[sdiy] Transformer question ( skin effect )

[Richie Burnett]

Hi Tom,

This doesn't really seem Synth DIY related but I used to design 
switched-mode power supplies in a previous life so was spurred to reply...

Each switched-mode power supply design is generally different and the 
magnetics are critical components to achieving an efficient design. 
Starting by gutting a PC power supply is not really the best course of 
action unless you only want to pull the output voltages by about 10% or so 
from the original spec.  You don't say what the specs are for input and 
output of your new supply, but if it's a significant departure from the PC 
supply you're gutting then you'ld be better starting by designing your own 
transformer from scratch.  In order to make significant changes to an 
already designed transformer you need a good knowledge of magnetics, 
electronics, skin-effect, thermal management and safety standards.  And to 
be honest if you have this knowledge it's much easier to do the calculations 
and then just purchase a new ETD-39 or whaterver core set you need with the 
right air gap to meet your requirements.  Trying to frig another design to 
do something it wasn't meant to do is always going to be a compromise at 
best, and at worst could be potentially dangerous if it were to overheat or 
has compromised creapage/clearance margins etc.

Regarding skin depth, the skin depth for a conductor isn't a sharply defined 
depth after which no current flows.  It is just an arbitrary point where the 
current density falls below a certain value, i forget the % current density 
at one skin depth but Wikipedia will surely have some data for various 
conductors.  You do get increased current handling ability by using single 
thicker conductors even when the diameter is greater than two skin depths, 
partly because of increased surface area, and partly because the inner core 
still conducts away heat even if it doesn't conduct much HF current.  You 
also need to be aware of proximity effect which causes current crowding in 
SMPSU windings that are close wound too.  A good magnetics book should cover 
these topics and give some rules of thumb.

In general, the copper and core losses are approximately equal for an 
efficient design.  Bundles of individually insulated conductors that are 
each 2 or 3 skin depths in diameter make a better overall conductor for HF 
transformers than one fat single strand wire.  That's because current flows 
on the surface of all of the individual conductors as they're electrically 
insulated from each other.  Copper utilisation is better.  You have to get 
three things in the winding window of the transformer:  The primary 
conductors, the secondary conductors, and the insulation.  You don't want to 
use wire bundles any thicker than necessary for the windings as this will 
waste valuable winding window space whilst giving little reduction in copper 
loss.  More turns of thinner wire on the primary would reduce the flux 
density and the core would run cooler as a result.  Everything interacts 
though so i'd strongly advise you to work through the maths of some trial 
designs with different core sets, conductor bundles and possibly air-gaps to 
see what works best.

You don't say anything about the specification or application of your 
supply, or your chosen topology so it's hard to give any specific advice.

I hope this helps,

-Richie Burnett,