[sdiy] Transformer question ( skin effect )

[Richie Burnett]

> ...Its a boost regulator.  A simple switching supply that takes in
> 12 volts from your battery and boosts it to 13.8 volts where
> equipment tends to operate most efficiently...

Are you sure the slight increase in efficiency from operating your transmitter
at 13.8 volts isn't going to be offset by the dissipation in the boost
converter.  A well designed boost converter will only be about 90% efficient, so
you'ld need an 11% efficiency improvement from the transmitter operating at
13.8V just to break even.

> In this particular case the design uses trasformer cores from old PC
> supplies and you rewind them.  It can just as easily be on new cores
> (amidon makes quite a few for example).  The original rewind spec on
> the transformer calls for 4x24ga wires on the secondary...

A boost converter doesn't contain a transformer.  It consists of a switch,
inductor and diode.  There is no isolation from input to output.

> The idea isn't to make the design handle twice the current All the
> time, its just that occasionally I'll exceed the design spec of
> 25amps by a few amps.  All the semis can handle it but the transformer
> and final choke both need some attention.

If there is a transformer and output choke then it sounds like some variety of
forward converter, not a boost converter.  If you are talking about output
currents in the range of 25 amps, i'd recommend winding the transformer with
bundles of Litz wire or possibly foil.  I'd also recommend powdered iron for the
output choke as it's more tolerant of the momentary overloads that you mentioned
without falling foul of saturation.  Micrometals make lots of suitable material
grades and have an excellent computer aided design tool for their cores.

> The fun part comes in when I have the original design working as
> I get to the part that is synth applicable for me and thats figuring
> out all the noise sources and how to squish them.

Making a SMPSU good enough to power an audio amplifier or synth is much easier
than making a SMPSU quiet enough to drive a radio.  It is harmonics of the
switching frequency which radiate and conduct away from the switcher.  For any
modern SMPSU the switching frequency is already well above 20kHz so wouldn't be
heard even if it did leak into the audio path in quite large amounts.  The
harmonics of the switching frequency will however potentially rake right through
the radio frequency spectrum of a sensitive receiver.  A decent radio will
easily detect a few micro-volts of ripple at harmonics of the switching
frequency.  This is the reason why most high-end receiver instruments still use
linear power supplies.

Filtering a SMPSU to feed a sensitive radio receiver can be a real challange. 
Good layout, screening and filtering are all vital.  Look up terms like
"common-mode filtering", "Differential-mode filtering" and "ripple steering." 
These are all techniques that are used to minimise RFI.

Good luck!

-Richie,