Odd G01 behavior in Mach3?

[OlfCNC]

No problem man.
You know with a plasma machine it is not really an option to just lower the acceleration because it burns the sheet on the corners if the acceleration is too low.

[eldata]

20 in/sec^2 is not what I would call "high acceleration" by any means. Also, since your last post reveals that you are using the demo version of Mach, the manufacturer documentation says the demo is limited to 25kHz. 25 kHz and 20,000 steps/in would only be 75 IPM. are you using external step generation or parallel port?

For a newbie like me who rolled off a turnip truck last night, 0 to 300IPM in .25 seconds is subjectively high for a mill. But assume I had access to YouTube and Google while on the truck so know what else is out there. Please don't get that ignoramus started on about his 2G machines.

Been using parallel port at 100KHz kernel frequency for the last 6 years or so without issues even though the demo limitation states;

•Kernel Frequency locked at 25kHz (only applicable if using the parallel port driver)



LinuxCNC limits me to around 65KHz based on latency test jitter. It's the main reason why I seldom use it. One would have thought that with its RT extensions it would outperform Mach3 in this area.

[eldata]

No problem man.
You know with a plasma machine it is not really an option to just lower the acceleration because it burns the sheet on the corners if the acceleration is too low.

No, I didn't realize that. Good to know.

[mactec54]

For a newbie like me who rolled off a turnip truck last night, 0 to 300IPM in .25 seconds is subjectively high for a mill. But assume I had access to YouTube and Google while on the truck so know what else is out there. Please don't get that ignoramus started on about his 2G machines.

6 years and you are still a newbie, must be a very slow learner, seems like you want everything for free and work perfect, it's not going to happen

Just for your smart ass comment/fail, 2G is common day stuff if you have any knowledge of this business, 4m/sec +/- 2 micron accuracy at that speed, most linear servo motors do this all day long, not just the one's I have and use

Mach 3 because it is not closing the loop with your setup, has very little to do with the Acceleration /Deceleration, your problem is your Servo Drive, settings or the servo drive it's self, not being able to handle the Deceleration, anyone who uses Mach3 with servos will tell you the same thing, your Servo is doing all the control of the Servo Motor, Mach3 is just throwing code at it

You have not said what servo's and motors you are using or Encoder count, this all plays the main part of the system

[ger21]

your problem is your Servo Drive, settings or the servo drive it's self,

No, the problem is with Mach3. It's been documented numerous times over the last 5-7 years.

[109jb]

@mactec54. Part of what you said is just not true.

Mach3 is not "just throwing code at it". Mach3 still has to send step and direction signals to the servo drive and it still has to send those step pulses such that the servo will accelerate smoothly and decelerate smoothly at the ends of the motion. if a controller doesn't do this, then the drive will fault, but it isn't a problem with the drive, it is a problem with the step signals that were sent to it.

So, it still makes no sense that a G1 F300 move will fault the drive whereas a G0 with a max axis speed of 300 will not. This problem is still inside mach3, not in the servo drive.

[mactec54]

No, the problem is with Mach3. It's been documented numerous times over the last 5-7 years.

Yes correct when running Steppers, you do not have this problem when running quality servo's

[ger21]

Yeah, if they infinite acceleration....

[mactec54]

@mactec54. Part of what you said is just not true.

Mach3 is not "just throwing code at it". Mach3 still has to send step and direction signals to the servo drive and it still has to send those step pulses such that the servo will accelerate smoothly and decelerate smoothly at the ends of the motion. if a controller doesn't do this, then the drive will fault, but it isn't a problem with the drive, it is a problem with the step signals that were sent to it.

So, it still makes no sense that a G1 F300 move will fault the drive whereas a G0 with a max axis speed of 300 will not. This problem is still inside mach3, not in the servo drive.

No when you have a closed loop Servo Drive the Drive is in control not Mach3, when you are using steppers that is the case, but not with a servo drive is closing the loop

That is why anyone using quality AC Servo Drives and Mach3 do not have a problem with this at any speed, some 3D high speed profiling I have had problems, as others have, but never with normal G1 moves

I guess I have to do another Video and show his same settings, and double them at that to show that there is not a problem with G1 moves

[OlfCNC]

No when you have a closed loop Servo Drive the Drive is in control not Mach3, when you are using steppers that is the case, but not with a servo drive is closing the loop

That is why anyone using quality AC Servo Drives and Mach3 do not have a problem with this at any speed, some 3D high speed profiling I have had problems, as others have, but never with normal G1 moves

I guess I have to do another Video and show his same settings, and double them at that to show that there is not a problem with G1 moves

That's true that with quality servos you will not loose steps, but your machine will still jerk and shake when this abnormal decceleration happens.

[mactec54]

That's true that with quality servos you will not loose steps, but your machine will still jerk and shake when this abnormal decceleration happens.

That jerk and shake you see is that you have the acceleration set to high, if you have this set correct it will be as smooth as silk

[ger21]

Not when Mach3 sends steps without the correct acceleration rate. You can't immediately jump from one position to another without acceleration, and that's what Mach3 does. You either get a jerk, or a fault. Not smooth as silk motion.

[eldata]

Recapping/clarification (probably useful for those with the attention span of a drunken wombat)......

There was never any issue with a single G1 move like this;

G1 Z-9 F300

There was no faulting or jerking whatsoever during acceleration or deceleration. The issue arose when there were two diametrically opposite G1 moves like this;

G1 Z-9 F300
G1 Z0

The drive faulted during the acceleration phase at the beginning of the G1 Z0 move as though the deceleration phase of the G1 Z-9 F300 move never occurred or didn't complete properly. In other words, the start of the G1 Z0 move was not synchronized with zero velocity. If the same thing happened using G0, instead of G1 commands, then I would have simply concluded that the acceleration setting was too high.

Furthermore, both the following correct the issue;

G1 Z-9 F300
G4 P0 ;dwell/pause
G1 Z0

or

G1 Z-9 F300
G0 Z-9 ;no physical move or pause
G1 Z0

(The second workaround is a better choice since there is no pause)

This, together with the fact that LinuxCNC had no issues using the same settings, has led me to conclude that something odd is happening during or immediately after the deceleration phase of the G1 command in this reversal scenario. This is the first time coming across a Mach3 G1 issue in over 6 years of use but it's also the first time attempting to use Mach3 as a programmable square wave sweep generator. I have no doubt that others have seen similar issues associated with the G1 command.

[mactec54]

Not when Mach3 sends steps without the correct acceleration rate. You can't immediately jump from one position to another without acceleration, and that's what Mach3 does. You either get a jerk, or a fault. Not smooth as silk motion.

A good servo drive will fault if the steps don't follow what is programed

The Video I have just done will show different, and I did it using the Parallel Port, on an old XP IBM $100 computer, it is a whole lot better to use UC100 or the Smooth Stepper, with Mach3, but I used the parallel port on purpose, as the worst case scenario, to show that it can run at those speeds without problems if the user knows what needs to be set up, and with a good servo system, anything is possible

Mach3 means very little once you add a motion controller like the UC100 Etc

Would I use Mach3 by it's self, most likely not, but if I had nothing else, and limited the speed that you run at, it works well for what it is, it's been a great work horse for thousands of user's all over the world, the Hobby world would not be where it is today, if it was not for Mach3

[ger21]

Your ignoring the fact that there are bugs in Mach3.

I still run my machine with Mach3, as do tens of thousands of others. And 99.5% of these people never see these bugs.
But there are serious bugs, and you can't ignore them, as much as you keep trying.

[mactec54]

Recapping/clarification (probably useful for those with the attention span of a drunken wombat)......

There was never any issue with a single G1 move like this;

G1 Z-9 F300

There was no faulting or jerking whatsoever during acceleration or deceleration. The issue arose when there were two diametrically opposite G1 moves like this;

G1 Z-9 F300
G1 Z0

The drive faulted during the acceleration phase at the beginning of the G1 Z0 move as though the deceleration phase of the G1 Z-9 F300 move never occurred or didn't complete properly. In other words, the start of the G1 Z0 move was not synchronized with zero velocity. If the same thing happened using G0, instead of G1 commands, then I would have simply concluded that the acceleration setting was too high.

Furthermore, both the following correct the issue;

G1 Z-9 F300
G4 P0 ;dwell/pause
G1 Z0

or

G1 Z-9 F300
G0 Z-9 ;no physical move or pause
G1 Z0

(The second workaround is a better choice since there is no pause)

This, together with the fact that LinuxCNC had no issues using the same settings, has led me to conclude that something odd is happening during or immediately after the deceleration phase of the G1 command in this reversal scenario. This is the first time coming across a Mach3 G1 issue in over 6 years of use but it's also the first time attempting to use Mach3 as a programmable square wave sweep generator. I have no doubt that others have seen similar issues associated with the G1 command.

Just ran what you posted, changed it so Mach3 would accept it and it ran correct, more than 20 times

%
G1Z-9.F300. ( G1 is modal you only need to use it once or if there is a change)
Z0.
M30 (M30 will rewind it to the start of the program)
%

Your 300IPM might be on the edge for your machine, back it of until it runs

[mactec54]

Your ignoring the fact that there are bugs in Mach3.

I still run my machine with Mach3, as do tens of thousands of others. And 99.5% of these people never see these bugs.
But there are serious bugs, and you can't ignore them, as much as you keep trying.

I have never said that Mach3 does not have Bugs, that will most likely never be fixed/ resolved,

[mactec54]

So here is the Video running 3 axes simultaneous, Z is not in the Video as it is being redesigned, it starts off at 100 IPM and then at 300 IPM

[eldata]

Just ran what you posted, changed it so Mach3 would accept it and it ran correct, more than 20 times

%
G1Z-9.F300. ( G1 is modal you only need to use it once or if there is a change)
Z0.
M30 (M30 will rewind it to the start of the program)
%

Your 300IPM might be on the edge for your machine, back it of until it runs

You didn't see this post...

What you said makes sense...If I reduced acceleration (and hence deceleration) from 20in's/sec/sec to 10in's/sec/sec the issue would go away with the same 300IPM feed rate. Feed rate was never the issue, it was deceleration all along. But I chose to keep that to myself until now. The fact that Mach3 G00 moves have no issues with 20in's/sec/sec, LinuxCNC G01 moves have no issues under the same conditions, and placing G00 "placebo" instructions immediately after Mach3 G01 move instructions have no issues, suggests there can be issues with the Mach3 G01 command when relatively high deceleration is involved. I was never going to spend $175.00 just to find out whether Licensed Mode had issues or not, so thank you for your contribution to this thread.

[eldata]

So here is the Video running 3 axes simultaneous, Z is not in the Video as it is being redesigned, it starts off at 100 IPM and then at 300 IPM

Is the video in slow motion?