Measuring a Compromise Antenna: What the NanoVNA Shows That an SWR Meter Hides
A low SWR at the shack can mean a lossy feedline rather than a working antenna. What R plus jX tells you, and how to measure it for under $90.
This article contains affiliate links. I may earn a small commission at no extra cost to you.
Disconnect the antenna from your feedline, leave the coax end open, and key up into it. If you are running 100 feet of RG-58 on 10 meters, the SWR meter in your shack will read about 3.4:1. Not infinity. Not a fault light. A number that plenty of operators would accept and start making contacts on.
Nothing is connected. The reading comes entirely from feedline loss.
That is the whole problem with judging a restricted-space antenna by shack-end SWR. The meter measures the feedline and the antenna together, and in a small install the feedline is usually long, thin, and routed badly. Every dB of loss in the run pushes the reading toward 1:1 regardless of what is happening at the far end. A vector network analyzer for $80 measures the thing itself.
Why loss flatters the reading
A reflected wave has to travel down the feedline, bounce, and come back. It pays the line's loss twice. The transmitter only sees what survives the round trip.
For an antenna reflecting everything, here is what your shack meter shows through various amounts of one-way matched loss:
| One-way matched loss | SWR seen at the shack |
|---|---|
| 0.5 dB | 17.3:1 |
| 1.0 dB | 8.7:1 |
| 2.0 dB | 4.4:1 |
| 3.0 dB | 3.0:1 |
| 5.0 dB | 1.9:1 |
Five decibels of loss makes an open circuit look like a decent antenna. And 5 dB is not exotic: it is what you get from a long run of RG-58 on the higher bands, or from a shorter run with water in it, or from a run that has been sharply kinked behind a baseboard.
The relationship runs the other direction too, which is the part worth internalizing. An SWR reading that improves over months is almost never an antenna getting better. It is a feedline getting worse. If your attic dipole read 1.8:1 last spring and reads 1.3:1 now, go find out why the coax is lossier.
What the extra number buys you
An SWR meter gives you one number, and it is a magnitude with no sign. The load could be too capacitive or too inductive and read identically. So people trim wire in whichever direction they guessed, watch the SWR get worse, and trim further the same way.
A VNA gives you resistance and reactance separately, written as R + jX. The sign of X is the instruction:
- X negative (capacitive): the antenna is electrically short for that frequency. Add wire, or add loading inductance.
- X positive (inductive): it is electrically long. Take wire off.
- X near zero but R far from 50: it is resonant and still mismatched. More wire will not fix it. It needs a matching network.
That last case is the one that eats weekends. A shortened, loaded vertical on a balcony can sit at 12 ohms with no reactance at all. It is perfectly resonant. SWR is 4.2:1 and no amount of trimming moves it, because the problem is a low radiation resistance, not a wrong length. You need an unun or an L-network at the base, which is a different purchase than more wire.
For the cases where trimming is the answer, resonant frequency scales inversely with length closely enough to work with. A 40 meter dipole resonant at 7.30 MHz that you want at 7.15 needs to be about 2 percent longer: roughly 8 inches added to each leg on a 66-foot wire.
Measuring your own coax, which is the real value in a small install
Most restricted-space feedlines are unknown quantities. Salvaged coax, an old run left by the previous owner, something threaded through a soffit at an angle nobody wants to think about. You can measure it in two minutes without knowing anything about it.
Disconnect the far end and leave it open. Connect the VNA at the near end. Read the return loss magnitude, S11 in dB, across the band you care about. One-way matched loss is half that number.
If S11 comes back at -5.2 dB on 10 meters, the run has 2.6 dB of one-way loss at that frequency. Compare against what the cable type should do. Fifty feet of RG-58 is around 0.9 dB at 14 MHz and 1.3 dB at 28 MHz; RG-8X is roughly 0.7 and 1.0; LMR-400 is about 0.35 and 0.5. A measurement that comes in at double the book figure means water ingress, a damaged shield, or a connector that was crimped in a hurry. Feedline choice and routing under constraint is covered separately in the coax routing article.
This measurement also settles the shack tuner versus remote tuner question with an actual number rather than a guess, which is what the tuner article asks you to compute.
Measure at the antenna, and measure it where it lives
Two habits separate useful data from noise.
Take the reading at the feed point, not the shack. Everything above explains why. If the antenna is in an attic and you can physically reach it, take the analyzer up there. The difference between the feed-point reading and the shack reading is your feedline, and now you know both numbers.
Sweep it in its final position. Nearby drywall, roof sheathing, wiring, and ductwork all add capacitance and pull resonance down. An attic dipole tuned on the lawn will land somewhere below where you cut it once it goes inside, often by 2 to 5 percent, and more if the sheathing is foil-backed. Cut long, hang it, then trim in place. The attic antenna article covers what those materials do to the antenna itself.
There is a third habit that pays for the instrument on its own. Sweep the antenna, then run your hand slowly along the outside of the coax and watch the trace. If resonance or SWR moves when you touch the feedline, the feedline is part of the antenna and common-mode current is flowing on the shield. Clamp a ferrite choke on and sweep again. If the trace changes, you had a problem and now you have less of one. That is the same failure mode described in the RF grounding and counterpoise article, made visible.
Bandwidth is a loss gauge
For high-Q antennas, the width of the SWR curve tells you about efficiency, because loss resistance damps the resonance and widens it.
A small transmitting loop is the clearest case. If a loop advertises 30 kHz of 2:1 bandwidth on 20 meters and yours measures 90 kHz, something resistive in the loop is dissipating power that should be radiating. The usual suspect is the capacitor contact. That mechanism is worked through in detail in the magnetic loop capacitor article.
The same logic applies in reverse to wire antennas, though less dramatically. A very broad SWR curve on a short wire antenna is not good news about bandwidth. It is usually news about loss.
Two things that will ruin your day
Calibration is per-span and per-cable. A NanoVNA calibrates open, short, and load at a reference plane, and the calibration is only valid for the frequency span you calibrated over. Change the span, calibrate again. Calibrate at the end of the test lead you will actually use, or you are characterizing the lead along with the antenna.
Never transmit into it. These instruments put out about -10 dBm and their bridges are destroyed by watts, not kilowatts. A single accidental key-down at 100 watts is the end of the unit. Build the habit of disconnecting the analyzer and reconnecting the radio as one motion, and keep the rig on a dummy load while you are working.
What to buy
| Instrument | Range | Approx price | Best fit |
|---|---|---|---|
| NanoVNA-H4 | 10 kHz to 1.5 GHz | $70 to $90 | Default choice, 4 inch screen readable in an attic |
| NanoVNA-H (2.8 inch) | 10 kHz to 1.5 GHz | $50 to $60 | Same capability, harder to read on a ladder |
| LiteVNA 64 | 50 kHz to 6.3 GHz | $110 to $150 | Better dynamic range, useful if you measure filters too |
| RigExpert Stick 230 | 0.1 to 230 MHz | around $160 | One-button field use, no calibration ritual |
| RigExpert AA-55 Zoom | 60 kHz to 55 MHz | around $350 | HF only, rugged, best interface of the group |
The NanoVNA-H4 is what I would buy first and what most of this article assumes. The 4 inch screen matters more than the spec sheet suggests, because a lot of this work happens crouched in an attic or leaning off a balcony with bad light.
Budget for two accessories with it. A calibration kit with proper open, short, and load standards, because the ones bundled in the kit are inconsistent. And an SMA male to SO-239 adapter plus a short RG-316 jumper, so you are not stressing the analyzer's SMA connector every time you attach a PL-259.
The RigExpert Stick 230 is the alternative if you want a reading without a calibration step. It costs twice as much and does less, and for someone who will use it three times a year that is a fair trade. DX Engineering stocks the full RigExpert line if you would rather buy from a supported dealer than a marketplace listing. The same marketplace caution from the used HF rig guide applies to NanoVNA clones: there are many, firmware quality varies, and the seller name matters more than the photo.
What I would do
Buy the NanoVNA-H4 and a calibration kit before you buy another antenna.
Then run three measurements, in this order. Measure the feedline alone with the far end open, and find out what your run actually costs you. Measure the antenna at its feed point, in place, and write down R and X on every band you use. Measure again from the shack, and confirm the difference between the two matches the feedline loss you calculated.
That sequence takes an afternoon and it usually reorders the shopping list. In a constrained install the answer is more often a shorter or better feedline, a choke, or a matching network at the base than it is a different antenna. You cannot tell which from a single scalar number, and the instrument that tells you costs less than one evening of guessing at wire lengths on a roof.