trying to use emc2 for 2x3 router control

[mediaconvert]

Is anyone willing to share their emc2 setup parameters for a system
similar to the plans published by Fine Line Automation and CNC Router
parts for their 2x3 router?
see Home : Fine Line Automation
I'm using their electronics package, a G540 stepper and their 380 ozin
motors. the details of my build are on my web pages.
see CNC Machining
I thought I had things working well but then I tried to drill the top
plate for a vacuum hold down and I realized I was loosing steps.
I've been running the diagnostic Gcode from the emc2 manual and
no matter how slow I go I'm still loosing steps...

Frank Pirz
Media Conversions

[chester88]

did you use stepconf to build the config?
What timing and acceleration settings did you use?
can you post your HAL and INI files?
are you sure you are losing steps or is it just not scaled right?
Are you losing steps in all axis or just one?
Is there any tightness mechanically?

[mediaconvert]

Chester88. Lots of questions...
Yes I used Stepconf. I'm running emc2 v2.4.6 on Ubuntu 10.04.
Jitter tests give me latency of around 20,000. I'm using 25,000 to be
conservative. that gives me a max step rate of 30,303 hz.
[I've also tried 30,000]. the stepper motors are 200 step/rev.
the feed screws are 2TPI. the gecko g540 is fixed at 10 microsteps.
translates into 4000 step/inch. theoretical speeds are 7.7 ips or 454 ipm.
reality matches in that objects cut come out symmetrical, square (or
round), and the correct size. but cuts are restricted to feed rates
limited by spindle speed, cutter size and material. I've used cutting
feed rates of 10 to 20 ipm. in other words, slow.

what i'm trying to figure out is what the acceleration and velocity
numbers should be. I keep hearing that systems like this should be
able to do 150 ipm rapids. so far I can't go above 20 ipm without
loosing my way. and yes, it occurs on all axis tested so far (z and x).

thanks for the suggestion about binding. I tore down the X axis today
but didn't find any problems. same things showed up before and after
the tear down.

for testing, I'm using an edited version of the gcode that's in the
emc2 user's manual, section 23.3.1 testing step timing.
basically, I start at x=5.5, pause to setup/read a dial indicator.
then move between x=5 and x=3.5 [change the numbers to suit]
say 20 times. then go back to x=5.5 to see if I get the same reading.
first test runs with a .1 sec pause between g00 moves. second
part of test runs more iterations (100) but without the pause.
results are that i'm off anywhere between 1 to 5 thousands.
[at 4000 steps/inch, 1 step is .25 thousands, so I'm missing between
4 and 20 steps]

if you still want to look, the stepconf, hal and ini files are in a zip
file on my website at
emc2
and attached here.

thanks
Frank Pirz
Media Conversions

[sansbury]

Things to try:

1. Crank each axis back and forth over its full travel by hand, just to make sure nothing is binding anywhere (esp. on a new machine)

2. Cut your acceleration values in half and run each axis back and forth at various speeds and see what happens. Try both G1 and G0 to make sure your max isn't set too high.

3. Disconnect one motor and do #2 again to see if your power supply is insufficient.

4. Double-check the current settings to make sure motors are being fully-powered.

Basically, what you need to do is simplify so you can isolate the problem. You should be able to send all three axes from one point to another and back at rapid speed without any problems. If you can't do that then start with one axis and work out from there.

[chester88]

your stepper timing numbers are horribly wrong.
step time should be around 1000 not 1
step space should be around 2000 not 0
dir hold should be around 200 not 30200
dir setup around 200 not 30200

Actually these are numbers that you can automatically select in stepconf.
These are minimum timing numbers they may need to be larger.
Try fixing these numbers.

Chris M

[mediaconvert]

Chester88
I'm selecting the G540 option when I run stepconf.
the values for variables you mentioned are
step time - 1000
step space - 2000
dir hold - 200
dir setup - 200
but they are grey'ed out so I can't change them anyway.

what else should I do - it seems like the defaults are
what they should be...

Frank Pirz
Media Conversions

[chester88]

hmm must be something going on.
in stepconf manually add the right timings or manually edit them in the HAL file.
If you manually edit them in the HAL next time you use stepconf they will be overwritten again.

look for lines like this for each axis:
setp stepgen.0.steplen 1
setp stepgen.0.stepspace 0
setp stepgen.0.dirhold 15200
setp stepgen.0.dirsetup 15200

This is a bug in stepconf.

[chester88]

I may have spoke too soon .... I forgot about the double stepping feature of stepconf.

Try the manual settings in stepconf and see if there is a difference.

[chester88]

Yes sorry I was on wrong track. After looking at the source code the timing is as it should be.

What voltage power supply are you using?

[mediaconvert]

The problem I was originally talking about in this thread has been
traced to the G540 current set resistor location.
Not documented by Gecko, that resistor needs to be at the G540
end of the motor cable. FLA, who supplied my electronics, was putting
it at the motor end, and using a 9pin 'extension' cable to connect
to the G540. The result was that motors got too much current and
after about 1/2 hr of use started to loose steps. The resistor may
also need a 100 nf bypass capacitor. Gecko has not confirmed that
but they did say that future versions of the G540 will have the bypass
cap built in.

I'm now running reliably at with vmax settings of 5 ips (300 ipm)
and acceleration settings of 15 ip(s*s) using EMC2.

Frank Pirz
Media-Conversions
R2D2