensation (i.e. not everyone uses PHD2 to guide, though likely the majority).

Keith

--

Derek,

I think you misunderstood what I said. Using the Allen key method, I measured the backlash at around 2sec by counting. I then found the appropriate TVC value to completely eliminate the backlash per what I could see. The Allen key method is very precise. Even a fraction of rotational movement is detectable visually. There is no involvement of PHD2. For the record, The 2.5s measured backlash in PHD2 was from last week before I changed out the thrust bearings and devised the Allen method to make measurements during the day and figured out the correct TVC value.

I'm think this argument that PHD2 cannot measure the backlash on the SLW is highly dubious. Backlash is backlash after all.??I would suggest to Losmandy they create a TVC compensation curve that precisely relates the measured backlash value to the correct software compensation value. This way folks can measure their backlash in PHD2 and set the optimal TVC value.

Why it's important and why you should worry about it.?

I could be wrong on this, but If your backlash is say 2.5s and TVC = 0, PHD2 sends out a large 2.5 backlash pulse + guide pulse. It takes the mount 2.5s to respond and take up the backlash. If your guide interval is say 1s, PHD2 will see no movement in Dec and send out another guide pulse. It will then send out a third guide pulse before it registers any movement. It has now sent out a guide pulse 3 time the amplitude of what was intended, and it will then look to compensate for this over correction in the opposite direction. I think this is why the Dec plot can often look so erratic. If say the backlash is under 1s, then it should register the guide pulse and not send out any further unnecessary pulses.?

If the TVC value is set to the correct value, there entire 2.5s backlash will be compensated for almost instantly (I assume because Gemini moves the Dec motor at a faster than guiding rate). However, I have often seen suggested a TVC setting of 10 or 20. Thus is too small a value for this backlash value. It more likely has to be set to between 120 and 150. Really, this has the be properly studied and should be the responsibility of the manufacturer.?

If you are happy with what works for you that great. I'm more focused on getting to the route cause of some of these problems. I do this so you don't have to?

Peter

开云体育

开云体育

Alan, I tried to search on “forward lash” and all I got was links to custom eyelash extensions !?

What physical cause are you describing with this term??
--
Edward

Allen,

You can set the step size on he brain/algorithm form. Default is 500ms.

Are you imaging or viewing with this scope. I have a 12" F4 truss Newtonian. I figure with all the added gear for imaging it has got the weight at least 65-70lbs. The focuser and motor weigh in at 5lbs or more. I can go OAG and drop the guide scope to reduce weight a bit, and improve center of balance. Would love to see a picture of your final setup.

Also, what TVC value are you using? I found that I can reduce backlash to the point of not being able to visually see any per my Allen key trick. If you are at 1s, I would think the value ought to be 50 or a bit more. A graph needs to be plotted perhaps to set the right value. I.e. Measure backlash in guiding assistant and set TVC accordingly per a graph.

Peter

Think I've cracked and eaten this nut. Here is what I did:

1. Renamed the observatory to Sheep Field Observatory.
2. Replaced the two thrust bearings. The ring gear now moves much more easily when the clutch is tightened down. ?Also, the center point of the ring gear was raising by about 1 thou relative the plate. The worm is now better centered on the ring gear. This change is a big deal, as very little effort is now needed to rotate the worm.
3. Set the worm mesh using a feeler guide while measuring the backlash. It's important to make sure there are no hi/low spots. This is a periodic error problem and can be felt easily by rotating the coupler back and forth. Alignment of the two bearings black is critical, see picture below.
4. Installed the motor, make sure it is aligned, and measured the backlash with TVC set to zero. I made this measurement using using an Allen key, see pictures. This is very precise way to do it, but I have gone over to an open worn design per the pictures. I counted roughly a 2s backlash, perhaps a bit less. I then set the TVC value. Set at 100 there was no? perceivable backlash.? I will measure using guiding assistant in near at first opportunity. It's quite a sensitive thing to set. I have a feeling that folks are using too small a value. There is a chap here (Dave A.) whose mentioned his backlash was about 3.5s. I would suggest he try a setting in the order of 120 to 150.?

John,

Alignment was not an issue with the third bearing. I estimate the shaft if 3-5 thou undersize. I think this was done intentionally by Losmandy for various reasons. Chrome plating might be an option, but I'm not going to spend anymore time on this as the slop as Alen137 has pointed out had no perceivable impact on backlash, but it would appear in the FFT analysis as a large low frequency component, which I think it does, but guides out.

Peter