Re: Backlash issue or something else?

Tom Carrico

Just got back from working on the scope. The large spur gear was too tight (thought I had checked it before finishing up yesterday...), took some force to turn.

Loosened the bolts on the motor and re-adjusted the position of the motor to minimize movement on the dec axis and make sure the spur gear turned freely. Took a couple iterations of loosen/tighten to achieve this, but I think I did a reasonable job. There might be some play in dec, but backlash seems to be preferred to anything else.

I will know more tonight, we are having a run of clear weather.

Thanks for the suggestions!

- Tom

On 8/28/21 9:59 AM, Roland Christen via wrote:

Have you enabled auto backlash compensation in PHD?
Unfortunately that will make things worse and result in heavy oscillation.
No software can counter a mechanical static friction, and will be really challenged by retrograde motion. There is no control algorithm that has ever been invented that can counteract retrograde. The opposite of retrograde is backlash delay, and of the two it is far more desirable to have a bit of backlash in the system. Backlash can be addressed in software and does not affect the stability of a control loop the way retrograde does.

Backlash will always be present in any non-encoder mount, whether it has a gearbox or is belt driven. Belt flex has a similar effect but can be lower than a spur gearbox, but there will also be some backlash in the worm to worm wheel connection. Adding precision encoders to the mount axes eliminates backlash entirely.


-----Original Message-----
From: Brian Valente <bvalente@...>
Sent: Sat, Aug 28, 2021 11:39 am
Subject: Re: [ap-gto] Backlash issue or something else?

Hi Tom

can you upload the guidelog that you took these screen caps from?

It sounds/looks like stiction to me

Have you enabled auto backlash compensation in PHD?

On Sat, Aug 28, 2021 at 12:29 AM Tom Carrico <tom@...> wrote:
I have an AP900 that I have had for 15 years. It has worked flawlessly. A couple weeks ago I noticed what I thought was a minor DEC backlash issue, so printed out the instructions and went to work. I could barely detect any movement in DEC with the mount in Park Position 3, but figured I would go through the process.  I verified that the set screw was properly tightened and followed the instructions to 'rock' the dec motor into place. I could not feel any backlash and figured things were going to be back to normal. I apparently made it worse...
This image shows what I now see in PhD 2. At some point it starts making a dec correction and things keep getting worse, almost like the corrections were moving in the wrong direction. I re-did calibration in PhD multiple times to make sure I had not messed anything up. In the PhD graph, each large vertical division is 25 points at 6 seconds per point, or about 150 seconds. The telescope is a Celestron Edge HD 8". The guide camera is off axis and is a ZWO ASI 290.

I changed the min move in dec to 5 to effectively disable dec corrections right after this happened. You can see that the polar alignment is fine, as there were no longer dec corrections. Up until I tried to make the mount better, I was getting rms guide errors consistently < 0.5", typically around 0.4". I have never seen anything like this.

I ran the PhD guiding assistant and it verified that my polar alignment was not too bad, but that there was some serious backlash

This is verified by the backlash graph

I ran this test a year ago, and the backlash graph was a perfect triangle. Clearly I am not doing the adjustment optimally. Do I just need to fiddle with it to get it right? When I try to move the dec axis I can barely feel any movement, am I going for no movement at all?
When I 'rock' the motor into place, I place firm pressure on it while I adjust the bolts per the instructions. I could certainly use more force (or a rubber mallet), but not really sure if that is the way to go.
Any advice would be greatly appreciated.
- Tom C


Brian Valente

Roland Christen

Join to automatically receive all group messages.