Why Do Transformers Hum?

Scientists in Switzerland have explained, for the first time, the microscopic cause of hum in those massive yokes of iron which help transform AC electricity from one voltage to another.

When current reverses 60 times a second the iron core of the transformer undergoes magnetetostriction twice during each cycle. In other words, 120 times per second induced fields cause the core to stretch slightly; a meter-sized transformer might stretch or shrink by only a micron but this would be enough to set up an audible 120-Hz hum.

(Earth itself is a magnet. Because of its own magnetic field, Earth's diameter is shrunk by about 10 cm. Turn off the terrestrial field and the Earth would spring back; its surface area would immediately increase by about 10 square kilometers.)

The new experimental work probes theories, going all the way back to Werner Heisenberg in the 1920s, about how the shrinkage arises from the magnetic interactions (spin exchange) among pairs of atoms (dimers), which share a common electron. The two magnetic ions want to be closer together.

For studying this effect iron itself is not the best test material and the Swiss scientists (ETH Lab in Zurich and the University of Bern) use another magnetic atom, manganese. Mn is a common ingredient in the magneto-resistance data storage systems found in most disk drives. Normally in a pure crystal, Mn atoms would be arrayed in endless straight lines. But in this experiment the Mn atoms are isolated, two by two, with plenty of intervening magnesium atoms. This allows the researchers to variably "dilute" the magnet interactions between Mn atoms.

The strength of these interactions (or to be more precise the energy levels of the excited Mn atoms) is measured by scattering a beam of neutrons from the sample, a process called neutron spectroscopy. The observed microscopic magnetostriction mimics the striction at the macroscopic level, but it does depart considerably from the predictions of the traditional Heisenberg model. (Straessle et al., Physical Review Letters, June 25, 2004; contact Thierry Straessle, Universite P&M Curie, 44-27-38-31, thierry.strassle@pmc.jussieu.fr, 33 44 27 38 81)