Audio Transformer Inductance

[abbey road d enfer]

I recently noted Pat Turner's comments that deducing shunt capacitance from resonance frequency and leakage inductance may have significant error. 

Indeed. Any attempt at describing the working of a transformer with a limited model is prone to error. Nominal inductance, leakage inductance, stray capacitance...are all indicative figures in the waiting to be disowned by reality 

Pat proposed that lumped shunt capacitance is best measured using an RC LPF -3dB frequency technique, where R is ~5x nominal winding impedance and is inserted in series with the signal generator driving the winding, with other windings open and connected to ground to suit the normal operating condition.

That's the method I suggested earlier (reply #200); ultimately the shunt capacitance determines  the asymptote after resonance. It can be qualified as a "almost-direct" method, which is intrinsically better than the indirect method that requires prior assessment of the leakage inductance.

 

[CJ]

what are you going to do with the number once you get it?

watch your leads setup, they can have as much C as the xfmr,

 

[trobbins]

That's the method I suggested earlier (reply #200)

Ahh yes,  I didn't see your sketch at the time I was looking through all the posts in this thread. 

The leakage inductance and the lumped capacitance play a dominant role in high frequency response, which is my purpose in trying to define those parameters. 

Just to extend on the #200 post description, when measuring the primary of an output transformer for example,  a simplified 'lumped' model could expand the L out to have series DCR and leakage inductance parts that are in series with the signal generator interface resistance R, and the lumped C is in parallel with the transformer winding (where the voltage being transferred is either coming in or going out of).

For a step-up transformer, as per phono stage, then the lumped model that illustrates high frequency roll off is preferably changed to placing the secondary side shunt capacitance at the output terminals, with the leakage inductance and DCR being in series with the transformer winding before that shunt C.

 

[abbey road d enfer]

The leakage inductance and the lumped capacitance play a dominant role in high frequency response, which is my purpose in trying to define those parameters. 

Indeed; however, one has to remember that it's a radical simplification (just like modelling a transmission line as a single RCL). Certainly, it is good enough to predict the asymptotic roll-off.

Just to extend on the #200 post description, when measuring the primary of an output transformer for example,  a simplified 'lumped' model could expand the L out to have series DCR and leakage inductance parts that are in series with the signal generator interface resistance R, and the lumped C is in parallel with the transformer winding (where the voltage being transferred is either coming in or going out of).

For a step-up transformer, as per phono stage, then the lumped model that illustrates high frequency roll off is preferably changed to placing the secondary side shunt capacitance at the output terminals, with the leakage inductance and DCR being in series with the transformer winding before that shunt C.

It is almost instinctive to place the lumped elements on the highest-impedance side, because it is clearly there that the effect is dominant. However, when modelling a 1:1 xfmr, one can see that the simulations are not identical between lumping all on one side and keeping the parasitics separate. It all depends on what accuracy is expected. I have found out that lumping all together produces pessimistic simulations. Refining the model (even splitting each winding in its individual layers) produces a more accurate simulation.