I hope the title is explanatory enough.

The following is the practical model of a transformer having a centre-tapped secondary (eddy and hysteresis losses, and parasitic caps are omitted):

![enter image description here

Img src: Me

If I were to find the leakage of the primary using an LCR meter, what would be the best practice?

  • Shorting top and bottom legs of the secondary and measuring primary inductance?
  • Shorting all three legs of the secondary and measuring primary inductance?
  • Or what?

My initial thought was the conventional method (i.e. ignoring the centre-tap and shorting the outer legs) should be enough, so shorting all shouldn't bring any (or significant) difference. The measurements show quite close numbers so I wanted to be sure:

Loc = 77 μH (open-circuit inductance of primary, all secondaries are open)

  • Outer Legs Shorted: Lsp = 3.2 μH
  • All Legs Shorted: Lsp = 3.1 μH

And, a related question:

Is it possible to have an idea about the secondaries' balance/symmetry, by measuring primary inductance while the other leg of the respective secondary is shorted to centre-tap, and comparing these two measurements? The reason why I ask this simple question is that the secondaries have really low inductance (way less than 1 μH) while the primary has high enough (around 80 μH), and the measurement device is terrible at measuring such low inductances. So, I'm trying to find a simple-yet-accurate way to test the secondaries' symmetry.

With this method I measured the inductances as follows:

  • Lsp1 = 5.7 μH (Only top leg is shorted to centre tap)
  • Lsp2 = 5.8 μH (Only bottom leg is shorted to centre tap)

Again, the numbers seem to be close, so one could easily say the secondaries are symmetrical. But considering the fact that the secondaries have way too low an inductance, even 20 nH of difference may create a significant current imbalance.

  • \$\begingroup\$ Try it, come back with numbers and, I for one will try an explanation (if I can). You could even do secondary to secondary leakage current measurements. \$\endgroup\$
    – Andy aka
    Sep 5, 2022 at 13:27
  • \$\begingroup\$ @Andyaka I already have numbers and they are quite close. It could be the measurement error but I wanted to be sure anyways. I'm updating the question. \$\endgroup\$ Sep 5, 2022 at 14:50

2 Answers 2


Since the centre tap does not alter the meaning of primary or secondary inductance, leakage is measured on primary with the usual secondary shorted. Shorting the centre with no voltage already shunted has no effect.

I can imagine a coarse method of centre balance mixing two 0.1% resistors of equal values with the tap grounded and measure the common attenuation vs the differential signal then compare with the DCR ratios.

A more elegant way might be use two INA's to measure the different in gain with a load.

Yet another way is measure the gain in reverse , commutating the legs under high current.

Either way even with almost perfect balance, DC coupled mismatched loads may integrate switched DC currents to cause core saturation, unless sensing is used to correct or balance this. (Guessing)

  • \$\begingroup\$ Either way even with almost perfect balance, DC coupled mismatched loads may integrate switched DC currents to cause core saturation that's my concern. I'm working on a HB LLC converter in which the output current is 67A. Although the percentages of mismatch seem to be low (less than 2%) even such low mismatch causes significant amount of imbalance (the difference of peak currents is about 10 Amps!!!). Plus, this imbalance gets reflected to the input and therefore the magnetising current has a DC component which "might" eventually lead to core saturation. \$\endgroup\$ Sep 6, 2022 at 7:03

The measurement of the inductances in question can be more accurate when measuring secondaries with primary SC or OC respectively. In regards to the the current imbalance in the HB LLC. The main reason in that case might be the difference in duty cycle on each HB leg. Especially if your operating point is close to the resonance frequency. Hope it all make sense.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.