I am interested in measuring the output impedance of an high performance DCDC converter together with its board.
Up to now I referred to this document, a very well written one.
I am referring to fig15, here it is for ease of reference (thanks Keysight!)
The DUT is the output of the DCDC converter.
How doest this work? A disturbance is injected on the DC output voltage of the converter by means of the rightmost signal generator. Voltage across the DUT is measured by \$V_T\$, while \$I_{DUT}\$ is measured by \$V_R\$ thanks to the \$1\Omega\$ shunt resistor. Of course \$V_x\$ are vectorial voltmeters, so when the division is performed to compute impedance the number you'll get is, in fact, the impedance and not just the resistance.
Now this all looks very good and nice to me, I understand that a Network Analyzer demodulates the measured voltages at its inputs, getting rid of the DC component inevitably present at \$V_T\$.
But what about ripple? Our DUT can be modeled by a DC generator, a somewhat random ripple generator, and an impedance, all in series. This ripple voltage will be measured by \$V_T\$ but as far as I understand no current would flow in R.
How does this phenomena influence the final measure?
One could argue that ripple has its power concentrated in a narrow bandwidth but some converters out there purposely spread this power over a larger bandwidth, because of EMI issues. How would you deal with that?