Issue with rotary-axis feedrate

[arca]

Hi,

I'm having some trouble with the feedrate of a rotary-axis (C-axis of a 5-axis setup).
The main issue is, that when I only have a movement of the rotary-axis and no movement of a linear axis, the rotary-axis will rotate at full speed (with the value that is set in the "velocity" field, in my case 9000 [°/min] )

I've acitavated the "axis is rotary" and the "roll over on 360°" settings.
When I deactivate the "axis is rotary" setting, the axis does rotate with the desired feedrate, but the rollover-function does not work anymore.

Summary:
"axis is rotary" and "roll over on 360°" are both activated
- G1 C10 F60 -> the C-axis rotates with the maximum velocity (which is set to 9000 [°/min])
- the rollover-function works, eg. when I rotate the axis from C=10 to C=350 it rotates on the shortest way

"axis is rotary" is NOT activated but "roll over on 360°" is activated
- G1 C10 F60 -> the C-axis rotates with the set feedrate of 60 [°/min]
- the rollover-function does not work, eg. when I rotate the axis from C=10 to C=350 it rotates on the "long" way. But as it's the C-axis, I need the rollover-behaviour

What I've tried:
- it's the same behavior when I use G94 or G93
- when I move a linear axis simultaneously with the rotary-axis, that works as desired
- always writing out the position of the linear axis does help (in G93 mode), but I'm not sure if I can implement that in my postprocessor:
eg. the machine is currently at X=0, C=0
-> with the code G1 C10 F60 the C-axis still rotates with maximum speed.
-> with the code G1 X0 C10 F60 the C-axis rotates with the correct speed

Do you perhaps have some advice?
I would need the behaviour, that the rotary-axis is moving with the set feedrate [°/min] in G94 mode and that the "roll over on 360°" function also works, when the axis is not set as rotary-axis

setup:
UCCNC version 1.2113
UC300ETH

[arca]

So, according to this post, the feedrate is calculated based on the distance that the linear axes move inthe XYZ-plane, except if there are only rotary axis movements:

https://www.forum.cncdrive.com/viewtopic.php?f=4&t=3064#p23233

When the axes are in rotary mode then the feedrate is calculated on the linear axes (except if there are only rotary axis movements), the rotary axes then only following the XY in a way that when the XY finishes the movement with the programmed feedrate on the XY axes plane the AB finishes the same time.
This is how the motion is planned when rotary axis are in the motion.
The only exception is if the A or B or both axes cannot finish as fast as the XY, then the XY plane feedrate will drop, because then the A or B or both cannot finish that fast and so it is nessessary to slow down the XY. This can happen in cases when the AB acceleration and/or velocity is too low compared to the XY and/or the AB path is too long and so they cannot run that fast even with the set acceleration and max. velocity.
So, this is a physical constrain due to the settings and/or the motion code.

But this is clearly not the case for me: when I only have rotary axis movement, the axis is rotating at max speed. This would happen, if the feedrate is still calculated based on the linear distance: linear distance in this case=0mm -> time to complete movement in xyz-plane=0s -> rotary axis also tries to finish movement in 0s -> rotary axis is running as fast as possible.
So I guess this is a bug? But I somehow can not imagine, that this was not discovered until now. I also tried an older version of UCCNC (1.2109, the first one where the "axis is rotary" function was implemented) with the same outcome.
Is there anybody else using a rotary axis who came about this issue?

[arca]

Still no feedback?

In the meantime I did test it with Mach3, and there the behaviour is correct: when I only use a rotary axis, the feedrate is correct. It's definitely a bug in UCCNC. Should I post this again in the Bug-Report section? (If I don't get a feedback until in 3 days I'll do it anyways )

You can easly test it (you can even use any profile in the Demo-version):

1. acitvate the A-axis, set it as "rotary", set "Velocity" to 1000 [units/min] execute the following code (start at position A0):

Code: Select all

G1 G93
A10 F6


This should rotate the A-axis to the 10° position in 10s. Watch how the A-axis-value in the position DRO is counting up and stop the time. It will be much less, as the axis is moving with the velocity, that is set in the "Velocity"-field of the axis settings.

2. now change the "Velocity"-value in the axis-settings to 10 [units/min] and execute the code again (start at position A0).

Now it will move with 10 [°/min] and take 60 seconds to execute the movement (although G93 is acivated! The axis is still moving with the [units/min] value and not calulating the execution time! That is somehow a "bug in the bug").

3. execute the following code (start at position X0 A0):

Code: Select all

G1 G93
X1 A10 F6


It will take exactly 10 seconds to complete the movement as it should

4. execute the following code (this time start at position X1 A0):

Code: Select all

G1 G93
X1 A10 F6


The A-axis will move with 10 [°/min] (the value in the "Velocity"-setting of the axis), as there is no movement of the linear axis.


This bug could also be related to this here:

In case anyone is interested, I did more testing and I'm sure this is a bug. Compared to mach3, which ran the same model with g93's smoothly, no issues with feedrate jumps or caps. If anyone tests the linked file, issue is most visible when at the largest (third) diameter. So I wont use UCCNC for this purpose until it is fixed.

https://www.forum.cncdrive.com/viewtopic.php?f=13&t=2469&start=20#p19607
These random feedrate jumps happen, if there are some sections in your code, where only the rotary axis moves.


For now, I'll just not use the rotary-axis function and treat the rotary-axis as a linear axis. Then I do only have the problem, that I can not use the "Rollover function". This means my code will include some 6-digit-position-values, which is horrible for readability and debugging. Really would appreciate some feedback, if you are going to fix this issue.

[A_Camera]

Seems like nobody knows the answer...? Or the silence is caused by summer vacation holyday.

[cncdrive]

Because now there are 2 forum threads about rotary axis problems.
My suggestion is to please make a new (3rd) forum thread in which the issues are summarized with description how the behaviour should be changed.
Then we will go through the reported issues, think all through at once, because first we have to see all possible issues to see the whole picture all in once.
I mean if we want to make changes in the rotary axis behaviour then we need to see all problematic points clearly before we doing any changes to it.

[hmnijp]

I mean if we want to make changes in the rotary axis behaviour then we need to see all problematic points clearly before we doing any changes to it.

I found one more problem, in addition to those described earlier - DTG shows incorrect data when the commands of the x and A axes are entered at the same time. Checked in g91 and g90 in demo mode, the problem is observed every second start.

you can add this to the list of problems with rotary axes.

[ Play Quicktime file ] 2021-08-05-19-24-29.mp4 [ 1.7 MiB | Viewed 6396 times ]

[frie]

I find the same problem with my 5 axis setup.
Feedrate for A,B and presumbale also C is not looked at when entered in MDI, it just moves to commanded position at maximum velocity as per axis setup. NOT GOOD!.
On first encounter, seriously worried about the rest to come.

[BKG]

viewtopic.php?f=2&t=3419&p=24979&hilit=rotary+axis+feedrate#p24979