Axis Scale not right. NEED HELP

I recently built a new control for my cnc router. The router has been working for over a year, but I decided to rebuild the control with better drivers. The current configuration is as follows

KL-5056 drivers
C-10 BOB
40v 10amp power supply
X axis -Green monster 425ozin motor from probotix.com- set to 16 microstepping at 4.3 amps- 25 pitch roller chain
Y axis - Green monster 425ozin motor from probotix.com- set to 16 microstepping at 4.3 amps- 1/2-10 5 start acme screw
Z axis 282 ozin motor from Keling inc also set to 16 microstepping at 4.3 amps 1/2-10 single start acme

I also changed computers when I set this controller up. The new computer is an AMD 2.2ghz single core with 2 gig of ram The latency test came back around 44000us if I remember right.

Now onto my problems All the parts that are coming off the machine are undersize. Tonight I cut some 3" circles that were coming out almost .020" under. My straight line travel is also about that much under. Im running about .012 of backlash comp on my X axis to compensate for the chain, which I also ran in my old configuration with great results.

Im at a loss on how to handle this now and really just want a working machine again. Do you guys have any ideas on what could be causing this?

Thanks for your time,
Joe

at a quick guess your loosing steps , 16 x microstepping will loose a lot of torque , take it down to 1/2 or 1/4 and you should see an improvement
re check your backlash and see if you can get rid of it , or reduce it , such as making anti backlash nuts if you dont use them .

check movement for any tight spots too . or slack in the chains

is the undersize consistant over parts cut or varying

any backlash is bad , so anything mechanicly you can do to reduce slack is good , but not to the point of any axis binding

Anti-backlast on all lead nuts. And you could bounce a penny off my chains

I just tried 1/4 stepping and got the same results

With simple shapes, squares and circles I am getting excellent accurate results, But with complex shapes, like a roller sprocket, the entire diameter is .020 small and the gap between the teeth is around the same. Im at a loss still.

Ok first, My latency was 24800 not 44000 like I said before. Im not sure where I got 44000 from sorry guys.

John,
With my latency score, I entered it into step config at the beginning, is there something more I need to do?
Im getting deeper into the config files and learning to mod them so this stuff is great to know

Backlash- I increase the STEPGEN_MAXACCEL to 200 with the MAX_ACCELERATION being 100.

Also, is there documenation on a step test and how to go about it? I did some things like measure 6" move 6" and check, cut a 3" square and measure.



Now I think I found one of the culprits but Im having a different problem now. Twice now while running long programs (5+ minutes) for testing Ive had the computer completely quit responding. The first time the program kept running and finished out but the screen was frozen and the machine did not respond to any keystrokes, my dedicated E-Stop wouldnt even function. The second time the screen went black, mouse arrow was still visable but not responding, again no keystroke or e-stop response. Both times I had to hard reset the pc.

Could this be a latency issue? Should I be on the look out for a different pc?

BTW The issue I found was V-carve pro producing bad code. I was cutting a roller chain sprocket so it was very difficult to check by hand, but I posted it from a v-carve install I have running in wine (from the same .crv file even) and got very different results. This one produced a great sprocket. I can see some marks where the backlash comp activated I think.

Anyway Id love input on the PC issue

yes v carve pro is known to produce bad code from some versons .

glad you managed to find a culprit ,
it's always a good idea not to have the Estop circuit in software , always hard wire it to cut power to your drives . and use the software just to monitor the Estop and as a secondary control if needed

Hi

Ref the software side as you requested, I think you need to de-tune and then adjust until you hit problems.

My latency was 24800 not 44000 like I said before.

What is your BASE_PERIOD set to in your .ini file? Try setting it to at least 75000 to start with.

set to 16 microstepping

As cncbasher said, try this down at 4 x.
Microsteps hugely effect the number of pulses required to move a given distance, which in itself causes lost steps if it exceeds the ability of your computer to generate then.
Even worse higher microsteps don't exist in reality, they are interpolated positions between physical pole alignments in the motor.
They are as good as the stepper driver. I never go beyond 4x if possible.

Backlash- I increase the STEPGEN_MAXACCEL to 200 with the MAX_ACCELERATION being 100.

From this I have to guess that these figures were at least 100 and 50 previously?
As you are from the States also assume that the figures are in inches? That is huge for a nema 23 powered stepper machine.

Too high acceleration is another big cause of lost steps, especially where an axis changes direction 180 degrees and there is a sudden ramp down and ramp up.

Take them right down to 10 and 5 respectively.
Then set MAX_VELOCITY down to something which equates to 3 - 6 feet per minute

Hopefully your machine will run smoothly albeit slower than it did and you can then experiment with the figures to get optimum tuning.

Faster is not always better, never missing a beat and always producing an accurate work-piece is best.

regards

Could this be a latency issue? Should I be on the look out for a different pc?

It might be a good idea to run the latency test again for an extended period of time while using the computer for some heavy tasks.

You could also try a slower base thread and see if the problem still exists. The base thread only needs to be fast enough to run your fastest moves.

Rick G