Problem with V2.6 upgrade and rotation axis not limiting to +-180 degree

[andrew...@gmail.com]

Was not having this problem with v2.4

The problem does not occur if I limit placements to one nozzle.

If I run a job with two nozzles being used some parts get put down at wrong rotation and the silicone tube get knotted up.

I looked at the logs and I get RED lines that say

"2026-03-19 21:12:58.553 AbstractMotionPlanner$SubordinateMotion WARNING: Subordinate Motion queue not empty: (C1:88.276618), drained."

I have the C1 and C2 axis set to limit 180 degrees.

What am I doing wrong?

[Mike Menci]

You need to use Vaccum rotating coupling at top of steeper motor shaft connecting to vacuum tube 

Dne 19. mar. 2026 ob 11:38 je oseba andrew...@gmail.com <andrew...@gmail.com> zapisala:



--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/openpnp/e1010e4d-a10e-4cc6-ad73-fd10100f9ca2n%40googlegroups.com.

<knotted_tube.jpg>

[andrew...@gmail.com]

I did buy some of those couplings but never used them because they didn't work great.  So instead I just used very flexible silicone hose.

The machine has run for many many years using the silicone hose without this problem.

OpenPNP version 2.4 did not have this problem.  If the nozzle was at 170 degrees and has to go too 190/-170 it would go the long way around.

I actually don't understand how it keeps twisting around as the Duet3D config sets the max rotation at +-190

BUT even if I did have a rotary coupling - It would not fix the wrong angle part placements that happen long before the hose gets twisted.

For example - I have a board that has two SOT23-6 USB-ESD-Arrays.

If I mark all small components (0603) on the board as disabled then these SOT23-6 both get placed correctly with the right-side nozzle.

If I allow the machine to run with both nozzles (placing 0603 Rs at the same time) the second SOT23-6 always gets placed 180 degrees wrong.  Not some random angle like missed steps/slipping-part would make.  Perfectly bottom-vision-aligned-and-180-degree-wrong

[Mike Menci]

Than you don’t have step/ mm - rotation set up right - remove the hose and test 360 deg - 1 turn 

Poslano iz iPhon-a

Dne 19. mar. 2026 ob 13:25 je oseba andrew...@gmail.com <andrew...@gmail.com> zapisala:



To view this discussion visit https://groups.google.com/d/msgid/openpnp/de4c9aad-6213-4370-ac2e-af04e075d802n%40googlegroups.com.

[andrew...@gmail.com]

Nope - I have steps per degree set up fine.  I know this because I spent 3 days trying to work out why I could not calibrate camera/nozzles after upgrading to the Duet Mini Mr Black gave me.

I had different steps per mm on Mr Blacks toolhead than the old Smoothie had for C axis.  So when commanded to do 360 degrees it was moving 720 degrees.  So I have already had a problem with steps/mm and fixed it.

But I also just checked it now.  If I go to OpenPNP jog tab and move 180 degrees a sharpie mark I placed on the pulley rotates 180 degrees real world.  If I try JOG to 190 degrees the head spins opposite direction to get to -170.

Also if I go to Duet control panel thing on my web browser and I jog past the hard limit in the config file - I get told

"Error. G1: target position outside machine limits"

So I actually have no idea HOW openPNP is managing to spin the C1 axis around 3+ full turns because the Duet hardware tries to limit it.

I thought "Maybe it is loosing C axis steps in one direction" so I ran a job at 10% speed and it still happened.

However if I run a 100% speed job split in two parts "All small 0603 parts for C1 axis / N1 nozzle" and then a 2nd half job with all parts for C2/N2 it motors along quickly and perfectly places all the parts.

It MAY be a strange interaction with Mr Blacks tool head which is also a fairly recent upgrade.  However I only change ONE thing at a time on my machine and it was running fine and placed 50 of this exact same board without a problem on V2.4

So upgrade path steps 

Built machine 10ish years ago using a smoothieboard and always had problems with X/Y drift with small C moves but dealt with it to make 500+ of this exact same board.

Moved from Dining Philosopher feeders to push pull feeders - Ran 100+ boards fine

Upgraded my bottom camera to the new coaxial light I made - Ran 80+ boards fine - much better placement with the new coaxial camera

Upgraded from Smoothie to the awesome Duet 3D Mini with Wayne Blacks CAN tool head - Ran 50+ boards fine (after fixing my step per degree problem)

Upgraded to OpenPNP V2.6 - have struggled to make 10 boards because the silicone hose that has been present for 700+ boards keeps getting twisted in spite of Duet hardware trying to limit rotation to +/- 190 degrees

When I upgraded from V2.4 to V2.6 I did not import my old machine file - I started from scratch with "Issues and Solutions" to remake the machine config.

So It is probably something dumb I have done in that step - I just can't work it out - or understand how C1 can spin 1080 degrees when Duet has a hard limit of 190 degrees.

[andrew...@gmail.com]

Installed the rotary couplers and added some anti rotation brackets.

The couplers have improved by sitting in a draw for a decade - they are neither as stiff or as leaky as when I first got them and put them away because they seemed too poor quality.

Ran 6 boards and did not get the knotted silicone because of rotary couplers - However still got the random 90/180 degree miss placements  if running full job.

I then changed my Duet Config from

M208 X0 Y0 Z-24 A-190 B-190 S1 ; Set axis minima
M208 X420 Y520 Z24 A190 B190 S0 ; Set axis maxima

to

M208 X0 Y0 Z-24 A-99999 B-99999 S1 ; Set axis minima
M208 X420 Y520 Z24 A99999 B99999 S0 ; Set axis maxima

Deleted my .openpnp directory and went through "issues and solutions" to build from scratch again

THIS TIME I allowed wrapping the rotary axis

Ran 6 boards and they all placed perfectly.

So it was either something I had done wrong the first time setting up V2p6 or the new version has some problem with no-wrapping rotary axis.

I still can't understand how the rotary axis could do 3 full spins whilst the duet firmware was trying to limit to +-190 degrees

[Mike Menci]

Hi,
1)  your top bracket holding the rotational coupling must be 100% in centerline of your axis to have a nice centric rotation - but that bracket is not needed at all - it is only adding you trubles !  
(to get good rotation from rotational coupling I would place them in a drill and add some oil to rotation joint first and than install tem...)
2) increase the current to your rotational steppers - to get more torque out of them if needed..... (this steppers can get hot to abt 50 deg to work properly)
GOOD LUCK !

[vespaman]

Hi Guys,

The brackets does not look to be fixed to the rig at all, only taking care of the rotational forces, which makes sense, otherwise the will just continue turning the flexible hose instead.

I have rotational couplings still on my machine, but I don't use them any more, since Jan added the better logic to pre rotate then nozzles. I prefer the flex tubes to .. flex.. since my stepper occasionally could loose steps the couplings, especially it the machine had been sitting unused for a while, and there's little to none performance benefit any longer.

Now, these steppers looks to be much bigger/stronger than my ones, so it'll probably be working fine.

I never ran my couplings though my drill though. 

   - Micael

[andrew...@gmail.com]

<quote>  must be 100% in centerline of your axis to have a nice centric rotation  </quote> 

I can not understand why centerline is needed.  The bracket is just a yoke to prevent rotation.  It is not a rigid coupling and only constrains 1 DOF.

<quote> increase the current to your rotational steppers - to get more torque out of them</quote>

The steppers are very torquey compared to the hollow shaft ones most people use - and they are geared down 3:1.  I can't stop the rotation of the nozzle with my fingers on the 5mm shaft.  I can not imagine they are missing steps unless the software is commanding some insane accelerations. 

<quote>I have rotational couplings still on my machine, but I don't use them any more, since Jan added the better logic to pre rotate then nozzles. I prefer the flex tubes to .. flex.. since my stepper occasionally could loose steps the couplings, especially it the machine had been sitting unused for a while, and there's little to none performance benefit any longer.</quote>

OK - well that is confirmation that v2p6 is working for someone else that has/had limited C axis rotation.  It must have been something dumb _I_ did the first time I tried to set up V2p6

My second attempt with V2p6 is working very well and I am happy now.  I will leave the couplings here for now until I do my new  "Lighter, Faster, Stronger" head unit.

[andrew...@gmail.com]

Further update.  Placed 20+ boards today.  Everything went perfect.  Definitely must have been a software/config issue and not hardware as I ran significantly higher speeds and did not get one incorrect rotation.

My only problems today were the new push-pull-feeders not peeling tape reliably on 3 paper tapes.

Also with my PnP machine running nicer than it ever has I am keen to upgrade a few things I never bothered too in the past.

First of these is I need to find a more repeatable Z axis end stop than the crappy aliexpress opto-interrupters that can be +/- 0.2mm dependant on ambient light.

If anyone has a suggestion of what they use, I am all ears.  Otherwise I am off to fight google search.

[Harjit Singh]

This is my plan - I haven't done this.

3D printers use a piezo sensor for detecting probe tip touching something and it is very very sensitive.

I was thinking this might work well for our application, too.

To view this discussion visit https://groups.google.com/d/msgid/openpnp/a42e4e03-7ff1-46bd-bbeb-3b7fa0c0b9f2n%40googlegroups.com.

[andrew...@gmail.com]

Thanks - Piezo sensor was one of the ideas I had - but reverse of 3D printer - Have sensor on the bed and nozzle can come touch it.

It would have to be a secondary home (like visual home) and still have poor quality optical sensor to do rough home.