Re: Electronics gurus please help

Hi,

Jon, I agree with you on the 10% ripple thing. The "magic" in all
this is what you do with the decay current.

If you non-recirculate it, i.e. return the current to the supply, the
decay slope will be about the same as the attack slope and a near 50%
duty cycle will hold current constant. Again, for a 24VDC supply, a
2.4 mH coil and a 50 uS switching period will yeild a 250 mA peak to
peak ripple, (24V / 2.4mH * (50uS / 2) = 250mA). This would be just
fine for a 2.5A rated motor.

If you recirculate the current i.e. short the winding during the
decay period, the decay slope will be much shallower. If the motor is
rated at 2.5A and 3V, then the decay slope V/L would be 1,250 amps
per second, or 1/8 of the attack slope. A 1/9 (11%) duty cycle would
maintain constant current. The current ripple would be much smaller
now, on the order of only 55 mA, (24V / 2.4mH * 50uS / 9 = 55mA).

So how come we don't use recirculating topology? It's because it
comes with its own set of problems.

The first is a phenomena called "current tailing". As speed
increases, the slope of the sine-cosine reference increases as well.
At some critical speed, (1/2 rev per sec or so), the current decay
slope cannot keep up with the rate of change of the reference. The
current left in the winding now is greater then the commanded value.
Eventually the piper must be paid and that payment is in the form of
a large current discontinuity at the ordinal step location. This
causes a significant motor "jump" at the full step location.

Allegro deals with this, (foolishly in my opinion) by switching from
recirculating (slow decay) on the up-slope of the reference to non-
recirculating (fast decay) on the down-slope.

The motor responds only to the average current. A 2.5A motor with a
55mA ripple will have an average current of 2.4725A (2.5A - 55mA/2)
just before sin 90 degrees. Switching to non-recirculating just after
sine 90 degrees results in an average current of 2.375A (2.5A -
250mA/2). This 100mA "jump" is equivalent to several microsteps!

The whole idea for microstepping is to evenly distribute the
microsteps evenly across the span of one full step. The measured
distortion of the G201 is just over 1 degree (1/90 of a full step)
and there are plenty of motors now (Sanyo Denki, Vexta, PacSci, etc.)
that have similar linearity. We have opted for accuracy over any
other technical goal.

Mariss

--- In CAD_CAM_EDM_DRO@y..., Jon Elson <elson@p...> wrote:
>
> Yes, at least for the example of the 2.4 mH motor winding. I don't
know if that
> is a reasonable approximation of the inductance of YOUR motors,
though.
>
> Normally, you don't want the Delta-I in the windings to exceed 10%
of the
> average current. So, for a 2 Amp nominal current, you would not
want it
> to drop below 1.8 A during the chopping cycle. There is a constant
energy
> loss due to hysteriesis of the Iron laminations every time you
charge and
> discharge the flux in the magnetic circuit. The part of this due
to stepping the
> motor is unavoidable. But, the part due to the chopping cycle
should be
> minimized to reduce motor heating.
>
> So, if the current did fluctuate more than 10%, you would want to
reduce the
> delays and so increase the chopping frequency. This will reduce the
> variation in current.
>
> Jon