Cannot dial in a Series II - any ideas?

[tomindallas]

Info: Series II with ballscrews, DC Servos and Gecko 320's.

I spent the entire day yesterday adjusting the steps per unit in Mach 3 mill. It's consistently off in the Y as seen in the pics. The aluminum part was cut before the additional adjustments, and the wood test was just cut this morning. It just seems off in the Y by 10 thou or so during the same run. Backlash was measured at .002".

The drives are not faulting and I performed the back and forth 4" movement test twice yesterday during the fine-tuning process with perfect results. I use a 4" gage block and it moves 4.000" each way. The part looks perfect on the tool path window in Mach.

This is my first part on a newly completed retrofit, so I'm very happy overall but getting frustrated since I can't get this problem figured out.

Any advice is appreciated!

Tom in Dallas

[keebler303]

When exactly does it get off target? Is the entire part cut with one program? If it moves 4.000 when you tell it to then your steps/in are right on. If the backlash is only .002 the only way I could see it being off is if you were losing steps from mach or your encoders were doing something funny.
What kind of computer are you running it on? You also might want to manully go through part of the g code to make sure the problem is not there even if it doesn't show up in mach. You could also make another program and see what happens with a new program.

good luck
Matt

[tomindallas]

Hi Matt -

Yes, one program. Currently it's drilling, then inside circle then outside circle. I just turned on backlash compensation in Mach and that was a mistake with the outside circle now hitting the holes on the far side of the part.

I'll change the code and cut both circles then come in and drill and see if the holes move toward the front of the machine as it should if it's losing counts or the steps are wrong (right?). If the holes are still in the wrong place then I already wish to claim G-code newbie rights for a hole location mistake.

Argh :-)

Tom in Dallas

[tomindallas]

For posterity:

Found the Y Gecko was jumping the motor around a little but was unable to get it to work perfectly by adjusting the pots - even min gain had it hopping around. Swapped it out with my A axis drive and it's fine now - I'll recut the part and post later today. I may have noise on the lines and one Gecko is unhappy about that more than the other - I'll learn how to use an o-scope and check this all out before I declare a bad drive in public.

Tom in Dallas

p.s. Had to run the program without the head running so I could hear the motor jumping - otherwise I never would have found it...

[tomindallas]

Ahhh, much better with a stable Y motor..

[NC Cams]

Cutting a perfectly round circle is one of the hardest things for a stepper or servo to do because of the instantaneous and diverging changes in axis motion velocity that are going on.

You might find some interesting insight here:

http://www.cnczone.com/forums/showth...t=18178&page=9 (lots of detail here)

http://www.thewarfields.com/MT/CCBacklash1.htm

http://www.thewarfields.com/MT/CCBacklash2.htm

http://www.cnczone.com/forums/showthread.php?t=25181

http://www.cnczone.com/forums/showth...078#post202078 (especially post #41)

And, yes, it does help if motors are behaving properly.

[wddanie]

Tom, what accruacy do you get from this BP retrofit? Did you consider using the Pico Systems bds and emc?

Thanks,

Doug Danielson

[DennisCNC]

What kind of wires do you have from encoders to drives?

I have a setup were a encoder is near 3 VFD's and using twisted pair wires for the A B signals with A and ground twisted and B and ground twisted, get a nice clean signal from the encoder.

[tomindallas]

Hi Doug - I'm limited by my physical backlash on the machine. I've not yet deemed the time to reduce the backlash per NC Cams links of high importance as the parts I put out all fit together fine. I'm not making aerospace parts here yet :-)

I was a Mach/Gecko guy from the beginning due to the huge user-support available.

[tomindallas]

Re: wires from encoders to drives. I bought the wires from US Digital to go with the encoders. The jumpy drive works fine on the A axis but I did reroute all my encoder wires and drive power wires to be separate as much as possible. Since the drive swap it's been a happy machine.