(Compiled from the TopBand Reflector and private communications
Opinions expressed don't always agree) 

note: Cick here to skip discussion and read my recommendation


Comments from VK3APN 9/1998

Thanks to everyone for the feedback regarding antenna transformers,
which is very encouraging. Several people have since asked about
transformers for Beverages, and since the answer is probably of general
interest, I'll post it here.

Why Beverages need good transformers
The requirements for a Beverage transformer are less onerous than for a
pennant or flag, because the antenna already operates in the unbalanced
mode. Therefore there is no other mode of operation to suppress, in
which the antenna might receive signals from all directions, and fill in
the pattern nulls.

However good common-mode isolation between the feedline and antenna
still pays dividends. Once again Larry W7IUV's observations are
pertinent:

   "After investigating, I came to the conclusion that I had a ground
loop occuring in the coax shield. Using the same transformer
construction as I did for the pennant, I floated  the shield connection
at the antenna end of the line. i.e.  coax goes to the primary, antenna
and ground to the secondary,  no local ground to the coax.

    This did not help the F/B of my poor unterminated buried Beverages a
bit, but it did reduce the power line noise and computer birdies a
bunch!"

Over the last couple of years, I have found on my own antenna (a
bandswitched elevated vertical), that a tremendous amount of local
electrical noise can be coupled into an antenna via the outer conductor
of the coax. This is despite baluns (ununs) at both ends of the
feedline, several clip-on ferrite suppressers, a separate ground stake
and thick ground lead at the antenna, and the shack equipment all
shielded and bonded to earth.

Now consider a low output antenna like a Beverage; just how much more
susceptible it will be to conducted noise! If a high output antenna like
a vertical is affected, a Beverage will be history unless common-mode
noise is eliminated.

I believe baluns and clip-on ferrites are ineffective simply because so
much inductance is needed at low frequencies. If one is unlucky, they
could even turn the coax into a resonant length, making the problem
worse. What is needed is complete isolation between the antenna system
and the coax/shack system, and a transformer with a separate primary and
secondary is ideal. To maintain this isolation, as Larry says it is
essential to NOT ground the outer conductor of the coax at the antenna,
but leave it floating.

Good isolation would probably benefit most low-band antennas, not just
Beverages. If the antenna is also used for transmitting, tuned link
coupling at the antenna should be seriously considered, due to the
limited power handling capability of available ferrite cores (remember
we are talking about normal transformers, not transmission line types).
This is definitely one of my next projects.

Transformer construction
 The highest permeability ferrite I would use for a Beverage transformer
is #43, with a permeability of 850. This is actually very high for a
Nickel-Zinc ferrite, most of which are around 50-250. It will work well
at 1.8 MHz, and should be usable at 3.5 MHz. The Q=1 frequency is 5.5
MHz.

For a 50:450 ohm transformer, the following turns are recommended:

   Core     50 ohm      450 ohm
   Type     winding     winding
 ------------------------------
 FT114-43      9          27
 FT140-43      8          24

The shunt reactance will be +j550 and +j690 ohms respectively for the
two types.

Materials
I know people often lean towards high permeability ferrites (>1000),
even at HF. Sometimes such ferrites work but often they don't,
contributing to the aura of mystery. The recent discussions about the
pros and cons of u=10,000 cores is a good example. We tend to forget
that the Q1 and Q2 materials, which remain so popular amongst amateurs,
were amongst the first ever ferrites to be commercially released. The
technology has since moved way beyond these early types, and I can't
think of any reason why one would want to use such old fashioned
materials these days, in preference to more modern ones.

There's really no mystery, and it's worth taking the trouble to select
the right ferrite. I strongly recommend the Amidon book, which apart
from a comprehensive set of graphs, has some good applications
information as well.

(A Postscript via email after I asked Peter about 7Mhz operation with #43):
Because it is a 'conventional' transformer, not
transmission line, there is a significant tradeoff between coupling and core
loss. A higher permeability would provide better coupling, but these grades are
also more lossy at the higher frequencies. Without doing tests, it is hard to
predict whether coupling or core loss would dominate at 7 MHz. My off-the-cuff
guess is you probably wouldn't notice much difference at 7 MHz, by changing to a
lower permeability.

However the grade definitely affects 1.8 MHz, where the coupling loss is already
a couple of dB with #43. A lower permeability grade would increase this still
further. At 1.8, given the low output level, you really don't want any
unneccessary loss. I'd stay with #43.


Comment from W1ZK

The AMIDON FT-114-75 works perfectly for 9:1 beverage transformers.Use 
ON4UN's Table 7-3 (page 7-13) for winding details. Ref.#4 in Table 7-4 is 
a misprint and should read FT-114-75 not as printed FT-144-75 (no such 
animal). If you have a Autek RF Analyst or MFJ antenna analyzer just 
attach a 500 (+ or - 50 ohms)resistor to the output of the transformer 
and the analyzer to the 50 ohms winding and you should read darned close 
to a 1:1 match. Be sure to use noninductive resistors for the termination. 
You can make them using parallel/series out of carbon 1 watt  resistors. 
Don't worry about exact tolerances, 450 - 550 final resistance is fine. 
Don't use potentiometers of any kind for the termination. It is guaranteed 
that if your the power from your transmitting antenna doesn't burn them out 
close lighting strikes will!

If you have that antenna analyzer, the final SWR on a full wave will be close 
to 1:1. 


Comments from K0FF 9/16/00 & 9/24/00

Well with the FT-140 J's that were on hand 1.4" dia x .9" tall and u=5000), I4JMY's 
recommendations were darned close. I wound up liking a 3T primary and a 7T secondary 
for the 50 to 470 Ohm Bev transformers and 3T/11T for the 75/900 for the Pennant. Both 
worked well on the lashed up test setup that I threw together this morning. 

In order to match the 470 Ohm Beverage antennas to coax, I use a conventional donut 
torroid, set up for 470/50 or 470/75 Ohms. When using the type 75 core material, 
u=5000, it can be done easily with 7t or 9t for 50 Ohms or 75 Ohms respectively. The 
windings can either be 3 or 4 wires depending on how you connect them. I happen to 
use 4 wires, and "waste" one, but that gives the option of floating the shield on 
the coax. See fig. 2 The 7 or 9 turns gives a good match thru 160 on the low end and 
around 9 MHz on the top. If it is desirable to also use the same antennas for BCB 
(as a random wire), you must have a lower bottom end than the 7 or 9 t will give. By 
increasing the number of turns by 1.414, the low end is halved. Increasing that number 
again by 1.414 will halve it again. The results for 50 or 75 come out to :
( 7 x 1.414) x 1.414 = 14 t and (9 x 1.414) x 1.414= 18 t. The 16 turns is a 
compromise between the two which gives good performance for either. As far as the top 
end frequency, there is at least 3 octaves of useful range in a transformer like this, 
more likely 4 or 5 octaves, so the top end still comfortably covers 80 and 75 Meters. 
40 M use of a Beverage is a moot point for me, since when in the HAM mode, there must 
necessarily be a BCB filter in line, and it cuts of before 40M anyway. The same 
quadrafilar windings can be hooked up several ways to give different ratios, and if 
you leave the taps exposed, or even centertap the top wire, you will have an even 
wider range of choices.


Comments from W8JI

Remember the ui of a core is expressed at dc, and changes with frequency.

Always look at the impedance slope of the core at radio frequencies, and pick a core
material that has the highest impedance at the lowest operating frequency desired!

Highest impedance often occurs where the loss tangent is sloping upwards sharply, and 
the reactance hasn't fallen off yet.

For example, a 10000 ui core will sometimes have less impedance than a 2500 ui core at 2 MHz.

It's best to check or look up the exact core selected, rather than selecting only by dc 
parameters. Good goal is to set the no-load winding impedance at the lowest frequency at 
about five times the RF impedance across that winding, although less can often be used.

Reflection transformers and high impedance transformers like for EWE's and Flags (etc) are 
the most critical.

The core Misek suggests (dc ui~10000) is actually not the correct core for 160 anyway. 
Ui values should be picked at the lowest operating frequency, never at dc! The dc ui 10, 000 
core has less permeability on 160 meters than the following core materials:

77 ui 1800, 73 ui 800, 43 ui 900.  

All of these materials are suitable for broadband transformers. They have better RF performance
and are more common. If you select a 73 or 43 material, you can use about 40% less turns on the
transformer (even fewer with the 77 core) for the same stack length. Fewer turns reduces unwanted
coupling from primary to secondary, and they saturate at high signal levels. They also work much
better at higher frequencies, making the Beverage very useful on 7 MHz and higher.

Ameritron has a binocular core 412-5250  (73 mix) for $.60 each. A two turn primary is more than
enough for 50 ohms at 160 meters, and the core saturates at about 30 watts. Alternately, a stack of
FB73-601 beads can be substituted. I did use beads, but now switched to the free (eerrr $.60)
412-5250 cores. I use #24 wire, and have yet to lose a core during lightning despite losing dozens
of two watt resistors on my 11 Beverages. 

I use a .5x.5x1 inch #73 binocular core wound with teflon insulated number 18 or 20 wire. If you use 
the correct core, it only takes about two turns on the primary and five on the secondary. I also keep 
the primary and secondary totally isolated, which will help with lightning and common-mode noise 
ingress from the feedline. I cut fine cuts in a PC board to isolate pads for wiring, and use the cuts 
as spark gaps. (6/27/02)

As for spacing the windings or using a Faraday shield, neither are necessary or useful. I have typically 
measured about 10pF or less of capacitance with one winding laid directly over the other. That is more 
than 9k-ohms of leakage reactance, and that would easily put any common mode well into the Beverage's 
noise floor.

If you are worried about feedline pickup despite using an isolated ground for the antenna, you'd be far 
better off to add a ground rod several feet from the Beverage ground; grounding the feedline shield to 
that separate ground while just using a winding-over-a-winding type transformer. A conventional style 
transformer would have less loss and better bandwidth. (7/16/02)

Terminate the transformer with a resistor twice the normal
secondary resistance and measure the SWR.  Repeat with 1/2 the normal
resistance.  The SWR should be 2.0 ideally.  Multiply the
two SWR reading together and take the square root, if it comes out
close to 2, you have a pretty good transformer.

W8JI on Reflection Xfmrs:

Basically you need either an autotransformer or a dual winding transformer with one
winding center-tapped on the far end. I used the dual winding in my Beverages, because
I could use the same transformer in multiple uses. Either one works just as well. It's
too long to post, you'll have to read the articles. By the way, pick a core material
with an impedance that peaks on 160 meters!!! If you use too high an initial permeability,
the core impedance will peak far below 160 and you'll need to use more turns!!! 

Most ferrite cores in the USA actually come from Fair-Rite Corp in Wallkill NY at 
(914) 895-2055. 73 Material has the impedance peak at about 2 MHz, while higher ui 
cores (initial ui is a dc value, not the actual ui at the operating frequency) have 
peaks lower in frequency. That means a higher ui core than 73 can actually require 
MORE turns for satisfactory impedance, not less as might be expected. Many people only 
consider the dc permeability when selecting cores. Binocular cores have higher 
impedance for a given length of winding through the core, and give better overall 
performance. I have a large stock of 73 mix binocular cores available is anyone gets 
"stuck" and can't find a core.  

Remember the ui of a core is expressed at dc, and changes with frequency. Always look 
at the impedance slope of the core at radio frequencies, and pick a core material that 
has the highest impedance at the lowest operating frequency desired! Highest impedance 
often occurs where the loss tangent is sloping upwards sharply, and the reactance 
hasn't fallen off yet. For example, a 10000 ui core will sometimes have less impedance 
than a 2500 ui core at 2 MHz. It's best to check or look up the exact core selected, 
rather than selecting only by dc parameters. Good goal is to set the no-load winding 
impedance at the lowest frequency at about five times the RF impedance across that 
winding, although less can often be used. Reflection transformers and high impedance 
transformers like for EWE's and Flags (etc) are the most critical. 


N1EU Postscript:
Over the past four years, I've probably wound 8 different transformers of my own and tried 
two commercial transformers.  This includes designs from ON4UN and Misek.  I finally found 
a design I am very pleased with, and it happens to be the easiest of all to wind.  It is 
based on information from W8JI and W7IUV.  I wound it on an  Amidon BN 73-202 binocular 
core with 26AWG enamelled wire - 2 turns for the low impedance feed with leads coming out
one side of the core and wound over it, 6 turns for the high impedance Beverage wire with
leads coming out the opposite side. In terms of signal level and SWR flatness, this transformer 
was the best of the lot and can be built for under a buck!  Add $4 for a RS case and coax 
connector.  I used brass machine screws and nuts to make wire terminals. 


 
Here is the mathematical derivation of the transformer windings:

The reactance of the windings should be at least four times the impedance the winding is designed 
to look into.  So if you have a 450 impedance, 4x450=1800 ohms of reactance. Using 1.8Mhz as your  
minimum frequency, the inductance would need to be L = 1800 / 2 x pi x 1.8Mhz. So L= .000159 henrys 
or .159mh.  To find the number of turns the formula is N=1000(sqrt(L in mh/A sub L)) Or 
N=1000(sqrt(.159/8500))= 5.24 turns on the primary. To find the number of turns on the secondary 
use Np/Ns = sqrt(Zp/Zs) or Np/Ns = sqrt(450/50) = 3, this is the ratio of primary to secondary  
turns. So Ns = 5.24/3 = 1.75 turns on the secondary.  (NOTE) Np=number of turns on the primary, 
Ns= number of turns on the secondary Zp=impedance of primary, Zs= impedance of secondary   So a 6 
turn primary and 2 turn secondary is ideal.  


 
For two wire Beverages, wind reflection transformers on the BN 73-202 core with 4 turns primary
and 2 bifilar turns for the center-tapped secondary (1:1).  For the RF Choke needed in the 
switching box, just use 10 turns on a 73-202 core.