Baluns - bifilar, trifilar et al
Graham Seale
It is a recipe for making effective broadband "transmission line" type
baluns. It contains some constructional tips, and a qualitative rationale
intended to give those who would normally seek a commercial device some
confidence to make their own.
I suspect that I could upload it to a suitable place on the Demon site
where it could live awhile, unsullied by the multiquote mayhem that you
chaps and your text editors seem capable of; but I am new to this mode,
and anyway, there may be those out there who would contribute to give the
recipe the benefit of their accumulated wisdom. I only ask that any
contributors avoid requoting texts that would involve messing up the ASCII
graphics as it gets pushed to the right to the point of wrapping.
Please forgive that I have left in the provincial style in conversation
with AA2UV. Life is too short to be polishing text when I know you guys
don't care much how it reads - so long as it works! We can rework it later
when the theme has stabilised.
___________________________________________________
>>>>>>>>>>>quote
From: Graham Seale <gra...@southlin.demon.co.uk>
To: r...@soho.ios.com
Subject: 9:1 Baluns
Reply-to: gra...@southlin.demon.co.uk
Date: (snip)
(more snips)
Bobs Balun:-
This recipe starts out for you Bob, but I have hopes that others at my
radio club can use it . If you like it, and think others may find it
useful, I may post it more widely. I am assuming you really want a 9:1
balun. I strongly suspect you may also want to have the recipe for other
ratios, but the good news is that a simple reconnect of the ends gives you
a choice to have 1:1, and a two-wire variant gives 4:1.
There is also a *good* technical rationale that goes with this design,
but I will try to just stick to the recipe this time, except to say that
this is of the class known as "Broadband Transmission Line Transformers",
and in this design, the ferrite cannot saturate even at high powers (kW!).
There are *some* things you need to know before invoking this device.
1. They work excellently transforming between impedances already matched
to be nearly resistive. If you operate with an unbalanced matcher with
a balanced antenna feeder on the other side leading to an indifferent
multibander (eg G5RV) then to simply QSY can lead to trouble. Despite
the caution, I have done exactly that, abusing them in all ways
imaginable - and they performed well.
2. There has been only one occaision when I have used the 9:1 connection
and it was in nearly ideal conditions. Be very sure you really need
such a high ratio!
3. These baluns will work excellently on 20m, 15m and 10m without ANY
ferrite rod in them. The coupling is is by the "transmission line
effect" of the capacitively coupled twisted conductors. As we get
to lower frequencies, (40m) the capacitive effect is less, and you
start to need the ferrite. The coupling is then boosted by the magnetic
(true transformer!) type coupling. This is why they are "broadband
devices. At 80m you are working a transformer only, and at 160m, you
need more turns (say up to 16).
4. I have found with a 6-turn balun we were pushing our luck at 80m,
but it tested as OK up to 65MHz. This is essentially an HF device.
If you want VHF and UHF devices, there are other tricks.
80m thru 6m is easy.
______________________________________________
Ingredients:
^^^^^^^^^^^^
3 pieces of enamelled copper magnet wire. AWG #16 (insulated dia 1.4mm)
is as thick as is practical. Go as thick as you can with good insulaton.
Help the insulation with PTFE plumbers tape or polyester transformer tape
if you like. PVC tape is too thick. For length about 50cm. (err..18inches)
is enough.
(Switch on inches mode!)
1 piece of ferrite about 1/2inch diameter. About 2 inches is enough
for a balun. This actual size is no big deal. One can adapt bits of
ferrite from old TV line transformers, or get rods for not many cents
from adverts in QST, or from Radio Shack etc. If its 3/4 inch, then
go ahead and use it. 1/4 inch is too small! To neatly snap it I score all
around with a warding file, then pad the wanted end with card and gently
secure it in a vise. A towel over the rest and a sweeping hammer blow to
shear it and it cuts off clean and brittle.
Heat shrink tubing + sundry scrape, bend + cut stuff.
The Method:
^^^^^^^^^^^
1. Temporarily clamp all three together at one end, and twist them
together carefully. Allow the free ends to "untwist" as you go. Dont clamp
both ends. You end up with a pretty 3-wire helix. Don't overtwist - 2 per
inch is enough. Put the heatshrink over and shrink it on... (paintstrip
heatgun?)
2. Now use the trifilar set as if it is a single thick wire. Use anything
you can find that is a bit smaller than the ferrite rod, as a mandrel. A
piece of dowell, an old felt marker pen body, whatever. Wind on more turns
than you need allowing that there has to be "ends" to connect to.
3. The turns will "spring out", hopefully to fit snugly onto the ferrite
rod if untwisted slightly to offer the rod, so it grips when you let go.
Unwind a couple of inches, using pointy pliers to kink the trifilar so it
comes away from the rod tidily. Count the 8 turns, then similarly bend the
trifilar to come away from the rod. Tie it to the rod with fishing line or
twine if you like. Identify the starts and finishes of each wire (easy if
they are different colours). You end up with something like the sketch
below. (switch on ASCII art mode!)
Trifilar means three twisted together and taped or put in heatshrink!
Sometimes, for bifilar, you might not bother to twist them.
|<--- 8 turns -->|
|__ __ __ _ |
______/\ \_\ \_\ \_\_\_\________
/ \ \ \ \ \ \ \ \ \ \ \
/ \ \ \ \ \ \ \ \ \ \ \
| | | | | | | | | | | |
| | | | | | |<----- 1/2" dia approx
| | | | | | | ferrite rod.
\ / / / / / / / / / / / About 2" long
\____/___/ /_/ /_/ /_/_/_/_______/
Triple | | /_/ / /_/ | |
set of wires --- |_| |_|
in heatshrink (_) (_) s = starts
or taped /|\ /|\
s1/ | \s3 f1/ | \f3 f = finishes
/ s2 \ / f2 \
Now it only remains to connect them up. To understand this bit we go to
the circuit for a spell.
The Circuit:
^^^^^^^^^^^^^
We connect all three wires into a series. The finish of one links to the start
of the next. The Unbalanced coaxial bit is put across one winding, with the
grounded braid bit connected right in the centre of the series - well isolated
from any balanced conductor. The balanced winding is across all three in series.
The turns ratio is 3:1. The impedance ratio is 9:1
Trifilar Wound
UnBal
Side ____________________________-->
|
It "shares" the |s1 _|s2 _|s3
middle part of the ) / ) / ) To Balanced side
winding ) / ) / ) Hi Z = 9 x Lo Z
) / ) / ) (Turns 3:1)
)__/ )__/ )
|f1 |f2 |f3
| | |___________-->
_________ | |
(low Z) _________(--^ | Note how balanced side straddles
T___________| three windings in series
_________________________________________________________________________
Back to practicals:
^^^^^^^^^^^^^^^^^^^
Look at the lower part of the sketch
\ / / / / / / / / / / /
\____/___/ /_/ /_/ /_/ /_/_______/
Triple | | /_/ /_/ | |
set of wires --- |_| |_|
in heatshrink (_) (_) s = starts
or taped /|\ /|\
s1/ | \s3 f1/ | \f3 f = finishes
/ s2 \ / f2 \
| |___|_____| | |________ --->
| |__T______| BALanced line
|__________|__T________________ --->
| |
| |
________ | | 9:1 Connection
Coax Unbal_______(--^ |
T_____|
The wire is good and stiff, and can be formed and cut. You need a file to
scrape the ends clear of enamel. Form the starts and finishes to fully
overlap, then bind them with some strands stripped out of regular electric
flex and solder. You can mount the whole thing across a dipoles centre
on a plastic strain relief, the whole thing suspended in a lightweight
used plastic Cola bottle hung upside down. (If you look, they have a stiff
lower part - easy to cut off, and refit with overlap + sealed with tape.
You *can* do a real pro job mounting the rod on P-Clips inside a metal
box, complete with coax socket, bal side insulation panel and sprung
push-grip terminal connect posts. Thats all bells and whistles. The core
functional part is as good as from any factory, and because it is *not*
a toroid, it cannot saturate to cause heating, distortion, bandspreading
etc. There you have it - a 9:1 Balun.
_______________________________________________________________________
"Other" Ratios - The "1:1" Connection
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
If you have come this far, it would be insane not to go the extra inch
and appreciate that a very *minor* change to the connections gives the
1:1 configuration. For 1:1. you have both the balanced AND unbalanced
sides each having the same number of turns. Since there are only three
windings, one is shared thus:-
Trifilar Wound - 1:1 Connection
UnBal
Side ___ ____________________-->
| | |
Only the middle | |s1 _|s2 _|s3
winding s2-f2 is | ) / ) / ) To Balanced side
shared | ) / ) / ) Bal Z = 1 x UnBal Z
| ) / ) / ) (Turns 1:1)
| )__/ )__/ )
| f1 |f2 |f3
| | |___________-->
_________ | |
(UnBal Z) _________(--^ | Note how both sides straddle
T______________| TWO windings in series
The 4:1 Connection
^^^^^^^^^^^^^^^^^^
The impedance ratio you want is the square of the turns ratio. The 4:1
has turns ratio 2:1. This requires only two winding wires. It *can* be
done by simply ignoring the third wire, but you get a much easier build
if you make it with two wires as a dedicated 4:1 version - like this...
Bifilar Wound
UnBal ____ ______________-->
Side | s| _s
_____ | ) / ) Balanced side current goes thru
(lo-Z) _____(--^ ) / ) 2 windings - hence 4:1
T ) / )
|________|__/ |______<-- s = start
f f f = finish
Yet more Ratios - "Non-Standard" ones
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Once you realise the principle that the unbalanced outer braid return is
always connected to a point in the middle of the windings, "stood off"
from the balanced connections, then we have the possibility of a portion
of winding inserted almost anywhere to give us non-standard turns ratios.
It can appear as a single conductor overwind, or perhaps a few turns extra
coming out of one end before going back to the beginning and resuming bifilar.
This latter trick is best seen with an example. Suppose we wanted 6:1.
^^^^^^^^^^^^^^^^^^^^^^
Consider the ratios:- Turns Ratio Impedance Ratio
2:1 4:1
3:1 9:1
3:2 9:4 = (2.25):1
4:3 16:9 = (1.78):1
5:2 25:4 = (6.25):1
Nice!... note that a 5:2 ratio gives (6.25). Scale this up to 10:4, and more,
to 20:8 (thinking you need sufficient turns make the self-inductance, and hence
unloaded impedance high compared to the cable). Too many turns will introduce
high frequency limitations. Too few and you cannot couple efficiently. If you
must use toroids, then you don't get fractional turns. (6.25):1 would have to
do! This would be good enough for me, but if you are picky, then use a rod
wound with (19.6) vs 8 turns, giving an impedance ratio of (6.0025):1.
The two windings in series gives 16 - you only need 3.6 extra turns for 19.6
The physical realisation of this balun is:-
Bifilar Wound
UnBal ____ ______________-->
Side | | _
_____ | ) / ) Balanced side current goes thru
(lo-Z) _____(--^ ) / ) 2 windings - "plus a bit" - hence
T 8 Turns) / )8 Turns 6:1
|________|__/ |_end bifilar
|
Extra 3.6 Turns (____ <--
Essentially you wind an 8 turn bifilar onto the rod, then give the balanced
winding an extra 2.6 (aw three-and-a-half!) turns on one end.
*** End of digression about "other" ratios ***
_________________________________________________________________________
One last point. I have never yet come across a piece of unidentified
ferrite rod that was not up to the job. Such "low frequncy only" stuff
does exist, so take care! Its nice to know the grade in use will work over
the required frequency range, but if you don't know - make the balun anyway.
Thats it - a long(ish) epistle. I tried to keep it to the point, but
I recognise some guys who read it may need some detail. Its my best shot
anyway, and I hope it helps.
73s G4WNT
--
Graham Seale