Fix poor capacitor, inductor, and DC/DC impedance measurements
When designing or optimizing a VRM (voltage Regulation Module), we need its output impedance data and impedance data for the filter inductors and capacitors for us to have complete simulation models. Unfortunately, vendor data on these components is often incomplete, erroneous, or difficult to decipher in terms of the setup involved to make the measurement. Thus, we often need to collect the data ourselves.
The measurements need to be performed over the entire frequency range of interest, typically from a few kiloHertz to about 1 GHz, depending on the application. Because of this very wide frequency range, we generally turn to S-parameter based measurements. High performance simulators can directly incorporate the S-parameter component measurements in AC, DC, transient, and harmonic balance simulations while including the finite element PCB models.
While extremely useful, standard S-parameter measurements frequently aren't sufficient. What's really needed is an extended range, that is, a partial S2p measurement. I'll explain why you need it and how to make this improved measurement.
S-parameters are a simple method of performing measurements over a very wide frequency range. The measurement is performed using a fixed resistance port rather than a high impedance probe. Two options are available for measuring impedance with S-parameters. One option is a reflection measurement and the other option is THRU measurement.
One port or two? Why partial?
The reflection, or 1-port, measurement is the simplest because it requires only one cable. But, it also requires a complex calibration, generally consisting of an OPEN calibration, a SHORT calibration, and a LOAD or MATCH calibration of the port used for the measurement. Most VNA's (vector network analyzers) include the transformation from the S-parameter reflection measurement (S11 or S22) to impedance, but it's quite simple. Using port 1 as an example, the reflection for a given reference impedance, Zref, (typically 50 Ω) the relationship between S11 and the device impedance is shown in Table 1.
Table 1. 1-Port Transformations
We can perform a 2-port measurement by placing the device to be measured either in series with or in shunt with the measurement ports. Table 2 shows the relationship between S21 and the device impedance for both the series and shunt configuration.
Table 2. 2-Port Transformations.
DC Ground Loop
An additional issue arises for the 2-port shunt-thru measurement due to a DC ground loop, which occurs as a result of the RF ground at the VNA and the series resistance of the interconnecting measurement cables. The Keysight E5061B VNA has a semi-floating input on the low frequency gain-phase ports, eliminating this DC ground loop for low impedance measurements up to 30 MHz. For the E5061B high frequency ports and other VNAs in general, the DC ground loop must be minimized using a common-mode coaxial transformer such as the Picotest J2102A. Otherwise, the low frequency measurements will be inaccurate.
The setup diagrams for these impedance measurement options are shown in Figure 1.
Figure 1. Basic Schematics for the 1-port and 2-port impedance measurements.
The simulations in Figure 2 show the S-parameter magnitude for each measurement technique as a function of the device impedance. All measurements lose sensitivity as the S-parameter magnitude approaches 1.0
Figure 2. S-parameter magnitude as a function of the device impedance.
Figure 3 shows a higher resolution view of the S parameter magnitude from 0.95 to 1.0.
Figure 3. Higher resolution view of the S-parameter magnitude as a function of the device impedance.
Setting the measurable S-parameter (either S11, S22 or S21) to a minimum of 40E-6 allows for a reasonable signal-to-noise margin and a maximum of 0.95. The ranges for each measurement are shown in Table 3.
Table 3. Measurement Impedance ranges.