On Tue, 31 Aug 2004 22:58:21 -0500, Don Foreman
wrote:

With some trepidation, I have to say that I don't think it's "just a
rotary transformer". Energy must be stored and released somewhere to
create a third phase, because a third phase provides power when there
is none available from the mains (during zero crossings).

One way to check this (and perhaps prove me wrong) would be to make
measurements of phase currents (with no "tuning" capacitors) with
various loads with and without additional mass (flywheel) on the rotor
of the idler motor.

It would also be interesting to observe rotor speed on the idler as
load is increased on the driven motor. If it decreases, that would
suggest that some exchange between mechanical power and electric
power is taking place in the idler, with the process of generating
the third phase creating countertorque that slows the idler rotor.

Fitch is *very* busy designing his new house at the moment, but he'll
be done with that eventually and have time again for thinking about
other things.





A slightly different way of looking at this vexed
problem may help.

It doesn't directly clarify the operation of a real
real rotary phase converter with all its losses and second
order effects but it does at least give a reasonably
convincing (to me anyway) insight into the way it works.

Consider a 2 pole 3 phase wound LOSSLESS squirrel
cage motor supplied with single phase power to one winding.
There is no mechanical load so there are no losses of any
kind. Once the rotor is spun up to operating speed, the
stable operating condition is with the rotor spinning at 2
pole synchronous speed.

This necessarily means that the rotor is
diametrically magnetised as a single N-S magnet as a result
of the induced current that is circulating in in the
superconducting rotor bars. Because these are perfect
conductors this current CANNOT decay so that the rotor
behaves exactly as a rotating permanent magnet.

The rotating magnetic field produced by this magnet
induces equal voltages into each of the three phase windings
and it is this rotating field that produces the true
balanced three phase output voltage pattern.

In the case of the energised winding, in this
completely lossles system, the induced voltage (the back
EMF) has risen to be both equal to, and in phase with the
supply voltage so that no current is drawm from the supply.

If a load current is taken from either or both of the
unenergised windings the current in the windings produces a
magnetic field that tries to slow down the rotor. It doesn't
succeed because the motor is operating in a synchronous
mode. What happens is that there is a slight change in the
angular position of the rotor so that the voltage maximum of
the induced back EMF is no longer exactly coincident with
the supply voltage maximum. This shift causes just enough
supply current to flow to cancel the drag forces generated
by the load currents.

In this lossless motor there are no voltage changes
resulting from voltage drops in the winding resistance so
the output remains at three voltages all equal to the supply
voltage and 120 deg apart.

Exactly the same correction mechanism applies if a
mechanical load tries to slow down the rotor. This means a
lossless motor delivers balanced three phase out independent
of variations in both mechanical and electrical loading.

It's interesting to note that this explanation does
not require any direct transformer action between the
windings, all the energy transfer is via the rotor.

A factor,so far ignored, is that the torque generated
by input supply currents fluctuates between zero and a
maximum value twice per cycle of the input waveform. Because
of this there must be enough mechanical inertia in the rotor
to continue the rotation through the low current parts of
the cycle. This is only matters if the inertia is so low
that the input torque fluctuations produce significant
instantaneous rotary velocity changes at twice supply
frequency. In practice the rotor inertia of the commonly
used motors so large that this is not a problem. Any
variation large enough to matter would show itself as a
modulation distortion of the phantom phase voltage(s)
waveform at twice supply frequency.

Jim