[Bdi4emc-help] Max speed

[Paul]

Hi Jim

On Friday 03 March 2006 15:50, Jim Fong wrote:
> I'm currently running BDI 4.38 on a Pentium2-400Mhz PC.  I recently
> installed a new 276oz-in stepper motor on my X-axis and trying to tune EMC
> for higher rapid speed.  The fastest reliable speed I can get is 25IPM
> (500RPM, 20TPI leadscrew, 13333 pulse/second) with my 8microstep driver.
> Anymore than 30IPM, you can hear the stepper motor getting rough sounding
> (jitter??) and motor stops spinning.  The period is set at 0.000007.  I
> have tested it set at 0.000005 (any lower PC locks up and resets) but I
> still can't get any faster or more reliable pulse out of the parallel port.
>  I played around with the P setting but just makes it worse if set over
> 1000.
>
> I currently set the MAX_VELOCITY = .41 (25IPM) since that seems to be the
> most reliable speed I can get for now.
>
> Anyone out there running a similiar speed PC and what is your fastest pulse
> output? 13,000 steps/sec seems rather slow to me.


Dug out my little Sherline and ran a few tests this morning. The computer is a 
Celeron 433 with a 440BX chipset, and a C&T 69000 graphics controller (2MB 
memory running at 800x600), currently with 256M of memory installed. Probably 
close enough in specifications to do a reasonable comparison of EMC's 
performance..

The stepper motors on the Sherline are fairly wimpy little NEMA23 rated at 
134oz-in running from a unipolar driver - Not the best combination for 
maximum performance.. About the best I can achieve is 25ipm with a 7.5uSec 
base period before the motors start squawking and I loose steps (some 
6.7KHz). Reducing period to 15uSec, and the maximum speed increases to 30ipm.
[Sherline uses 16000 steps per inch when using their driver box].

Hooking up a scope, I see a pulse train right up to 240ipm or 64KHz with a 
7.5uSec period. There is a fair bit of jitter which could explain the lower 
max speed. At 15uSec, the pulse train is much smoother with little noticable 
jitter, but maximum frequency is limited to 33.3KHz.

I would suggest hooking up a scope if you can get your hands on one, 
quantifying the jitter and selecting a period to suit - I can run at 7.5uSec, 
but jitter is a problem, the same at 10uSec. 15uSec produces the better 
results when using quickstep, but your milage may vary.


Regards, Paul.



-- 
From the Klingon book of C:
Klingon function calls do not have 'parameters' - they have 'arguments' - and
they ALWAYS WIN THEM.