Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: GytarrMann19 on March 13, 2012, 02:37:12 PM
-
Hello all,
Starting to get my lathe alive and running parts, and having an issue. I seem to be losing position occasionally. I played with the PID settings on the Granite servo drive, and got my following error limit down to 15 counts (20,000 per inch), so I think that should run pretty tight. So I'm trying to run multiple pieces. I'm using 5 tools: face/turn, center drill, drill, cutoff, and barpull. The facing and turning seems to go fine, but occasionally the shoulder I'm turning comes up .005-.010" too long or too short. The bigger problem is when cutting off the part, it can be .05-.100" too long or short. As if Mach is losing position during the tool changes or rapid moves. This doesn't happen every time. The shoulder is off in 1 part in about 10, and the overall length has been of maybe 1 part in 5.
I've tried rehoming the Z axsis, and it generally ends up way off (.100-.200) after doing so.
So where am I losing counts? I guess I need help understanding the failure modes that are possible.
Thanks for reading!
Kevin
-
Could be the active state or you step pulse is wrong or even the pulse width is needing widened.
If using electronic gearing in the Granites that may be the issue, no experience with the Granites but I do know other non industrial drives that have electronic gearing can have issues with it.
Hood
-
Thanks for the reply Hood.
No electronic gearing in use. I increased the step pulse width to "5," and lowered the output frequency in the smoothstepper config from 1Mhz to 512khz. Seems to be working better than it was.
-
Oh sorry pulse wdth will have no affect if using the SS.
Hood
-
What effect does the pulse frequency in SS config have?
-
If you are meaning the controller frequency then it should have no affect on the issue you are having. Where it would have an affect is if you were running out of data being sent to the SS, ie getting that message and hangups etc.
Hood
-
No, it was the maximum output frequency that I was looking at. The instructions say if it is say to high, you lose some resolution.
-
Are you meaning the frequency for each axis? If so set that to what you need or the next nearest above. If you go higher than you need then you will be using overheads that you do not require but you should not lose any resolution
Where it may become a problem is the pulse width may become to small for your drives if you have it set too high.
Hood
-
How do I know what I need?
-
what is your steps per unit and velocity?
Hood
-
20,000 steps per Inch, and 200 IPM
-
20,000 x 200 = 4,000,000 steps per min
4,000,000 /60 = 66666.666666666 steps per second or Hz so divide by 1000 to get KHz so 66.666666666KHz is what you need so pick next nearest above which for SS is 128KHz
Hood
-
Thanks Hood.
-
Tried all that and was still having issues. I switched from constant velocity to exact stop mode and the machine runs much much better. Is that normal?
Kevin
-
The only thing ES does is decelerate to a stop at the end of each line of code before accelerating again for the next line. CV blends the two lines together to try and keep the velocity constant.
Why it would help I dont know, slow acceleration can make sharp corner rounded but shouldnt affect dimensions on straight parts.
Hood
-
What counts as "slow" acceleration? What would a "good" number be?
-
Just spent some time playing with a dial indicator on my Z axis. Run the axis up to zero, then using the Inc jog buttons, back it up. Strange thing- back the axis up .001", but it "jumps" about .004". Jog back the other way, it jumps again, back to zero. Only does this on the first button push each way, I.E. only when changing direction. Each successive move in either direction moves .oo1" as it should. This might say something about my drift issue. I wonder what the problem is- obviously not backlash as that would cause it not to reverse at all, especially not that small of a move. Maybe a gib too tight? "Stick-ion?"
Hmmmm, I'm making progress.
-
Sounds like you have backlash comp enabled.
Hood
-
You will always have some backlash in a system just depends on how much. If the thrust bearings are not preloaded right you can have varying
backlash. Usualy it is not repeatable and varies some. Since it's part of the of the backlash it's difficult to isolate. ie; ball nut +screw+bearing play
What is slow acceleration?
Slow could be an acceleration value which low as compared to say the max acceleration one could use for their system. How much or percent
of max is relative. Acceleration is just a number which one can relate to in how fast the axis can change from current velocity to a higher or lower velocity.
The value for a system is directly related to motor torque since it is the motor which must provide the power / force to overcome the current inertia of the system. Acceleration is specific to your machine, thus you have the capability for some desired rate and it's acceptable or need to change something to get an increased acceleration. The value you use should be reliable ie; you can only have a reliable velocity which would be below it's max and likewise would set the acceleration accordingly.
RICH
-
Backlash comp was enabled. Disabling did not change the drift issue. However, re-enabling it and change the value from .005" to .001" works like a charm. I wrote a test program to to run through a series of arcs and angles, rapids and slow feeds, and its been running through the program for two hours without losing position one bit.
-
Good to hear, having too much in backlash will make it travel too far on reversals. Best to measure what backlash you actually have with a clock and enter the correct value if you want to use comp.
Hood