When I characterise ultrasound transducers, I normally do it with a network analyzer in a shunt configuration - ie, I connect the transducer across port 1 of the network analyzer. This gives me plots of phase and resistance across a range of frequencies.

Now I want to see if the impedance changes when the stimulus voltages are higher than what the network analyzer can supply (I need ~15Vpp). To do this I will boost the output of the NA using a 40 dB power amplifier. I came across this document which describes different ways you might protect the NA. The second configuration in this document is the easiest to implement and uses an attenuator to dissipate the excess power. But I don't think this is what I want as it will also reduce the current going through the DUT:

Original configuration in document

I am wondering if instead of an attenuator I can use an RF sampler between port 1 and the DUT, whilst connecting the sample port to port 2 on the NA. The sample connection will have a 40dB attenuator, and the RF connection will have a very low impedance path to the DUT. Like this:

Proposed configuration for protecting NA

My questions are:

  1. Is this a good way to protect the NA?
  2. Will I get accurate measurements?
  3. What information will I not get doing it this way? The document mentions that I cannot get reflection measurements, but I don't see why the NA can't deduce that from the forward measurements going into port 2. (presuming my amplifier is fairly linear in phase and gain).

By the way, the nominal resistance is 70 Ohms in the region of interest. I care most about 100 kHz - 1 MHz range.

  • \$\begingroup\$ What does it say in your network analyser documentation? Have you tried contacting the supplier in case you need to ensure voltage levels are below a certain amount (to avoid smoke)? \$\endgroup\$
    – Andy aka
    Jan 9 at 16:17
  • \$\begingroup\$ What was the source for your 2nd diagram? Without being able to see the "documentation" you refer to it's hard to respond to that part of your question. \$\endgroup\$
    – The Photon
    Jan 9 at 16:57
  • \$\begingroup\$ @ThePhoton addressed most of the issues. But you asked about using the coupler "Is this a good way to protect the NA?" The answer is yes it is. Couplers like this are commonly used to measure/monitor the transmitter outputs of things like radar transmitters. Couplers like these are usually entirely passive and so are very reliable. \$\endgroup\$
    – SteveSh
    Jan 9 at 17:06
  • \$\begingroup\$ I learned today that an RF sampler term can mean two completely different things - a time domain sampler, and an energy pickup/coupler. Good to know. I looked at that RF sampler schematics and for a while was perplexed of how would it work as a time domain sampler :) TIL, so thank you! \$\endgroup\$ Jan 9 at 17:16
  • \$\begingroup\$ @Andyaka My network analyzer manual (helpfiles.keysight.com/csg/e5061b/index.htm) only says that I must not exceed 20dBm. \$\endgroup\$ Jan 9 at 19:15

1 Answer 1


The diagram you showed is for measuring a 2-port device, not a 1-port. Each port is shown as a single point on the DUT block rather than as two distinct terminals, and each transmission line is shown as a single line rather than as distinct signal and return conductors.

The set-up you want, (from the same Agilent app note) is this one:

enter image description here

I've erased a line from the original drawing because you are measuring a 1-port device rather than a 2-port.

This uses a directional coupler to pick off the reflected wave from the DUT's input port and return it to the 'R' port of the network analyzer. The 'R' port is available on a certain style of VNA that requires an external test set. If your VNA doesn't have an 'R' port, you could connect the reflection signal to the '2' port instead, and treat your DUT plus the directional coupler as a single 2-port DUT. You would then need to do some calibrations to determine how to scale the measurement to accurately determine the reflection coefficient of your DUT.

The difference between this arrangement and your proposal is that the directional coupler preferentially directs the reflected wave to its coupled output, while your "sampler" circuit doesn't distinguish between the forward-travelling wave and the reflected wave.

I should add one other comment:

Now I want to see if the impedance changes when the stimulus voltages are higher than what the network analyzer can supply (I need ~15Vpp).

If your device's behavior changes when the stimulus amplitude increases, then it is a nonlinear device. When you provide a stimulus at frequency f, the reflection will have components at 2f, 3f, etc. The basic VNA is, by design, only meant to measure linear devices, and won't correctly deal with the harmonics of the reflected wave.

You may have a "special" VNA designed for this purpose, but if you don't, you need to be very careful about how you interpret your measurement results...or consider doing the measurement with a different instrument.

  • \$\begingroup\$ The nonlinearity implied by impedance change can still be characterized using impedance measurement in narrow band like VNAs do it - no need to worry about harmonics, as long as the VNA filters them out well enough. \$\endgroup\$ Jan 9 at 17:18
  • 1
    \$\begingroup\$ @Kubahasn'tforgottenMonica, But you can't just read the impedance off the VNA screen. That's what I meant by "be careful how you interpret your measurements". \$\endgroup\$
    – The Photon
    Jan 9 at 18:09
  • \$\begingroup\$ @ThePhoton Thanks for your answer, it helps me a lot. (1) In your configuration it presumes that the amplifier does not distort the phase in the forward direction (or it must be calibrated out), correct? (2) My VNA has an R port, but none of the measurement modes (helpfiles.keysight.com/csg/e5061b/quick_start_guide/…) allow me to use just R + port 1, I must also use T - should I connect the forward here? Alternatively can the DUT be converted to a 2-port somehow? \$\endgroup\$ Jan 10 at 2:12
  • \$\begingroup\$ @ChrisAdams, an S11 measurement should be giving you the measurement on the R port...but read the manual for your instrument and the app note to be sure. \$\endgroup\$
    – The Photon
    Jan 10 at 7:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.