#9 🚀 Herbert Core Losses, Frenetic AI vs Experimental results

The Newsletter of High Frequency folks

👋 Hello! Dr. Molina here! 👨‍🔧

🤗 Thank you for reading and welcome to the show my Newsletter. As I said last week, I will continue talking about Product Design in the first place.

In the second place I will comment a great presentation of Ed Herbert, one of the most important engineers in the Magnetic industry. 🔝

Finally, I will check the measurements of the samples of my design and compare them with the AI prediction. 🧐

🤖 Product Design

⭐️ This week has been the first real week of 2022 and we were very lucky to attend two very interesting lectures about Products from a great Spanish company, Genially, and a great Product Manager, Patricia de Loro.

I would like to highlight some of my notes on Patrica’s lecture:

“Product is Business and Business is Product”, Patricia de Loro

She described something very interesting: The product should be aligned with the company business and all the individual incentives should be aligned too.

Sounds very basic but, I’m sure if you analyze it in detail, 90% of the organizations don´t have 100% every task aligned with the business.

Designing a product in a very noisy environment is a real challenge. We are receiving infinite number of insights from users’ feedbacks, competitors, Twitter, Facebook, news… it’s really difficult to isolate the noise to the great solutions.

I have spent several weeks preparing the product plan for Frenetic and still, I find it very complex how to communicate effectively without stopping everyone for infinite meetings.

Today, I would like to introduce a Genius who I discovered in a great exposition at Madrid during 2021, Buckminster Fuller . I’m sure the contemporaneous of Fuller didn´t understand him 🤓.

🔥 Herbert Core Losses

Using Specific Core Data instead of Volumetric Losses Data

I met Ed Hebert the first time I went to the Magnetic Workshop in 2018. He made the introduction to the event. A year later I started talking with him to participate in the Magnetics Committee, where he has been a great Co-Chairman until now.

As you can read in his introductions:

Ed was a champion of the core loss studies at Dartmouth and is the champion of the present drilled cores study at SMA. 

Today I will comment on very interesting research about core losses he has been conducting.

 Generic Specification for Ferrite Cores

The classic core losses estimation is coming from these graphs where Losses correspond with a magnetic field applied to the core for a given material. There are lines for each frequency, where engineers take the losses according to the magnetic field at the specific frequency.

Using this approach, we are assuming; Losses are the same for each core shape and volume if we keep constant the magnetic field, frequency, and magnetic field. This is the main focus of these studies. I include his literal comments about this type of graph because I don´t think there would be a better way to describe it 😊:

However, this graph is practically useless for finding the optimum operating frequency and applied excitation to minimize core loss.

Core loss decreases with increasing frequency for a given excitation voltage, to a point, then, as eddy currents become dominant, loss increases. 

Can you see that in this graph?

The answer to the last question is clearly, NO, you can´t.

We can conclude. With this information is not possible to find the optimal frequency for a specific shape unless volumetric losses are equal for all the shapes, are they??

According to Ed Herbert's research, losses in a core are not constant for all the shapes and sizes and each specific core has optimum frequency for a given excitation.

In the slide below, you can see the Herbert graph. In this graph are represented losses curves with respect to Volts/turn applied and frequency for a specific core. Therefore, you can find the optimal frequency for a specific core, playing with the Volts/turn, keeping losses under control.

In the next two slides, he changes the axis, including the losses in the y axis and including curves for different volts/turns.

To conclude the affirmation that each core has different properties, Ed add this graph of Glenn Skutt’s thesis, presented in 1996. In this graph, we can see the losses estimated based on the material values and the results of the losses using the specific core values.

To conclude here, I add a picture where we can see the performance factor of different toroids of an experiment presented by John Lynch, where it is shown that different sizes affect the losses for the same conditions.

Do you wanna have all the slides of this presentation of Ed Herbert?

Send me a message through Linkedin or write a comment in this edition with your email and I will send you the slides (with Ed’s permission).

🚀 Frenetic AI Prediction Vs Measurements

In the previous edition, I did a complete design to replicate the transformer of a TI reference design, which was impossible to find in the market. A Full-Bridge Center Tap 400V/12V 600W transformer.

If you are interested in samples, I still have a few of them.

I received the measurements of the components in Frenetic Online and I would like to compare the AI predictions with the AI.

The Leakage inductance measured is 4 uH and 4,4 uH.

The Artificial Intelligence Prediction was 3,5 uH, not too bad men!!

The measurement of the Magnetizing inductance is 2,47 mH (2,42 mH is the prediction) but this measurement doesn´t use AI.

Regarding the tests, the component pass all the isolation test.

I would like to highlight the work of the AI Team of Frenetic, specially Carmona & Noe, who has been working on the AI models really hard in the last few years. In the coming year, we have a lot of new projects for using AI in the losses prediction and we have a new guy in the office, Juan Aguarón, who is spending a lot of time in the lab.

That’s all for this week. I would like to find time to design a DAB Transformer but I can´t promise… :(

Top finds🧑🏽‍🌾

Series: Halt and Catch Fire

Book: From Zero to One

Articles: A Simplified Winding Design Procedure For Transformers (by my friend from How to Power)

Thank you again for reading and don´t forget to subscribe.

Sincerely,

Chema 👋