Accuracy troubles

[binfordw]

Hi,

Ive been fighting this problem for about a week now and still havent figured it out. Ive asked on other forums with little response, I'm working with the manufacturer of my cnc kit but am about ready to go nuts here. Im looking for any and all possible advice :confused:


My cnc plasma table is losing zero- i.e, starting at 0,0 on a part, when its finished it can be .5" or more off in both X and Y. This makes it impossible to cut larger, decorative parts.


I've got things narrowed down to the electronics somewhere, The bare stepper motors are all not returning to zero. I put tape on the motors and mounted them securely, and ran a simple program from 0 to 5", over and over. The longer this runs, the farther off from "zero" the motors get. Feedrate and accerleration settings dont seem to have any effect on the amount of error. All 3 motors, X and A, and Y, all seem to have the exact amount of error when checked.

Im using a Bladerunner cnc kit and Mach3 software. Does anyone have any ideas as to what could cause this issue?

Thanks in advance

Ben W

[djc]

I'm using a Bladerunner cnc kit and Mach3 software. Does anyone have any ideas as to what could cause this issue?

Slow the acceleration and velocity way on down, by 80% or so. Try again.

What computer are you using to drive it? If a laptop or a small form factor desktop, it's possible that the parallel port isn't meaty enough (I had this problem on a mini-ITX system I tried). Try a different computer.

There's no possibility of electrical interference from the plasma is there? Try with the torch off and disconnected.

I notice you mention X, Y & A. I think I saw a post on the Mach forum that said that rotary axis work requires a lot of number crunching by Mach. Make sure the computer is well-specified, a fast processor, plenty of RAM and switch off _everything_ in Windows that you don't need (screensaver, power management, networking, disconnect all USB devices, check what is run at start-up).

Can you drive the table with TurboCNC on a DOS machine or EMC if you are smart enough for Linux? If it works OK on these, the problem is the Windows PC end of the parallel cable; if not the problem is the table itself.

[binfordw]

Thanks for the quick response,

I had tried my mill pc, with the same result. I had also slowed the feedrate and acceleration way, way down, with no change. Apparently there is something wrong with the g241 driver board, the manufacturer requested I return it for testing.


One odd thing I noticed while testing the motors, was when running them from 0 to 1", with a mark on the pulley (pulley was tight) for 0 and 1, was that it would go to 1 and stop briefly, but it wouldnt go all the way to 0 before reversing and going back to 1. When i continued to let it run, it would visibly get farther and farther off of the marks. Not going all the way to 0 seemed wrong to me. Watching in Mach3, it wasnt reaching 0 either, it would get down to .200-.300" or so before reversing back to 1. It should go completely to the coord and stop briefly right?

[macona]

I cant imagine it being the drive. Disconnect the motor from the screw and see what happens.

What is the code you are using to move it?

[binfordw]

The motors were removed- there are no screwdrives- its a timing belt/pulley drive. The errors I am having are visible on the bare motorshafts themselves.

I was using a simple x0, x1, m47 program to test for error.