Much is written about ATU efficiency, about the need for them or not, and often in subjective terms like “lossy ATU”, and most of it lacking quantitative detail.
The little quantitative detail is almost entirely for purely resistive loads… as if that is typical of real life conditions.
The most common configuration used today is the ‘high pass T match’, but a range of other configurations are seen as being superior… though usually without quantitative evidence.
More Hams use MFJ-949s than any other antenna tuner in the world! Why? Because the worlds leading antenna tuner has earned a worldwide reputation for being able to match just about anything.
… so let’s make some measurements with a reactive load on a MFJ-949E. Capacitive loads tend to be very common for antenna systems at lower HF, so let’s choose a load of 50Ω with a 100pF silver mica cap in series at 3.6MHz. The reactance of the cap is -442Ω, so the load is 50-j442Ω, and the 50Ω part is a RF power meter (RFPM1).
The test setup then is:
- a standard signal generator (SSG) on 3.6MHz with 20dB precision attenuator so that we are confident that Zs=50Ω (important to the adjustment of the ATU for maximum power as indication of 50Ω match);
- 100pF silver mica capacitor (low loss);
The SSG was adjusted for -10dBm out directly into the RFPM1, then the ATU+cap inserted and ATU adjusted for maximum power indication. Power indicated was 1.4dB lower, so InsertionLoss and TransmissionLoss are both 1.4dB.
Above is a simulation of the T network in RFSim99, component values are adjusted for a match and inductor Q is calibrated to the measured loss of 1.4dB.
Above is an estimate from T Match efficiency estimator using the values from the simulator, but which also reconcile with previous calibration of the ATU controls. This calculator is a simple approximation and is not as accurate as RFSim99, however it is within a tenth of a dB. The measured and simulated values confirm that approach taken by the T Match efficiency estimator .
Above is a screenshot from W9CF’s tuner applet adjusted for zero capacitor loss, Ql=100, and matched for the B switch setting of 17µH. Again the loss reconciles within a tenth of a dB.
A similar test was done without the 100pF capacitor, ie for a purely resistive load of 50Ω and the measured efficiency was 0.6dB, so the effect of the series capacitive reactance was to reduce efficiency a further 0.8dB (17%).
All of these methods can be used to estimate the efficiency of an ATU on a given load.
Above is the W9CF simulator with a load of 10Ω with 100pF in series. It doesn’t quite tune using the 17µH B setting, so needs the 26µH A setting. Manually tuning the capacitors for perfect match, the efficiency is -4.5dB (35.6%). With the ATU dissipating 65% of the input power under these conditions, you can understand why hams damage them.
Have a look at A look at internal losses in a typical ATU for thermographs of the internal heating of a MFJ-949E under quite good efficiency.
Above is a simulation of a similar ATU on 80m, the worst losses commonly encountered are with low R or low X (meaning large negative values) or worse, both. The graph shows how inadequate a pure R load analysis is as it is just a slice through the X=0 plane from this graph.
Another perspective is to consider the variation in loss with feed line length for a given VSWR.
The above chart shows that the location of the ATU on the standing wave can make a large difference. The extreme values of VSWR are often encountered in multiband antenna systems with ‘tuned feeders’. The jagged lines illustrate the compromise of a switched inductor, but that does not damn the very popular roller inductors which tend to have lower Q and therefore lower efficiency.