Re: Newbie Q, tuning for max speed

[Jon Elson <elson-at-pico-systems.com>]

Damon Gentile wrote:

>Hey folks, a couple of newbie questions here. Hopefully nothing too
>dumb, I've been struggling with these for a little while now.
>
>BDI 2.18 & PII 333MHz 64MB ram, running EMC_MINI with freqmod.o
>with four NEMA23 116oz/in steppers. Only 12V for the moment.
>
>I've tried a bunch of combinations thus far and think I'm doing
>pretty good.
>
>When I was running steppermod, I could get to about 42IPM (motor
>sitting on the bench) before the motor would lose a step. Follow
>errors were rare. I've since upgraded (?) to freqmod.  I've turned
>down CYCLE_TIME to 0.004 and PERIOD to 0.00002.
>What neighborhood would I want to be in for cycletime & period ?
>but now I get follow errors at >35"/min. (about my comfortable
>speed w/ steppermod). The follow error will trip before the
>motor skips (within the resolution of my eye, at least) so this
>looks a little better, not getting lost.   
>
>What INI parameters are the biggest knobs to tune speed ?
>
>How about FERROR & MIN_FERROR ?  P/I/D/FFxxx ? Still having some
>trouble getting stripchart up and running.
>  
>
First, freqmod should never give sollowing errors.  this means that the 
step output is not
following the commanded position.  You can crudely fix it by increasing 
FERROR and
MIN_FERROR, but this just allows larger error between actual position 
(based on
steps output) and the desired position based on the toolpath. 
 Increasing P, and possibly
FF1 should help.  I think you can also increase the MIN and MAX_OUTPUT.

Depending on your step resolution and desired max_velocity, you may need 
to cut PERIOD
to allow more steps/second.  I think you can go down a bit, but you are 
already at 20 uS,
I don't know how much more you can get out of a 333 MHz Pentium.  20 uS 
allows
a step pulse to be output every 40 uS, which is 25 KHz.  But, this means 
the last speed
jump is from a step pulse every third PERIOD tick,  or 16.67 KHz, to 
every other at
25 KHz.  That's a jump of 33%, no motor could follow that.  Your 
acceleration of
5.0 may also be too high.  I usually use 2.0 on my systems.

Jon

>To judge lost steps, I have it make a 200" move in one direction and
>see if the shaft ends up in the same rotation that it started at.
>
>Thanks,
>-Damon
>          
>[TASK]
>CYCLE_TIME = 0.004
>
>[EMCMOT]
>EMCMOT = freqmod.o
>COMM_WAIT = 0.010
>PERIOD = 0.00002
>
>[TRAJ]
>CYCLE_TIME = 0.010
>DEFAULT_VELOCITY = 0.0167
>MAX_VELOCITY = 1
>DEFAULT_ACCELERATION = 5.0
>MAX_ACCELERATION = 5.0
>
>[AXIS_0]
>P=1000
>I=0
>D=0   
>FF0=0
>FF1=0
>FF2=0
>FERROR = 1.000
>MIN_FERROR = 0.010
>
>
>  
>