G540 + EMC2 = missing steps

Hi,

I cut 3 circles with 20mm dia and came out 19.54 to 19.62 using 1.5875 dia mm end mill(1/16). I have played with the accel and velocity so many times no luck and also I have checked all the spindles and they look tight.

Thanks

Hi

You don't seem to be answering the questions I am asking, so I should probably be more specific.

Run my altered code, it should run without overcutting, as it did before.

Then remove the G61, still run OK?

Then combine the X Y and Z moves for the first circle into one move as per the original code, does it overcut now?

If it does, try changing the direction of the tangental plunge, by starting the program with
G0 X20 Y20 instead of G0 X0 Y0

Does that stop the overcut?

... and so on through the program to isolate the moves that are the problem.

Regards the circle diameters, they will be as good as the maths used to calculate the the G3 moves (without cutter compensation), plus the amount of spindle run out.
So long as they are round and not oval, I should not worry about that right now

regards

Andy,

Good morning. I took all 3 spindles out and push and pull x ,y and z no slack or movement at all. I even check all the coupling between spindle and the lead screws.

Thanks

Andy,
problem with circles

I finally solved my problem by just adding Lead in and Lead out . ( I called the K2cnc and advised me to add lead in and lead out.)

Thank you very much for all your help.

If adding lead in and lead out fixed your problem. Then it almost certainly proves that your machine has to much flex (not stiff enough) and or back lash. Probably most likely in the Z axis. Lash and loose ways in the Z can be difficult to detect because the weight of the axis masks it until you try to plunge into some material.

Hello Everyone, tagging on to an older thread with a similar issue...

I am running a small 3-axis gantry router, G540 10 microstep driver, LinuxCNC 2.6.something. I am having troubles with the Z axis becoming lower and lower as I run. X and Y are generally fine.

Last weekend I ran a 3D program and a feature that should have been .165" deep from the top of stock ended up more like .230" deep. Program was 22k lines of code and much of that is parallel finishing a convex curved form so X-Y-Z are all moving at the same time. The longer the program the more I notice the issues. And some of my stuff is two-sided so I get hit double with errors.

My stepper motors are NEMA 23: Max Torque 282 oz-in, 3A
Power supply is 12.5A 48VDC

Normally I am cutting wood with 1/8" end mills, 12krpm and 60-75 IPM. I tried some FR4 too and the cutter seemed to pull into the material until it broke.

The motor and coupler on Z seem tight but the machine as a whole is not as rigid as could be desired, Y is the worst and allows Z to flex as well.

Currently X and Y are set to 3 inches per sec max velocity and 30 inches acceleration. Z is 1 and 10 respectively.

If I run the Z axis test alone things seem good but real world performance is not.

I have the G540 Stepconfg set up as X-Y-Z-A. Could I be wasting power supply headroom maintaining the A axis position?

Would slowing down acceleration even more on the axes help? Disable the microstepping?

Or do I need to stiffen the machine first?

Thanks!

this is a common problem , slow down your z axis speeds will help enormously , your simply running at higher rates than the stepper motor can cope with , ideally you dont want a 10 microstep driver on the z .
as a start i'd lower them by 50% , or untill you have them running fine and then increase slowly untill they fail and then drop back about 10%

the faster a stepper goes the less torque it has
( and why you find z axis tend to have a 2:1 reduction pulley incorporated )

running 4 steppers could be dragging the system down , i'd tend to run the Z axis on a seperate supply ,
if the machine is not ridgid , then thats the first port of call to fix, electronics can only do so much , it cant make a machine better than it's frame can support

dont forget a stepper uses more current when holding a stationary position ,

Thanks for the tips Cncbasher!

I reran Stepconfg and turned off the A axis of the G540 and also found one loose linear rod on the Z axis and snugged it a bit.

I ran the same chunk of program, 22800 lines, over again twice changing the velocity and acceleration.

The first time I did not notice a big difference in the run time of the program but the Z 'rundown' error reduced from .070 to .040". I should have wrote down the setting but did not. I want to say I was at X &Y velocity 2 IPS & acceleration 10 IPS^S, Z velocity 1 IPS acceleration 5 IPS^S or half of X and Y.

The second time I dialed down the acceleration even more. I could really hear and see a difference in how things ran. The machine would start, speed up, run, slow, stop. When I had a combination XY or XYZ move I would not reach my programmed speeds of 60 (roughing) or 75 (finishing) IPM, indicated was from 20 to 45 IPM.Once I got to some straight/flat stretches the machine would open up to the programmed 75 IPM. Overall Z Error was reduced dramatically, to somewhere between .004 to .010" (hard for me to measure depth with a caliper on something that small plus I was routing on a pressed fiber style material) which is more than good enough for my projects and close to the 'spec' of the machine. Settings the second time were X and Y velocity 2 IPS & acceleration 2 IPS^S, Z was set at 1 IPS velocity and 1 IPS^S acceleration. All the other settings are the stock ones from the preloaded G540 config. The only downside was the program ran quite a bit slower than before, so longer runs on the router and vacuum to heat up the work space.

It looks like my next step would be be to creep the settings up to gain a little speed without adding too much error.

This is a huge breakthrough, I thought I was doomed to have to reprogram or redesign pieces to get them to rout decently.
Thanks again!