RE: Ini file

["Tim Goldstein" <timg-at-ktmarketing.com>]

Freqmod.o is a step and direction output just like steppermod.o

The problem with steppermod.o is that the way the pulses are formed
causes them to vary in spacing (frequency) which leads to resonance and
stalling problems. Freqmod.o was created as a way around this problem by
creating a programmed pulse generator task in the realtime side. You can
speed up the rate of this task by decreasing the period value. The way
the pulse is uses is that every nth pulse triggers a step with the n
value increasing for slower motion rates and decreasing for faster
motion rates. The pulse train is much more evenly paced than with
steppermod.o The problem comes in when you are pushing to very fast
motion rates. The faster you go the greater the % jump in speed between
the successive available frequencies. In a worst case imagine you are
stepping every other pulse and when you command to go faster the next
available option is every pulse which is a 100% speed increase. To get
around this problem you crank the task up as fast as your computer will
let you and still have some response to the mouse.

Where steppermod.o has become a great solution is with the step and
direction type servo drives and with the Gecko G210 steppers that
internally generate a pulse train.

Tim
[Denver, CO]


> -----Original Message-----
> 
> On Wednesday 27 March 2002 01:38 am, Tim Goldstein wrote:
> > A 133 is pretty anemic for running steppers with EMC. To 
> get a smooth 
> > output you need to use freqmod.o Unfortunately you need at 
> least a 350 
> > or better to get the task set fast enough to get good rapid speeds. 
> > With a 133 you will be much better off running steppermod.o 
> and accept 
> > that you will have premature stalling as the ragged pulse 
> train causes 
> > the motors to instantaniously accelerate at some point.
> 
> Tell me more about this ragged pulse train problem.  I use 
> steppermod.o and have been running into the following situation
> 
> I have the max velocity set to 6.0 for each axis.  
> (determined empirically
> by trial and error)   However, for moves that involve more 
> than one axis
> the velocities are some fractional amount on each axis to maintain a
> given feed rate.   What I find is that for certain angles I 
> will loose phase
> on one of the motors.   I had assumed this was caused by stepper 
> resonance.   My work-around was to write a program to filter 
> the G-code
> by adding appropriate feed rate commands where required.
> 
> Does freqmod interface to the stepper motors the same way 
> (step & direction)? Or does it control the motors via each 
> phase and requires a different 
> interface/controller?   
> 
> 
> BTW, I am using a 450 mhz machine.
> 
> 
>