On Thu, 02 Sep 2004 04:04:47 -0400, Gary Coffman
wrote:

On Wed, 01 Sep 2004 11:19:05 -0500, Don Foreman wrote:
I like your explanation of induced EMF in the load's wild leg
varying with load motor slip speed, Gary. That clearly and concisely
shows how the load motor draws additional third-phase power under
load.

Thanks. If you look a bit closer, you can see that the back EMF
determines the load current drawn by the windings of any motor,
not just the wild leg of a 3 ph motor running off a converter.

Yup.

I'm sure not inclined to dispute Fitch's observations from data. .
I will, however, try to understand what that data tells us.

He may have found that *additional* flywheel makes no observable
difference. I'd bet that he did his experiment with an idler at
least 1.5 times as large (HP) as the load motor. I rather doubt that
he built an induction motor with a rotor of negligable mass.

No, he didn't, but he did put a *heavy* flywheel on it, and noted no
improvement. In fact, the converter can't respond as quickly to changes
in load with a heavy flywheel attached. (And we want it to, since the
majority of the energy in, and all of it that is passing through, the
system is electrical, not mechanical.)

OK with the first sentence. Did he observe and report what you
assert in the second sentence? You note that he noted no
improvement. Did he note any degradation?

The mere fact that an induction motor draws many times rated load
current to reach speed in a period long compared to 1 cycle suggests
that the kinetic energy in the rotor is considerably greater than the
integral of rated line power over one cycle or fraction of a cycle.

True, but the lighter the rotor, the less power needed to change its
speed (and hence its slip). Since that slip is mechanically coupled for
all bars of the squirrel cage, more power can be immediately drawn
from the primary feed if the rotor mass is low. Unless we can draw
this power, we can't pass it along to the load.

In electrical terms, a lower mass rotor lowers the source impedance
of the RPC, and a lower impedance source can provide a stiffer output,
ie less sag under changing load.

Variation in rotor speed (kinetic energy) over 1/60th of a cycle would
therefore already be difficult to observe. The effect of adding more
mass would then be even more difficult to observe. One would need a
high resolution (fraction of RPM) speed sensor with a very high
sample rate to observe the ebb and flow of kinetic energy over each
cycle.

Or simply watch the voltage and current on a dual channel scope,
then, if you have a good scope, integrate the two to get instantaneous
real and reactive power.

Some clarification of the term "rotary transformer" might be helpful.
Simply stating that it is a rotary transformer doesn't clear anything
up for me! What exactly is a rotary transformer? If the rotating
mass is irrelevant, then how is rotation relevant?

Well, you have 3 sets of stator coils, and one rotating squirrel cage
made up of a number of bars. The latter are all energized in parallel,
so the electrical phase is the same on all of them at any instant. But
their positions aren't the same, and the whole thing is turning. This
creates a mechanically rotating B field which sweeps past the fixed
position stator coils inducing a back EMF in them (all of them, including
the ones hooked to the wild leg, ie the "transformer" secondary).

*Some* of the stator coils (primary) are also energized by line current
from utility power. This is electrically time varying, producing its own
electrically rotating B vector. This vector is what induces currents in
the squirrel cage in the first place to allow it to produce its own rotating
B field.

The rotor is just an intermediate between primary and secondary which
has the interesting time varying property of mechanically displaced
(phase shifted) synchronization with the primary field.

If the wild leg is generating power during periods when little or no
power is available from the mains each cycle, where does that power
come from if not from kinetic energy stored in the idler rotor? If
that *is* where it comes from, then I contend that idler motor rotor
mass is indeed relevant, though I could sure see how adding more
mass may not make any noticable difference.

It is important to realize that a RPC (or more conventional transformer)
has mostly *reactive* currents circulating in it. These are inductively
reactive, so the current lags voltage. (If the coils were lossless, the
lag would be exactly 90 degrees.) These reactances consume no power,
but they do store considerable energy in their magnetic fields. It is this
energy which is transferred from input to output of any transformer,
whether rotary or not, as the fields rise and collapse. Negligible
mechanical energy is exchanged.

A conventional transformer may have relatively very little reactive
current when operating at full load. In fact, this is a necessity in
high-current high-freq switchmode power xfmrs and great pains are
taken to achieve very tight coupling so as to minimize stored energy.

A fully loaded induction motor has power factor considerably below
0.5, which says that power converted from electrical to mechanical is
more than power being stored in the magnetic fields.

I'm not saying you're wrong about RPC's, mind you. I don't know. It
would be interesting to measure the inductances of a typical 3phase
motor to see if the stored energy is consistent with your assertion.

Your assertion *is* consistent with (but not proven by) the
observation that idlers should be at least 1.5 times the rating of the
load, since induction motors run at very low power factors at light
load.

What's different about a rotary transformer is that the coils are in
different spatial relationships at different points in time. This means
that the geometry of the stators and rotor are such that inducing and
induced currents are of differing phase (time). So even though the
primary is going through a voltage zero, the secondary is seeing a
rising voltage induced by the collapsing rotor field, and vice versa.

In other words, the B field induced into the rotor doesn't instantly
die when the primary goes through a zero. That field is still collapsing
since it is inductively lagging, as it mechanically approaches a
secondary coil. The collapsing field induces a current in the secondary
which lags the current in the primary by the *mechanical* phase (time)
difference between stator locations.

A non-rotating transformer can't do this phase shift, but a rotary
transformer can, and does. That's how it is able to produce 3 ph
from 1 ph. But all the energy being transferred is *electromagnetic*,
the mechanical rotation is just there to provide the phase shift.

The ratio of real power to reactive (imaginary) power in the system
varies with load. But the system stored energy is constant (at least
until you overload it enough to drive it into saturation). In the steady
state, ie after sync speed is achieved in the idler, energy out equals
energy in less system losses. Same as for any transformer.

All true ... and all based on your initial thesis that all (or nearly
all) energy storage is in magnetic fields. None of the above
argument addresses that question but builds on an assumed answer to
it.

Again, I'm not saying your wrong.


Gary