Hello, Sotiris here! 👋

 🚀 Today I’m going to be talking about some background story to a PFC prototype, probing tips and we will check out a different core material inductor, testing it under real conditions.

👉 I have found from experience as well as talking and watching young engineers evolve, that when it comes to gaining experience in a power topology there are real struggles😵   .

👉 Of course, it’s going to take to time to get familiar with every step, but every now and then there are a couple of important 🔑 design points that only experienced professionals know and pay attention to.

Story time 👪:

3+ years ago I got involved in a project where I designed a Flyback +PFC+PSFB (Phase shifted full bridge) board which would be used to charge heavy duty lead batteries.

My experience up to that point was more focused on high voltage analog electronics R&D for the mass spectrometry industry. With the experience level I had back then, power electronics design was a viable path to explore, but I was still a greenhorn for the power electronics world…

I tried hard to get it right the first time, focusing on theory a lot, then moving on to app notes/evaluation boards, excel calculators, YouTube videos, verifying equations and, eventually trying to analyze the equations themselves. I was a quick learner when it came to mosfet driving, signal integrity rules, common mode noise issues and other stuff I had previous experience with.

❌  The real surprise was that the easiest concepts to understand, in each topology, where the ones the most problematic during debugging.

❌  High frequency magnetics was a real struggle from the get-go… that was not a surprising…

End of story 👪

▶️ The CCM boost PFC looks easy right ? Just an inductor, a mosfet, a bulk capacitor, a diode bridge in the front and a 8-pin controller IC. I though to myself, “with the amount of free material out there, this will be easy to figure out and debug”

😈 HAHAHA - NOOOT 😈

⚡ Let’s talk about an easily ignored trap, that will turn heads around in your company or home. Basically, if you fall into the trap you will end up shutting down power in parts of your building.

It’s has to do with the PFC output power rails. They output a fixed +390V DC for use in the SMPS topology latter.

🥱 Yeah, what about it?

🔹 Can you hook up your scope probe to see that 390V rail ?

🤔 Emmmm, sure?!

🔹 WRONG! (90% of the time)

Where is the catch?

The problem is that when you start feeling confident 😎 enough to put you PFC circuit directly to mains then the output ground is “floating” and develops a +325Vpk (230VAC main) potential between it and your scope’s earth connected ground lead (talking for passive probes only). That means the lights will shutdown…😳

Take a look here:

If the N connection (Mains Neutral) wasn’t connected to earth in your building, as it is the case with an isolation transformer, in between the PFC and main plug, direct probing is fine. But otherwise in every period the line is connected though the diode bridge to the GND for half a period.

So we get LINE ->GND ->EARTH-> NEUTRAL    =   🔥LINE connected to NEAUTRAL🔥

🔑 Solutions🔑

1. Use an isolation transformer 👍

2. Use an isolated probe 💲 👍

3. Isolate the oscilloscope 👎 👎

💥   Let’s test a KoolMu(“Sendust”) core material inductor 💥  

Here is the new PFC inductor rated at a maximum of 3kW, I’ll be testing for this week’s newsletter.

👉 Inductor design project 😊

▶️    How did it perform under 3kW PFC load for 1hr+ (natural cooling)? 🧐

🤔 So 23.9% Hmmm….

This time I considered the end temperature at the room and input it in the simulation tool, so no much error there. Also let’s ignore the infrared camera 1-2% possible error source.

Was it me doing something wrong in the test? Actually not…

Well, I was looking at the previous week’s inductor performance (with temperature error prediction less than 5%) and this one.

Let’s compare them:

💡 Looking at the pictures of the inductor above I’ve noticed something!

🧐   I noticed that this inductor was supposed to be a 2-layer inductor. Instead, if you look closely the 1^st -2^nd layers get mixed to each other especially at the outer diameter. Are losses greater than predicted, because of this? Also, the fill factor is different. In this design you can barely see the core. Does this play a role too?

▶️  The first question is if the losses are greater than predicted thus the higher temperature rise?

▶️ The second question is why the lower temperature prediction if the losses are indeed correctly predicted?

There is some investigation going on with this result. Currently talking to other engineers about it, as well.

🤞I have a couple ideas myself, but I must verify things first🤞

👉 If you have a test idea let me know at sotiris.zorbas@frenetic.ai

By the way, this is just 5% of what we cover in our Frenetic Training, which includes one month of Frenetic Online.

🚀 We have successfully started (20th of April) the first session of our Training Program with a group of electrical engineers around the world 🌎 

We have opened the pre-booking for the next session, here´s the link to get the Frenetic Superpowers: training program

 👋  Until next week 👋