Abstract
The author first insists that magnetic scaling tests of electrical machines are not valid according to Maxwell equations but only in the restrictive area of quasi steady state approximation. Looking for practical limits, he measures the hysteretic cycles of two homothetic soft iron torus up to 200kHz. The cycles remain identic all over that range when the level of excitation is sufficient. But, for medium levels, he has found a lack of scaling behaviour. Using the saturated layer model, it is shown that skin depth is a complex function of both frequency and excitation level. It is proposed to relate high frequency anomalies to surface pollution of samples and to use this test method to investigate magnetic structures. (Author/publisher)