6 phase drive of a 3 phase motor using two controllers

I'm starting this thread to help explore the idea of running a 3 phase 12 pole RC outrunner motor as split phase or dual stator. Luke brought up the idea in the "attaching hall sensors to RC motors" thread.

In the paper by Lyra and Lipo, they describe a type of 6 phase machine like this in the introduction:

A particular case of split-phase or dual-stator machine, the
six-phase machine can be built by splitting a three-phase
winding into two groups. Usually these three-phase groups
are displaced by thirty electrical degrees from each other.
This arrangement composes an asymmetrical six-phase
machine since the angular distance between phases is not all
the same [6]. The analysis of an induction machine for
multiple phases and arbitrary displacement between them is
presented in [2] where the six-phase induction machine is
used as an example and an equivalent circuit has been
derived. The dq0 model for a six-phase machine was
developed in [7].

Reliability is one of the advantages in using six-phase
systems. In the case of failure of one of the phases, either in
the machine or in the power converter, the system can still
operate at a lower power rating since each three-phase group
can be made independent from each other.

and then later under System description:

An experimental setup was built to test the proposed
technique. Fig. 7 shows the setup configuration. A common
DC link is used for two three-phase inverters connected to
each winding group of the six-phase machine. The neutral is
connected to the mid point of the DC link capacitors. An
additional inverter leg is provided for future investigation on
the neutral current direct control.

Ok, wow, that's quite a mouthful. But, I think that it applies to the high pole count 3 phase motors that we are running!!

Here's a link to the thread that got me going on this:
viewtopic.php?f=28&t=9061&start=60#p147898

and the paper

http://www.ece.wisc.edu/~lipo/2002pubs/2002_12.pdf

Ok, so to spur this on, I bought two of the Turnigy 12S 120V units, and another HXT 130 kv.
I bought the 130 kV vs. the 180 kV version because, I was thinking about the windings and
what splitting the stator would do to the resistance. They claim a 32 ohm winding, and if you
split the phases into two parallel 6 pole sets, then you'll probably end up with two 16 ohm windings.
Well, the 180 kV motor runs 17 Ohm windings....so this should keep coil resistance high enough.

Luke, you said you'd be able to help if we have a game plan, so here goes....

1. Figure out how to unwind/wind the stator in two 6 pole 3 phase sets (abc delta,xyz delta)
2. Come up with a load cell. This will be the most mechanical work, i.e. poor mans dyno
3. Instrument up the current from batteries/phases....measure voltage and current at all times.
4. Obtain batteries... Lead acid banks for cost, or 12S LiPo @ 2.5 AH, 30C cells?
(my 6S Lipo 5 AH Zippy cells are still in backorder!!!!)
5. Try not to saturate the stator or demagnitize the magnets!! (i.e.nonlinear output at peak or diminishing output after every run)

(plan needs more development)

Fun, fun, fun!!!

I asked a few people about this. I was told the two sets of windings would confuse the back EMF to each controller. I think Bob from Astro called it "Talk back"?

Matt

Why 36 degrees between the sets of phases, and not 60?

They mention reliability as one advantage, are there others?

Re: Talkback

Yea, I've been thinking about that. After I sort out the windings, I'll spin the motor
and monitor the phases two at time, since I don't have a four channel scope.
I may buy an A/D USB data aquisition box for monitoring lots of channels
at some decent speed. It seems that the problem is related to the virtual
neutral point of a delta wound motor. If you wind them in a Y then, you can
get an absolute neutral point that can be used to detect the crossing.
So, I don't know what kind of algorithm that the controllers are running, but I'm assuming
it's a zero crossing algorithm....

If the BEMF gets heavy distortion due to the split stator and causes cross talk due
to magnetic coupling between poles and prevents detection of a good zero
crossing, well then so be it. It's pretty easy to make some voltage divider
circuits (resistors in series) that will knock down the voltags to something that can be
monitored. Then we can look at the effects. Like I mentioned, I'll just spin the motors
to see what the EMF looks like when they are ran as generators and try to reconstruct
the neutral point of each delta wound split phase.

I'm still going to pursue it, I consider to be cheaper
and more educational and more fun than a class at a university !

I'm not questioning the guys at Astro abilities or knowledge, but I wonder if they
actually tried to do it!!!!!

Anyway, it'll be well worth the effort (for my own satisfaction anyway) to see what effect a dual stator
will have on the BEMF.

It's pretty interesting how you can wind the motor for delta or Y and a sensorless controller can still
deal with the differences in the BEMF. I'm gonna read up on this topic and see what I can gleam
from it.

Maybe you can compensate for the distortion with some sort of differential circuit that
will subtract the offending phases cross talk before you feed it to the controller.... I dunno ?

Lawson, didn't you rewind an RC motor from delta to Y ?

Re: 30 not 60.....Not sure why it's 30 not 60...
But, They due mention the ability to run two 3 phase drives as a means to overcome drive limitations.

from the conclusion of the paper:

Multi-phase systems are broadly used in industry to
achieve higher power levels based in limited range power
converters.

My understanding of the electrical theory behind multi-phase AC is next to nothing, but I have a working knowledge of the basics and the differences between different motor types.
If I understand correctly, you want to try to run a motor designed for 3-phase input with two separate controllers by dividing the poles into six phases, half for one controller and the other half for another. For our applications, I don't think this would present any real advantages and would add considerable complexity and cost. The only advantage I could think of would be reduced loading on each controller, allowing for higher power. But, like you said, cross-talk may become an issue.

Every brushless DC motor I've ever tried has produced a (near) perfect sine wave output. It wouldn't really make sense if it were anything else IMO.
As for your lack of oscilloscope, I bought mine for $51 on eBay, Georgia Tech surplus. It even still has the school property sticker from 1960 something.
Keep scouring eBay for a good deal.

Meanwhile, you can watch this short youtube clip I took of my scope on one of the motor phases while being driven by a Castle HV85:

[youtube]FuLfNrDlZaI[/youtube]
You can see the sinusoidal waveform from the back-EMF on top of the controller's PWM trapezoidal output. Kind of neat.

oofnik,

Thanks for posting that....

But, I disagree on weather or not it's suitable for the application.
If you can get a 30% increase in torque, that is a good reason.
and........this is fun for me!!!!!
The added complexity is manageable.

Yea, with a delta wound motor, the BEMF should be sinuisoidal.

I've got an old 2 channel scope.......the problem is that it's only 2 channels!
I'd like to be able to monitor at least 6. 3 scopes sync'd together would do that...

I think that Y wound stators will give a trapezoidal BEMF.

I just occurred to me that as part of the testing, that I can split the stator, and run the motor on 3 phase on one split. Then
on the dead split, you can monitor the BEMF. This should show the effects of one split on the the other.


I found some more information in a Lipo patent that could help to solve the theoretical controllability problem due
to the split phase rotor:

A dual stator winding induction machine in accordance with the invention has two polyphase windings with input terminals available to be supplied separately with drive power. The two stator windings are wound with a different number of poles to essentially eliminate the magnetic coupling between the two stator windings and to decouple the torques produced by each set of stator windings. In addition, circulating harmonic currents encountered in conventional dual stator winding machines due to the so called mutual leakage coupling are completely eliminated. Since the output torque corresponds to the algebraic sum of two independent torques, the stator frequency is no longer determined uniquely by the rotor speed and the slip frequency, but also by the added variable of a second torque component, adding an additional degree of freedom to the system for greater control flexibility.

United States Patent 6242884

For the sake of arguement, let's say the an even split would work (2-6 pole 3 phase sets).
Wouldn't the stator frquency would be cut in half for a given rotor speed compared to the full 12 pole motor.
This lowers upper frequency requirement of a controller.

Now, if we follow Lipo's ideas in his patent (damn, now we'll have to license it for sale) you'd
have two controllers running at a different frequency. The resistance of the split's may become more important....
or you might end up with 4 ohm and a 28 ohm splits, which probably need to be rewound proportionally to the number
of poles in the split. I guess in the end, the amount of torque each split and controller provides would need to be setup
so that the torques always add (unless low speed controllability is desirable, where you might want them to subtract)

anyway, the patent is a good read for those that have a stomach for this stuff.