Magnetostriction - Transformer Noise

[tttman]

From the viewpoint of a transformer manufacturer ....

In the process of manufacturing strip wound silicon steel toroidal cores (to be used in making 60Hz power transformers) what are the major factors required to make a low noise (low magnetostriction) core? How are the following factors tied in ...

winding tension of strip wound steel
annealing process
grain / rolling direction
steel thickness, steel material, steel insulation coating
varnish impregnation of core & baking to cure
transformer exciting current (hysteresis loss, eddy current loss, interlaminar loss)
flux density level

???

 

Here are my opinions. Have lots of experience with cut cores at 60 hertz, not so much with toroidal cores.

Winding tension. Probably not much effect.
Annealing process. Usually optimized for lowest core loss. Probably not much effect
Grain/rolling direction. Always used in direction with lowest core loss.
Steel thickness, steel material, steel insulation coating. Have no idea, probably not much effect
Varnish impregnation of core and baking to cure.
Lowest noise when everything solidly connected together.
Transformer exciting current. Mostly effected by first 4 preceding items. No direct effect.
Flux density level. This is the most important cause of noise in transformers. 15 KG is typical for transformers used in an office environment.
If use cut cores, place a soft material between cuts, such as silicon rubber.
Manufacturing torodial cores requires a significant investment in equipment. The cores have to be wound, heated for a critical time to a critical high temperature in a controlled gas atmosphere. Most transformer manufactures purchase their cores already built.
Good Luck
Carl

 

[tttman]

Thanks for the response Carl.
Indeed, flux density level does seem to be an overriding transformer design factor in minimizing transformer magnetostriction noise. To be competitive as possible, one is always trying to push the limits on levels of flux density operation. We do not manufacture our own cores. These transformers would be used in an audio application requiring no dicernable audible noise at no load condition in a very quiet room.

It has been our experience that a transformer running 16200 Gauss at no load condition has resulted in exciting currents of 20 to 40 mA. When sample cores were excited to 10% above this level, we saw a majority in the 30 to 65 ma range, the rest were higher, the maximum value seen being 90mA. Anything up to 65mA was "acceptable" in noise level. In comparison, the same core with one grade lower Si Steel (11 mil M4) was producing 160 mA exciting current at 16200 Gauss and very significant noise level. The material change is the only change we are aware of. We are monitoring any other variations in core charcateristics at the moment.