[Amps] how to wind an HF broadband 10:1 transformer

[Manfred Mornhinweg]

Hi Jim,

>> I didn't know they had so much variation with frequency
> 
> All of this data has been in Fair-Rite's printed catalog for many 
> years, and that catalog has been on Fair-Rite's website as a pdf for
>  at least 5 years.

Could you give me an exact URL for some page that gives this information
for the 61 material? I couldn't find it, and after a few hours searching
I gave up. Living in a rural location, my internet connection is over
the cellphone network, with dismal performance, so it's not conducive to
efficient web browsing!

Also, if you find some place that gives loss curves for ferrite
materials, that would be very useful too! I could find only very sparse
single-frequency information.


Carl,

> Terms such a "should", "could", "most cost effective", do not give me
> a very warm feeling Manfred.

Maybe I misused those words. English is only my third language. When I
wrote "should be able to work at 1.5kW", I meant that when I set out to
design this circuit, one of the design goals was 1.5kW operation. Later,
when the system was ready, I could confirm through years of daily use
that actually the goal was met.

And when I wrote "on a transmission line that could have high SWR", what
I meant is that this thing had to drive an open wire line connected to a
random antenna, with completely unknown impedance. No design can
be guaranteed to work over "any" impedance to be found in the real 
world, so here the goal was to be able to work with good performance at 
SWR levels up to 5:1 or so, and acceptable performance at somewhat 
higher SWR. In practice I have operated into antennas that show an SWR 
so high that the meter reading is undistinguishable from infinite. In 
those cases of course the efficiency must be lower, but still the 
transformer has not blown up despite that use. It does get warm though, 
with some extremely bad loads.

> However if you can show 1.8 - 30 MHz performance statistics at the
> 1500W level they would go a long way to put a sense of engineering
> reality on the subject.

I have neither the instrumentation nor the time to do a scientifically
valid investigation of the transformer, so I'm sorry, I can't provide
precise data. I can only say that in practice the ferrite material has
worked very much better than powdered iron, and that theory and data
extracted from the relevant sheets beautifully explains why. So my
assertation is a relative one: Type 61 ferrite performs much
better than type 2 powdered iron, in this broadband high power
application, both in terms of loss and in choking performance, but I'm
not able to provide precise absolute data on transformers built with 
each of the two materials.

> His book went thru a few revisions as he was faced with reality and
> not a lab test at low level RF.  At one point there were some rather
> heated on the air discussions that I partook of strictly as a
> listener.

That must have been interesting! Unfortunately, I rarely find anyone to
talk about technical things on the air. That's why I vent here! :-)
In any case, my ferrite balun has been used for several years, under
many different conditions, at legal limit power and all bands, and so
far has worked well.

> My own experiments with a FT240-61 at 1200W was rather dismal with
> excessive heating that resulted in tuning drift as well as TVI.

The three symptoms all sound like you drove the ferrite to a much 
excessive flux density. What flux density did you use?

> At that time I did not own a spectrum analyzer to look for non
> linearities nor a network analyzer. I now own both.

I'm still at that stage of not owning these instruments! And believe me,
I would love to get a spectrum analyzer! But it should work at least up
to UHF, it should have a dynamic range not under 90dB, and it should
not cost as much as a car. So far this has kept me without one.

> I then tried T225-2A thru T400-2A powdered iron at various power
> levels from 500W to "well above" 1500W; the tests used from one to 
> three cores wrapped with Scotch #27 HV tape and #14, 12 and 10
> stranded Teflon wire. The result was several balun 4:1 kits for 500
> to 3500W that were sold for about 10 years thru a part time business 
> that I owned.

You must have used many turns to get enough inductance, and as a result 
there must have been reduced performance at the higher bands.

> As expected they were not perfect due to winding capacity and trying
> for a 1.8 to 30 MHz compromise.

Exactly. On powdered iron cores you can make baluns that work well over 
a reduced range of bands, but 160 to 10 meters is asking too much from 
them. There have to be more severe compromises at both ends of the 
range, than if you used ferrite.

> However they worked well, did not
> overheat when used within their stated power ratings and did not
> cause any TVI/RFI at home or reports from neighbors. I did not
> attempt to check for feedline radiation as they were only accessible
> for a very short part of the total runs.

Then it's OK. Still, a side-by-side comparison with ferrite baluns would 
have been interesting.


Peter,

> I did that when I got my copy of V.O.Stokes book many years ago,
> where I found his wideband transformer concept. Using the largest
> ferrit toroids I could get, they were similar to Amidon #61, I
> sandwiched them between 5mm thick aluminum plates 20x10cm for 
> cooling, used 2x12 toroids and wound a 50:600 transformer with 3mm
> teflon covered wire to feed my terminated V-Beam from 160-10m. That
> worked perfectly well with all the power my modified L4B (3.5KV)
> could deliver.

The fact that you needed heat sinks makes me think that you ran those 
ferrites at pretty high flux density. That was probably necessary to get 
an acceptable wire length. How much wire did you have to wind on these 
transformers?


Jim,

> Were you building a TRANSFORMER (commonly called a voltage balun) or
>  a CHOKE (commonly called a current balun)? 

It was a common mode choke, incorrectly called a current balun. The 
current is balanced on both sides, so this is really a current balbal! ;-)

 > They are VERY different.

I know. Very different in some regards, and very similar in others.

>  #61 is a good choice for a transformer, but it is a lousy choice for
>  a choke.

Why? It works very well for me! I don't see why we should need different 
kinds of core material for voltage baluns of common mode chokes aka 
current baluns!

> See http://audiosystemsgroup.com/RFI-Ham.pdf

I saved that file, to read it in detail when I have time.

David,

> To improve the range of choices, could the non-linearity introduced
> by the core be improved by a feedback system ? 

Yes, that's possible in many cases when a transformer is used in an 
amplifier. But with increasing wire length, the phasing problems become 
harder to manage, and that causes difficulty with feedback.

My approach is to start from the most linear transformer I can make, and 
then perhaps use feedback to further linearize the entire amp.

> Perhaps the goal of 160 to 10 (6=dream?) might have be managed in
> more than one lump.

That would make it much easier, but also more expensive. The first prize 
would go to someone who covers the entire range with one simple, elegant 
and inexpensive circuit!


And now for all people interested: I dug up whatever information I could 
find about Micrometals #2 powdered iron, and 125-permeability ferrites, 
to make a comparison in the application of a high power broadband 
transformer used as "current balun".

First some basics: The best magnetic material for transformers is one 
that has the highest permeability, and lowest loss. If the permeability 
is really high, variations of permeability with temperature and other 
factors become less critical. Also, nonlinearity of the magnetization 
curve is then less critical, because the magnetization current will be a 
smaller part of the total. Instead in materials with low permeability, 
both the instabilities and the nonlinearities become more important.

Also, a good magnetic material should have a high saturation flux 
density, but with present day materials used at RF this is irrelevant 
because anyway the maximum practical flux density is dictated by the 
losses, and not by saturation.


Now let's see how the two materials under consideration fare:

The permeability for the powdered iron material is 10, pretty stable. 
For the ferrite, it's 125, varying significantly. So, from this point of 
view the ferrite is 12.5 times better for transformer use, but 
unsuitable for resonant circuits.

The volumetric loss depends on flux density and frequency. It is really 
hard to find enough data at the manufacturer's web sites, to do a 
meaningful comparison. The ONLY comparison point I could find, with firm 
data provided for both materials, is at 500kHz and 50mT. The ferrite 
considered here is the K type, which is Ferronics' equivalent to 
Fair-Rite's #61, for which I couldn't find the data. At these levels, 
the loss of the powdered iron is 2100mW/cm^3, while that of the ferrite 
is only 80mW/cm^3!  Which means that the ferrite from this point of view 
is a whopping 26 times better that the powdered iron!

Now, of course, the losses need to be extrapolated to the frequency 
range of interest, 1.8 to 30Mhz in our case, and then adjusted to the 
actual level of flux density during design. Amidon gives an equation for 
their #2 material on their web site. Using it, it turns out that we have 
to wind the powdered iron cored transformer so that the flux density at 
1.8MHz will not exceed 7mT, in order to keep the core loss below 
200mW/cm3. This 200mW/cm3 would result in a power loss of about 3W in a 
T-200-2 core, or about 30W in a T-520-2 core. If the builder thinks he 
can tolerate a larger loss, then he can apply slightly higher flux 
density, but SLIGHTLY higher only, because the loss increases very 
rapidly with flux density.
The transformer designed by the above guideline would run at only 0.42mT 
at 30MHz, which would result in 127mW/cm3 of loss. So, designing for the 
lowest frequency is OK from the core loss point of view, as on the 
higher frequencies the core loss drops due to the lower flux density.

Now with the ferrite, I don't have the formula to calculate the loss at 
arbitrary frequencies and flux densities. If anyone has a link to some 
site that has it, please let me know! So I can only guess at this 
moment, which of course is not scientific, but at least might give an idea.
Power loss in cores comes mainly from two sources: Hysteresis effects, 
and eddy currents. Hysteresis loss increases mainly in direct proportion 
to the frequency, while eddy loss increases roughly with the square of 
frequency. For this reason, the rate at which powdered iron loss 
increases with frequency is something between linear and quadratic. RF 
Ferrites instead are essentially nonconductive and thus free from eddy 
losses, and for this reason I would expect their loss to increase only 
in direct proportion to the frequency! If this is true, then the loss at 
1.8MHz would be 40 times lower for the ferrite than for the powdered 
iron! And it means that we could drive the ferrite core to about 40mT of 
flux density at 1.8MHz, for the same core loss of 200mW/cm3! I stress 
again, this is based on an assumption, and I would gladly to the math 
again if anybody can provide a link to the relevant ferrite loss data. 
But my practical experience with ferrite and powdered iron confirms that 
ferrite does have very much lower loss, at given frequencies and flux 
densities.

Now what does this mean in practice? I will do the exercise of designing 
a plain simple common mode choke, aka current balun, for 50 Ohm, 160 to 
10 meters, for 1.5kW, using either #2 iron powder or type K ferrite. I 
will assume that the choke will see half the total line voltage across 
it, and that 90% choking action is the minimum acceptable.

Let's try first with T-200-2 cores. They have a cross section of 1.3cm2, 
a volume of 17cm3, and a specific inductance of 12nH/turns^2. For the 
given choking action at 1.8MHz, we need 22uH. That puts us at 42 turns, 
not very practical. Anyway, the core would run at 3.2mT, lower than 
necessary. But the loss in the wire would be huge, and the total wire 
length would be more than a half wavelength on 10 meters, which is not 
acceptable at all.

Let's try stacking three of these cores. Now 25 turns would be enough to 
get the required 22uH. The flux density is now lower, 1.8mT, and the 
wire length is still much too long for 10 meters, well over a quarter 
wave!

I tried a T-300A-2, a T-520-2, and several stacks, but always too many 
turns are required to give the desired inductance, which results in 
phasing problems on the higher bands. This material simply has too low 
permeability!

It is tempting then to compromise the low end, and use much lower 
inductance than really required. For example, let's try to wind that 
stack of three T-200-2 cores for the acceptable flux density. At 7mT at 
1.8MHz, we would end up with 6.4 (or rather 7) turns. The inductance 
would be 1.7uH, and the total wire length would be about three meters... 
still not good. But you can look on the web and buy baluns and balun 
kits that are just this! On 160 meters, choking action will be 
essentially nil. On 80 meters it's very poor, on 40 meters its decent. 
On 20 meters everything is fine, but above that the performance goes 
down again due to the excessive wire length. Still, this current-mode 
balun won't burn up, it won't cause interference, and the typical ham 
buying it will be happy, never noticing that he is getting decent 
performance only on the center bands!

Now let's try ferrite. Let's take a single F-240-K core. It will deliver 
22uH with just 11.4 turns on it. Let's make that 12 turns. The flux 
density will be 9.1mT, at which this core will run stone-cold due to the 
very low resulting loss. The total wire length will be just slightly 
over 1 meter, which is fine up to the 10 meter band. The result? A 
current mode balun, (or more correctly a common mode choke) that will 
perform beautifully from 160 to 10 meters.

And this is how it went in practice. All baluns (voltage and current 
type) I made on iron powder materials were problematic, showing good 
properties only over a few bands, sometimes only on a single band; 
instead, the ferrite baluns work very well over the whole MF-HF range, 
as long as the correct type of ferrite is chosen. I still have my 
T-200-2 cores here, as I never found a good use for them. In 
transformers higher permeability is needed, and for resonant inductors 
such as mobile antenna loading coils, an air-cored coil still performs 
better.


Concluding thoughts:

It seems that ferrite has gotten a bad reputation with some people, 
because they experimented with the wrong kinds of ferrite. I did those 
mistakes too in my younger and more desperate years. I remember a case 
when a balun wound on an old TV flyback core just plain exploded and 
send shards all over the place, when run at moderately high power on 10 
meters. I have also seen countless hams complaining about poor balun 
performance, when the poor guys bought commercial baluns wound on 
ferrite RODS!!! And there are many of them. Of course, no thinking 
person would wind a wideband transformer on an open core, such as a rod, 
when closed cores are available, but still these rod baluns are plentiful...

As a final remark on the suitability of powdered iron and ferrite for 
broadband HF high power transformers, I would just like to point out the 
following thing: Has anyone, and specially Carl, ever seen a 
professionally built power amplifier using powdered iron for the output 
transformer? Or has anyone seen a power combiner using powdered iron 
cores? I haven't. All I have seen use ferrite. There ought to be a 
reason for that...

Manfred.

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