Yes, the best single reference for transmission-line
transformers is Sevick. If you have the first
edition, though, get a newer one. Sevick himself
acknolwedged that he missed the most important class
of such transformers. For basic material properties,
the Fair-Rite catalog is the most comprehensive, but
its application notes do not contain enough
quantitative data to actually design a transformer. I
think the entire Fair-Rite catalog is available
online. The Amidon catalog, additionally, contains a
needed but rarely published equation for core heating
as a result of power dissipated. Those three items
together contain everything you need to know about
cores and transmission line transformers.
One of the most-quoted papers on transmission line
transformers was written by Ruthroff in the late
1950s. I believe the paper's title was "Some
Transmission Line Transformers". The paper contains
one of those wonderfully elegant and complete
mathematical analyses which produdes a complete
equation for the transformer's frequency response.
The low frequencies are dominated by the magnetic
loading, and the high frequencies are dominated by
transmission line length phenomenon. However, the
transformer which Ruthroff analyzed is one where the
incident voltage is caused to be replicated through a
1:1 transformer, and then the two are added together.
This is not significantly different than a standard
auto-transformer. The transformer experiences a
complete zero in transmission at the frequency where
the transmission line is 1/8 wavelength long. This is
because you get additive double-travel phase delays in
the primary and secondary. The ground node is common
to input and output.
If you add a second transmission line, whose purpose
is to delay the ground return by the same amount that
you delayed the "hot" signal, then you eliminate the
transmission zero at 1/8 wavelength completely. This
is often called an "equal delay" transformer.
The equal delay transformer was evaluated
significantly before the Ruthroff paper, in 1946
(within a few years) by Guanella. It is Guanella's
paper that Sevick neglected in his first edition of
the book.
When using Guannella-style transformers, as long as
you're using the correct impedance of transmission
line, the upper frequency limit of the transformer is
based purely on how much loss the transmission lines
themselves contribute. The ferrite loading is not
being used to couple energy from input to output. The
magnetization of the ferrite cores is due purely to
the current generated from the primary side of a
winding to the secondary side. If you can load the
transformer with enough ferrite to make this a very
high impedance, then the current levels are very low,
and you are generating very little magnetic field -
even at high power levels.
Sevick points out, correctly, that transformers are
not "transmission line transformers" just because you
wind them with transmission line (coax or twisted
pair). They become transmission line transformers
only if the incident signal actually drives the wires
in a transmission line mode, without needing for a
delayed version of a voltage to energize the "second"
conductor in a transmission line. If you're not
operating in transmission line mode, then you will
place high magnetic fields in the ferrites.
The equal delay transformers are inherently in
transmission line mode. As are 1:1 baluns constructed
by placing ferrites around the outside of a piece of
coaxial cable.
In a 20kW solid state amp that I helped develop some
years ago, the ferrite cores used in our final
combiner's transformer were only 0.6" diameter, 1"
long and had RG-142 (teflon version of RG-58) cable
passing through them. They barely got warm to the
touch at 20kW.
Most of the transformers used in Helge Granberg's
series of app notes for Motorola are Guanella designs,
but he does occasionally use "flux coupled"
non-transmission line transformers. The transformers
described in Krauss, Bostian and Raab are a mix of
Guanella designs and a push-pull feed choke, which is
a special version of a Ruthroff transformer that
facilitates feeding a push-pull amplifier, achiving a
1:4 step-up, and forcing balance at the same time.
Also, take a very careful and close look at the Q
curves (or, equivalently, the reactance/resistance
curves) for the ferrite/powdered iron material you
propose to use. At frequencies above 100kHz, it is
extremely rare to operate these materials anywhere
close to "saturation". The reason is that the linear
loss caused by the finite Q causes far too much
heating to even consider running them close to
saturation. That's the reason why you don't have to
worry about ferrite cores causing IMD. Linear losses
lone will prevent you from operating them anywhere
near saturation.
To determine the maximum power level at which you can
operate a particular material at RF, you must first
determine the power dissipated by the material. If
you know the inductance and Q (or reactance and
resistance), and you know how much RF voltage you'll
impress across the coil - or how much RF current
you'll feed through it - then you can calulate power
disspated in the resistance. The Amidon catalog
contains a relatively simple equation for predicting
the temperature rise of a core versus power
dissipated. It's based on the surface area of the
core and little else. If you use this technique to
determine the power level at which you can operate a
core, limiting it to, say, a 50 degree C rise, then
re-calculate the actual magnetic flux density you're
creating in the core, you'll discover that you're
probably limited to operating the core at a fraction
of a percent of the actual saturation value.
Another results of this is that, since you can't
operate these things anywhere close to saturation
anyway, the losses you measure at low levels with a
network analyzer ARE the losses that you'll experience
when the unit is operated at full power - at least,
expressed in dB, or percent. It doesn't change much
if you're feeding RF and DC at the same time, as long
as the DC doesn't move the material too close to
saturation.
With magnetic materials, the way you apply them has a
vast impact on what you can do with them and how they
perform. They're quite possibly the most complicated
of all the passive elements that we use, and the ones
most likely to be applied incorrectly.
73,
Dave W8NF
K5PRO wrote:
>The designs in Sevick's book and prior transmission
>line transformer
>design papers and books work well, as long as you pay
>attention to
>details such as winding the turns as tightly coupled
>transmission
>lines. Magnetic flux density in the core can be low,
>and the core
>itself extends the LF range. I built a 1:4 unun for a
>test setup at
>0.3 - 5 MHz at work, using some commonly available
>Fair Rite
>material, where I needed to drive 200 Ohms with a kW
>of CW power.
>Using a calibrated Werlatone directional coupler with
>Hp/Agilent 437
>power meters on the 50 ohm side, and capacitive
>divider and Pearson
>current transformer on the 200 Ohm side, i measured
>the loss in the
>transformer while under full drive. It was similar to
>what was
>measured with a network analyzer at a milliwatt. The
>transformer runs
>cool and does the matching that i built it for.
>BTW, Sevick wasn't the first to describe these types
>of transformers,
>as there are papers from the 1950s from Phillips,
>from Herb Krauss at
>Va Tech, in Solid State Radio Engineering by Krauss,
>Bostian and
>Raab, and in many other references. These people
>didn't just 'dry
>lab' data.
>73
>K5PRO
>John
__________________________________
Do you Yahoo!?
Yahoo! Calendar - Free online calendar with sync to Outlook(TM).
http://calendar.yahoo.com
|