Latest Posts
Paul O’Grady
29 March 2023
Radiodays Europe 2023
28 March 2023
Podcast Series Made by AI
23 March 2023
21 March 2023

How you can achieve accurate loss and power measurements


Keith Rawlings G4MIU explains how you can achieve accurate loss measurement in coaxial cables, and he takes a look at some sturdy vintage RF power measurement devices.


Introduction | Accurate Loss Measurement | Assessment | Other Methods | Forums | RF Power MeasurementThe Bird Termaline | Watts


I know that readers use a varied range of different coaxial cable types with their stations, ranging from the expensive high-quality, low-loss types, and the more ‘common’ RG cables, such as RG58/213, to just lengths of unknown-specification 75Ω TV cable which happen to be to hand.

All coaxial cable exhibits the phenomenon of loss, and the amount of loss will vary depending on the specification of the cable being used, its length, the frequency it is operating, and its condition.

Estimating losses on a cable run of known type, in new and good condition, is quite easy, just visit a coax loss calculator, such as this here.

If the cable is rather old, has been physically damaged, or if it is suspected that it has suffered water ingress, this will all make estimating losses quite difficult. This is where an aerial analyser of some sort can prove invaluable; and, of course, the Nano fits in here.

Accurate Loss Measurement

We have already covered the setup needed to measure the loss of a run of coax when we tested the low pass filter last month (RadioUser, February 2021: 54-56). This was a ‘through’ S21 measurement. To measure a length of coax cable for loss, we need to set a frequency span to cover the range that is of interest. If we are interested in VHF/UHF monitoring, for example, a span from 100 to 500MHz may suffice.

For HF, a maximum of 30MHz could be selected.

For this demonstration, I decided to try the Nano on a 10m length of unknown TV coax and chose to run the test over a span from 1-500MHz, which should cover frequencies many readers are interested in.

All I can tell you about the cable is that it was fitted to a cheap quality TV aerial of the type intended for use on a caravan. It has a reasonable-quality outer braid with a sturdy solid centre conductor, and the inner dielectric (isolation) looks like polythene.

So, the setup is as before: CH0 has to be calibrated using an SOL (Short-Open-Load) and CH1 calibrated with Isolation and Through. CH1 has to be selected as LOGMAG so we can see our S21 measurement in dB.
My calibration was made at the end of two short lengths of RG233 cable, which were fitted with SMA plugs.
To make this test on the Nano I soldered a pair of SMA sockets to the end of the length of the TV cable to be tested, see below. 

Below you can see the result of the test.

The reference level has been set to mid-screen, the trace has been set to read 5db per division, and you can see that the response gradually drops away to the right, indicating increasing loss with increasing frequency. There is quite a large amount of ripple on the trace. At a frequency of 435 MHz – midway in the 70cms band – the calculated loss(with the 50/75Ω mismatch) is 7.6dB.

This indicates a signal loss of over One S-Point at this frequency.

This is with just a relatively short run of 10m long, so it may be OK for an attic or bedroom-based shack close to the aerial; needless to say that, if this length were increased to run over a greater distance, the losses would also increase. 

On HF, the losses are much lower but on longer runs, and these will still mount up. Therefore, the NanoVNA is a quick and convenient way to check and see if a run of coax is up to the job. 

A Comparative Assessment

I next moved the measurement over to my SDRKits VNWA. If we take a look at the image below, we can see a comparison made between the white TV Coax (Yellow trace ) and a 40ft run of old but very good quality RG213 (grey trace).

When I say old, I mean it: I bought the reel this cable was taken from in the late 1970s. This section is in good condition and it has never been used outside. I used the VNWA for two reasons, One, I wanted to make comparisons between the Nano and the VNWA and Two, it is easier to make screengrabs with the VNWA! 

As can be seen from the grey trace, the RG213, as expected, returns better results. At 1MHz, it is displaying a quarter of the loss of the TV coax (Yellow Trace), and at 500MHz the loss is around 1.6 dB, bearing in mind the run is 3m longer. The loss measured on the RG213 may be a little higher than might be expected but due to its age, I am not surprised.

The image above also demonstrates that the NanoVNA gives reasonably similar results to the VNWA. What discrepancies there are, maybe due to the calibration kits used. For the Nano, I used the supplied kit, and for the VNWA a Rosenberger 12GHz SMA kit. The measurement at 435MHz in Fig 2 is taken in a trough in the ripple, thus giving a slightly higher loss reading. The ripple is probably caused by the quality of the cable. Also, being coiled up might not have helped it.

Another Method

I made this measurement with the cable inserted between the two ports of the Nano, but there are occasions where it may not be possible to get both ends of a cable to a point where this type of measurement is possible, for example, when the cable is at the top of a mast.

Content continues after advertisements

Assuming access is possible, there is a method where an S11 single port method may be used, by measuring Return Loss, with the far end of the cable either open or with a short across it.  Any aerials will have to be taken out of the circuit before these tests can be undertaken. This method may not be as precise but should return an acceptable accuracy. The VNA requires a single-port-calibration and is connected to the shack end of the cable.  Using the ‘LOGMAG’ feature on the NanoVNA, a CH0 reflection sweep is made, and the measured return loss is divided by two to get the result.

One benefit of owning a Nano is that it can be used to make tests on cable runs at any time (and anywhere), and the results can be kept for future reference. Therefore, if a cable fault is suspected in the future, it will be a relatively easy matter to compare results with your recorded data.

Some info on these methods may be found here.

A New Forum

By the time you read this a new user forum for the users of MMANA-GAL/GAL-ANA aerial simulation software should be up and running. You can find out more here

RF Power Measurement, Vintage-Style

A friend of mine recently asked what kind of equipment I used for power measurement. I replied to him that the kit I used now was quite mundane, in that it just consists of a Bird Through Line model 43 or a Bird Termaline model 6104. “Don’t you have anything more precise”? came back the reply, and my answer was “No”.

The simple truth is that what I have is more than good enough for what I need.

It seems he has ‘blown up’ his Hewlett Packard 435 Power Meter!

Although now quite old, these are still good power meters, but if the remote Power Sensor gets damaged they are not much use because the sensor heads are not interchangeable. Nevertheless, I believe that HP still supports them, so the meter may be repairable, at a cost.

From time to time, readers sometimes ask me what equipment I use, and this correspondence prompted me to think there may be interested to learn a bit more about these Bird meters. 

While not strictly ‘aerials’ they do pug into a socket usually marked as such so that’s good enough for me. 

The Bird Termaline 

This instrument is a Dummy Load/Wattmeter, see below. It has a 50Ω cylindrical film resistor as the load. This is immersed in a dielectric coolant, the whole of which is housed within a finned enclosure. RF power is applied to the Type-N socket located on the front of the enclosure. A rubber diaphragm allows for any expansion of the coolant due to heat. A IN79 Crystal diode rectifies the applied power. 

This is read off of a meter mounted on a detachable housing. A length of about 1m of RG58 coaxial cable between the dummy load and meter housing allows the head to be detached for semi-remote use. There is a toggle switch on the front panel. It determines the scale used. Each scale has two ranges, which are selected by swapping the crystal diode between the two sockets provided on the front panel. This gives measurement ranges of 2-6-20-60W with a frequency range (on my model) of 25-512MHz. 

The LOW scale marked in Black reads 2 or 20w depending on the position of the toggle switch, and the HIGH scale reads 6 or 60W again dependant on the switch position. Modes are AM, FM, CW. The instrument’s accuracy is claimed to be at 5% of the full scale. 

Incidentally, AM modulation may be monitored directly from the DC meter circuit. I have never needed to use my meter for power measurements greater than 50W, but I do have some power attenuators that can be used to measure higher levels if need be.

There is a latch mounted on the meter housing to enable it to be removed. One thing to watch out for is with the dummy load section. Being filed filled, as it is, with a coolant (about 380ml) there may be leaks. I am not sure what this coolant is made from, but it might be worthwhile handling a leaking meter with care and avoiding contact with the skin.


The meter works by the simple expression W=E2/R. 

Where E is the voltage across the resistor R and W is power in W.

This equation need not bother us, as the Bird simply displays voltage E on the meter as Watts. 

The use of the meter is simple. Connect a transmitter to the front connector using as short a length of cable as possible. If you have to use a long run of the cable allow for its losses (perhaps using the Nano as above).

If you are not sure of the transmitter output power, select the higher ranges first, and then change ranges/scale to get a reading as close to FSD as possible. When swapping the diode it needs to be rotated in its socket to obtain maximum deflection on the meter.

There is one thing to remember with wideband loads such as the Bird: By being wideband they will also measure any harmonics or spurious signals present. Harmonics may be suppressed with a lowpass filter and spurious signals need to be eliminated anyway.

The load itself is good to 1GHz if used without the wattmeter.