On Tue, 28 Oct 2008 00:52:28 +0100 (CET), Grumpy The Mule
wrote:

Hmmm!

Bolts through the laminations should have insulators
(usually fiber board washers) under the heads. Just
one end will do, there's no need for them under the
nuts. But shoulder washers are best so the bolts don't
short any laminations together deeper into the core.

If uninsulated the bolt can form a poorly coupled shorted
turn and that in itself can cause noise, heating of the
core and the hardware as well as spray flux around where
you wouldn't have expected any. How bad it is depends
on the locations of the holes in the core.

A bead of weld across the ends of the laminations won't add
to the eddy currents significantly. Steel isn't a great
conductor. The weld's cross section is small and it's very
poorly coupled to the primary.

Why would a weld across the laminations let line spikes
though?


It's not just the laminations that make noise. The windings can
vibrate if they're not locked in place with transformer varnish or
epoxy. No amount of tightening the laminations will help if the
windings are free to vibrate. This is probably more likely in larger
higher power transformers (multiple kW) than in a filament or screen
supply transformer, but it can happen.

Also, depending on how much magnetic leakage your core has you can
couple to the steel walls of the cabinetry the transformer is housed
in if you don't maintain adequate spacing from the walls, in which
case the steel wall and the transformer can vibrate and make noise
even though no parts on the transformer are loose.