other reflection coefficient, using a separate VNA reflectometer; an example measurement is shown in Figure 2.11, where the internal source is set to 1 GHz. Here, the internal VNA source frequency is shown as a large spike at 1 GHz on the source output‐impedance measurement. This is again a case where the coupler in the R‐channel used to sample the incident wave provides better match (lower trace) than a two‐resistor splitter (upper trace), at least at low frequencies where mismatch after the reference splitter is minimal.
Figure 2.11 Measured source output impedance away from the source frequency: a trace using a power splitter in the R channel (upper) and a trace using a coupler (lower).
2.2.3 VNA Test Set
2.2.3.1 Test Set Switch
In some VNAs the source is switched between ports using a test set switch, which can come before or after the reference channel splitter. The termination of this switch provides the load match of the port when the source is not active on that port. This load match is not the same as the source‐match (ratio or power) and so some advanced calibration techniques that rely on the port match being consistent whether the port is a source or a load must be modified, as discussed in the next chapter. If the switch comes before the reference channel splitter, there will be a reference channel receiver for each port (four‐receiver VNA, Figure 2.2 lower). If the switch comes after the reference channel splitter (three‐receiver VNA, Figure 2.2 upper), the reference channel is shared between ports. It samples the source signal only when the source is active. This three‐receiver architecture does not support some calibration methods, such as thru‐reflect‐line (TRL) and so modifications and compromises to the calibration methods must be made.
The short explanation for the difficulty is that TRL calibration methods require measuring the load match of port 2, when port 1 is active. To do this measurement, the ratio of a2/b2 is acquired during the thru step. But there is no a2 receiver available in the three‐receiver architecture. Modifications can be made that assume the source‐match and load match of the port are identical, but this case is not common unless attenuation is added after the reference channel split. Attenuation added reduces the difference between source and load match at a port by twice the attenuation value. Pre‐characterization of the difference in source and load match, called the delta‐match, can be performed and removes the need for characterization at the time of calibration. This allows three‐receiver architectures to support the same calibration as four‐receiver architectures and is found in some of the more modern low‐cost analyzers.
2.2.3.2 Step Attenuator Effects
In some VNA designs, a step attenuator is added between the reference coupler and the test coupler to allow a greater change in the source power setting beyond that which the source ALC circuit can produce. This step attenuator has the additional benefit of providing a good match to the test port. In cases where the power source‐match and ratio source‐match are not the same, the step attenuator reduces the difference between these values based on twice the attenuation value. Reducing the difference between power source‐match and ratio source‐match allows one to compute the error in source power from the ratio source‐match, which is determined as part of the normal calibration process. In general, the power source‐match is not characterized during any normal calibration.
Another issue to be concerned about with step attenuations is their effect on the quality of the measurement when the attenuation value is changed. In most newer VNAs, the nominal attenuator value is known, and the effective value of the reference receiver is compensated for when the attenuator value is changed. The source ALC power is also changed, so that changing the attenuator value causes only a slight change in the value of the power coming from the test port; the nominal attenuator value is usually within 0.25–0.5 dB of the actual attenuator value. Since the port power stays the same, the internal source power must be raised by the amount of increase in the step attenuator. Since the reference receiver comes before the step attenuator, it will see a larger signal value; as it is desired to have the reference receiver power display the same value as the port power, its reading is also decreased by the value of the step attenuator.
Placing the step attenuator at this point in the block diagram has a distinct advantage in that it allows a large signal in the reference channel even when a small signal is needed at the test port, providing a low noise signal.
The loss of the step attenuator is well compensated for, but its effect on the match is not. If the preset condition has the step attenuator set to 0 dB, the match of the port is terminated back in the test port switch. When even a single stage of test port attenuation is used, the predominate source and load match characteristic is set by the match of the attenuator, which is typically quite good. Thus, it is good practice to use some source step attenuation if the maximum test port power is not required for the measurement. And since the raw match is better for any attenuator setting other than the 0 dB step, the effect on a calibrated measurement if the attenuator is switched to a different, non‐zero, value is smaller if the calibration is performed with some attenuation applied.
Some older VNAs did not allow changing the attenuation value after calibration and did not compensate for the nominal value of an attenuator change; error correction in these VNAs is often turned off for a step attenuator change.
In general, changing a step attenuator will change all the raw error terms on the port that has the step attenuator. Techniques are discussed in Chapter 3 that can compensate for much of this change.
2.2.3.3 Test Set Reflections
In addition to the source‐match effects produced by the source impedance, power source‐match, and ratio source‐match, reflections from within the test set of the VNA will also exist, as well as from the test port cables and from any fixtures that provide an interface from the VNA to the DUT. These sources of mismatch are common to all of the previously mentioned source‐match effects and will add to them in a similar way. However, since they are common, their effects on port power and gain are also the same.
The reflection and mismatch between the reference channel split and the test port coupler affect the incident signal, a1, but are not monitored by the reference channel receiver. Reflections after the test port coupler also affect the a1 signal but will be apparent in changes measured on the reflected signal, b1. However, their composite effect will add to the overall source‐match, and their effects on measurements can be compensated provided they remain stable. In addition, mismatch and loss after the test port coupler can be characterized in such a way that changes to these values, such as due to drift in a test port cable, also can be compensated in some cases. Mismatch correction in power measurements is discussed in detail in Chapter 3.
2.2.4 Directional Devices
One vital VNA component is the directional device used at the test port to separate the reflected wave from the incident wave. This is most often a directional‐coupler or directional bridge, although simpler structures have been proposed as well. These devices are characterized by their main‐line loss (the attenuation of the a1 signal), the coupled‐arm loss (the attenuation of the b1 signal), and their directivity (the ability to separate the b1 signal from the a1 signal). In addition, any mismatch in the directional device will contribute