Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: mhdale on June 01, 2006, 04:28:46 PM
-
Hello Everyone I have an issue I am trying to fix. I have a system which appears to be working well but during a large job it lost position on the Y axis by .2108 Inches. WHich by My Stardards is Waaaaay Out there. I have been trying to trouble shoot where exactly it lost position and what may have caused it. To date I havent noticed any other jobs losing position by that much so I am baffled why it would do so now. I am running some tests trying to re-create the problem so I know If got it.... I dont feel like risking another $500 chunk of aluminium to cross my fingers. But It seems that sometimes My Verify Does not function like it normally does. Worse case I want to be able to verify I am not out of position by pausing and verifying my position.but sometimes... not allways when I hit Verify it travels up to top of Z and then just goes back down and I dont get any Relavent feedback.... I havent noticed a pattern as to why or when this occurs so I am really at a loss... Anyone out there with any suggestions?
Mike
-
What do you have for a drive system (drives, motors, type of drive)? This will start to tell us where we should be looking for the missed steps...
-
Brian,
I am running a Custom Built Stepper system running Pacific Science stepper drives & Motors with Mach 3 running on a P4 2.4G with 512 Ram and nothing else. I did have the screensaver on (Dont know why).
My big concern is that I have been unable to duplicate the problem, I.E. I have run a condensed version of the same program with less depth cuts and at a much faster feedrate to try and see if there was a motor setting or something and I cannot get it to lose position....also I cannot see any part of what it did machine where it might have gotten caught up .. everything looks perfect except that it gradually lost position on the Y axis....
Can you shed any light on the verify position behavior? If I can count on the verify position I can step through the job one section at a time.... painfull but at least I can get it done.
Thanks in advance
Mike
-
What do you have for a pulsewidth under the motor settings? You may need to send a longer setp pulse for the drives...
-
Pulse width is set to 4. But would that not show up all the time? What sort of symptoms would having a pulse width to short have on the overall accuracy of the system? This one has me stumped and very nervous about doing large jobs.....
I do appreciate your assistance and insight and anything I mean anything you can think of which might explain why it would lose position over a long job as opposed to same job cutting air with much higher feeds rates and less depth cuts...which I would think would tend to exacerbate the loss of position.. but did not.
Cheers
Mike
.
-
Brian,
If you Interrupt a program can you do a verify position without screwing anything up? Is that possibly whay it seems to not want to run the verify?
Mike
-
Hi:
The only condition I can think of where verify will lift and immediately drop again, is when the system senses that there was home ref done since turnon.
The Z lifts to prepare fo rthe move to home. If it then senses no legal home position has been set, then it will simply dop again. Is it possibel that the homing was not done in that session, perhaps the program was started up and zeroed to the work for a run, but hjoming to the switches was not done?
I agree with the position being way to far off, .2 inches is way to far in anyones books. :) , but looking in the code, the only possability woudl seem to be that the homing was determined to be not sone prior to verify, or the home switches sensed as inactive.
If in you rposition, Id run some tests to see if I could track it down. First, single axis tests.. 100 rep's of G0X10, G0x0 and makie sure it returns to zero. Repeat on each axis. Then I do 100 rep's of G12I10 with the tool zeroed in the center of the table. Does it come back to perfect zero?
These tests usually help track this type of error down. Let me know how you do.
If the program senses it woudl be wrong to do a verify, you wont get the Z lift in preperation, so I think your in an OK condition when your trying to do it.
Thanks
Art
-
Hi, Mike
You mentioned that, " I did have the screensaver on (Dont know why)" this has cause problems for me in the past, Turn it off.
If it goes to sleep the wake up can cause lost steps.
Thank's, Chip
-
Hi, How about stepper resonance problems at certain feedrates?
Kind Regards
CJ
-
CJ Note sure I know exactly how to tell if it is a stepper resonance problem....
Just an update on this problem I think my problem went away for a while, or I just havent had time to really delve into things, because of issues with the cnc we have not been using it as much as I would like.... However I have had some projects recently which have forced me to re-examine things again, so Im back at it and I have noticed the machine losing position a tad, I am just running a pretty long file and I will check at the end to see what I have lost. Now I am pushing my machine and I am sure I can slow everything down so that I never have to worry, but whats the fun in THAT!? There is NO reason I should need to run my system super slow... is there? In response to the inevitable question what type of machine it is a modified Red/Cam gantry CNC with Pacific Scientific nema 34 Steppers and drives the CNC 4PC breakout baords and a P4 2.8 G with 1 Gig of ram running mach .. spindle is a columbo 5hp ATC I am runing the MACH QUANTUM and I am loving the cutting speeds I can get with it. I am just concerned by the system seeming to loose position. I have double and triple checked motor settings and tried varous kernel settings. I seem to loose more steps with the kernel running at 45K as opposed to 35K without changing any settings. And when I try doing tests as ART suggests i cant seem to get it to lose position unless I ask it to do crazy things which will stall the motors..... is it possible its a PC problem?
Cheers
Mike
-
Just an FYI position check came back with x-.0069 y-.056 z-.0012 I can live with the Z and on a bad day the x but the Y is getting bad..
-
Ok, Another stupid question if you Home your machine and then go hit verify position I get a bit of variance usually around half a thou, but If I keep hitting verify that number just keeps getting bigger and bigger... is this "normal" behavior is that what other systems do?
Mike
-
Mike:
Does it lose at 25Khz, you should never go higher than 25Khz unless you have a perfectly workign system.. ?
Art
-
Ok, I will try it at 25 K. Just another not I did run the tests and the results I get are inconsistant, i.e. at 45K I run the same test twice, One one test it will be within tolerances and on the next it will loose 17 thou..... have you ever come across that?
M
-
Alrighty, I have my kernel set to 25K Max feed Rates are 180 ipm
I Have adjusted my accel rates from 28 to 16 and I seem to get pretty consitent creep in my x and Y axis, my z seems to be staying within tolerance. I have no idea what tolerance for this type of machine (Gantry) should be but I am taking a WAG at .005 or +/- .0025 The Limit switches definitely are within that as now at the 25K kernel speed I can hit verify till the cows come home and it stays under .001 If I do 100 repeats of a 5Inch Move in either the X or Y Axis I definitely get higher numbers out on that axis up to around .014. Any suggestion on dialing this in any further?
Mike
-
That sounds about as good as you can get for a router... Other then that you are going to have to move to servos and use the index to home the machine.
-
Brian, Actually when it is working correctly the system has been able to maintain under .005 the problem is is that the number just keeps getting bigger! The larger the toolpath the worse it is at the end.
M
-
Sounds to me like you have a noise problem... are you running steppers or servos?
-
Hi:
Have you tried running in Sherline mode as yet? Id be interested ot know if it loses position in that mode..
Art
-
Running Steppers, and I have not tried the sherline mode. I too suspect that it is noise in the system but I am uncertain as to where to start. I guess I am just putting off the inevitable and should just get down to re-wiring the whole control box as I think noise has been a problem for a while...
I was hoping and wishing it was anything BUT noise.....
What exactly does the sherline mode do ?
-
Ok. When faced with the daunting task of RE-Wiring the Frankenstien that is my CNC Machine.... I BALKED and tried a few other settings... I did scope the outputs to the stpper drives and it seemed not too too noisey but without hooking everything up to the other computer to record the scope in realtime I am only seeing a fraction of the pulse train. I Changed the STEP LOW Active setting on my motor outputs and that seemed to cover most of the noise I was getting.....I ran a long test which was showing a .064 deviance before and this one came in under a thou.... so I think I have solved the issue.... untill the next modification.... Thank you Brian and Art for your patience and understanding! BTW QUANTUM ROCKS!
-
Ok..... (Looking Sheepish) I guess I got ll excited about a one time fluke thing because after a few tests of working flawlessly, my system started acting up again....... Sigh I guess its time to turn this ugly duckling into a swan.... Or maybe time to dust off the G-Rex that has been sitting on my shelf waiting for that perfect time......
Thanks again Guys...maybe the next fix will stick ;)
Mike
-
Hi, I'm George from Cedar Creek 3d,
This is my first shot at forum, so please bear with me.
I have a TAIG micromill that worked perfectly for many months. Recently, the x-axis started bouncing, mispositioning, and failed to reverse. Used Mach2 initially, switched to Mach3 and still the same. Switched the x and y cables, y-axis now malfunctions. Sent drive unit back and was returned stating nothing wrong. They told me to check my computer, couldn't find anything wrong.
Any body have any ideas? It's driving me nuts.
Thanx, George
-
Hi:
When you say you swapped cables and the problem followed the cable, what cables? The motor power?
Art
-
I have had some lost positions on some long runs cutting foam and soft woods and haven't found the culprit except maybe the OS from Microsoft.
Has anyone tried the REALTIME function available in Task Manager?? I stumbled across the function by right clicking on Mach3 in the Process Manager and a drop down appears where you can set the priority to Realtime. You also get a message indicating some bad things can happen. BUT, if this stops MS from interrupting Mach3 's functions during a prolonged run .........
JWI
-
Hey JWI,
I have no idea, this is the second time I have seen the same question from you. If you don't get a response tomorrow, I'll ask some of the guys that may have an answer. Sorry I can't help.
Brett
-
Hey guys, I would be very interested in learning more about this realtime parameter as well. The problem with a problem like this is that it is so... vague... unpredictable. I think that it is unlikely that mach would be loosing steps on long runs because I suspect many more poeple would have issues. It is possible it has some relation to our PC's or more likely the whole package. In my Case I suspected my PC Power supply as being a bit weak and causing some possible issues with the ports maintaining a clean signal. Add to that a frankenstien creation which has been re-wired more than my house. I have seen bad noise from my system in the past so it would be my first pick....if and when I get around to re-wiring the darn thing you will all be the first to hear wether or not it clears it up. I am VERY interested in what solves other peoples issues so please let me know if you track down your own.
Mike
-
JWI, is this a dedicated PC? Have you gone through the Optomisation File in the downloads section? Does it have a network connection?
Brett
-
Hi Brett,
I have optimized the unit per the guide, but only succeeded in stopping all communication from the PC to the control box. The PC is dedicated and the control box holds 3 Gecko 320's, an opto BOB and a very good Power supply with a flow thru cooling system. The problems in the past have been in very long runs < 3 hrs. where one or all of the axis' will loose touch with reality. I haven't had a long run come up lately, but when I do I will try the Realtime function to see it it cuts off the extra stuff that the MS OS produces. I'm interested in this as the EMC program seems to thrive in a realtime environment
Thanks for your reply.
JWI
-
JWI, How does your out of position show up? Does it just creep out a little on certain axis depending on the program? Some of the programs I have run are 6-12 hrs, the really wierd thing is that after discovering the problem I went back through and basically ran the same program step by step and verified position in between and it was fine..... (Sheepishly) However I did have the screensaver turning the monitor off after 2 hours of idle time... which MAY have caused some of the problem. I have since had it show up with no screensaver... But again... could be noise....
I would be very interested in any info and insight you learn from your own issues.
Mike
-
Hi Mike,
My position problem is not just a few thou. What happens is one (or more) of the axis will drop a large amount of positions, sometimes to the point of crashing the entire machine. It is rare, but still a pain. I have not had a program run over 6 hours yet, so can't comment on the precision after that period of time. I do not think it has anything to do with MACH's interpretation functions. It could be something in the computer system such as non-ECC memory or something weird like that. Anyway I haven't had a problem since turning off "Automatic Updates" and never allowing the monitor to go to sleep or screensave.
BTW - When setting the priority of Mach to Realtime, I have lost a USB wireless mouse function. I was only manually moving the gantry, but had to reboot as not even the 3 finger salute would break into the system. So beware. I am going to try some higher priorities just below Realtime.
Regards
JWI
-
Thanks for that info. JWI. Every little bit helps. I have sent you a personal message.
Brett
-
JWI, Sounds all too familiar, I had been running fine for about a year with my system and then on one large job it lost .268 on the Y axis.... and .06 on the x but it happened over time because you could not tell by looking at the job untill it went back and did some re-machining with a diff bit..... then it became pretty obvious. I will look into the realtime setting in Windows... I have not optimized my system as the first time I tried, it ended up not booting up...and then it complained about everything when I finally did get it running... so I have been a bit shy about trying that again...
Mike