Sorry about the cumbersome subject line.
I have a DC-DC converter with a transformer with a center-tapped primary & secondary. I'm trying to simulate it with Spice to try to get reasonable agreement with what I see on the bench.
To represent my transformer in Spice, I have 4 inductors, L1 & L3 for the primary and L2 & L4 for the secondary. I couple them with a Spice statement
K L1 L2 L3 L4 .999
That simulates fine but does not give very good agreement with experiment.
I'm guessing that the coupling constants among the 4 inductors are not the same. For example L2 is wound on top of L1 and L4 on top of L3, so I would expect
K12 > K14 and K34 > K32
where K12 is the mutual coupling constant of L1 & L2, etc.
So I tried
K12 L1 L2 .999 K34 L3 L4 .999 K14 L1 L4 .997 K32 L3 L2 .997 K13 L1 L3 .998 K24 L2 L4 .998
This simulates but still does not give good agreement.
If I try making the K'2 too different, I get a Spice error "The inductors L1, L2, L3, L4 comprise a transformer with a non-physical winding possibility".
So what this is leading to is: what are the constraints on the K's of a bunch of inductors? For N inductors, there are N(N-1)/2 mutual inductances each with a K. But they cannot all be independent. How many are independent and how many are dependent? In my transformer, there are 4 inductors and 6 mutual inductances. How many can I set independently and what are the constraints on the others?
There must be a theorem on the subject. Any experts?
For your enjoyment, I attached the simulation file. It uses the free Linear Technology simulator.
Thanks for your help.
Geno