Spice model for line input transformer

[brian92fs]

I was tinkering in LTSpice with adding a Jensen line input transformer to the B1K preamp. As I tried to educate myself on audio transformers (largely from materials on Jensen's site), I realized an idealized transformer in LT Spice didn't really seem to represent an actual transformer. (shocker )

I'm interested in a 1:1 Jensen line input transformer (JT-11P-1HPC). I tried contacting Jensen to ask if they had simulation models available - no response. So here is my attempt to model this particular transformer in LTSpice.

I used the "Test Circuit 1" from the transformer's datasheet. I used these steps to create the model:

1. Inductance imputed from datasheet primary series resistance and low frequency corner (0.25Hz).
2. Leakage imputed from datasheet secondary series resistance and high frequency corner (80kHz).
3. Parallel resistance imputed from datasheet source resistance and voltage gain (-3dB).

The frequency response and voltage gain of resulting model seems to conform to the datasheet with one exception. The deviation from linear phase at the high end of the audio frequency range is accumulating faster than the datasheet.

I've also noted that the imputed parallel resistance forms a voltage divider to arrive at the voltage gain stated on the datasheet. As a result, the gain of the circuit seems to be highly dependent on source resistance.

Any thoughts on this approach? Does it have any validity, or is this a misguided effort?

 

[brian92fs]

I spent some time reading and re-reading the "AN008: Audio Transformers – Bill Whitlock Chapter 11" pdf from Jensen's website. I found a section on calculating ideal and real voltage gains and was able to use these formulas to calculate the gain of the JT-11P-1HPC transformer and match the datasheet values. I realized from this math that they use an ideal (zero source impedance) source for this figure. I'm also now able to model the impact of source impedance on voltage gain. I've attached a spreadsheet with my math in a zip file if anyone is interested.

I revised my model using only datasheet values along with imputed inductance and leakage. I still match the datasheet for everything except deviation from linear phase at higher frequencies in the audio band.

So, I can match the datasheet for the high frequency corner or phase deviation through 20kHz (by altering leakage), but not both.

 

[davidsrsb]

How well does this model low frequency core saturation?

 

[brian92fs]

It doesn't model for core saturation or hysteresis. You would need to add parameters to the inductors to model them as non-linear. See the LTWiki article here for additional info.

I'm not skilled enough to determine how to spec these parameters. In addition, I assume they can't be readily derived from the datasheets, so I'm not sure how you would identify the correct values.

In addition, these advanced inductor models don't seem to be supported when mutual inductance is specified (i.e. - a transformer). The modeling get much more complicated to work around this. See the LTWiki entry here for details.

This is why I originally reached out to Jensen for support. Tried emailing a few times, but never got a response. Perhaps providing a model would reveal too many trade secrets? Maybe I'll give them a call to see if I can get some type of response.