Rollover is wrong. Incorrect coordinates of the rotary axis.

[hmnijp]

Faced the problem of programming a rotary axis.
When the work goes with angles <180 - everything is ok. But with turns> 180 degrees - there are problems - they cannot be programmed in one line.

I read this thread - http://www.forum.cncdrive.com/viewtopic.php?f=4&t=2142- it mentions this issue even though there is logic in your words too.

Yes, there is ambiguity in the rotation, for example, from position A0 to position A250. But this cannot be in the G91 mode, even with the rollover turned on, working as it is now!
There can be no disagreement: A250 should turn 250 CCW, A-720 = 2 turns on CW
This is problem №1 - In the G91 mode, the path should not be shortened by multiples of 360 and should have a clear direction.


I will return to the previous topic from the link - problem number 2 - I also think that rollover should not work in the G1-G90 mode
You write -

What you saying makes no sense.
When an axis is rotary and is set to rollover on 360° then it can move 2 ways:
1.) Shorter path
2.) Longer path

Because G1 does not define which way to move, not like G2/G3 which defines CW or CCW therefor the control has to select one of the ways, because it is simply undefined.
So, moving "as it is programmed" makes no sense at all, because it is not programmed (cannot be programmed) with the G0 or G1 which way to move.

Rotary axis now works the same way as in Mach3. (If that makes any sense?)

- this is not true. RS274, mach3, and other systems do not work the same way as UCCNC..
The direction of rotation is clearly identified by the "-" sign!

If from position А10 - I want to go to position А300 using the long way - I will write А300 and it should make a CCW rotation of 290 degrees. If I write G1A-300, it will make a short turn 70 CW.
And you can't do a rollover in this place! It should only work in fast travel mode.

Using this, you can also say exactly what the axis should do in the A3600 or A-3600 command, excluding ambiguity.

[cncdrive]

Test this in Mach3 and you will see that it goes the shorter path in G90 mode.
And as I wrote earlier you can't define the direction of rotation, it is always the shorter path.

In G91 mode it might have to work differently as it is now and maybe you are right about that, but I will have to check it to make sure.

[hmnijp]

Test this in Mach3 and you will see that it goes the shorter path in G90 mode.

Not. I am absolutely sure, and I checked it out - when the "rollover" is on - mach3 selects the shortest path only in G0 mode. there is a checkbox in the settings for this.
in G1+G90 mode - there is a choice to make a short or long path, this is determined by the direction of rotation using the "-" sign.
from A0 to G1A90 - short path to A90, from A0 to G1A-90 - long path to A90.

Another industrial machine(Fadal vmc) works in exactly the same way - this is consistent with the manual attached above.

And as I wrote earlier you can't define the direction of rotation, it is always the shorter path.

Is very sad...
UCCNC is deprived of this choice - it always makes a short path. There is no possibility of choosing a "long path" even in G91. (although I think this is wrong even for the G90 - each rotation has to be split in the postprocessor into rotation angles less than 180 to avoid ambiguity - this is an unnecessary action when you can simply tell in which direction to rotate. And naturally, this greatly complicates manual parametric programming.)

In G91 mode it might have to work differently as it is now and maybe you are right about that, but I will have to check it to make sure.

Yes, there should be such an opportunity - first of all

[arca]

Is there any update yet?

Test this in Mach3 and you will see that it goes the shorter path in G90 mode.

Not. I am absolutely sure, and I checked it out - when the "rollover" is on - mach3 selects the shortest path only in G0 mode. there is a checkbox in the settings for this.
in G1+G90 mode - there is a choice to make a short or long path, this is determined by the direction of rotation using the "-" sign.
from A0 to G1A90 - short path to A90, from A0 to G1A-90 - long path to A90.

Another industrial machine(Fadal vmc) works in exactly the same way - this is consistent with the manual attached above.

I can confirm that the implementation in Mach3 is different compared to UCCNC (although I tested G1+G90 only). And I also think that the implementation in Mach3 is the "correct" implementation. At least it is the one that is commonly used. The postprocessor of my CAM software (Siemens NX) also has exactly the rollover behaviour as described in the first post from hmnijp: the direction of rotation should be depending on the sign.

Any plans to correct this in the next versions?

[cncdrive]

Nobody else reported this as a problem so we did not plan on changing how it works.
But I will talk to one of my collegues about this to see what he says about it.

[hmnijp]

Nobody else reported this as a problem so we did not plan on changing how it works.
But I will talk to one of my collegues about this to see what he says about it.

I think you need to look at how a rotary axis is programmed in other known CNC machines.

In all cases, there are several important things -
1) Sign - determines the direction of rotation
2) G90 and G91 work in different ways
3) there is a choice to shorten the G1 path or not, (even in G90, regardless of the rollover in G0)

Next, I will give fragments of programming manuals that were the first to be found in Google, in addition to what I showed earlier. It is known that other systems have these items, for example, mach3, mycnc, linuxcnc...