Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: macavity on October 01, 2009, 05:22:45 AM
-
Hi,
i have a problem. Mach does not maintain constant speed (feedrate) when executing my gcode. Differences are rather significant, ranging from 1800 to 1400 mm/min (you may imagine the jerking). and appear when i cut corners using g02/g03. my machine is plotter for cutting glass (thus high speed) but i observe this on 3 different pc, including C2Quad 2,4Ghz.
Has any of you came across similar phenomenon?
-
Are your curves G02/G03 moves or a series of lines (g01)?
Steve
-
Are you using ExactStop mode (G61)? Would probably be better using Constant Velocity (G64) if you are.
Hood
-
Doh, I should open my eyes ???
So much for scim reading.
Steve
-
Doh, I should open my eyes ???
So much for scim reading.
Steve
No your post is very relevant, there is a good chance the arcs are made of short segments and that would just make things even worse if in G61.
Hood
-
that's why I said it but then lloking again he said he's tried it with G02/03 moves.
Aonther one is maybe power supply, does it supply enough to run 2 axis at once?
Steve
-
Thanks for replies,
The problem is i test it on my pc, unplugged from machine, and i observe "deviations" ;) ,what i observe is also reported by the owners of the machine, so it;s not machine issue. I suspect it is something with mach configuration or mach algorithm for generating motor steps. My gcode is composed virtually only of arcs, but they are very short (i take curve made of straight lines and then round the corners). gcode is correct. We tried to "walk around" the problem setting different speeds for arcs and straight lines (which are rare, but then they are very long so machine jerking is not a problem, speed is constant on a line). It helped a little for diminishing differences between speed for line and speed for arc, but when my path is composed of only arcs ( ie 10 arcs one after another making halfcircle) the speed differs while cutting such a path. ( i hope you follow me )
-
It helped a little for diminishing differences between speed for line and speed for arc, but when my path is composed of only arcs ( ie 10 arcs one after another making halfcircle) the speed differs while cutting such a path. ( i hope you follow me )
Why are you using 10 arcs instead of one? Have you tried it with one? I think that even in CVmode, the single arc will cut faster than the 10. What is the radius of the arcs?
-
Are you using G61 or G64?
If the moves are short and you are using G61 then at the end of each line the machine will stop then start the next. This means it has to decelerate then again accelerate for the next move and with the moves being short it is likely you will not be getting up to the programmed feedrate before its again having to slow down. If you use G64 then Mach will attempt to keep the velocity constant and will not slow to a stop between lines of code.
Hood
-
I use g64 ( it appears in the line with other codes in the line MILL->G15 G3 (...) G64).
i unfortunately must compose my arc from ie 10 arcs because i import PLT file and if there is big arc in the file it is made of straight lines (ie half circle is made from 20 lines) so in order to avoid stopping machine and turning knife in the air, i round corners between these short lines, so that i do not need to stop the machine and the resulting path is smoother.
Part of my file:
g90
g64
G01 X1775.295 Y-15.820
g17 G02 X3.923 Y-0.757 I-5.576 J-39.439 G91 A -5.749
G90
G01 X1782.358 Y-17.348
g17 G02 X3.835 Y-1.122 I-10.880 J-44.309 G91 A -5.020
G90
G01 X1789.157 Y-19.480
g17 G02 X3.716 Y-1.468 I-13.443 J-39.453 G91 A -5.494
G90
G01 X1795.702 Y-22.227
g17 G02 X3.562 Y-1.810 I-17.932 J-39.696 G91 A -5.258
G90
G01 X1801.910 Y-25.538
g17 G02 X3.384 Y-2.126 I-21.989 J-38.757 G91 A -5.140
when i type one big arc, the speed is constant for this arc, it is not a problem when speed differs for a line which is long ( either straight or in arc) but i only can generate little, short arcs and between these arcs speed differs most and couse machine jerking
thanks for your support :)
-
Can you attach your xml and maybe also the compete code so that I can simulate it here.
Hood
-
thanks for the effort, really :)
BTW, Could you tell me what mach does when generating toolpath? Because CPU load is the same as when Mach does nothing. Effect is that i wait for ages when i load file with many arcs.
-
Can you describe how your machine is set up, first glimpse at the xml and I see you have an A axis active but no pins or port assigned to it. Also you are sharing Pin6 with spindle and an Output.
Also 20 steps per mm seems rather low but could well be correct if you have 10mm pitch screws and your drives are full stepping.
Hood
-
I can't tell you how the machine is set up because i am just the programmer of the application which converts plt files into gcode.A axis rotates the knife. I run demo version of mach3 and got xml from machine manufacturer. however, even on my pc which is not connected to anything i see that speed varies( label units/min. ). Meybe it is because of the changing direction of the movement? even if i round the corner and movement should be smooth, i observe dramaticac speed changes.
edit:
i think the reason why A axis is not associated with pins is that my pc is not connected and xml is from my pc, not the one from the plotter.
-
I think this problem may be related to the A Axis. Brian is currently working on code for that as there has been quite a few issues recently. A new revision with the fixes will hopefully be released in the next few days, if that doesnt clear things up then we can go from there.
Hood
-
Thanks for the support Hood. I greatly appreciate it :)
-
Hi Hood,
Do you have anymore info about the tangentail fixes?? Like a release date?
Jason
-
Jason,
afraid I dont, just when Brian gets it done but its probably not tangential fixes as such if you are meaning the tangential knife control. The fixes are going to be for the way the velocity is calculated when doing an A move I think.
Hood
-
Thanks Hood,
I to have converted all my shapes to arc's to get rid of all the stop/starting i tangentail control. I have done way with all the G0 codes so that the knife will not spin in the bed. The next wil fix for me would be to look ahead to the G2,G3 codeafter my rapid G1 so that the A can orientate itself for the next cut.
I'm awaiting a fix. I have just go a quote in for a commerical cutter at $82K. I have been patient however now that I have staff running the machine all the quirks rear there head which see's me spending a lot of time tweeking it so the bugs do not tear up the material.
Can't wait for a fix.
-
I may have misunderstood the problem entirely but... why are you programing the A axis rather than using Mach's tangential knife (TK) control? I took your code and removed all the A moves and just let Mach do its thing and movement (feedrate) seemed pretty good to me - definitely not showing the stop starts that the code you posted does.
-
Ok,
thanks for your replies. Update solved the problem. Thanks for guidance :)