Hello Guest it is December 02, 2024, 12:04:10 PM

Author Topic: Formula axis correction limitation  (Read 5172 times)

0 Members and 1 Guest are viewing this topic.

Formula axis correction limitation
« on: June 08, 2013, 01:55:24 PM »
Hi, I have been trying to set my z axis to shift about 17º relative to the x axis.  I use the formula z + x/3.4 (also found that 100x/310 doesn't work) which works unless the z axis is given a g0 z0 command; such is used for a go to zero command or within a gcode, in which case the z stepper squeals briefly while mach3 thinks it is moving, causing the reported z location to be wrong.

I have played around a bit (after hardwiring motors, lubing the rails and trying two different computers) before realising it was the axis correction formula applied that is the cause of this problem.  I have tried slowing the axis acceleration down which does shorten the squeal time, but it doesn't stop the squeal and position loss.  I also tried setting a shallower angle to z + x/100 which seems to be about the point which the squealing disappears.

I have tried a workaround by turning off the formulas before I go to zero, then turning them back on when home.  However this is the point where I found that any g0 z0 command may cause the motor squeal evidenced by a hefty gouge in my workpiece.

I have observed that the acceleration is way above my settings with no audible wind up in the motor squeal.  It seems that the axis correction is occurring without due acceleration and it is only by chance that the axis occasionally elevates at all prior to the next move.  This would explain why small axis corrections work while larger ones don't.

Putting an angled hole through a flat workpiece is very useful to me.  The particular workpiece is set at the 17º angle, meaning that without the axis correction, the toolpath cuts a lot of air.  The cutting depth is also limited to the highest depth the spindle can encounter around the hole, rather than allowing the z axis to follow the angle of the workpiece thereby enabling the maximum cutting depth around the entirety of the hole.

It was a bit disappointing seeing the lovely toolpath make little waves as the z axis lowers, yet being prevented from using this strategy successfully due to a small detail in the software.

Is there a setting that can be manipulated to enable large axis corrections?  Otherwise, is there a possibility that this feature can be implemented properly into mach3?  I have a g540 controller which works perfectly otherwise and I am reluctant to try the tb6540 or whatever those cringeworthy toshiba boards are called (I have received two out of three duds from China).

Thanks
Re: Formula axis correction limitation
« Reply #1 on: June 08, 2013, 02:31:46 PM »
"any g0 z0 command may cause the motor squeal"

Radguy,

This sounds like your velocity/acceleration settings may be too high on your Z axis.  Try reducing them until the G0 Z0 command doesn't cause lost steps.

John Champlain
Re: Formula axis correction limitation
« Reply #2 on: June 08, 2013, 07:56:51 PM »
Thanks John.  I have tried reducing the acceleration down to 50 and the velocity below 10.  The squeal only happens when formulas are on and set to a large correction.  The machine works perfectly otherwise.
Re: Formula axis correction limitation
« Reply #3 on: June 08, 2013, 09:17:05 PM »
"I have tried reducing the acceleration down to 50 and the velocity below 10."

Is this right?  I don't know for sure, but it seems having acceleration so much higher than velocity might cause what you're describing.

What Mach3 versions have you tried?
John

Offline ger21

*
  • *
  •  6,295 6,295
    • The CNC Woodworker
Re: Formula axis correction limitation
« Reply #4 on: June 08, 2013, 09:48:58 PM »
There are bugs when using Formulas. Formula's aren't in effect 100% of the time, and when they kick back in, they can cause stalling.
Gerry

2010 Screenset
http://www.thecncwoodworker.com/2010.html

JointCAM Dovetail and Box Joint software
http://www.g-forcecnc.com/jointcam.html
Re: Formula axis correction limitation
« Reply #5 on: June 09, 2013, 03:29:07 AM »
Thanks John, I am using 3.042.032.  I see they're up to 3.043.066.

Sorry, I should have mentioned that I changed the velocity to 10% of 3000mm/min, not 10mm/min.  I posted just as I was racing out the door.

I'll give the newer version a whirl and post my results.

Thanks Gerry, yeah, it seems that way.  I tried some circle machining as I do to flatten glued sections and the formulas didn't seem to work, but I think that has to do with the start and end points being the same.  I plan to try half circles at some point with start and endpoints in line with the axis correction.

Regarding the bugs, I think it is just a simple omission of adherence to the acceleration when formulas are activated or deactivated that causes the stalling.  Would I be correct in stating that all g0 moves are not affected by the formulas?  This would explain the gouge in my workpiece that happened mid program following a g0 z0 command.

Offline ger21

*
  • *
  •  6,295 6,295
    • The CNC Woodworker
Re: Formula axis correction limitation
« Reply #6 on: June 09, 2013, 08:04:41 AM »
Quote
Would I be correct in stating that all g0 moves are not affected by the formulas?

I don't think so. I believe that the formulas should be working during G0 moves.
I recall someone in the past posting some exact sequences that would cause the stalling, and I was able to duplicate it.
I think that what happens, is that in certain situations, the formulas are ignored, and when they start working again, one axis is out of position (per the formulas), and it tries to instantaneously jump back into position, which isn't possible.

I'd try to find a workaround for using the formulas, as the vast majority of Mach3 bugs like this will never be fixed. When it get released, I'd expect Mach4 to be much better at things like this.
Gerry

2010 Screenset
http://www.thecncwoodworker.com/2010.html

JointCAM Dovetail and Box Joint software
http://www.g-forcecnc.com/jointcam.html
Re: Formula axis correction limitation
« Reply #7 on: June 09, 2013, 12:17:57 PM »
Thanks Gerry, that's exactly what it is doing.  I have found that the squeal phantom distance is the value of the x modifier for f(z).  At x / 500, the squeal is short displacing the reading just under 1.5 mm with the x out at 700 or so.

It's a shame, as it is obviously trying to do the right thing but missing the crucial consideration to f = ma.

I am still a bit weirded out by the behaviour of the formulas, as I was able to run the program with the second computer which didn't gouge the work in a situation that the first computer gouged.  I am still suss on the first computer which seems to have caused problems with another machine.  It has however worked perfectly for a while with the new machine that I am working with now.  The only problem with the first computer is that it is hypersensitive to noise and kept being tripped by the spindle.  This computer still causes motor stalling when go to zero is pressed.  I am curious to run this computer with some better filters, because on test runs, I didn't hear any stalling through the first 15 or so minutes of the program.  I didn't actually confirm this with a location check.  I am pretty sure I also checked the last lines of code through this computer which did cause a stall.  I only started on a workpiece when I had established some trust in the formula functions, but the spindle interference problem annoyed me after two z increments.  That's when I saw how pleasing the toolpath was, albeit briefly.  When I tried the old computer, my jolliness was abruptly extinguished as the z axis failed to rise on a z0 command, exactly where it had done this move correctly on the second computer. It did show me though how impressively heavy a cut my 800W Chinese water cooled spindle can do.  Struth!

I will update again once I have played around with the second computer.  If I only have to delete the last risky lines, that'll do.  I can turn the formulas off for g30.