AXIS 1.4a0 & EMC2 2.0.4 (stock iso + all updates)

This story takes a while to set up...

While cutting out a cam that involves lots of fiddly 
motions, my (mostly stock) Sherline mill gradually forgot 
its origin position and would have, had I not popped the 
Esc key, gnawed its way right through the fixture.

I set up another blank and, dang, the same sort of thing 
happened. A bit different positions, but equally bad.

So I took everything apart, cleaned & lubed & tweaked, made 
a great improvement in the overall mechanical goodness, 
chucked up another blank, and -blap- same story. I hate it 
when that happens.

The failures are similar, but not identical: the XY position 
gradually drifts in off into the +X/-Y quadrant and the Z 
position drifts downward. The final G0 move away from the 
(ruined) blank should be vertically upward, but typically 
goes vertically downward and thunks the fixture.

The AXIS preview looks fine, so the g-code is correct. 
Admittedly, it took a while to get there, but it's OK.

After fiddling with numerous dead ends, I reinstalled the 
whole Ubuntu/EMC thing from scratch. Starting from the 
stepper_xyza base, I set the velocity & acceleration values 
to drive the mill at well under its ratings: 0.25 ips max 
velocity, 1 in/sec^2 max acceleration. This is achingly 
slow, but it eliminates motor dynamics from the problem: I 
can -watch- the motors ramp up to their peak nose-picking 
speed!

This is on a Dell Dimension 4550, 2.4 GHz P4, 1 GB RAM, so I 
set the base period = 25 us. I then added steplen / 
stepspace / dirsetup / dirhold statements to the HAL file 
to put 4 ticks between each edge going to the controller 
box. That eliminates noise and pulse speed from the 
problem.

With a minimum pulse length of 4 ticks, the maximum motor 
step rate should be 8 ticks = 200 us. That's well under 
actual rate required by the max velocity, so it's not 
producing following errors even with the slow pulses.

Now, under manual control with the "jog speed" dialed back 
to 1 ipm, I have a solid failure. Tapping PgUp and PgDn (or 
the other axes, as well) generally does what I'd expect. 
However, -sometimes- the motors turn (slowly!) the wrong 
way; start correctly, jerk backwards, then run backwards; 
run correctly for a short time and then stall; and so 
forth.

I attached a digital 'scope to the Z axis step & direction 
signals. They're clean digital pulses, good amplitudes, no 
noise worth mentioning. The step signal shows the expected 
ramping-up and ramping-down frequencies at about the right 
rates.

However, the direction signal misbehaves!

Erratically, occasionally, but reproduceably (if I'm 
patient), the direction signal changes while the step 
signal is pulsing. For example, it will change from 1 to 0 
to 1 (high - low - high) while I'm holding down the PgUp 
key (not calling for a direction change). The other axis 
behave similarly.

The changes seem to be in the tens-to-hundreds of 
millisecond range and generally occur at the leading or 
trailing ends of the motion, not in the middle, but that 
may just be my observations so far.

While the direction signal "glitches", the step pulses 
continue to ramp up or down, so it seems as though 
something is losing track of which the motion direction for 
a while . The resulting transients whack the motors pretty 
hard; they're obviously not carefully planned direction 
changes!

Note that the parallel port is disconnected from the driver 
box, the spindle motor is off, and this is pure software 
from the PC. Nothing mechanical at all!

This is the most complex part I've made, so the mill was 
running longer and doing more motions than ever before, 
which would cause the misbehavior to add up. However, I 
-think- it's the first part I've done since the most recent 
EMC updates, although I'm ashamed to admit I haven't been 
logging what I've been doing against those changes.

It -definitely- didn't misbehave like this back in the BDI 
days. I -think- the first few EMC2 versions were OK, but I 
can't be sure because it's intermittent enough that I might 
have not noticed slight errors on the simpler parts. It's 
not like I'm doing really high-precision stuff here...

I did manage to make one good cam before the latest updates, 
so it -did- work correctly a week or so ago.

Anyhow, I think I've eliminated all the things that could 
cause the problem outside of the EMC2 code itself, but if 
there's any other information I can provide or debugging I 
can do, I'm willing, because it flat out doesn't work for 
me right now!

Thanks...

-- 
Ed

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Emc-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to