Hi Folks,

At the time of this posting we had the following votes:

#1: 2 #2: 7 #3: 3 #4 : 1 #5: 1

There's some element of truth in each of the answers (even #5), but as the "judge & final arbiter" I'm declaring Answer#3 as the only fully correct statement

Pressing into service a non-trapped, single-element vertical on several bands seems to be a relatively recent phenomena - I don't recall it being used widely years ago. It has the virtue of simplicity, but it also has some disadvantages as we shall see.

Here's my reasoning to support Answer#3:

------------------------------------------------------------------------------------------------------

"Brain Hurt" level: *

The presence or absence of the UnUn cannot affect the radiation pattern of the vertical, so we can ignore that as a factor, although we may return to it in another Puzzle. Nor can it affect the feedpoint impedance at the base of the vertical, and hence the ground losses. What the UnUn does affect is the impedance "seen" by the coax, and therefore the coax losses and the tuner losses; and of course the UnUn itself may be lossy.

Firstly let's look at the situation if we don't include an UnUn: We can model 34ft of vertical wire over average ground, include some modest ground losses (10ohms) to allow for the radial system not being perfect, and check the feedpoint impedance on every band 160m thru 10m. As expected we will get a very wide range of impedances from 12-j1250ohms on 160m where the vertical is electrically very short, through 47+j0ohms on 40m where it is a resonant quarter-wave, to 2651+j0ohms on 20m where it is a resonant half-wave.

[Don't be put off by the "j" stuff! It's just a shorthand for separating the resistive and reactive components of the impedance. So 12-j1150 is the equivalent of a 12ohm resistor in series with a capacitor with 1150ohms reactance. A plus sign in front of the "j" would indicate an inductor.]

Of course these widely varying impedances cause widely varying VSWRs on the coax. Here's the plot from EZNEC:

The vertical dashed lines indicate the various Amateur Bands. Notice how the VSWR is a very low 1.1:1 on 40m, and also below 1.5:1 on 15m, but on all other bands it is over 10:1. If we now calculate how much loss we incur in 100ft of RG213 for the different feedpoint impedances, we get:

160m 20dB, 80m 11dB, 40m 0.5dB, 30m 4.4dB, 20m 7.4dB, 17m 6.9dB, 15m 1dB, 12m 5.4dB, 10m 7.5dB

Now let's introduce a 9:1 UnUn and see what happens:

Notice that we have ruined the match on 40m and 15m, but we have improved the match to a greater or lesser extent on the other bands. Here are the new coax losses:

160m 20dB, 80m 5.5dB, 40m 2.2, 30m 1.5dB, 20m 1.8dB, 17m 2.8dB, 15m 2.5dB, 12m 1.8dB, 10m 1.9dB

Of course, on 40m and 15m the introduction of the UnUn has increased the coax loss - by 1.7dB and 1.5dB respectively. On the other bands it has reduced the loss by amounts varying from 2.9dB on 30m to a very significant 5.6dB on 20m and 10m.

In a nutshell, introducing the UnUn has taken us from a situation where some bands were very low-loss and some were very high, to a situation where all bands now have moderate losses. Unless you had a particular interest in 40m and/or 15m work, you might think that the penalty suffered on those bands was worth the improvement gained on the other bands.

As you would expect, using a 4:1 UnUn rather than a 9:1 degrades the 40m/15m performance less, but does not improve the other bands as much.

So on balance, introducing the 9:1 UnUn appears to be a good move. However, that's not the whole story .........

------------------------------------------------------------------------------------------------------

"Brain Hurt" level: **

The analysis so far assumes that the UnUn introduces a perfect 9:1 impedance transformation at the feedpoint. Let's see how valid that assumption might be.

A perfect impedance transformer reflects at its primary the impedance placed across its secondary, modified by the turns ratio. For that to happen, the impedance of the primary and secondary windings must be high enough that they do not significantly "shunt" the load, or the transformed load. A good rule of thumb is that the winding reactance should be at least 5 times the load impedance they are handling. So, for example, if we placed a 450 ohm load across the output terminals of our 9:1 UnUn, expecting to see 50 ohms at the primary, we have to ensure that the secondary windings have a reactance of at least 5*450=2250ohms and that the primary windings have a reactance of at least 5*50=250 ohms.

Now let's consider a typical UnUn design published on the Web; the design I have in mind specifies a T200-2 Micrometals toroidal core and recommends 14 turns as the primary winding. We can look up the data for this core and see that 100 turns on it would produce an inductance of 120uH. Knowing that the inductance varies according to the square of the number of turns on a toroid we can calculate our UnUn primary winding inductance as:

120uH*(14/100)^2 = 2.3uH

Let's suppose we now tried to use this UnUn to transform a 450ohm load down to 50ohms at a frequency of 3.5MHz. The reactance of the primary winding is:

2*Pi*f*L = 2*PI*3.5MHz*2.3uH = 50.6ohms

Instead of our required primary reactance of 250ohms we have a very low 50.6 ohms; this appears in parallel with the 50 ohms of the transformed load and will severely shunt it. Working backwards, we can say that this UnUn will only perform satisfactorily as a 450ohm/50ohm impedance transformer at frequencies above 17MHz where its primary reactance is at least 250ohms.

Out of interest I just wound 14 bifilar turns on a T200-6 core (I didn't have a T200-2) and placed a 200ohm load across the secondary. Instead of the expected 1:1 VSWR at the primary, I measured 5:1 on 160m and it didn't drop below 1.5:1 until I got above 12MHz. Type 2 material would have been better, but not much better - it's 120uH/100 turns vs 100uH/100 turns for Type 6. I then tried 10 bifilar turns on a ferrite core and got a VSWR <1.2:1 from 1.5MHz through 35MHz.

But hang on! In our application we're not placing a benign 450 ohm resistive load across the secondary of the 9:1 UnUn - it's "seeing" the widely varying feedpoint impedance of the vertical wire. For example on 20m the antenna impedance is 2651 ohms, and that requires a primary winding reactance of at least 5*2651/9=1473ohms; our UnUn manages just 2*Pi*14MHz*2.3uH=202ohms. Let's get this in perspective. We have transformed a secondary load of 2651ohms down to 294ohms at the primary, and then shunted it with 202 ohms of primary winding inductance.

The effect of this low shunt inductance is that the impedance at the input of the UnUn will be significantly different than we would expect with a perfect 9:1 transformer, particularly on the lower frequencies. That places a big question mark over the accuracy of our earlier analysis about coax losses, and therefore the perceived advantage of the UnUn. There may be some feedpoint impedances where the low shunt inductance actually reduces the VSWR on the coax; there will be others where it makes it worse.

The point is that it's difficult to do the analysis when the UnUn is not behaving as a 9:1 impedance transformer.You might think that increasing the winding inductance by increasing the number of turns would help - it would, but a new problem arises. Suppose we increase the number of turns to 40 in an attempt to get the UnUn to work well on 20m. Apart from the difficulty of winding that many trifilar turns of a reasonable wire gauge on a T200 core, we find that the wire making up the windings is now long enough that it represents a transmission line which is a significant fraction of a wavelength long. In fact it will transform our 2651ohm 20m load down to 11-j128ohms solely from transmission line action, before we ever consider the effects of the transformer.

It should be clear by now that the design of broadband transformers is non-trivial. We need plenty of inductance without many turns, and that means using a ferrite material with its much higher permeability -

Type 2 or Type 6 dust-iron material is a poor choice because it cannot deliver high inductance with short winding length.------------------------------------------------------------------------------------------------------

"Brain Hurt" level: ***

We now need to consider the power-handling capability of the UnUn. As a worked example let's calculate out what the power dissipation and temperature rise of the core will be on 80m running 100W.

EZNEC predicts the antenna feedpoint impedance at 3.5MHz as 17-j500ohms. That means we need to get a current of 2.4A into the antenna to radiate 100W, and that means we need 1211 Vrms across the secondary of the UnUn.

Peak AC flux density in the core B = (E.10^8)/(4.44*A*N*f) = (1211.10^8)/(4.44*1.3*42*3.5.10^6) = 143 Gauss

Core loss = 8.86.10^-10*f^1.14*B^2.19 = 1,342 mW/cc

According to the Amidon data, 436 mW/cc is enough to raise a T200 core temperature by 40 degrees centigrade. Their charts don't beyond this, so I leave you to imagine what 1,342 mW/cc might do!

For fun, I re-ran the calculation for 160m. The peak voltage across the UnUn secondary was 5kV and the power dissipation was 29,000 mW/cc - that's 66 times the level that Amidon's table goes up to. In that respect, Answer#5 was correct ... it should go like a bomb on 160m

------------------------------------------------------------------------------------------------------

"Brain Hurt" level: *

So far we've ignored losses in the tuner. They vary from band to band, with and without the UnUn, but are typically less than 0.75dB except on 160m and 80m. The total losses on these bands (coax+tuner) exceed 10dB before we include the ground losses, making them marginal to say the least. So in this instance tuner losses do not argue strongly in favour of, or against, the UnUn.

------------------------------------------------------------------------------------------------------

"Brain Hurt" level: *

The original Puzzle included a suggestion that an UnUn might reduce common-mode current on the feedline. It can't! It does nothing to change the common-mode impedance looking into the coax braid from the antenna feedpoint. If that common-mode path looks low impedance compared to the impedance of the radials, significant braid current will flow.

------------------------------------------------------------------------------------------------------

Summary:

* A 9:1 impedance transformation reduces the coax loss on a majority of the HF bands, but increases it on 40m and 15m. Unless these band are your primary interest it would probably be worthwhile including an UnUn.

* Losses on 160m and 80m, with or without an UnUn, are unacceptable.

* Type 2 or Type 6 dust-iron mix is a poor choice for a broadband 9:1 transformer core

* The T200 core size has limited power-handling when dealing with low resistance reactive loads

* An UnUn does nothing to help common-mode problems.

Is there a better way? Yes, the UnUn is very much a compromise solution. A better engineering approach is to match the antenna at the feedpoint, keeping the VSWR on the coax low on all bands and minimising losses. Of course this requires either a remote autotuner or switched fixed networks for each band. It's certainly the solution the keen contesters use; to them picking up a couple of dB is important.

Finally, please note that the loss numbers we've looked at are unique to the particular length of vertical and the length of feedline. The UnUn would show little advantage if the feedline was much shorter. We would also get a different set of figures if we considered a 43ft vertical - another popular length.

if I were in the market for one of these popular antennas I would want to know that the manufacturer/designer understood what would be the range of impedances presented by the vertical, and that they had designed the UnUn to handle the more extreme impedances at full rated power. If I saw the characteristic red colour of a small Type 2 micrometals core showing through the UnUn windings I'd be walking away

Contentious perhaps, but I'd be interested to hear your views.

Steve G3TXQ