Kirk Wallace wrote:

> I scoped the driver board inputs again and corrected my signal diagram
> on the schematic. The gray traces are the result of the coil input and
> it's inhibit. I confirmed this with scoping the far side of the input
> optocoupler. I played with the axis speed and noticed the inhibit gets
> smaller and goes away around half speed. The trace leads me to believe
> the designers were trying to soften the low speed steps, with a brief
> half step, so that the motor would not overshoot and degrade the surface
> finish. If this is the case and not for some sort of electrical reason,
> I can switch over to EMC2 right away, and not worry about blowing out a
> board. Of course Stepgen doesn't have this inhibit signal available, so
> I may need to find another way, maybe using selective and adjustable
> delays in HAL.
It would be real easy to rig an edge detector to detect when the 
phases changed, and enforce a blanking on that phase for a set 
time.  A one shot could set the blanking time and allow you to 
adjust to the same as the Bandit controller produced.
  If step overshoot is a common stepper problem, is there a
> common way to fix it?
> 

Do you mean jumping steps?  The worst cause of this is mid-band 
resonance, where the (full) step frequency matches the rotor's 
pendulum-like natural frequency due to rotor mass and the 
magnetic pull of the motor's poles.  The Gecko drive detects an 
electrica phase shift in the motor's waveform and counteracts 
it.  Otherwise, you can have the controller skip over the 
resonant frequency, install viscous dampers, or try to come up 
with other electronic damping solutions (RC dampers, etc.)
None of these schemes is perfect,  I think the latter RC dampers 
are almost certainly a disaster due to component size.  The 
Gecko also greatly solves the problem with microstepping, as it 
reduces the excess energy supplied at the full-step frequency so 
as not to excite the resonance.

Jon

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to