Mach3 tuning (cuts small stuff fine, but is off on big stuff)

[jdbaker]

We recently changed the configuration of our machine (added another motor to the y-axis etc...) and tuned the machine, and just finished our first test cut.

I have programmed in some 1 1/8" circles that cut perfect (those small circles are dead on), but they are part of a larger piece that was programmed to cut out at 40", and it is almost a whole inch off (just under 41"). We can tune it to be exact on the larger items, but when we do that, the smaller items (like those holes) cut incorrectly (they actually cut as slight ovals, with the y-axis correct, but the x-axis off).

It doesn't make sense to me that it would be precise for one but not the other. I found an old thread that someone else started with the same problem (http://www.cnczone.com/forums/showthread.php?t=32297) but there were no replys to it (I revived it with my reply, but it is in a sub category, and I wanted to bring it to the attention of the main category in case some of your experienced guys don't ever visit the sub forums).

Any help would be very appreciated, this is a very frustrating problem that just doesn't make any sense to me.

[SORCHEROR]

are you using two motors on the y axis?what size motors?what power supply and drivers,you have a pic of this machine,could help answer some questions,maybe your binding the further you travel?you have ball screws on all axis?give some info maybe i can help
steve

[jdbaker]

Thanks for replying. We just recently changed the configuration to use 2 1200oz motors for the y-axis (before it used only one 1200oz motor in the middle, and we had a steel bar that ran along the underside that connected the 2 sides of the y-axis together. We had too much binding with this method, that is why we switched to 2 seperate motors with their own ball screws on each side). It doesn't seem to bind any more, and the cuts look nice and straight, it's just they are not accurate.

Funny enough, I don't have a nice picture of the whole machine (I have pics of pieces when we were putting it together). It's a hobby built machine. We have a few computer power supplies linked to power the boards (we have gecko drives (G202 for the x and y axis, just one on the x and two for the y). We have a G201 for the smaller 300oz motor on the z-axis (doubt that matters, but thought I would mention it).

If you think it will help, I will try to go after work and grab some pics of the machine (let me know what parts you would need to see).

[SORCHEROR]

you are using copmuter power supplies?12 volts?that not enough power for the big motors,you should run at least 48 volts,i use 54 volts supply with my geckos and 1050 motors,runs great,and what size motor is on the x axis?usually its the motor tuning thats off,are you using mach2?mach3?check the timing setting in the motor tuning they are defalut set to 1 and zero i think
but ill be honest,the supply is your biggest problem,what feed and rapid you set at?with that low voltage supply you cant go higher that 15-20 ,you will stall and loose accuracy
steve

[Torchhead]

You need to run some calibration test of the bad axis. Put a small bit or pointer on the chuck and run it out an exact distance by the DRO couple of feet and peck the spoil board then run it back to 0. Do it 10 times (you can write a small g-code script to do it). Does the mark always end up at the same spot? measure from 0 and see how far off the mark is from your actual command.

Lost steps will show as inconsistant distance moves. Tuning errors will show as being consistantly off in the distance but will hit the same spot.

You can run the tests at different distances.

What are you driving the Gecko's with? do you have them on separate channels and slaved or did you parallel the Step & Dir off one channel?

1200 oz-in motors are harder to drive than smaller ones. It might show up in longer runs at higher speeds versus short moves.

Another good tests are circles of varying diameters. You will need to either let it cut a piece of test material or fit a marker and big piece of paper or cardboard.

I suspect something else here besides the Steps per unit. You may want the reduce the max velocity and acceleration numbers as a test to see if the problem goes away.

I understand you have several 12V computer supplies in series. That should work but the increased current demand of the two big motors could be causing some imbalance current wise across the supplies. Each one will have current limit, and if they don't share the load equally, could be causing voltage fluctuations. It might help to have a large cap across the combined DC outputs (1000 to 2000 mfd per amp of draw) to act as a voltage "flywheel". Computer supplies are swtichers and don't work as well to supply highly inductive loads. The Caps on the combined output will help with the non-linear loads.

Tom Caudle
www.CandCNC.com