[Amps] Real time tests to see if an RF transformer is saturating?

[Manfred Mornhinweg]

Chris,

it's very easy to calculate the flux density. Easier than measuring it. 
So, let's do the maths:

You have 2kW applied to a 7 turn winding. Assuming the impedance at that 
point is 50 ohm, 2kW is 316V.

You have three FT-240-77 cores. That core has an effective cross 
sectional area of 1.57cm². Three of them are then 4.71cm², or 0.000471m².

This is all we need to apply equation 4 from my web page

http://ludens.cl/Electron/Magnet.html

316V / 7 turns / 4.44 / 0.000471m² / 136000Hz = 0.159 tesla

That's the peak flux density in your cores, meaning that the total flux 
swing is between 0.159 and -0.159 tesla. This is comfortably below 
saturation, which begins roughly at 0.3T, but it's in a range where core 
loss is high.

Let's consult manufacturer's data for core loss. Extrapolating the data 
given in a graph in the material's datasheet, it looks like the 
volumetric loss is around 600mW per cubic centimeter of ferrite. Each of 
your cores has a volume of 22.6cm³, so you have a total of 67.8cm³, and 
that means a power loss of slightly over 40 watts in those cores! That's 
clearly FAR too much.

So, saturation is no problem, but power loss in the cores is excessive. 
You need to either add more turns, or design a different transformer, so 
you get lower core loss.

You might start looking at the material. Type 77 is really not so good 
at 136kHz. There are newer materials that have lower loss at that 
frequency, at the roughly the same permeability and other data. The 
question is what you can buy...

If you want to keep the same cores, you need to increase the turns 
number. At 136kHz the wavelength is enormous, so you won't run into 
trouble from excessive wire length.

Let's take 5W as an acceptable maximum core dissipation. That would make 
it warm, but probably not too hot. So you need 8 times lower core loss, 
which is roughly 75mW/cm³. Consulting the loss graph given by the 
manufacturer, the allowable peak flux density is roughly 50mT. You have 
to triple the current number of turns to achieve this: 15 turns primary 
and 21 turns secondary should work.

A lower-loss ferrite would allow you to get away with fewer turns for 
the same core size, but never as few as you have now!

And a pot core, RM core or even EC core would probably be better than 
stacked toroids - if you can find one large enough!

And at that frequency a bundle of thin wires has lower loss than a 
single solid wire of the same diameter, and is far easier to wind.

Manfred



> Hi. Two off 1kW quasi Class D push pull FET amps on 136khz feeding a
> Wilkinson combiner via the amps own output transformers designed for a
> 50 ohm load. Combiner built from plans for one to combine two 500 Watt
> amps. Combiner's output transformer is three stacked FT-240-77 ferrite
> toroids, 5 turns 12 AWG enameled (2mm OD) wire primary. Secondary is 7
> turns of the same wire. The combiner feeds a big LPF bank. At full
> power the combiner's toroids get hot quite fast with WSPR 2 signal
> applied. Needs a small fan to stay within the realms of sensibility,
> even then it creeps over 70C if left too long. How can I tell in real
> time, with measuring instruments, if it's saturating, and if so what
> cores might be more suitable please? Hopefully the schematic of the
> combiner is linked. Many thanks! Chris 2E0ILY in the UK.
> 
> http://www.w1vd.com/137-500-500WCombiner.pdf
> 

-- 
========================
Visit my hobby homepage!
http://ludens.cl
========================