EZ Trak Drifting

[Mechie3]

We have an EZ Trak (maybe an SX II?), 2 axis CNC in our engineering prototype shop. It worked well for a few months then started to drift while machining. If we cut foam, it cuts a perfect circle. If we cut metal, or anything with resistance, it drifts. The same circle cut in the foam becomes an ellipse slanted up and the left.

I replaced the encoder on the x-axis with a brand new Heidenhain replacement, but it still drifts. I need to check the Y axis to see if it is loose/damaged.

Any suggestions?

Thanks.

[machintek]

The first question is does it return to zero? Try rapid X moves and see if it returns to zero as well as Y moves and check for zero return.
Check motor belts to make sure teeth are not worn. Check pulleys to make sure they are tight on shafts.
The key is that it appears to be load related, thus something is slipping. That is why I am not going to cables and such.

George

[Mechie3]

It does return to zero in both X and Y after rapid moves (using jog button). I had the machinist push against the table as we jogged it as well to see if the small amount of resistance we could put on it would affect it, which it did not. The X belt looks good, haven't had a chance to take apart the Y belt.

The pulleys are held on with 4 cap screws and sandwhich 2 split ferrules and a spacer between the pulley and a back plate. Is there a specific torque value for these or a method to check that they aren't worn?

[machintek]

I remember on the Y axis there is a aluminum spacer that pushes on one of these split taper rings and it gets chewed up but then you start to lose clearance between the dial and housing.
I am starting to believe you may have a brush issue in the motors and/or a tuning issue. In theory, these SEM motors should be examined every 4000 hours of operation. If the armature brushes get worn down, the resistance goes up, as the contact pressure goes down and the result is that the current cannot be put through the motor, less power therefore and the position deviation goes up.
This can be verified by running a program in PFM.EXE in DOS. In the tuner section you can run stuff and watch the position error for each axis.

George

[Mechie3]

Dials feel tight. Checked the gibs and they are good too. I also found that if you cut that same circle twice, on top of each other, the second one (with minimal load) follows the true pattern while the first cuts the ellipse.

I found in another thread how to access the PFM, but I can't do anything. I have a keyboard plugged in but when I click run program (2?) and type in 98765.pgm (generic single circle program) it says "error loading program".

Doing rapid moves X is off by .068" during the move and Y is off .032" (moving 15" and 7" respectively) but the error zeros out after the move is complete. Since I can't run a program, I'm unable to watch errors during a cut. Am I missing something?

Thanks.

[Jukka Honkanen]

HI,

Check the lag display while running the axes. The values shuold be same at both axes (X and Y) while running at the same speed. Usually I run my tests at F1000mm/min.
Depending on your control parameters the LAG display should be either zero (or as close as possible) or depending on the feedrate and KV-factor some value.
(if the speed is F1000mm/min and the KV is 1, then the display should be 1.000 (mm) on LAG display while running at a constant speed.
If the value is not ok, you should be able to adjust that on the servo.
There should be a potentiometer called tacho, speed or something like that.
This is a normal basic speed adjustment of a servo drive.

Jukka

[machintek]

I have to be in front of a machine but as a guess, PFM is running in DOS so try C:\pgmfiles\59765.pgm to enter a location of a file to run.
To tune drives there was a default folerror or something like that that did 45 degree moves of the x and y.
Besides balance there is a tacho pot and a gain pot. If gain is too high it will go into oscillation and sing/growl at you.

George