Messages

4071

Mach4 General Discussion / Re: Mach4 problems when moving.

« on: July 10, 2018, 06:40:53 AM »

Hi.

I think there should be two braking modes: with the saved position (FeedHold) and when the current exact position is not important (CycleStop).

This is the same conclusion I come to. If you wish to retain exact position the <FeedHold> then <CycleStop>.  If you wish to stop with the maximum
possible deceleration obtainable with your servo then <CycleStop>

At the current time when you <CycleStop> the deceleration is so abrupt that the servo faults which is less than desirable. You do know however that it can
mange the acceleration in your motor tuning page because otherwise it would fault under acceleration as well.

I'm going to guess that you can program a maximum current in your drive which will in turn limit the acceleration. If however you turn the
current limit down eventually the servo will not be able to even accelerate as quickly as your Mach tuning. If you have the current limit just above that threshold
then your servo will accelerate at the max Mach tuning rate and should decelerate at the same.

There are three fault conditions that could be happening when you execute <CycleStop>:
1) Overcurrent....the drive is trying to decelerate so quickly that the drives current limit is exceeded
2) Overvoltage.....the servo after run is generating voltage which has to be absorbed by the drive and regen resistor if fitted. If the voltage gets too high the drive
     will fault rather than have the DC link capacitors blow up
3) Following error....under normal operating conditions we want the following error window to be small and so be assured that the servo is following its commanded
    position as closely as possible. Under <CycleStop> conditions the commanded position stops whereas the encoder runs on as the servo decelerates with an ever
    widening error and eventually the following error will be exceeded. One possibility it to set the following error window much wider so that it doesn't fault under
    <CycleStop> condition or program Mach to ignore a following error under <CycleStop> condition.

Can you post some details about your servos and drives? A manual would be great.

Craig

4072

Mach4 General Discussion / Re: Mach4 problems when moving.

« on: July 10, 2018, 05:13:55 AM »

Hi,
what sort of servos and drives do you have?

I'm thinking of my Allen Bradley AC servo and with that it is possible to limit acceleration and decelerations. I'm wondering if it is so for your
drivers.

I understand you seek to have the highest acceleration you can. Currently however your maximum acceleration on your motor tuning page must be less than
the capability of your servo or otherwise it would cause a fault, either over current or following error on acceleration. When you hit <CycleStop> however the
cessation of pulses implies a much greater acceleration than your maximum allowed under Machs tuning and in fact greater than your servo is capable of and
that does cause a fault.

I think there is some room to move there.....you could limit the servos acceleration somewhat by imposing a current limit but would still allow a slighty greater acceleration
than Machs tuned max acceleration.

What I am proposing is that you limit your servo/drive acceleration to be just a bit more than your max tuned acceleration. Your max tuned acceleration would become the effective
maximum acceleration AND deceleration.

Craig

4073

Mach4 General Discussion / Re: Mach4 problems when moving.

« on: July 10, 2018, 04:07:07 AM »

Hi,

If you need to keep the exact position - then I use a Feed-Hold.
In any other situation - CycleStop. But it should not cause errors in the servo system of the machine.

My misunderstanding, you are not concerned with keeping exact position? You just want to stop?

If that's the case just program your servo to some maximum deceleration. You will have to disable the following error also as there will
be an overrun that will cause a fault.

Craig

4074

Mach4 General Discussion / Re: Mach4 problems when moving.

« on: July 10, 2018, 03:58:47 AM »

Hi,

I'm sure it's wrong to stop the impulse step instantly.

I agree, if the pulses were to stop immediately it would imply an infinite acceleration. But isn't that exactly what you have described as the fault?
That the deceleration is that violent that the servo faults?

I use other CNC systems, different manufacturers

That may be so but remember that Mach is not a feedback control system, it can't because Windows is not a realtime environment.

Lets say for example that a move involving the X and Y axes is planned. Mach will plan the trajectory and commit pages of motion data to the motion controller,
one page of position data for every 1 ms of the move. During this move you hit <CycleStop>. If Mach aborts the remainder of the move Mach no longer knows
where it is. It knew where it was at the start of the move and it knows where it was supposed to be by end of the move but it has no idea how far along the move it
got when you hit <CycleStop>. I suppose if the motion controller reported the position of its axis motors then Mach could recover its position, after all this is
exactly how G31 probing works.

Mach is not a feedback control system so the motion controller does not report back to Mach in the normal course of events which includes TO MY KNOWLEDGE
ONLY <CycleStop>.

Lets say that we wanted <CycleStop> to introduce a graceful deceleration phase. First mach would have to abandon the remaining portion of the existing move
and then recalculate the pages of position data required to decelerate the axes. The communication delays  between Mach and the motion controller and back again are
way too slow for that to work.

If you use a FeedHold, the machine will stop after a long distance, this is not correct.

I agree it is not desirable but I believe it has to work that way. I reason it that because a deceleration phase is built into the planned move and given that Mach cannot go back
and recalculate the move once its already been committed to the controller the ONLY way to decelerate is to allow the current move to complete and THEN stop.

A I said earlier this is based on my understanding of how Mach works, I am by no means sure. I'm interested how you think it works. Its interesting that you claim that
you have had issues with Mach3 and Mach4....could it that this constitutes an explanation why that should be so?

The laser cutting machine and milling machine are different machines and work differently

I am not convinced this is true. What is true is that Laser requires very VERY high accelerations which make a problem that happens with a mill way WAY worse in a Laser,
but they are both affected by the same phenomenon.

Craig

4075

Mach4 General Discussion / Re: Mach4 problems when moving.

« on: July 09, 2018, 11:20:16 PM »

Hi,
I think the correct way is <FeedHold> followed by <CycleStop>.

FeedHold allows the co-ordinated move already committed to complete including its graceful de-accel
phase.

Craig

4076

Mach4 General Discussion / Re: Mach4 problems when moving.

« on: July 09, 2018, 10:27:04 PM »

Hi,
yes you could  set a max de-acceleration in your servo but with a Cycle Stop command where the
pulse stream stops immediately the servo must overrun a little bit and you end up with a following error
rather than an over current or over voltage fault.

I don't think CycleStop is the correct way to stop Mach, the sudden cessation of movement is effectively
an infinte acceleration and any servo will fault. If there is a means to cause Mach to de-accelerate gracefully
in response to a CycleStop I don't know what it is.

Craig

4077

General Mach Discussion / Re: Help needed, limit swtch trip when turn on spindel

« on: July 09, 2018, 10:03:23 PM »

Hi,
yes, its possible the high frequency current may damage the line filter so don't go
spending up bigtime. Even a modest line filter with a comensurately small capacitors
will help the noise situation, hopefully without the noise current wrecking them.

Craig

4078

General Mach Discussion / Re: Help needed, limit swtch trip when turn on spindel

« on: July 09, 2018, 09:42:55 PM »

Hi,
thats a DC motor and the ouput of the controller is PWM, rich in high frequency harmonics
just the thing for screwing up sensitive signal lines!

I would recommend a decent line filter on both input and output.

Something like this:

http://nz.element14.com/schaffner/fn2020-16-06/filter-16a-chassis/dp/1191368#

Note the filter on the output side of the controller could come under serious stress due to
the high frequency PWM, but still worth a go to tame the noise.

Craig

4079

General Mach Discussion / Re: Help needed, limit swtch trip when turn on spindel

« on: July 09, 2018, 08:58:33 PM »

Hi,
you should first try to eliminate the electrical noise emanating from your spindle.

Shielded wiring and low pass filters are all well and good, but the noise is still
there, eliminating the noise should be your primary goal. Shielding and filtering
reduce the suseptability of the wiring to noise only.

What kind of spindle do you have and how is it controlled?

Craig

4080

General Mach Discussion / Re: Help needed, limit swtch trip when turn on spindel

« on: July 09, 2018, 08:04:10 PM »

Hi,
electrical noise on the limit switch line. RC low pass filters.

Craig