[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Three turns, superposition wasRe: OLTC update - primary IGBT loss



Original poster: "Jim Lux by way of Terry Fritz <twftesla-at-qwest-dot-net>" <jimlux-at-earthlink-dot-net>

At 01:40 PM 9/3/2002 -0600, you wrote:
>Original poster: "Terry Fritz" <twftesla-at-qwest-dot-net>
>
>
> >
> >I'm having trouble modeling your triple parallel turn primary - the
> >interaction between the three is not modeled by the tssp software and
> >they are too far apart to model as a single turn.  This is why I
> >haven't sent any V/I profiles.
>
>The three turn primary was nice in that it gave lower Lpri.  But from a
>theoretical and modeling point of view, it sure is a pain.  I wonder if
>coupling in three spread out rings hurts the waveform on the secondary.  A
>point source would be best.  Three rings may be like trying to set up a
>standing wave on a rope with three people holding onto it shaking in unison
>but in different places.  Maybe it could be modeled by superposition where
>you do each one and then just add the numbers.  Not sure...  A messy
>problem.  You don't need to worry with it unless you really want to.


It's a bit trickier than that... you have 4 coils to worry about P1,P2,P3 
and S.

P1 induces a current in S, which then induces a current in P2 and P3.
P1 also induces a current in P2 and P3.

Sort of like with antenna analysis, you wind up with a symmetric 4x4 matrix 
of mutual impedances (or couplings, or admittances, or somethings, 
depending on how you do your analysis).

The overall matrix will give you something like [E] = [Z] [I], where [E] 
and [I] are 4 element vectors, and [Z] is the 4x4 impedance matrix.

Fortunately, once you've figured out this matrix, it will drop right into 
Spice, etc., because that's how it does the linear analysis in the first 
place (by building up a big admittance matrix that is NxN (where N is the 
number of nodes in the circuit), and then solving it).

And, the coupling matrix is a linear combination, so you can model and/or 
measure pieces of it; much like measuring parameters on an unknown 
transformer with windings shorted and open.

As I see it, you need to model 6 things (4 things, taken 2 at a time):

P1 : S
P2 : S
P3 : S
P1 : P2
P1 : P3
P2 : P3

from which you should be able to build the overall matrix.