Re: stall detection without encoder?

Alan,

Stall detect on a step motor consists of seperating the "back-EMF"
current component from the motor winding current.

Without this component, the winding current is entirely inductive,
meaning it is a triangle wave at medium and higher speeds.

The problem is the amplitude of the inductive component increases
while the the "back EMF" component decreases with decreasing speed.

The stall detect method I investigated was to use two identically
driven bridges; one driving an operating motor while the other one
drove the same type motor except its shaft was super-glued so it
couldnt turn.

A dual-channel 'scope probed both bridge's current sense resistors
differentially, (same V/div both channels, channel "B" inverted,
scope set to "ADD"). This cancelled the inductive component, leaving
only the "back EMF" if the operating motor was turning.

The phase angle of the "back EMF" component relative to the step
pulse gave an accurate "pecentage of stall torque" being delivered by
the motor. This would have been very useful as feedback to the step
pulse source to slow down if the motor load were to approach stall.
Not only would you have stall detect but stall prevention as well!

A practical circuit was designed where the cancelling signal was
derived from a current source driven capacitor ( 1/C*di/dt = L*dv/dt).

I ultimately abandoned this enquiry because below a cetain speed the
signal simply became too small and was buried in the noise. Put
another way, "back EMF" goes to zero at zero speed, so motor load
cannot be detected for a stopped motor.

I suppose missed steps could be detected due to the generated pulse
of "back EMF" but I didn't feel this was enough to justify the
additional circuit complexity.

Mariss

--- In CAD_CAM_EDM_DRO@y..., Alan Marconett KM6VV <KM6VV@a...> wrote:
> Hi Peter,
>
> Interesting idea. I don't think it's new. I believe Mariss and I
have
> discussed monitoring V/A's of the stepper motors, for this or
similar
> purposes. The Software would have to read A/D's, and look for the
> signature of a stall in the data. I believe Mariss' hardware might
> sense? Not sure. If we knew what to look for... "missed" step
> detection would be valuable!
>
> Alan KM6VV