Abstract

This dataset aims to provide more insight into magnetic core losses at high frequency (MHz range) with superimposed DC bias in ferrites. The tested MnZn ferrite has a nominal relative permeability of 1500 (PC50) and 800 (PC200), which was tested at frequency from 500 kHz to 3 MHz. The DC bias appears to create a shift in the Steinmetz parameters, in particular the ki and β are affected the most. Higher DC bias creates higher losses gain and higher operating frequency makes the relative losses increase lower. Finally, the Steinmetz Premagnetization Graph (SPG) + iGSE method and Artificial Neural Network (ANN) method are used to create a core losses prediction model. As the result, the maximum error of 20% and 37% were found for SPG and ANN method respectively, while the average errors are 5.4% and 7%.

Instructions:

* How to use the dataset ?

There are 2 measurement series results provided in the dataset: PC50 and PC200.

Their raw data is located in [Measurement], while the processed data is in [Processed].

Nevertheless, the user can add and process more measurement results, given the data structure is compatible.

The general procedure to process the data is the following.

- Collect measurement results with the same frequency and DC bias in 1 folder

- Adjust the measurement parameters in [def_meas.m]

- Adjust the Device Under Test (DUT) parameters in [def_dut.m]

- Fill in the User Input section in [main.m]

- Run [main.m]

The results will be saved in a new *.mat file.

Some demo files are provided in [demo] folder. Please copy to root folder before running.

* Compatibility

The dataset and the scripts are developed with Matlab R2020a.

Funding Agency:

Danmarks Frie Forskningsfond

Dataset Files

dataset.zip (243.69 MB)
scripts.zip (6.87 kB)

Documentation

Attachment	Size
README.txt	1.38 KB

Datasets

Standard Dataset

Magnetic Core Loss with Superimposed DC Bias

Abstract

Dataset Files

Documentation

QUESTIONS?