EMC2 loosing position

hi all,

I'm using this machine to cut out a mold for a new wing section. I'm using
CUT3D to generate the Gcode and I have noticed a lot of Z axiz direction changes instead of being smooth. The end result is that eventually I will loose a step or two several times during the cutting of the finish path.

Lets see:
I changed the HobbyCNC board to a commercial Pacsci driver 6410- same problem
I changed the stepper motor - Same problem
I changed the computer and installed the latest version - same problem
I changed the Accel rate for the Z axis (it is now 3"/s^2 Very Slow) - same problem
I ran the stepper motor all by itself - same problem
I even took the axis apart and rebuilt it - same problem

If I run any test program I can't get a path tough enough to cause it, however any and all parts that have any level of detail are destroyed.

I've run the stepper configuration so many times that numbers are running together in my head

I'm running out of things to try. Please help!

John Lindsay

Hi John

Did you try in .hal to delay the wait time to change direction...

setp stepgen. n .dirhold xxxxxxxx
setp stepgen.n.dirsetup xxxxxxxx

making your xxxxxxxx values bigger?


Mike

Hello,

Is your program running in exact stop mode? If so look at G64 Continuous mode with optional path tolerance.

linuxcnc.org/docs/html/gcode_main.html#sub:G61,-G61.1,-G64:

Also you may want to post the program that is causing the problem.

Which post processor did you use?

Rick G

by what you describe it's probably more a missing steps problem than the gcode , or misstuning
can you give details of your stepper drive and stepper motor used and at what voltage and amps you are running the z axis at

remember the zaxis is working against gravity , as well as the work load etc , so it needs to be more capable of some of the old ' grunt '
to be able to lift etc . if your using microstepping drives try dropping the multisteps by half , steppers give more torque at lower speeds than running fast

Dave

By your post I realize you have tried many adjustments.

But to clarify ...
Have you tried running the program that losses steps at a slower speed?
Reducing EMC max velocity and the max velocity of the z axis?

Rick G

Whew,

to answer all of your questions and responses,

Yes I've tried all that!

g64 p0.01
g64 p0.05 looks smoother but still looses steps

had 1/4 step, went back to 1/2 step with same results and lots more noise.

I have successfully produced a carved mantle previously using this system.

my current project is a very simple wing, without radical z motions.

I'm using the HobbyCNC controllers with 250 oz-in steppers I also tried a pacific scientific drive with a vexta 120 oz-in with exactly the same result

I have run it at 40 i/m 20 i/m and 4 i/m the accel rates from 10 all the way down to 3.

I'm using vectrix software Cam3D to generate the Gcode

Now for the kicker. It used to work using this exact hardware. I even ran the same code that had worked previously. I think I upgraded from version 8 to version 10 of EMC but there was a bit of time between when it worked and now, so I don't know when the problem showed up if I had upgraded the software.

I do know that putting the system on a brand new computer didn't solve the problem.


I put some paint on the coupling so I could see if the position had changed, and removed the router from the system. very lite, very easy to move ...... same problem!

any other ideas PLEASE!!!!! I'm going nuts

John

it performed that test successfully

John, I also rebuilt the entire Z axis, and then I marked the coupling mounted on an isolated stepper. NO LOAD, and it still lost position.