Antenna analyser accuracy on extreme impedances

Hand held antenna analysers have become very popular, a result of newer components making instruments powerful and less expensive.

It seems that a very common response to a newbie posting an antenna question on an online forum these days is "get an antenna analyser". One can't help thinking that most people who offer that advice are not very competent in exploitation of such instruments.

To exploit any measuring instrument, indeed to make sensible measurements requires an understanding of the instrument, it characteristics, its limitations and and understanding of the thing being measured. None of these "come in the box", they are acquired through research and experiment, and often not made easy due to the lack of supplier information.

Wise practitioners spend some time with an instrument measuring known loads to become familiar with an instrument and to expose limitations.

Principles of operation

Most of these instruments contain some type of directional coupler that may provide signals that permit calculation in the onboard processor of the magnitude and possibly the phase of the reflected wave at the instrument's reference plane, and display of a range of parameters based on those measurements.

The first generation of these instruments measured only the magnitude of the reflected wave, they were little more than a self excited VSWR meter.

The second generation determined phase of the load impedance to some extent, the methods sometimes did not determine the sign of the phase but merely its magnitude in the range 0-90° so they could not identify capacitive loads from inductive loads.

The third generation determined the phase angle of the load unambiguously.

Along the way, features like synthesis of the oscillator where added, ability to sweep a range of frequencies and display the results graphically, and facility to download the data to a PC for further analysis.

To some extent, the bells and whistles of sweeps and exported data are great productivity aids and add significant value to the instruments, but the principle of Garbage In - Garbage Out applies and none of these facilities materially extend the instrument's accuracy other than saving transcription errors.

The implementation of the directional coupler, the conversion of analogue voltages from the detectors to digital are key factors that determine accuracy of the instrument, especially away from the nominal impedance for which he directional coupler is optimised (often 50Ω).

An example:

If you have an instrument that has a 50Ω directional coupler, and you measure an unloaded mobile whip at 3.6MHz that is 20+j1000Ω, Γ (the complex reflection coefficient) is approximately 0.998∠124.1°.

Example: Rigexpert AA-230

Fig 1:

The Rigexpert AA-230 above is one of the third generation instrument with lots of features. Importantly, it uses a 12-bit A/D converter where many others use 8 or 10 bits.

The user manual makes no claims about accuracy of the instrument. Nechitailov from Rigexpert explained in an email:

The AA-230 is optimized for measurements nearby the zone of SWR=1. This means that loads with high R or X can only be measured with very low precision. Remember that AA-230 is designed for measuring beams on the tower, not for the lab work on the table.
Fig 2:

Fig 2 shows measured and expected values of R and X from a sweep of a o/c RG58C/U (Belden 8262) cable. Expected values are calculated using the model used for TLLC. The modeled expected cable length was increased beyond the measured physical length by 30mm to calibrate for the resonance at 95MHz, a compensation for the fact that the measurement plane is inside the instrument and there is an unavoidable length of transmission line to the DUT.

It should be noted that this is a very challenging load for these instruments, but a very relevant test as this type of instrument is often used to measure parameters of transmission line stubs such as this, and capacitors and inductors that have fairly high Q. In all these cases, the magnitude of Γ (the complex reflection coefficient) is almost 1 (0.94956 - 0.99981 in the case above).

Fig 3:

Fig 3 shows the same data as Fig 1 with the Y scale expanded to show low values more clearly.

The X curves are in fairly good agreement below 80MHz, but there is some departure above that frequency, partly due to the extreme impedance around resonance, above that possibly degrading frequency response of the coupler.

The R curves have large errors at some frequencies, whilst not too bad at others.

In summary, this is an extreme load though a quite practical case and the X measurements are fairly good though the cable measures 30mm longer than its actual length (measured electrically with a VNA and VIA). The R values are not dependable.

This doesn't make the instrument worthless on these high Q loads, just that while the reactance reading is ok, pretty good at the lower frequencies, the R value is not dependable and whilst you could use the instrument to measure X of a coil or capacitor, don't infer Q from the measurements. If using it to cut a tuned length of transmission line, allow for the fact that the measurement plane is not quite at the instrument jack.



Version Date Description
1.01 27/04/2013 Initial.