Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: dlaery on December 15, 2010, 02:54:01 PM
-
a few days ago I was taking off some material (aluminum) 8" x 2.5", DOC was .053 and stepover was at .2188.
I am using a carbide 7/16 em at 6500 rpm feed rate somewhere around 10. This is on a Taig cnc mill. I turned it on and it had made several passes and everything was fine. I left and came back 20 min later and now the machine had moved over 1.5" in Y The tool was where it should be acording to the Mach3 but it was not where I had set it. I uninstalled the mach software and reinstalled the latest version. and now I have cut 2 of these pieces with no problems. today I recut a pocket that needed to be deeper and everhthing worked as should. I changed to a ball mill, set my zero and started to cut and it was wrong. It had moved X .060. I rechecked zero and it was off .060. It did not throw a reset. I have had similar instances of this but was not for sure.
It looks to me like it is loosing steps somehow. I have not changed anything except in the motor tunning, the Step pulse. 3 months ago when I was having a reset probelm, the Step pulse was at 1 and it was recommended that I change to 9 and the reset problem went away. When I thought I was loosing steps I changed from 9 to 5 and didn't have any problems until just now.
I have no other programs running or installed on this computer. I have done all of the recommendations (found here) to get a computer ready for the Mach software. I am kind of lost at this point.
Any idea what might be going wrong or what I can do?
Thanks,
-
Have you tried setting the pulse width back up again?
Hood
-
Set the step pulse back to 9 and check you max speed. You want to test the drives at max speed when they are hot as that is the time when they are most likely to lose steps. Cool drives will generally allow a bit more current than will warm drives, so you could be a hair to fast on rapid moves.
-
With the spindle on a high speed, do you hear or can you feel a slight clunking noise from the table.
-
I have set the pulse back to 9, but when you say "Check your max speed," do you mean run some code? Right now it is cutting at 90 to see if it going to fail. I have ran this a couple of times at 30 and hasn't done it again.
On the slight clunking, I don't think I hear that. Do you mean while cutting or just running through the motion?
I have ran through this code 3 times, one to finish the job when the problem occured and 2 more cutting air to see if it changed orgin and has not.
-
The thing about steppers is most people run then close to their max and the slightest thing such a dull cutter can tip them over the edge and they will miss steps. It is much better to slow the Vel and Accel a bit in motor tuning, especially on a small mill like you have as rapid speed is not so important as it is on a machine with large travels.
Other things to check for are binding axis, this could be gibs too tight or oil needed on scrws or ways.
Hood
-
The thing about steppers is most people run then close to their max and the slightest thing such a dull cutter can tip them over the edge and they will miss steps. It is much better to slow the Vel and Accel a bit in motor tuning, especially on a small mill like you have as rapid speed is not so important as it is on a machine with large travels.
Other things to check for are binding axis, this could be gibs too tight or oil needed on scrws or ways.
Hood
I was to the understanding that this one would not miss steps because of its
"utilizes advanced Digital Sync Lock Servo (DSLS) control technology. "
When I 1st got this machine about 10 months ago I had a terrible time with throwing th reset. so I loosened up everything trying to stop the reseting because I thought it was binding up. then someone suggestion to change the step pulse and that seemed to stop the random resetting. Now I have been retighing everything back to get rid of the backlash. IF i get it too tight then it throws the reset. then I back it off unti it runs smooth without binding. But I didnn't think it would miss steps if its binding. I don't think its binding but it could be.
-
David,
Usually if you miss a step you know it/ can hear it and the result can vary from a minor position loss to the table just going off with a mind of it's owne.
I will echo what has already been posted.
I don't know what size steppers you are using, but as a rough guide :
- See where the axis will just start skippping on a rapid move, and reduce the velocity to 50%. ( 90 ipm goes to 45 ipm)
- Now do rapid moves for a short distance of say .2" start to finish and see if you lose position, if so reduce the acceleration ( at 45 ipm try accel of 5 )
- If all is well at 40 ipm, then just cut some material at varing depths in increments of say .050" in Al and see what happens.
BTW, IPM (Feed Rate) = # end mill flutes x chip load x rpm
ie: 26=4 x 0.001 x 6500 and chip load becomes important with smaller HP equipment
RICH
-
Are they steppers or servos?
Hood
-
I thought it was a servo. Maybe I shouldn't have said missed a steip. Where I use to work we had a plaxma talble and it had stepper mortors and you could tell when it messed a step.
This link tells what it it.
http://www.microproto.com/DSLS.htm
-
Ok it is steppers but there seems to be some sort of closed loop similar to servos. Problem with steppers is if they are loosing steps its because they are stalling and have no more to give so they cant really be corrected and really the only thing that can be done is to stop them and fault the control. It may be possible to catch them just as they are starting to stall and try and rectify the problem but personally I cant see how that will work and it may actually be where your out of position is coming from. It will all depend how small/large the error allowance is in the loop.
Hood
-
So not knowing how the loop actualy works ,then the only thing one can do is adjust the machine to get it as good as possible, and then find the appropriate velocity and acceleration settings that
provide for reliability.
It's tough to adjust to get "good as possible" unless you measure. If you have a torque indicator you can adjust the gibs, bearing preload, etc. and see the affect of the adjustments.
ie; tight may mean it takes say 60 in oz to rotate the screw as compared to loose which may mean 20 in oz required. Now when running fast and your at the upper end of the stepper speed
there is not much torque left , so say you only have 60 available and if tight, you have nothing left for machining, and the stepper is likely to skip no matter what control is used.
Just making the point that, without an adjusted machine , software settings or controls are not true to their intended use.
RICH
-
I'm still trying to absorb all of this and how to proceed. According to who I bought it from, it is not suppose to skip steps, it is suppose to stop. I don't know but it did. This has happened since I have been re-tighting things back up. So will try to back off a little. I don't know enough about "velocity and acceleration settings " to change those. Thanks for the help. I appreciate all of your guys help.
-
The way I am seeing things is if the closed loop is sloppy it may take quite a bit of missed steps before the error is thrown. Is there a way to adjust the loop?
As for Accel and Vel they are in motor tuning, from Config menu. Lower the Vel and Accel settings by half and save axis and see if it helps. If it does then you can try increasing one at a time until it becomes a problem again then back off a bit.
Hood
-
Is there any way to completely disable the "advanced Digital Sync Lock Servo (DSLS) control technology."? because I'm thinking this might be causing your problems. Certainly your 0.06 inch out of pos is 480 steps out if my maths is correct - which if nothing else says your "advanced Digital Sync Lock Servo (DSLS) control technology" is as much use as a chocolate tea pot. BTW is this 480 steps missed or added? - because if it's added, then it's definitely the DSLS that's the problem.
Also (again) if my maths is correct then I make your "High power electronic chopper drive amplifier" (goodness - are these suppliers over egging their pudding or what?) half step drives - so I'm wondering if Sherline mode might be worth a try - but others will know more about this than me.
Cheers
Ian
-
Yes Sherline may work better, especially as pulse width of 9 helped before it would suggest the wider pulse is required. Just make sure your kernel is 25KHz if you try sherline, then again maybe selecting sherline automatically puts it to 25KHz.
As for the DSLS I am thinking it could be a couple of things,
First its not tripping out because the loop is set so slack it is allowing position to be off by a lot before it trips. The reason it is getting out is due to the motor missing steps ,either due to tight axis or could even be the drive is not seeing the steps, due to pulse width being to low.
Second is like Ian says and its the drive putting out extras because it sees the motor not getting there but then its overshooting.
So definitely worth disabling it if possible as it will definitely help track down the issues.
Hood
-
BTW is this 480 steps missed or added? - because if it's added, then it's definitely the DSLS that's the problem
When I checed my zero in X, it showed minus .060 and shoud have showed zero.
I am going to try to cut the part today.
-
When I checed my zero in X, it showed minus .060 and shoud have showed zero.
OK but it's not just where you ARE (X-0.06), it depends on where you've COME from.
If you'd come from (say) X1 and commanded X0 and it's at X-0.06 then that would be 480 ADDED steps.
But
If you'd come from (say) X-1 and commanded X0 and it's at X-0.06 then that would be 480 LOST steps.
-
Are you loosing position on both axes or just the X?
Is it random and not repeatable......ie; may be happening at anytime while running the program?
Are you using backlash compensation?
Can you post the Gcode for the file your running?
RICH
-
It ran through "bottom.NC" first. Then I had to make make a modification that resulted in moving my starting z down .005. then I modified my drawing and repost the code as "bottom_redo hinge face.NC"
ran the this code and had the problem. The time interval between codes was just en ought time to change the tool. Now when It started, I didn't pay attention to where it was cutting because it was going to cut some air first, which would have been on the left side of the part, then it move over in x approx .500 and that it when I noticed that it was cutting the wrong spot.
This happened 1 time when it moved in y approx -1.5 during a face cut, was recommend to un-install and re-install Mach. did that and a few days later this moved in x happened. I was trying to see what was wrong before I ruined another part. I'm hoping the problem was the gibbs were too tight.
-
Dave,
Are you using backlash compensation?
Exact stop or CV mode? Maybe just post the XML file.
What version of Mach are you using?
I am just attaching a pic of the hinge for others to see that your contouring the hinge and there are a lot of moves ( some of them .005 to .010" ).
With a lot of moves the chance to skip steps goes up and there is a fair amount of change in direction. ( One file has 20000 lines of code compared to 1000 in the other )
I would have thought that since the drive is meant to stop on some range of / or an error your axis would have stopped moving ( i think that's what i read above......) so that would have given you
some idea on where in the code it happened and provide a way of maybe trying to hone in on the problem.
The problem, the way i see it is it can be the drive control loop or the machine is creating the problem. There are a number of things that you could do, such as:
- Air cut only with all axes, with the just the X axis, with just the Y axis
- Bypass that controller board if possible. If the stepper is going to skip it will skip with or without some loop
- Place a pause in the code every so often so you can see if the error is cumulative ie; say every 100 lines in the 1000 line code ( G4 P5 gives you 5 seconds to check position )
- Swap the drives and do similar to the first suggestion
- Just write a little subroutine that repeats some of the moves thus is easier to monitor for postion or try to create a repeatable axis fault
Sometimes it's easy and sometimes it's not...........
How do I know ?........... Well i have a similar problem with my mill when running some 3D contouring. I know it's is not the Gcode, nor the drives, nor Mach, don't think it's a setting
and believe it's machine related. It is a PITA. Since the machine never skipped in years. I have some thoughts on mine, but am not going to muddy this thread up with my problem as i am interested in solving yours.
BTW, Now you don't need to feel like your alone! ;)
RICH