Problem- weird cutting (Missed steps?????)

[MonoNeuron]

Pentium 4 3GigHz, 1 Gig RAM, 120meg HDrive,
Mach3
2 x Vexta 5 Phase UDX 5114 drives (X and Y axis)
1 Vexta CSD 5814 (Zaxis)
Y = 0.750" wide 0.200 pitch belt drive geared down 3:1
X = 0.750" 12 start acme thread with antibacklash polyacetal nut
Z = Thompson linear guide with ball screw

hey all.

I have finished building my 3rd router and I am having problems with the axis moving during a cut. If I cut a square then it comes out like a parallelogram but with the Y axis sides straight as they should be. I wasn't sure if I was losing steps because the cut length is the same, unless I lose steps right at the end of the cut length. Each line of cuts starts a little behind the previous one so it seems I am losing steps.

My last job finished about 80mm away from the zero point on the Y axis when homed after finishing and about 4mm off on the X and 1mm on the Z.
This is repeatable and was exactly the same after doing the job 5 times after zeroing.
I thought it was mechanical but everything is tight with no loose pulleys or belts and I double checked this as it seems to point to a mechanical fault.

I tried the following to see if it made any difference.

I separated the main power wires for the router from the motor wires and ran it overhead instead of next to all the stepper motor wires.
I put Ferrite cores over the motor wires and power wires.
I rewired the motor wires with twisted pair wires and grounded the shield on one end only.
I rewired the step and dir wires with shielded cable and grounded the shield.
I rerouted all the wires going to the motors.
I deselected the limit switches in Mach3 (I was getting a reset that I thought was being induced into the Home and Limit micro switches before I rewired with shielded cables).
I swapped the 2 UDX5114 drives for X and Y.
I set the lookahead in mach3 to 500
I connected to the internal parallel port 0x378 instead of the 2nd parallel port card.
I used another parallel port cable from the plasma cutter that I know works ok.
I cut at 50% of the cut rate = 50 ipm. (default 100 ipm)
I cut at 120% of the cut rate = 120 ipm

I still get the same results everytime.
It seems mechanical but I have ruled that out.
The first job I cut on the router was with the majority of the cutting done on the cross X axis with the step over on the long Y axis and it turned out ok. This new job was with the majority of the cutting on the long Y axis with the X axis used for the step over. I am trying to carve some designs on a table leg. i.e long and skinny.

Do you think it could be a software problem? I am thinking of using CNCpro but I will have to install Win98 for the DOS command line interface because I don't know if I can do that with WinXP Pro.
I hope I have explained the problem good enough for you to get a grip on what's happening.

Any suggestions?
ANY help will be soaked up with much enthusiasm.
Thanks in advance.
Richard

[CarveOne]

Look in Start/All Programs/Accessories/Command Prompt for a DOS window in Win XP Pro. Don't know if you can run CNCPro from there though. Worth a try.

There's no reason that Mach3 won't work properly when the motor tuning parameters are set up correctly. I would suspect binding in the mechanics somewhere. The direction that the parallelogram is skewed should help figure which axis to suspect.

CarveOne

[LaserImage]

I had almost the exact problem with my machine and I went through hours and hours of troubleshooting, to no avail. Finally, I happened to notice the kernel speed was 100 instead of the default 25. As soon as I set it back to 25 it worked fine.

Gary

[MonoNeuron]

Thanks guys. Any help is a big help.
I seem to remember that CNCpro won't run in a DOS window. I will try and make an install disk and try to run DOS through that setup.
I will check the kernel speed today and get back to you.
Cheers.
Rich

[MonoNeuron]

Well,
It only took 3 days and a complete rewire a few times but I found the problem.
I reinstalled Mach3 with the latest lockdown version and still had exactly the same problem with lost steps at the end of each cut causing the tapered ends instead of them being square. The cut followed exactly the old one every time.

I thought it must be a problem with the step and dir pulse stream so I went-a-looking for something awry between the parallel port and the drives again.
As it turned out the latest C10 breakout board I am using has 2 jumpers on board whereas the old one only had 1. ( I didn't realize this when connecting it up )

The jumper that is selectable for the +5 volt and Ground had the jumper attached to only 1 pin and wasn't shorting out one of the settings. I placed the jumper over the 2 pins for ground and the problem went away........finally.

Seems that the step signals were close to the threshold level and would lose steps when changing over from the direction pulse.

Thanks for the help and taking the time to look at my problem.
This one will go into the memory bank along with the other hundred problems I have encountered.
Richard.