On Wed, 07 Mar 2012 02:11:08 -0800, Robbie Hatley
wrote:

Yabbut, Phil Allison left a lot out.

These are called variable frequency drives. They produce a PWM output
that induces a frequency and voltage controlled sinusoidal current in the
motor. The typical voltage limit is related to motor rated voltage and
for 480 V system rated (460 V nameplate) is about 8 V per Hz at any
frequency that produces less than full rated voltage. If you use properly
burdened current transformers to measure phase current you will see a much
more informative display on your 'scope.

That said, 400 Hz switching frequency is uncommonly low and would take
excessive load reactors to smooth out. Modern units are a few to several
KHz PWM base. IGBTs tend to be rather slow, but they are getting better.
Of course they pretty well better, there are super power fets with ratings
that would make early IGBTs shudder.


Someone named Phil Allison apparently replied to my message about
variable-speed motor drives, but I never got that message direct.
(Lost in the ether of Usenet?) I did get someone named "Jamie"
replying to it, though, so I can reply. (Lemme strip off one
layer of "". Done.)

Please do NOT strip the "". They help others keep track of who said
what.

Phil Allison writes:

The technique is called PWM

While it's true that the pulse widths change, this is kinda
far removed from what I normally think of as being PWM.
I generally think of PWM as feeding variable-duty-cycle DC pulses
into an LC filter so as to get a DC voltage proportional to
duty cycle.

http://en.wikipedia.org/wiki/Pulse-width_modulation

Ok, lemme read and see if there's a variant of PWM that acts
like what I'm seeing in these motor drives.

::: read, read, read :::

Yes, you're right. Thanks for the link! First page has a
picture that shows why I'm seeing 400Hz where I expected 2Hz:

http://upload.wikimedia.org/wikipedi...2C_3-level.svg

Basically uses the inductance of the motor windings and the magnetic
reluctance of the core to convert the 400Hz electric pulses into
2Hz or 20Hz or 38Hz (or whatever) magnetic sine waves.

I also see a link on W'pedia's PWM page to the particular type of
PWM used for 3-phase variable-speed AC motors:

http://en.wikipedia.org/wiki/Space_vector_modulation

Very cool chart there showing 8 different valid states that
a 6-IGBT bridge can be in (and hinting at the various catastrophic
*unallowable* states as well -- I've experienced some of *those*).

Turns out, 6 of the 8 states line up with the axes of the 3 phases,
and the remaining 2 (000 and 111) correspond to "off" states.

The duty cycle of the 400Hz rectangular wave is slowly modulated
to create low frequencies in the 3 phases required.

Apparently so.

A steady 50% duty cycle wave at 400Hz creates no torque in the motor
as the average value is zero

Yep, if all 3 phases are in-phase, speed will be 0RPM and current
will be very low (10ma? 1ma?) even if the voltages are 320V p-p,
because the voltage across each winding is near 0.

and the motor's inductance at thatfrequency causes little current
to flow.

I think any current at 0RPM would have to be due to parasitic
capacitance. With 0V across each coil at all times, nowhere else
for current to flow to.

Shallow modulation depth creates the low voltages needed for
slow running.

I think it's the brief periods of time that one phase is "Hi" while
another phase has switched "Lo" that creates the pulses of differential
voltage necessary to ramp up magnetic flux. If these "overlap zones"
are then rotated around the 3 phases at 10 rotations per second
clockwise, the motor should spin clockwise at 600RPM.

I think the overlap durations would need to be proportional to
actual power draw, in order to keep the speed from bogging down
under load. Perhaps that's why these devices all have current
sensors: not so much to protect from overcurrent (though there
is that), but mainly to calculate phase overlap durations.


Anyway, Phil, thanks for the link; it answers my questions and
then some.


PS: I'll also reply briefly here to the other 2 responders:

Jamie: Actually, Phil's right.

Sylvia: Yep, the physics described in W'pedia's PWM and SVM pages
does match what I'm seeing on the bench.