Hello everyone, Sotiris here!

While working in my lab in the past weeks, I’ve been testing some parts I had previously designed, and a couple of interesting things stood out during the analysis. So, I’ve decided to dedicate this Newsletter to what I call “Lab traps”, sharing my personal experience with you and focusing in particular on the importance of parasitic inductors/capacitance in the secondary side of transformers.

🧑‍🏭  Transformer Testing

I started with the testing of a 60W flyback transformer, similar to the one I analyzed some weeks go in the Newsletter nº40. You can check the basic specs below in Table 1.

Table 1. Transformer basic specs under test

I used a flyback PSU capable of delivering 60W. Because the board transformer had a different pinout, I used some round 0.7mm solid wires to extend the pins of my sample. This is how the transformer looked like at the beginning of the experiment:

Figure 1. 30mm-0.7mm wires used to quickly mount the transformer to the flyback circuit

⚠️  Then I started the tests, raising the power at the electronic load to reach 60W @ 20V. But, in fact, I couldn’t do it… the flyback controller was shutting down after about 40-45W! My first reaction was “🤬- That's the Murphy’s law!” (…and some other French words that every self-respecting engineer starts barking out in a situation like this)

When wondering what had just happened, these were my first ideas:

❌ I failed selecting the proper turns ratios in the design stage.

❌ I got confused specifying pinout in Frenetic Online.

❌ The guys in the lab that built the sample missed something.

After checking the validation sheet that each sample gets and performing some more testing, neither of the ideas mentioned above were true, except one spec! In the validation sheet that I had in front of me, the leakage inductance didn’t match the one in my measurement. To be exact at 60-100kHz range the leakages measured are listed below:

Table 2. Differences measuring leakage inductance for the same sample

💡 Note that the leakage inductance is the one of the secondary, as it is measured reflected back to the primary. That is easily done by shorting the secondary and taking an inductance measurement of the primary. The two values are measured at 2 different frequencies, so I couldn’t make a good comparison.

Then I remembered that the controller board mentioned could tolerate up to 5uH of leakage. The original transformer had a leakage inductance of approximately 4.7uH. With 6-7.5uH the design was probably triggering the controller protection, and thus the shutdown after 40-45W.

🤔 OK, that might be the reason why. But what’s really happening with leakage inductance? What’s the true value? Pfff, an engineer can’t have a relaxed day, we have to face challenges every day….

🙌 Time for action!  

I took the transformer completely apart and rewind it using round wires. This time I didn’t care for isolation, as functional isolation would be enough. My focus was to make a revision to get the leakage as low as possible and to see if that was the limitation of the test, but also make sure that no mistakes were made in the assembly.

🫣  Don’t blame me for not trusting other people in debugging. In these situations I don’t trust anything (not even myself…) that I can’t prove, so I had to make a new sample to eliminate every other possibility. I also ditched the LCR meter and brought the big guns: my frequency response analyser is more than capable to give us a clear idea about the leakage-frequency dependency.

As Archimedes had his “eureka!” (“Εύρηκα” in Greek ) moment, I had mine while setting up the equipment testing! I asked myself: "How did I shorted the secondary winding?". And my answer was: "Bending those 30mm wires you can see in Figure 1".

  🫵 Aha! Got you!

Look at Figure 2. That’s the difference in leakage measured using different lengths of shorting wire at the secondary transformer pins.  

Figure 2. Difference in leakage inductance for 2 shorting wire lengths

☝️ There is a 8.32-5.72 = 2.6uH difference between the two measurements. Let’s prove this!

Figure 3. Leakage model of an n:1 transformer

 The measured leakage inductance shorting the secondary and taking a measurement on the primary side is: 

Where:

• L_lkg is the total primary reflected leakage inductance of the secondary (and the primary leakage) – also called lumped inductance.
•  L_lkp is the primary leakage inductance.
•  L_MAG  is the magnetizing inductance.
• n is the turns ratio of the transformer.

We can simplify (1) with negligible error, noticing the fact that the magnetizing inductance L_MAG is 50-60 times bigger than L_lkg  , then:

In (2) the effect of different lengths of the shorting wire (L_lks) result in different L_lkg, since the effect of these inductances is magnified n²  = 7² = 49 times. For the 60mm shorting wire:

For the 8mm shorting wire:

Subtracting (4), (3):

Which gives us:

The round copper wire with an OD of 0.7mm has an inductance of 1nH/mm.

Less than 2% error in our analysis!

⚙️ What about testing the transformer in the circuit?

As you might have guessed already, when I modified the leads to make sure the secondary pins were soldered directly to the board, everything was working ok, as seen in Figure 4! Extra leakage inductance on the primary side is in the order of 50-100nH because of the wires. Compared to the 5.7uH of the actual transformer inductance that’s an extra leakage ind. of less than 2% (see equation (2)).

Figure 4. Soldering the test transformer secondary pins directly onto the board

⚡ Verdict?

Flying leads on the secondary side when we are talking about a stepdown transformer, as is the case usually with flybacks, can actually “kill” you!

In our case study, just 52mm of extra wire leads of ~52nH of inductance added a 2.6uH inductance looking at the primary side. In flybacks were a leakage goal of 5uH of leakage is reasonable, to minimize the losses on RC snubbers across the primary winding, 2.6uΗ of inductance is a big deal!

Although, if that was a step-up transformer, then the extra inductance appearing at the primary would be close to nothing. In Table 3 you can see a summary of problematic cases. 

Table 3. Problematic cases with parasitic inductances-capacitances

😎 I hope you've enjoyed the Newsletter! See you next week.

🚀​ Free access to Power Electronics and Magnetics design exclusive content?

Here you have it! Check our YouTube channel and discover our Webinars, Tutorials, Lectures and much more!

Sotiris Zorbas, MSc 

Power Εlectronics Εngineer