Compensation for axis drift?

[justrack]

I just finished installing my first copy of EMC2 and connecting it to a homemade plotter made from an old printer that used stepper motors. Two Gecko G201 drives power the steppers and are connected to the parallel port. I was able to get everything connected and in working order. After making some test drawings with it, I noticed that the Y-axis (what used to be the paper feed on the printer) will gradually drift off of where it should be after making many back and forth movements. After doing some more tests, the results were very repeatable with the axis shifting about 3/10" after moving 1" back and forth for 70 times. I'm pretty sure the problem lies with the design of the rollers (or just the fact that it has rollers as they rely on friction and have no positive engagement of the paper).

I don't really expect to be able to fix the problem on the hardware side and was wondering if it would be possible to have EMC adjust and compensate for this. After looking over the documentation I didn't really see anything off hand that would apply. I suppose it would require something like having a separate backlash compensation for each direction of the y-axis and I don't know if that is possible.

Thanks for any help,
Justin

[SWPadnos]

A couple of things to check:

1) Set the direction setup and direction hold times very high. Heck, set the step duration and step spacing high too, like 3000-4000 each.

2) Jog a ways in one direction, then, using the smallest jog rate (0.0001 I think), jog the other way, and count the number of keystrokes you need to start the paper moving again. You can get a good estimate of backlash this way. Try it in the other direction too, just for grins.

If the paper is always off in the same direction, and especially if it's off by the same amount, then it probably isn't slip - that should be more random IMO.

There isn't a way to make EMC2 compensate for random offsets, or for problems that happen at reversals (actually there is, but you have to use servo motors and feedback ). It's best to find the problem and fix it, or get used to re-homing often in your process.

- Steve

[justrack]

Thanks for the reply and suggestions. I had tried messing with the timings earlier, even adding an extra zero on the end of all the values recommended for the geckos, but unfortunately the results were the same as before. I had previously set the backlash by estimating the play in the gears and then fine tuned it by drawing small circles until they looked good. I just tried your method see if it was different in each direction but couldn't really tell when the paper started moving as the motor has about 3000 microsteps per inch (after going through the gears).

I suppose the motor or controller could also be missing steps, but i think that would be more random also. I think I'm going to just try to write a simple java program to edit the gcode and manually offset the y values depending on what direction the pen should be moving before opening it in EMC. Although I probably won't be able to do too much too soon seeing as I have Finals coming up in a couple days.