2.4 Example of a VNA source block diagram.
Generally, there is a fundamental oscillator that provides a swept frequency response over one or more octaves. Previously, the output was often switched or split to use as an local oscillator (LO) to a low‐frequency heterodyne source stage; many modern VNAs use direct digital synthesis to get the low‐frequency response and divide the fundamental oscillator to get the higher range of the low‐frequency path.
The swept frequency oscillator is typically phase‐locked to a lower‐frequency fractional‐N (F.N) circuit or direct‐digital synthesizers (DDSs). In older VNAs, the phase locking was accomplished through the reference receiver, which reduced cost but required that the reference path signal remain present for all measurements. Modern analyzers separate the frequency synthesis for sources from receivers for greater flexibility and no longer require the any signal be present in the reference path. Because of this, the RF signal may be pulsed modulated without losing phase lock on the synthesizer.
The output may also go to multiple stages of division or multiplication, followed by amplification and filtering. The final output signal is combined from each of the input signals, resulting in broad frequency coverage. Typically, this common output has some RF detector on it to provide for ALC loop operation, maintaining constant source power over the different frequency bands and compensating for amplifier flatness. But often the ALC attenuator drive can be operated in an open‐loop mode to extend the power range beyond that of the detector diode.
The level control circuits often use an amplitude modulator before the amplification chain to complete the ALC loops. In some modern VNAs, a pulse modulator is added as well to provide for high‐speed pulsed RF measurements. When a pulse‐modulator is used inside the ALC loop, the ALC function must be disabled as it will attempt to respond to the pulsed signals. In this open‐loop mode, more sophisticated calibration or digital control must be used to control the source output power. Recently the use of a reference or test receiver as the ALC loop power control, rather than the internal diode detector, has become more common. This form of receiver leveling provides accurate results if the receivers are calibrated, a much wider range of leveling than the diode detector can support, and the ability to be programmatically controlled to correct for any external path loss or to provide a prescribed power profile.
2.2.2 Understanding Source‐Match
One of the most confusing issues with respect to VNA measurements is the idea of source‐match. Source‐match also affects non‐VNA measurements when signal sources are connected directly to a DUT, but in such cases it is almost always ignored as its affect cannot be determined; but in VNA measurements, the port 1 reflectometer can help to determine the exact effect of mismatch between the source and the DUT and correct for these errors. In fact, there are three different and distinct source attributes that are often confused as the source‐match of a VNA, and each affects different measurements in different ways.
2.2.2.1 Ratio Source‐Match
The ratio source‐match is that match that will affect the results of a ratio measurement, given a DUT is not perfectly matched to the reference impedance. The value of the ratio source‐match, most commonly called the raw source‐match or uncorrected source‐match, is derived from a combination of the quality of the reference channel signal‐separation device and any mismatch between this device and the input port of the DUT. This is always the value used to compute the uncertainty or accuracy of a gain or return loss measurement. But this match applies only to parameters that have a ratio of some receiver to the reference receiver. In this case the reference receiver is used to measure the change in source drive power incident to the DUT, and the error‐correction math removes the ripple at the output of the DUT associated with the measured incident ripple in power. Thus, for linear measurements such as S21, where the output signal is linearly related to the input signal, the ratio is not affected by changes in incident power to the extent that the reference receiver properly measures the actual incident wave at the DUT.
Errors in the measurement of the actual incident wave are mostly attributed to the ratio source‐match of the VNA. The ratio source‐match can be determined during the calibration process and is shown in Figure 2.5 for two cases of reference channel signal separation: the upper trace is using a two‐resistor power splitter, and the lower trace is using a directional‐coupler. While the detailed response is different, the overall quality is quite similar between the two cases.
Figure 2.5 Ratio source match: trace when using a power splitter (upper) and trace when using a directional‐coupler (lower).
When a splitter is used, since the splitter uses equal 50 Ω resistors in most cases, the input match to the splitter (as it appears from the source) is nominally 50 Ω, and the loss through the splitter is about 6 dB. The mathematical process of taking the ratio has the effect of creating a virtual ground at the common node of the splitter, so the ratio source‐match is a measure of the quality of the internal 50 Ω resistor.
Interestingly, for the case where the reference comes from a coupler, in the absence of other sources of mismatch after the coupler, the ratio source‐match will be identical to the directivity of the reference coupler. This makes sense as directivity is a measure of the reverse signal leaking into the coupled port, and this signal adds to the reference receiver reading even though it is not part of the incident signal, thus causing an error in ratio measurements.
2.2.2.2 Power Source‐Match
The power source‐match is the value that describes how the output power of the source varies with the applied load. If the power match is zero (perfectly matched), then the output forward wave, a1, would not be affected at all by the load. If the reference coupler is not ideal (that is, it has some leakage), it is possible have an ideal power source‐match and a non‐ideal ratio source‐match. Conversely, if the reference coupler is perfect but the source has a mismatch, the ratio source‐match may be ideal, but there is a larger power source‐match error. In some ideally constructed S‐parameter measurement architectures, where all components are ideal, the power source‐match is not zero. Consider the block diagram of Figure 2.1 simplified in Figure 2.6, where a two‐resistor power splitter is used, and the source impedance is also 50 Ω.
Figure 2.6 Simplified diagram of source power match.
From the test port one sees a series of 50 Ω resistance (of the splitter), behind which is the 50 Ω source impedance in parallel with 100 Ω (50 Ω from the splitter, 50 Ω from the reference receiver, in series), to generate a power match of
(2.1)
as the Thevenin equivalent impedance. From this it is clear that the for the two‐resistor splitter case, even in an ideal case the power source‐match cannot be Z0.
In the case where a directional‐coupler is used in the reference channel, the nominal