开云体育

When designing a broadband transformer, there are several parameters
which can be adjusted for an optimized frequency response:

Increased coefficient of coupling, decreased leakage inductance, can be
achieved by placing the primary and secondary windings closer together.
While keeping winding to winding capacitance at a low or optimal value.

Ferrite core loss is reduced by increasing the primary winding
inductance. While keeping secondary inductance and wire length in range
for acceptable leakage inductance and bandwidth.

A set of compromises are demonstrated here:


Core size, stack of two FT114-43, and primary turn count, four turns, is
selected for a Ferrite core loss of about 0.3 dB in the HF band.

Two secondary ratios are shown. The secondary turn counts are selected
to maintain minimum leakage inductance.

The primary windings are interlaced within the secondary windings to
increase flux coupling between the primary and secondary. Increasing
coefficient of coupling.

The location of the primary winding with respect to the secondary
winding is selected for optimal winding capacitance. The capacitance
will aid in canceling the leakage inductance.

The frequency response with the above adjustments is very flat through
the 3.5 to 28 MHz HF band. The device enters resonance conditions above
about 40 MHz.

John KN5L

John, have you tried either of these with a real antenna? Most interested on 80M halfwave with the 4T/28T ratio version.
-Steve K1RF

Hi Steve,

I have not tried with an antenna. So far all bench measurements. It is a
balancing act, the 16 uH primary, 0.3 dB core loss, limits the ratio to
about 3k Ohm secondary while still keeping transformer reactance
controlled. High impedance ratios requires a lower primary inductance
with increased core loss.

John KN5L

Hi Steve,

Antenna measurement is added to:


John KN5L

John, thanks.
- Steve K1RF

Temperature rise measurement is appended to:


The temperature rise measurement was performed at 20W CW for two
minutes. The 2xFT114-43 4T primary Unun should be good for 50 CW or SSB
operation.

For perspective, a three turn primary FT140-43 Unun has about 0.5 dB
loss, which computes to 2.2 core dissipation at 20 watts. The FT140-43
3T is good for about 25W CW or SSB.

Assuming a 20 degree F temperature rise as the operational limit. The
measurement was performed in the open. There may be additional derating
required for a Unun mounted within a sealed enclosure. Even more if the
enclosure is black on a hot sunny 100F+ Texas day.

John KN5L

Added SimSmith loss models to:


Towards bottom of page is a SimSmith model using Fair-Rite complex
permeability file for computing transformer core loss. Using this model,
Unun core loss at 14 MHz is 0.34 dB.

Also added is an EZNEC model integrated with SimSmith which demonstrates
similar SWR curve as measured antenna.

John KN5L

John,

For matching a EFHW its important to know that while there are all manner of magic ratios
the actual feed point resistance is the only real one. That varies with height and orientation.

The other piece of magic is if you stick a stake in the sand can call it 2000 or 3000 ohms the
2:1 SWR circles are:
2000 nominal is 1000 to 4000 ohms
3000 nominal is 1500 to 6000 ohms

So even if we are wrong the range for OK is still rather wide. I tend to use lower values for antennas
like 80 and 160M as they tend to be close to the ground, and higher for 20M antennas. Its rare that
I see a SWR greater than 1.3:1 which suggests its likely around there somewhere and more than
adequately close. For longer multiband antennas the resonance is likely lower as you go up due
to total wire length having a lower impedance.

Why? All antennas that are horizontal exhibit impedance where ever they are fed, even in the
middle, that varies with height. I've included a often seen graph that can be worked for any
frequency and height.

An added note is that height affect total length for low antennas more than high or verticals
so a half wave can be something other than a numerical half wave due to height.

FYI measuring a real antenna is a bit tricky as 50 ohm gear gives wild results at 3Kohms. A simple bridge
with one leg of 5000 ohms (variable pot) and the antenna as the other can be easily used to measure
the actual under resonance. The RF source can be a few milliwatts of RF and the detector for the bridge
is a diode detector and a simple 50ua meter as the goal is a null. Then measure the port and you have
the real part in ohms to typically for most digital multimeter better than 3 places (maybe more but three
is more than enough).

Allison

Hi Allison,

I'm somewhat confused by your response.



A Unun ratio was selected for a specific test antenna. An EZNEC model,
which matches test antenna dimensions, is also included. Is there an
issue with test antenna measurement or EZNEC model somewhere?

Two Unun ratios are presented, demonstrated 56:1 and 49:1 for antennas
with lower impedance.

John KN5L

Eznec rarely get the ground effects right. You need Nec4.2 engine to model that correctly.
too many years of modeling in the office and then going out on the grass and finding out
they dont match other than generally.

If you didn't measure the ground for the frequency your using an educated guess.
I've found that create errors and also the myth of the magic combination when its specific
to a installation. Also it tends to vary with season (wet/dry).

Also did you actually test the unun with a real antenna?

As to the unun doing what you intend that's not in question. It looks good. What is
the questiondoes the wire really behave at 2450 or 2800 ohms (nec model) which
is less than 14% difference. What if the wire is really 3300 ohms obviously the
1:56 will be better but not perfect. Yet 3300/2450 is only a swr of 1.34:1 and an
acceptable match.

I also said even if your wrong, to get to 2:1 you need to be wrong by a factor of 2
and that's a huge miss. What I didn't add is if you tweak the wire length you can
get to an acceptable SWR in some cases (usually with a higher complex impedance
and more RF on the shield). The trick for endfeds is a transformer that is in the
middle of the likely range or best exact which is not easily done.

For testing I use a bridge, and other instruments or an adjustable L network
(the later can match the wire. Then I put a variable resistor from the load
port and ground and adjust the resistor for SWR of 1:1 and that was the
feed point resistance of the wire. Why do both? often I have to lower one
end to measure and the other I can use a adjust and elevate (step and repeat)
to see the difference and often height is a factor.

Its more for those that don't understand that an end fed wire like a dipole has the same
variation in feed point impedance with height above a real ground.

I've included a white paper on how to make a resistive SWR bridge. fairly stock design for 50 ohms
but if you want to measure a real antnena you may want the bridge to be adjustable to find
out the antennas actual impedance. the design is a GQRP design and R1 determines the
"characteristic" impedance. Making R1 variable (about 5k for EFHW) and using a RF source
you can both find the resonant frequency and the actual resistive portion of the antenna
at its feed point. Note R5 should be in series with a larger choke to get a RF resistance
of about 15k ohms. [its to get the metwe out of the RF circuit.] Failure to do that will have
1K across the antenna terminals by default and create errors. Power to test in in the less
than a 100 mW range and depending on meter sensitivity 10mW may be enough. First
set the R for about 2 to 3K then adjust the RF source for null response (very low indication)
then adjust the resistor to minimize the null while adjusting the RF for the best null. The
result will be the resistor is the load resistance and the null frequency is the actual
resonance of the wire. The pot needs to be non inductive.

Usually standing on a ladder doing that, less then fun if you like me and height adverse.
Reason is to get the antenna up and also a short length of coax. a decade of antenna
design work and testing left me with a large bag of tricks to get the data and confirm
the result.

The common Tayloe design using a led may work were the R1 resistor is adjustable but the
led only lights with a lot of RF and is insensitive for this use. A meter could be added for
better null sensitivity.

When measuring high impedance or low I use the HP4191A and a bridge like the
one described to get answers as accurate as the HP (to about 4 digits.). I also
have a 50mw DDS RF source as the way to drive measurements. A SA like the
Rigol 815T or any with tracking generator can be used to make that measurement
using a resistive bridge. Experience says better than 8357A PNA without a
calibrated matching transformer and math to get the real impedance (most PNAs
do poorly at high resistive loads without fixtures).


Allison

Hi Allison

On 5/20/19 8:24 PM, ajparent1/kb1gmx wrote:

Also did you actually test the unun with a real antenna?

Look under the title: "Antenna Measurement."

John KN5L

Hi John. That UNUN appears to have very good performance without the need for a capacitor across the primary as used in many commercial units. That implies to me that it has relatively low leakage inductances. It would be interesting to measure relative coupling from primary to secondary by measuring primary inductance of the UNUN with secondary open and then measure primary inductance again with the secondary shorted. From that, a lot of data can be derived.

Thanks,
Steve K1RF

Hi Steve,

This Unun design uses winding to winding capacitance for leakage
inductance compensation. Capacitance to leakage inductance ratio tuned
by moving primary windings within secondary windings, as shown in the
Unun photo. k increases as primary windings are physically closer to
secondary windings.

I have found it's almost impossible to use classic inductance
measurements for k measurements approaching one. My preferred method is
matching a SimSmith transformer model to measurements, as shown in the
web page under the two titles starting with "2 X FT114-43 4T:30T 3300Ω
Load SimSmith."



SimSmith model k for this Unun is .996.

John KN5L

Wow that k value is excellent. Can you confirm with open/short test just to
verify that actual measurement is in agreement with the Sim Smith model?
Thanks

Hi Steve,

I've tried measuring k near one using inductance measurements with
limited success.

Confirming SimSmith matches inductance measurements is a good question.



documents the process using a transformer with measurable k value.
Process compares SimSmith matching method, open short method, and
Series-adding Series-opposing method.

After years of using SimSmith, I'm rather confident with the transformer
model.

John KN5L

Also, open short method is for a conventional two winding transformer.
Unun demonstrated is an auto-transformer. It is impossible to fabricate
a matching two winding transformer with same k value as an auto
transformer. A matching two winding transformer will always have k value
slightly lower than an auto transformer. This is because an auto
transformer has a k value of one for the common winding. Replacing
common wire with two wires, to form a conventional transformer will pick
up k < 1 for the two wire replacement.

John KN5L

I wonder what are the losses of such EFHW UNUN if faced with real world conditions.

Let's say, I cut the end fed wire to length at my back yard, where the center of the wire is fixed to the top of a crappie pole. Then I go to the fields, where the ground has different conductivity, and I do not take the crappie pole with me, but I use the trees (lower or higher than the crappie pole). I may decide to hang the center of the end fed higher than the top of the crappie pole if there are such trees available to lower losses. I may hang not just the center, but also the far end of the wire as high as possible to lower ground losses. Certainly the complex impedance at the end of the EFHW will change and I would bet my shoes that it will be out of the 1:2 SWR range of my back yard setup the wire has been cut to and the transformation rate of the UNUN was tuned to.

I may decide to use your end fed UNUN design to precondition the antenna to an Elecraft T1 auto tuner or similar. What will be the losses in your UNUN then? I may model couple of EFHW installations with NEC2 to get the end fed impedances to match.

Wouldn't be wiser to use a resonated EFHW transformer tuner with a tuning capacitor and taps? This way the tank could be tuned slightly off resonance to provide the required inductance or capacitance.

Or wouldn't be more efficient to just use a high Q LC tuner? Even just providing a high Q coil external to an Elecraft T1 auto tuner should lower the losses of the LC circuit significantly.

By the way, I spent quite some time toying with 4nec2. What I found out (and I just hope I got it right) when modeling a 40m antenna using 7m crappie poles as support is following:

1) Using two 7m crappie poles for an 40m antenna is always better than one. Two crappie poles allow one to bring more wire further from the ground, thus minimizing ground losses. The modeling shows that one can achieve roughly half the losses with an antenna over two poles than over a single pole. The losses below contain the ground losses, as the efficiency numbers are derived from the radiated power accumulated at the far field. Most of the losses of an inverted V antenna are due to the close proximity of the ends of the wire towards the ground.

2) A full size dipole between two 7m crappie poles reaches radiated efficiency above 60% over an average ground, but the full size dipole may be too long and it may be too heavy for two 7m crappie poles, as the tips of the crappie poles are very thin and they will bend. Bringing the two 7m poles to just a 7m distance (quite a difference compared to 20m distance for the full scale dipole) and hanging the ends of the dipole along the crappie poles leads to a radiated efficiency of around 40%. The trick here is to keep the ends of the dipole around 2 meters above the ground to minimize ground losses. The hanging ends act as capacitive hats, thus increasing antenna standing wave current between the two poles high and keeping the dipole impedance reasonable, so it could be matched with an ATU.

3) Very similar wire topology (a dipole over two 7m crappie poles at 7m distance, with one end hanging down along one crappie pole as in the previous case up to 2m above the ground, while the other end going to the ground and end fed offers a radiated efficiency of around 37%, very close to the center dipole in the previous case. Now contrary to the popular belief I think the key to the end fed efficiency without very dense ground system is not the high end fed impedance, but the horizontal polarization of the antenna, which minimizes ground losses. The more of the end fed antenna wire is vertical, the more vertical polarized the radiated signal will be and the more the antenna will be sensitive to a good ground, making it not much different from any other vertical.

3) A T antenna strung between the two crappie poles (wire between the two pole tips, a vertical strung from the center of the horizontal wire) and with two radials strung at the height of 1 meter between the two crappie poles, makes quite a nice 40m vertical with around 20% radiated efficiency and reasonable impedance to be tuned with an ATU. Again the two crappie poles are cheap and they will make a much better vertical antenna than any other single crappie pole exercise with loading coils or top hats.

I am looking forward to a fruitful discussion. The 7m crappie poles here in Prague are around $17 at the Decathlon sports equipment chain. They are not very rigid at the tip, so it does not really make much difference whether they are guyed 1m, 2m or 5m above the ground, 1m above the ground is just fine. I have seen a lot of discussion on EFHW antennas on a single pole, but not much on field antennas on multiple crappie poles.

73, Vojtech OK1IAK, AB2ZA

Hi Vojtech,

An EFHW antenna with high impedance, fixed ratio, Ferrite transformer
Unun is but one antenna solution. As you mentioned, deploying portable
antennas may result with varying antenna resonant solutions. A fixed
ration Unun EFHW antenna can be tuned by adjusting length of the
antenna. Length can be adjusted by looping free end over to shorten
antenna wire. Should be a two step process, measure resonant frequency,
compute length ratio for desired frequency, adjust length.

A tuned Unun is handy if the Unun is located at a convenient location
for tuning. Down side is that tuned Unun convenient height suggests
lower antenna resulting with possible antenna pattern or gain compromises.

Powdered Iron core tuned transformer Unun will have lower loss than
Ferrite core transformer Unun.

On 6/16/19 2:44 AM, Vojtech Bubnik wrote:

I wonder what are the losses of such EFHW UNUN if faced with real world conditions.

Ferrite transformer Unun loss is dominated by Ferrite core loss and can
be represented with parallel resistance.

SimSmith plot under title "2 X FT114-43 4T:30T 3300Ω Load SimSmith
Approximation With Fair-Rite Complex Permeability Model"
R1, 677.9 Ohm at 14 MHz, represents ferrite core loss. If match
impedance is reduced, ferrite core loss will decrease, if match loss is
increased, Unun core loss will increase. The assumption though is
antenna being deployed as a resonant match EFHW, not an End Fed Random
Wire antenna.

John KN5L

In line with the end fed theme, but done differently...

Mono band 10M EFHW about 16.3ft long (4.98m) fed with a L network.

L network 2uh air wound for power handling and the cap is a roughly
6" of PTFE coax (RG316). Wire for the element #22 stranded insulated.

Tune up was with 2500 ohm resistor on the bench and MFJ269B
to get the cap (rg316 coax) to the right capacitance (1:1 match).
then the resistor is removed and wire installed and outdoors the
wire is then adjusted for 1:1 match at 28.4mhz.

It will be used as a vertical (hanging) 2:1 bandwidth is below 28 to
above 29mhz. Since it is short it can also be suspended using a
single 20Ft z(6M) Breem/Crappy pole.

Testing concluded with FT817 and a few SSB contacts on 10M in 4 land
(Florida, Georga).

Monoband as its important to have a clean pattern for RF on the horizon
when suspended vertically. An air core inductor was chosen as lower loss
and adequate power handling at 100W. At 2uh it is a relatively small coil
with low weight. The end result is a small 4x2x1 inch plastic box with
SO239 connector and a #8 stud to connect the antenna. The finished
antenna gets the wire coiled up around the box for storage. The #22
PVC insulated hookup wire works well for two reasons I had it and it is
not being subject to any great weight.

Use will be for the 10M station at Field Day and to generally have handy.

Allison

John,

For the 14mhz model, try using mu 125 material (FT240-61)
as you getting near the point with hard ferrites (31 and 43)
where their absorption properties are becoming significant.

An image borrowed from wa7ark, Mike is something I find more
important. The antenna is a 80-10 End fed in harmonic modes.
The transformer losses are less relevant when you look at the
patterns as its not whats the loss in the transformer but where
is the RF going? The broadside 80m pattern is an end fire
20m pattern with broadside nulls.

Allison