Making a Shortwave listening antenna

[Guineafowl]

TL;DR Summary

While working through the ARRL antenna book, I’d like to start putting some theory into practice.

Thanks to @berkeman for the ARRL book recommendation.

While trawling through the theory, I’d like to build an antenna to help cement theory with practice. The receiver is an old valve/tube radio, covering the SW band from 6-18 MHz, or 49-16m.

For simplicity, I thought a small, tuned loop antenna would be a good starting point. That is, one under 0.1 wavelength long, 0.085 suggested. (5-3 in ARRL book). So the design will be:

Transmission line: RG6 coax, directly into the radio’s banana sockets.
Antenna: centred on 12MHz, 25m wavelength, (x0.085=2.125m conductor length). Single loop, triangle. Conductor will be PVC-coated 2.5mm2 copper wire, although could be changed for, say, 3/8” copper pipe. As I understand it, a larger diameter conductor has a lower Q, and so a broader response.

There is an equation on page 5-5 to find the inductance of such a design. I assume I then calculate the inductive reactance at 12MHz, and spec a capacitor to suit, and place across the feed point.

I would also consider a physically smaller, multi-turn loop, essentially duplicating what’s already inside the radio, but at least understanding how it’s designed. It could also be placed away from interference.

As you can see, I’m at a very early stage, so any comments would be welcomed.

 

[Guineafowl]

For a single triangular loop, of length 213cm and side 71cm, I get an inductance of $2.452 \mu H$.
This equates to $X_L=j184.9\Omega$ at 12MHz.

Using $C=\frac {1} {j\omega X_C}$, I get a tuning capacitance of $72pF$.

Checking backwards, $F_0 = \frac {1} {2\pi \sqrt{LC}$ gives a resonant frequency of 11.9MHz.

 

[Baluncore]

The receiver is an old valve/tube radio, covering the SW band from 6-18 MHz,...

Out of interest, what is the make and model of the radio? The internal loop antenna may be tuned with a variable capacitor, and you may find your tuned loop reacts with that, depending on the degree of coupling.

You are wise to go for a magnetic loop antenna in these days of switch-mode power supply noise. Build and test a single turn triangular loop, but don't invest too much money, or make it look too nice.

It takes more wire to go around the corners of a triangle, and the corners do not add much area to the loop. Corners also get in the way when you rotate the loop to null the interference or peak the signal. Your next loop will be a two turn, square, hexagonal, or octagonal loop.

I would also consider a physically smaller, multi-turn loop, essentially duplicating what’s already inside the radio, but at least understanding how it’s designed. It could also be placed away from interference.

While you are listening, rotate the loop about the vertical axis to null the interference.
Better rejection of electrical noise is with a two turn loop, (any even number), since you can ground the mid-point of the loop, and electrostatically screen it, (screen is open at the top).

Start looking now for a two-gang variable capacitor, (from an old radio), to tune your next balanced loop antenna. A two turn loop has four times the inductance of a one turn loop, so is best applied to smaller loops, or to lower frequencies.

 

[sophiecentaur]

You are wise to go for a magnetic loop antenna in these days of switch-mode power supply noise. Build and test a single turn triangular loop, but don't invest too much money, or make it look too nice.

That makes sense. For reception at HF in the presence of local interference, it's not so much a case of Signal to Noise Ratio that counts; rather Signal to Interference Ratio.

 

[Guineafowl]

It’s a Cossor 501 transformer-type set. Schematic below, where you can see the frame aerial hooked in to banana jacks ‘A’ and ‘E’. These can be accessed from outside the set, unplugged and substituted for the new aerial.

A 375/210 pF ganged capacitor is on the way.

Since I haven’t built it yet, I’ll change the design to a two-turn square loop; calculations to follow.

Is there a reasonably priced gadget that I can attach to the aerial to check its point of resonance? My original plan was to use my old AVO signal generator to send out a 12MHz signal, with 1kHz modulation, and peak the aerial to that.

 

[Guineafowl]

For a square loop, where N=turns (2), s=side length(26 cm), l=coil length(213 cm):



I get an inductance $L=1.9245\mu H$
Reactance $X_L=j\omega L=j145.1\Omega$ at 12MHz.
Capacitance $C=\frac {1} {j\omega X_C} = 91.4pF$

For square multi-turn loops, distributed capacitance is given by $$C(pF)=60S$$ where $S=side~length (m)$

$$C=60(.26)=15.6pF$$ Which gives $75.8pF$ when subtracted from the result above.

 

[Baluncore]

Is there a reasonably priced gadget that I can attach to the aerial to check its point of resonance?

As you tune the antenna, you should be able to hear the resonant peak in the received audio of the receiver.

There are HF antenna analysers available when you need one. Expect to pay US$50.
Search eBay for NanoVNA-H Vector Network Antenna Analyzer.

 

[sophiecentaur]

Is there a reasonably priced gadget that I can attach to the aerial to check its point of resonance?

If you dive in and identify volts on the AGC line then you have a fairly reliable (numerical) way of finding a maximum, not relying on your ears, as you tune your capacitor. No need for a signal source as you can use a handy off-air signal.
Does that document, part of which you posted, help to find the AGC part of the circuit?

 

[Baluncore]

The full circuit is shown in post #5. The only candidate I could see for AGC, is the horizontal line drawn below the ground line. I expect capacitor C7, connected to ground, sets the time constant.

 

[sophiecentaur]

Cheers. If the OP has a meter and can avoid a shock on a scary valve circuit, it may be useful for him. It would be easy to see whether it works - it should.
That line seems to affect the bias of V1 and V2. My eyes go fuzzy when I look at valve circuits; too many connections!

 

[Baluncore]

My eyes go fuzzy when I look at valve circuits; too many connections!

The symbol of a vacuum tube is sufficient to remind me of the smell of the dust burning off the glass envelope. That smell triggers a memory of the tone of The Carpenters.

 

[Guineafowl]

The full circuit is shown in post #5. The only candidate I could see for AGC, is the horizontal line drawn below the ground line. I expect capacitor C7, connected to ground, sets the time constant.

Yes, the AGC line is via R8 and decoupled by C7. I did a restoration and alignment of this radio a few years ago, so quite happy to open up and probe the AGC line if needed.



Well, I built a two-turn, shielded loop (no cap yet). Interestingly, it doesn’t seem to do anything - the 10ft or so of coax going to it gets a reasonable signal on its own, through the centre conductor. Disconnecting the shield also makes no difference. Connecting the antenna up doesn’t improve the strength much.

Nano VNA is on its way, so hopefully that might shed some light.

 

[Baluncore]

- the 10ft or so of coax going to it gets a reasonable signal on its own, through the centre conductor.

Don't expect a good result until the tuned antenna is independently resonant. Tuning performance will depend on how lightly you can couple the tuned two-loop antenna to the line.

For a feedline I would consider a higher-impedance twisted pair of thin wires, rather than a low-impedance coaxial line with higher capacitance.

 

[Guineafowl]

Don't expect a good result, until the tuned antenna is independently resonant. Tuning performance will depend on how lightly you couple the tuned two-loop antenna to the line.

Thanks. Ok, I’ll await the tuning cap and VNA. By coupling, do you mean an impedance matching network, and does light coupling refer to good matching?

For a feedline I would consider a higher-impedance twisted pair of thin wires, rather than a low-impedance coaxial line with higher capacitance.

Would a twisted pair from a cat 5 cable suit?

 

[Baluncore]

Would a twisted pair from a cat 5 cable suit?

For a receiver, yes.
But not as a part of the tuned circuit.

 

[Guineafowl]

I had a small amount of time to try the nano VNA on the antenna,

Sweeping from 6-18Mhz, in S11 mode, it shows $\infty$ SWR all along, and is acting as an open circuit on the Smith chart.

Changing the sweep to full range (50kHz-900Mhz), there are several dips in the SWR, but all above the 700 MHz mark.

More work needed! Tuning cap yet to arrive.

 

[berkeman]

Well, I built a two-turn, shielded loop (no cap yet). Interestingly, it doesn’t seem to do anything - the 10ft or so of coax going to it gets a reasonable signal on its own, through the centre conductor. Disconnecting the shield also makes no difference. Connecting the antenna up doesn’t improve the strength much.

Sweeping from 6-18Mhz, in S11 mode, it shows $\infty$ SWR all along, and is acting as an open circuit on the Smith chart.

Hmm, that's not good. Can you upload a sketch of your antenna showing the connections and such?

 

[Baluncore]

More work needed! Tuning cap yet to arrive.

Without the capacitor, the untuned loop is an inductor.
It will appear to be open circuit at RF.

There may be bumps in the response, due to the capacitance of the cable, transformed by the length and impedance of the transmission line.

 

[Guineafowl]

There is a rough sketch in pencil, showing the double loop, and copper pipe shield connected to the left-hand side.

SWR (Yellow), 50/div.
Smith plot in blue.
210pF capacitor plates open.
Sweep from 6-18MHz.

Now with plates closed, presumably 210 pF.

So we appear to be stubbornly skirting the outer circle, where reflection coefficient is 1.

 

[Baluncore]

No pencil sketch ?

Please show us a photo of the whole copper loop, so we can see the insulated break at the top.

Ideally, the capacitor should be balanced with two identical gangs, so the chassis can be grounded, with the tuned loops being balanced.

 

[DaveE]

Please show us a photo of the whole copper loop

Post #12

 

[Guineafowl]

Post #12

Indeed. There is also a pencil sketch on the wooden board pictured above, but here’s a better one:

 

[Guineafowl]

Looking at the top VNA plot, we appear to have a point of resonance somewhere between 6 and 12Mhz, but the real resistance is very high. Could this be due to skin effect at these frequencies?

The conductor diameter is only 2.5mm (0.1”).

 

[Baluncore]

Could this be due to skin effect at these frequencies?

Probably not. It is more likely that the asymmetry of the earth is causing the problem.

The mid-point of the two turns should be earthed, not one side. One gang of the capacitor should tune each side of the coil, with the chassis of the two-gang capacitor being earthed. Some form of isolation between the tuned antenna and the coax is needed, such as a 22 pF capacitor from one side capacitor-coil junction, to the coaxial centre conductor.

 

[Baluncore]

Here is a schematic for a 2 turn loop antenna with a screen.
Note that the centre of the two loops are connected to the screen and earth.

 

[Guineafowl]

Thanks for taking the time to draw the schematic.

One problem that occurs to me, is that the radio tuning cap I have is not symmetrical. It’s also hard to access the wire loop, now it’s enclosed in the copper pipe.

With that in mind, I made a second loop without the shield, and hooked up as you’ve drawn. The VNA readout from 6-18MHz showed a spiral in the outer third, indicating several points of resonance.

Out of interest, I attached the VNA to the on-board antenna of the radio, and it showed an arc at the $\infty$ side of the plot, much like I had before. This is a 6m loop with 6 turns.

Am I right in thinking that, in order to tighten the spiral response, and so improve the response of the antenna, I need a larger conductor for the loop?

 

[Baluncore]

Am I right in thinking that, in order to tighten the spiral response, and so improve the response of the antenna, I need a larger conductor for the loop?

For receive, you can use thin insulated wire. For transmit, thick wire is better, and the screen is not needed.

One challenge you will have, is coupling and matching the antenna to the coaxial cable. A Smith chart is centred on the impedance of the transmission line, 50 ohm for the VNA. You may need to use a broadband TV antenna balun, between the antenna and the analyser, to get the trace to approach the centre of the Smith chart.

The crossover at the top of my schematic is symbolic. The two turns of the loop, inside the screen, can be twisted together so they are closely coupled. That should reduce the unbalanced tuning capacitance. If you twist the two wires together before threading them once through the screen tube, you will have four terminals accessible at the base, to connect to the earth and tuning capacitors.

An easy way to make a screened, two turn loop antenna, is to use twinaxial cable, or screened network cable, attached to a frame. Remember to cut all the screen/braid at the top of the loop.
https://en.wikipedia.org/wiki/Twinaxial_cabling

 

[sophiecentaur]

Looking at the top VNA plot, we appear to have a point of resonance somewhere between 6 and 12Mhz, but the real resistance is very high.

If a resonance is a parallel resonance then wouldn't the parallel R be very high? How have you measured this R?

 

[sophiecentaur]

If a resonance is a parallel resonance then wouldn't the parallel R be very high? How have you measured this R?

Ahhh. You are measuring the DC resistance somewhere, with a DMM. That can easily be infinite if there's a capacitor in series but that won't be the real part of the impedance at your operating frequency.

 

[Baluncore]

The cable used between the analyser and the antenna will rotate the Smith chart diagram due to twice the line transit time, changing the phase. You should recalibrate the analyser, with a short coaxial cable, dedicated to the analyser. Use the analyser as an S11 reflectometer.

At resonance, the antenna reactance will change sign, so it will cross zero. That will place the antenna resonance somewhere along the horizontal axis of the Smith chart on the analyser. Ideally, I would expect it to be at the infinite R, but just where on that axis will depend on the matching transformer and radiation resistance of the antenna. Maybe you should practice with a lumped LC parallel resonant circuit first, to see how it performs.

 

[sophiecentaur]

There are HF antenna analysers available when you need one. Expect to pay US$50.
Search eBay for NanoVNA-H Vector Network Antenna Analyzer.

That's seriously cheap. The display on the images implies all sorts of things about what it does. I remember using an h.p. analyser that cost a couple of grand. What does the nanoVNA do different for £50?

 

[Baluncore]

That's seriously cheap.

The difference is that the Nano VNA is NOT calibrated to a reference standard.
You can afford to buy one and play with it.
When you are finished, gift it to a mathematical HAM radio operator, one who is interested in antenna design.

 

[sophiecentaur]

The difference is that the Nano VNA is NOT calibrated to a reference standard.
You can afford to buy one and play with it.
When you are finished, gift it to a mathematical HAM radio operator, one who is interested in antenna design.

OK - no calibration but what about the directional coupling, complex voltage measurements and display? £50?!!!!! I bought a cheap VHF SWR meter for about the same amount and that was £30. It hardly had more than a half dozen components. Very useful though!
The years have made such a difference.

 

[Guineafowl]

As above, I’m getting an idea of real resistance from the Smith chart display. I can move the marker along the sweep, to get a readout of the impedance at a particular frequency, in $R+jX$ form.

The nano VNA comes with open/short/$50\Omega$ terminations which can be attached to the end of the little test cable in calibration mode. I assume this takes care of the extra cable length.

Twinax cable on order.

 

[Baluncore]

The nano VNA comes with open/short/ terminations which can be attached to the end of the little test cable in calibration mode. I assume this takes care of the extra cable length.

I believe that is the case.

Twinax cable on order.

I would have used some old cat x, network or USB cable, and made a loop from that. Or taken the braid off some old coax, and fed a twisted pair through the centre. You need to get creative and scavenge for things to make antennas. It is both cheaper and quicker than waiting for delivery.