Best Method of modeling LT-spice transformer

Hello All,

I am wondering what is the best method for me to model a three-winding transformer in PLECS that will results similar behavior as in LTspice model as attached in the image.

In the modeling, assuming all I know is the k factors between each winding and the inductance of each winding, but I don’t know the physical turning ratio, the leak insurance, and so on.

Thanks in advance!

Also, I guess it would be nice if the modeling could work for k-factor that are less than 1

The “Mutual Inductor” component can be configured in a similar fashion. You can enter a square matrix of inductances to model the self (diagonal) and mutual (off-diagonal) inductances.

Note that the coupling factor is not entered directly but is used determine the mutual inductances. See the component help for more details on the relationship between coupling factor and mutual inductance.

1 Like

If you want the ‘best’, I advise to use the Extended cantilever model as described by Erickson and Maksimovic. It is the best because you can measure all parameters directly and very easily, and from these measurements you can construct a canonical model that everybody can verify. Interestingly, the cantilever model does not need coupling factors. Every interaction is described purely by inductors. It is therefore intuitively clear how to tune it for specific results.

The model in the PLECS help is fine, but it is possible to construct non-physical transformers when tweaking the parameters. That is not possible with the cantilever model. The only disadvantage of the Extended Cantilever model is that I never can remember its name.

Cite: R. W. Erickson and D. Maksimovic, “A multiple-winding magnetics model having directly measurable parameters,” PESC 98 Record. 29th Annual IEEE Power Electronics Specialists Conference (Cat. No.98CH36196), Fukuoka, Japan, 1998, pp. 1472-1478 vol.2, doi: 10.1109/PESC.

2 Likes

Thank you for suggesting the paper.

I was able to simulate the transformer equivalent model in LTspice. However, when trying to simulate the same diagram in PLEXIM, I got the error regarding “algebric loop” in the circuit, as shown in the attached pictures. Thus, I wonder is there any build in component that can do the job I did in LTSPICE?

The ‘wooden core’ idea has problems with stability, even in SPICE (that’s probably why you have those tiny resistors there…)Try moving L3, L4 to the other side of Vm2 and Vm3, in series with the output diodes.

I have a quick question regrading the mutual inductances.
I wonder under the hood, does it also automatically calculate the resistance matrix between each inductance component under the hood as shown in the figure, or does it only model the inductance matrix that I input it?

Thank you!

The mutual inductor represents only the inductive portion of the equation you entered.

@Bryan_Lieblick
You are stating that mutual inductance parameter of Plecs transformer model can also accept 2x2 matrix, I guess 3x3 for the 3 winding model. Unfortunately this isn’t mentioned in online help. Do you have a complete specification?

Rearding relation to LTspice coupled inductor model, LTspice and also generic SPICE documentation explains the relation to physical transformer parameters. As far as I remember, the three windings setup, as shown in first post doesn’t work for arbitrary couplings, you need to split the transformer into multiple coupling terms.

[EDIT] I realized that you are talking about ideal mutual inductor while I looked at Mutual Inductance with mutual resistance component. But apparenetly it also accepts matrix parameters for mutual inductance.

Frank

The Controlled Voltage Source has a parameter Allow state-space inlining that is off by default. If you set this to on, PLECS will substitute the source variable in the underlying state-space equations with the meter variable multiplied with the Gain. This should be the equivalent of a voltage-controlled voltage source in Spice.

1 Like