¿ªÔÆÌåÓý

ctrl + shift + ? for shortcuts
© 2025 Groups.io
  1. MachCNC
  2. Topics

stepper stall (to Art)

Isak Levinson
 

Hi,
When I enable backlash the steppers sometimes stalls.
The problem arises when a continous move is made on one axis (suppose X) and
in the middle of the move the Y axis changes direction.
This situation could be in an arc.
When the Y changes direction, the X axis is halted without deceleration and
then started without acceleration (so it seems).
Maybe that there is some problem related to the backlash algorithm.

When I disable the backlash there is no problem

Thanks, Isak
Art
 

Isak:

This is probably because you have Constant Velocity turned on. When you
run with backlash, CV is not possible on all systems. Usually, you have to
turn it off, then the acceleration and deceleration will occur.

Thanks,
Art
www.artofcnc.ca
Art
 

Isak:

This should only happen in CV mode, or if your acceleration is too high.
Lower your axis acceleration to a ridiculous level to see if it does indeed
decelerate and accelerate. Then make sure the backlash is set to a proper
speed for the correctrion because backlash is not ramped.

Thanks,
Art
www.artofcnc.ca
Isak
 

It seems than when there is a direction change inside a single command (e.g.
G02), there is no effect to the exact stop setting.
For axample:
G1 x0 y0
G2 x0 y0 i10 j0
In this case there is a stall posebility in 4 points.

I used G61 and G64 and the effect is the same.
When the direction changes the other axis stop moving for a brief period of
time (enough to stall it).

Thanks, Isak.

----- Original Message -----
From: "Isak Levinson" <isakl@...>
To: <mach1mach2cnc@...>
Sent: Saturday, August 23, 2003 1:46 AM
Subject: [mach1mach2cnc] stepper stall (to Art)


Hi,
When I enable backlash the steppers sometimes stalls.
The problem arises when a continous move is made on one axis (suppose X)
and
in the middle of the move the Y axis changes direction.
This situation could be in an arc.
When the Y changes direction, the X axis is halted without deceleration
and
then started without acceleration (so it seems).
Maybe that there is some problem related to the backlash algorithm.

When I disable the backlash there is no problem

Thanks, Isak






To unsubscribe from this group, send an email to:
mach1mach2cnc-unsubscribe@...



Your use of Yahoo! Groups is subject to

Levinson Isak
 

It seems than when there is a direction change inside a single command (e.g.
G02), there is no effect to the path setting (exact stop or CV).


For axample:
G1 x0 y0
G2 x0 y0 i10 j0

In this case there is a stall posebility in 4 points.

Could you please check it up.
Thanks, Isak.
Art
 

Hi Isak:

Can you descibe this better. Is CV mode turned off, does the unit stop at
each 90 degree boundry of the circle? Have you tried setting acceleration
very low to see the effect better?

Art
www.artofcnc.ca
Isak Levinson
 

I've set the acceleration to 10 (which is rather small).
Then I issued a "G2 x0y0 i10 j0 f300" command from a x0y0 position.
It was very obvious that there is a sharp stop at each 90 deg.

There was no difference if I issued a G61 or G64 before the command.
Neither was no difference if I set CV on or off in the state menu.

If I turn off the backlash then all is ok.

Thanks, Isak.

----- Original Message -----
From: "Art" <fenerty@...>
To: <mach1mach2cnc@...>
Sent: Monday, August 25, 2003 9:49 PM
Subject: Re: [mach1mach2cnc] stepper stall (to Art)


Hi Isak:

Can you descibe this better. Is CV mode turned off, does the unit stop at
each 90 degree boundry of the circle? Have you tried setting acceleration
very low to see the effect better?

Art
www.artofcnc.ca





To unsubscribe from this group, send an email to:
mach1mach2cnc-unsubscribe@...



Your use of Yahoo! Groups is subject to

Art
 

Isak:

Let me check to see whats what. I'll get back to you...

Art
www.artofcnc.ca
Art
 

Isak:

. You need to decrease your acceleration to about 1 or less to do this
test. Does the axis slow down to stop, then backlash correct, then
accelerate again? I'm just trying to figure out if the algorithms are
working correctly... An acceleration of 10 is NOT small, it means 1 sec to
get to 10 inches per second. While this is not overly fast, it may be too
fast to tell me what I need to know.

Thanks,
Art
www.artofcnc.ca
Isak Levinson
 

Ok, I tried with acceleration of 1. I'm working with mm so it is 1sec for
1mm/sec. It is VERY slow.
The problem persists.
This is what happens:
The 1st axis accelerates as it should.
When the 2nd axis changes direction, the firts axis stops emidiatly.
The backlash compensates for the 2nd axis (I guess that the compensation
process dose not accelerates and decelerates but startst and stops emidiatly
with the correction velocity).
When the backlash is done the 1st axis imidiatly returns to its previous
velocity.

I think that you may have a problem with the algorithm. Maybe you should
break some moves (which have a direction change) into several seperate
moves.
As I anderstand the CV algorithm defines the way in which the path is
generated at the end of each command.

Thanks, Isak.

----- Original Message -----
From: "Art" <fenerty@...>
To: <mach1mach2cnc@...>
Sent: Tuesday, August 26, 2003 1:20 AM
Subject: Re: [mach1mach2cnc] stepper stall (to Art)


Isak:

. You need to decrease your acceleration to about 1 or less to do this
test. Does the axis slow down to stop, then backlash correct, then
accelerate again? I'm just trying to figure out if the algorithms are
working correctly... An acceleration of 10 is NOT small, it means 1 sec to
get to 10 inches per second. While this is not overly fast, it may be too
fast to tell me what I need to know.

Thanks,
Art
www.artofcnc.ca




To unsubscribe from this group, send an email to:
mach1mach2cnc-unsubscribe@...



Your use of Yahoo! Groups is subject to

Art Fenerty
 

Isak:

The results you describe are exactly what will happen if CV mode is
on, was it on for this test, and if you turn it off does the same
thing happen?

Thanks,
Art
Isak Levinson
 

I tried r1.0 and I'm sorry to say that it has the same problem.
Thanks, Isak.

----- Original Message -----
From: "Isak Levinson" <isakl@...>
To: <mach1mach2cnc@...>
Sent: Tuesday, August 26, 2003 2:30 AM
Subject: Re: [mach1mach2cnc] stepper stall (to Art)


Ok, I tried with acceleration of 1. I'm working with mm so it is 1sec for
1mm/sec. It is VERY slow.
The problem persists.
This is what happens:
The 1st axis accelerates as it should.
When the 2nd axis changes direction, the firts axis stops emidiatly.
The backlash compensates for the 2nd axis (I guess that the compensation
process dose not accelerates and decelerates but startst and stops
emidiatly
with the correction velocity).
When the backlash is done the 1st axis imidiatly returns to its previous
velocity.

I think that you may have a problem with the algorithm. Maybe you should
break some moves (which have a direction change) into several seperate
moves.
As I anderstand the CV algorithm defines the way in which the path is
generated at the end of each command.

Thanks, Isak.

----- Original Message -----
From: "Art" <fenerty@...>
To: <mach1mach2cnc@...>
Sent: Tuesday, August 26, 2003 1:20 AM
Subject: Re: [mach1mach2cnc] stepper stall (to Art)


Isak:

. You need to decrease your acceleration to about 1 or less to do
this
test. Does the axis slow down to stop, then backlash correct, then
accelerate again? I'm just trying to figure out if the algorithms are
working correctly... An acceleration of 10 is NOT small, it means 1 sec
to
get to 10 inches per second. While this is not overly fast, it may be
too
fast to tell me what I need to know.

Thanks,
Art
www.artofcnc.ca




To unsubscribe from this group, send an email to:
mach1mach2cnc-unsubscribe@...



Your use of Yahoo! Groups is subject to





To unsubscribe from this group, send an email to:
mach1mach2cnc-unsubscribe@...



Your use of Yahoo! Groups is subject to

Art
 

Isak:

Is this with CV disabled? It still halts hard, then corrects???
Thanks,
Art
www.artofcnc.ca
Isak Levinson
 

It has no difference if CV is enabled or disabled, via the menu or the
g61/g64 commands.
The effect is the same.

As I see your algorithm works as follows:
The G file is parsed and converted into a tool path list.
The kinematics algorithm calculates the motor speeds.
I think that your backlash correction is done in the RT (mach.sys)
service.
This R service does not have knowladge of the kinematics (only the final
speeds).

If I migh suggest an alternative that algorithm:
After you parse the G file you can scan it for direction changes.
Now you can insert the backlash corrections paths into the path list.
And now calculate the kinematics algorithm for the motor speeds on the new
path, which include the bcklash corrections.

That way you will be able to use baclash correction reguardless of CV
setting (If CV is on there could be a slight drift from the optimal path)
You won't have to deal backlash in a special way.
The backlash corrections would comply with the same
acceleration/deceleration as all other moves.
You won't have to define a speed for backlash correction.

I think that all that has to be done is to scan the desired path list for
direction change, and insert the backlash compensation segments in those
points.

Isak.
Isak Levinson
 

Hi, Art.
Are there any new about the backlash correction problem ?
Thanks, Isak.

----- Original Message -----
From: "Isak Levinson" <isakl@...>
To: <mach1mach2cnc@...>
Sent: Wednesday, August 27, 2003 1:51 AM
Subject: Re: [mach1mach2cnc] stepper stall (to Art)


It has no difference if CV is enabled or disabled, via the menu or the
g61/g64 commands.
The effect is the same.

As I see your algorithm works as follows:
The G file is parsed and converted into a tool path list.
The kinematics algorithm calculates the motor speeds.
I think that your backlash correction is done in the RT (mach.sys)
service.
This R service does not have knowladge of the kinematics (only the final
speeds).

If I migh suggest an alternative that algorithm:
After you parse the G file you can scan it for direction changes.
Now you can insert the backlash corrections paths into the path list.
And now calculate the kinematics algorithm for the motor speeds on the
new
path, which include the bcklash corrections.

That way you will be able to use baclash correction reguardless of CV
setting (If CV is on there could be a slight drift from the optimal path)
You won't have to deal backlash in a special way.
The backlash corrections would comply with the same
acceleration/deceleration as all other moves.
You won't have to define a speed for backlash correction.

I think that all that has to be done is to scan the desired path list for
direction change, and insert the backlash compensation segments in those
points.

Isak.





To unsubscribe from this group, send an email to:
mach1mach2cnc-unsubscribe@...



Your use of Yahoo! Groups is subject to

Art
 

Isak:

Still checking, but I cannot repeat it. When I set backlash on and command
a circle, for example, it rotates to the 90 degree points, ramps down,
backlash corrects, ramps back up and continues....
I continue to test...
Art
www.artofcnc.ca
 

Hi Art,
Maybe I didn't explain myself correctly.
The problem is with the other axis (the one that dosen't need
backlash correction).

In a full circuit the axis which need the correction is naturally
deceleratase and accelerates.

The 2nd axis runs at full speed, then stops, then the 1st axis
corrects, then the 2ns axis starts at full speed.

Thanks, Isak.





--- In mach1mach2cnc@..., Art <fenerty@a...> wrote:
Isak:

Still checking, but I cannot repeat it. When I set backlash on
and command
a circle, for example, it rotates to the 90 degree points, ramps
down,
backlash corrects, ramps back up and continues....
I continue to test...
Art
www.artofcnc.ca
Art
 

Isak:

Is this in a G2,G3 move? On mine , both axis slow to a stop before
correcting, then both axis accelerate. (Unless CV is used in which case it
works as you describe.) Is the CV light off when doing this?

Thanks,
Art
www.artofcnc.ca
 

It is in a G2 full circuit move (G2 x0 y0 i10 j0).

It seems that the CV light has no effect on the behaviour.

I changed it by G61/G64 MDI command, and confirmed that the light,
infact, goes on and off.

I've tried with CV off and with CV on.
In both cases there is a sudden stop and then sudden move.

Maybe there are some files which are not replaced during the upgrade
process.

Thanks, Isak.


--- In mach1mach2cnc@..., Art <fenerty@a...> wrote:
Isak:

Is this in a G2,G3 move? On mine , both axis slow to a stop
before
correcting, then both axis accelerate. (Unless CV is used in which
case it
works as you describe.) Is the CV light off when doing this?

Thanks,
Art
www.artofcnc.ca
Art
 

Isak:

I will look deeper and get back to you.

Thanks,
Art
www.artofcnc.ca

----- Original Message -----
From: "sisakl" <isak@...>
To: <mach1mach2cnc@...>
Sent: Tuesday, September 02, 2003 4:55 PM
Subject: [mach1mach2cnc] Re: stepper stall (to Art)


It is in a G2 full circuit move (G2 x0 y0 i10 j0).

It seems that the CV light has no effect on the behaviour.

I changed it by G61/G64 MDI command, and confirmed that the light,
infact, goes on and off.

I've tried with CV off and with CV on.
In both cases there is a sudden stop and then sudden move.

Maybe there are some files which are not replaced during the upgrade
process.

Thanks, Isak.


--- In mach1mach2cnc@..., Art <fenerty@a...> wrote:
Isak:

Is this in a G2,G3 move? On mine , both axis slow to a stop
before
correcting, then both axis accelerate. (Unless CV is used in which
case it
works as you describe.) Is the CV light off when doing this?

Thanks,
Art
www.artofcnc.ca


To unsubscribe from this group, send an email to:
mach1mach2cnc-unsubscribe@...



Your use of Yahoo! Groups is subject to

1 - 20 of 24