planar transformer design

Discussion in 'Electronic Design' started by [email protected], Jun 17, 2007.

1. Guest

Hi, I have to design a planar transformer with the following
specifications:

1:1 ratio
12A continuous RMS current in primary and secondary
frequency 150kHz - 400kHz
input/output voltage - 8Vp-p square wave (4Vrms)

I don't know where to start with the designing process.

I have calculated my S as 48VA and I chose a 3F3 ferrite double E core

Now I seem to get stuck in the next step, do I first choose a core
and then calculate my maximum flux density?

Johan

2. EeyoreGuest

Planar transformers are no different to any other other aside from the method of
construction. All the sums for the magnetics work just the same.

Philips (Ferroxcube) had a good application note on them.

OTOH if you don't know to design a transformer at all, that may not help you.

Graham

3. Don KlipsteinGuest

Whether the transformer is "planar" or not, what I do is:

1. Pick a core.

2. Pick a maximum flux density that does not produce saturation or
excessive heat at the frequency in question, at the maximum temperature
that the core is anticipated to achieve.

3. Figure out how many primary turns are needed to keep the maximum flux
density that low.

4. Figure out how much I-squared-R loss you get at the current in
question at the frequency in question at the maximum wire temperature
foreseen.

Keep in mind that the skin effect is significant at frequencies in the
10's of KHz and higher for any wire size likely to carry 12 amps.
Resistance of copper wire is also roughly proportional to absolute
temperature, but the variation with temperature is less (probably closer
to proportional to square root of absolute temperature) once the skin
effect has boosted resistance a lot.

There are charts and formulas for predicting resistance of a given wire
size at a given frequency.

I would suggest Litz wire with a high strand count for 8 volts 12 amps
at frequencies in the 10's of KHz and higher, since that is likely to
allow substantially smaller wire that will allow a smaller core. And I
would try to find some data on the resistance of the Litz wire in question
at the frequency in question.

And after that, test and verify that the transformer does not overheat,
since most published skin-effect-related resistance increase figures are
for isolated straight wire and the figures can be different in a winding.

- Don Klipstein ()

5. Guest

I will use the formula V = 4.44 N A e f B 10 -8 substituting 4 for
4.44 as I will be working with square waves. I don't know what to make
of the turns though.

Would it be wise to start with the windings?

I can't just select a core? Obviously I need to calculate a suitable
core volume?

Thank you for all the feedback.

Johan

6. Guest

Thanks Don. I will give that info some thought and do some
calculations tonight. Will post my results.

7. EeyoreGuest

Planar transformers use the copper foil of multi-layer pcbs for the windings.

Graham

8. Guest

What would the duty cycle for a 1:1 transformer be? I suppose it would
be one?

Also, the minimum input voltage? Considering its a 4Vrms square wave?

9. Guest

I don't understand how the ferroxcube cores work. There is a core (E)
and a plate (PLT). When I calculate my transformer's core loss density
do I take the Ve of the E plus the Ve of the PLT into account?

10. EeyoreGuest

You do it exactly like any other EI core. You can also use them in an EE configuration too.

Try this to calaculate losses.
http://www.epcos.com/web/generator/Web/Sections/DesignTools/Ferrites/Page,locale=en.html

Graham