How to: SNA Measuring Reflection - Return Loss

RF Explorer product family offers advanced features when combining Spectrum Analyzer and Signal Generator working together as a Scalar Network Analyzer for 1-port and 2-port RF devices.

In this tutorial we will describe how to use RF Explorer to measure the <return loss> S11 / S22 of a device in dB. The return loss can be described as the fraction of signal reflected back to the source, and therefore we will use the term <reflection> in this document to make it easier to read by non-expert users. The reflection in dB is a parameter used to describe how good the impedance matching when compared to specifications is. In most cases we want to know the reflection of a 50ohm impedance, the higher the reflection, the worse the impedance matching.

The reflection is a measurement of how much energy incident to a device is being reflected back and, therefore, not entering the system. When measured in dB, the following values can be used as indicative:

  • Reflection 0dB: all RF energy is being reflected. This is an ideal open or short circuit, equivalent to VSWR=infinite.
  • Reflection -3dB: half of the RF energy is being reflected, and thus half of the energy is being received by the device. This is equivalent to VSWR 5.8.
  • Reflection -10dB: 1/10th of the RF energy is being reflected. Usually this is the threshold when most devices are considered to be tuned and have a reasonably good impedance matching. This is equivalent to VSWR 1.9.
  • Reflection -20dB: 1/100th of the RF energy is being reflected. This is a very good matching, expected for good designed and matched filters. This is equivalent to VSWR 1.2.
  • Reflection -30dB or less: 1/1000th or less of the RF energy is being reflected. This is considered exceptionally good matching, and usually found in lab grade devices such as precision attenuators and filters. This is equivalent to VSWR 1.07 or less.

As an example of real world reflection, a narrow band antenna is expected to reflect/reject the RF energy for frequencies that are far from the antenna center tuning frequency, intended by design so the RF input circuit is not overloaded by undesired signals. On the other hand, the narrow band antenna should not reflect but actually be a good receiver for the specific frequency it has been designed for.

Similarly, a good Low Pass Filter (LPF) should offer very small reflection (-10dB or less) for frequencies lower than cutoff frequency, and large reflection (very close to 0dB) for frequencies higher than cutoff frequencies.

With these examples in mind, we will learn how to use RF Explorer SNA to measure reflection in dB.

Items needed

  • RF Explorer Signal Generator models
  • RF Explorer Spectrum Analyzer model for the required frequency
  • 2 or 3 high quality RF cables. We recommend these ones.
  • A good 50ohm termination. We recommend this one.
  • The device to test (antenna, filter, etc)
  • Windows computer with latest RF Explorer for Windows installed.
  • And last but not least, a good directional coupler. We recommend Mini-Circuits ZHDC-10-63-S+ as a good price/performance device for frequencies range 50MHz - 6GHz with good directivity. If you do not need full wideband support but a more restricted frequency range, you can consider a lower cost limited bandwidth coupler. Select one with 30dB directivity or more for best results, and never use couplers with directivity lower than 20dB. Look at this article for other options.

(click on images to get a larger view)


This step is required so the components are calibrated for use with the SNA and all signal references are internally recorded by RF Explorer for Windows. All measurements done with a SNA start with a normalization step, and reflection measurement is no exception.

The most confusing detail for novice users is how to correctly connect a directional coupler for reflection measurement. See example below on how to connect for normalization. Note the open (unconnected) INPUT port in the directional coupler.



  • Connect RF Explorer Signal Generator to the OUTPUT port of the directional coupler
  • Connect RF Explorer Spectrum Analyzer to the COUPLED port of the directional coupler
  • Keep the INPUT port of the directional coupler OPEN for normalization step. If you need a cable to later connect the antenna with the right orientation, then add the same cable to the INPUT port but keep the cable end OPEN.
  • Connect RF Explorer Spectrum Analyzer to the computer USB port
  • Load RF Explorer for Windows. The RF Explorer Spectrum Analyzer should be automatically connected.
  • Connect RF Explorer Signal Generator to the same computer, different USB port
  • Using the Signal Generator tab in the RF Explorer for Windows tool, find the right COM port and click on Connect. At this point both Analyzer and Generator should be connected and therefore the SNA functionality is available.
  • Select Power=-30dBm for directional couplers with coupled port of 10dB or less, as the one recommended. If you are using a coupler with a coupled port of 20dB or more, you may need to increase power up to -10dBm.
  • Select Start/Stop frequency you are trying to characterize response for. Select steps in the range of 25-50 to start with, you may want to increase to 100-200 for ultra-detailed graph, at the cost of slower trace updates.
  • Click on [Normalize SNA…] button, wait till the process completes.

To confirm the normalization works as expected, click on [Start SNA…] button using the same connections, you should see an almost flat line, with small noise like the graph below, usually in the range of +-0.5dB or less.



Measuring reflection for an antenna

An antenna is a 1-port device and therefore easy to connect. And that is the only easy thing we can say about measuring reflection in an antenna.

As a matter of fact, antennas are very tricky devices because while it is being measured, is influenced by all sort of items in the environment:

  • External unintended RF sources will interfere with measurements, as the antenna is receiving at all times
  • Walls and metallic objects will produce external reflection which are bounced back to the antenna while the SNA is performing measurement, distorting performance
  • Most antennas, such as monopole and dipoles, are influenced by ground. An antenna designed for a handheld 2-way radio, for instance, is adjusted for best performance when a human is holding it on his/her hand and, therefore, connected to a bench setup will not be the exact same environment where it is expected to work best.

Due to all this, high precision antenna measurements must be done in special RF chambers, where all external and internal radiations are dramatically reduced.

For most users, an anechoic RF chamber is not available and so only imperfect measurements can be done. That said, the closer we get to an environment where the antenna is expected to work in real life, the better. Keep it far from metallic objects, set it vertical with the help of a good semi rigid RF cable, and minimize external RF sources as much as possible (don’t measure antennas any close to your cell phone, WiFi AP, etc).

Connect the antenna to the coupler INPUT port, and click on [Start SNA…]



An example of the connection and graph for a 216MHz antenna is shown below.



Measuring reflection of a LPF

Filters are easier to measure than an antenna but, being 2-port devices, is important to remember it must be correctly terminated. In other words, one of the filter ports must be connected to a SMA 50ohm load, and the other port used for reflection measurement.

In RF terminology, one filter port is 1 (input) and the other is 2 (output). Most filters are bidirectional and therefore port 1 and 2 are expected to work the same. When you terminate port 2 with a load and measure port 1, the resulting measurement is S11 return loss. By doing otherwise, you measure S22. Most filters will exhibit S11 and S22 identical responses, so there is no real need to measure response on both ports except if trying to find problems in a faulty filter.

Below is a sketch of all items you need to first normalize, then measure your filter


Connect the filter (terminated with 50ohm load) to the coupler INPUT port, and click on [Start SNA…]



An example of the connection and graph for a 200MHz LPF is shown below.


5 minutes video tutorial

This video tutorial is only 5 minutes long and drives you through the steps required to measure VSWR of a 800MHz antenna.

Video Best viewed in HD 1920x1080 full screen


Additional notes and special cases




# FrankT 2015-03-17 07:49
Thank you for this step by step.
Reply | Reply with quote | Quote
# Ashley Booth 2015-03-18 20:14
Where can one buy a Mini-Circuits ZHDC-10-63-S+ directional coupler in the EU?

Reply | Reply with quote | Quote
# Webmaster 2015-03-18 20:16
Mini Circuits online shop has UK office, just doing normal registration to buy in EU they will price you in EUR or GPB and will ship from UK.
Reply | Reply with quote | Quote
# Ashley Booth 2015-03-19 10:22
Thanks! I've ordered one. Had to answer this: Is the end use a nuclear, chemical, biological weapons, or ballistic missiles application?

Reply | Reply with quote | Quote
# FrankT 2015-03-19 16:42
And it was a yes :-)
Reply | Reply with quote | Quote
# Z 2015-03-24 04:30
Quoting Ashley Booth:
Is the end use a nuclear, chemical, biological weapons, or ballistic missiles application?

When I was enquiring about purchasing one from the US I too was given an Export End User form which had these questions amongst others. However, when I decided to order locally for a SEDC-10-63+ on a demo board, no such questions. Maybe they figure the SEDC-10-63+ could not possibly be robust enough for any self respecting dictator or meglomaniac with aspirations of building an WMD arsenal :)

Reply | Reply with quote | Quote
# Z 2015-03-20 13:35
Thanks for the article, it has inspired me to try it out.

As this is purley a hobby for me I was hesitant spend the money on the suggested wideband directional coupler so I'm looking at buying the surface mount directional coupler (SEDC-10-63+) mounted on a demo board (TB-605+).

Looking at the graphs, the performance characteristcs are not quite as good but certainly compareable. The biggest difference is having to forgoe the robustness due to the lack of it being encased. However it works out to be about 50% of the cost of the ZHDC-10-63-S+, based on pricing through my local (Australian) distributor, and for a hobbist like me that makes it a winner.

Reply | Reply with quote | Quote
# Webmaster 2015-03-23 17:27
Excellent - thanks for sharing, other users may find this interesting lower cost solution.
Reply | Reply with quote | Quote
# Neutron 2016-11-02 21:56
Thanks Z for the suggestion of the SEDC-10-63 mounted on a TB-605+ demo board. I also couldn't afford the ZHDC-10-63-S+ and hoped this would be good enough.

But did you get it to work satisfactorily? Unfortunately I can't seem to get the figures quoted for directivity (worst case -23dB). I only get between -20 and -18 dB at 2400-2485 MHz for example. Plotting a graph like the one in the SEDC-10-63 data sheet, mine is consistently 5dB or more worse!
Reply | Reply with quote | Quote
# Kris McLean 2015-06-19 08:58
I've had trouble with antennas that have fairly low SWR but don't work that well in the field, in fact some of them do a very good impersonation of a dummy load. The problem is compounded by the very short on air time of the RF gear I'm trying to deploy.
Can I measure the radiated power of these antennas at say 10m range using the RF explorer with a known whip as the receiver and your sig gen as the transmitter & infer antenna gain that way?
Reply | Reply with quote | Quote
# Webmaster 2015-06-19 12:35
Yes, absolutely. Separate antennas a minimum of 3 wavelength and normalize with good known antennas, probably using max Signal Generator power (0dBm). Then replace the antenna on the spectrum analyzer side (receiver antenna) with the one you want to characterize and check insertion loss.

You will get then a good idea of how well the antenna radiates/receive, regardless VSWR.
Reply | Reply with quote | Quote
# Jimmy 2016-04-23 09:52
Can this method be used to test the antenna together with the antenna cable? As a Pro audio systems Integrator, we often need to install long runs of antenna cables to remotely situate the antennae. So I'm wondering if this method of testing can be used to ensure low return loss and thus hopefully indicate a proper installation of the antennae and their cables. Or is my thought process skewed?
Reply | Reply with quote | Quote
# Webmaster 2016-04-25 11:37
Yes, you can include your cable and that will check VSWR of the cable+antenna setup. If you find the VSWR is not the one you wanted, then you can separate cable and antenna to find which one is the offender. If you test a cable alone, make sure to terminate it correctly with a 50ohm load.
Reply | Reply with quote | Quote
# Bob 2016-07-30 02:26
Hi, I just acquired an Omnispectra DC PN 2049-6020-00. No data was supplied, as I acquired it via surplus for $45 plus shipping. attatached is link to DC image:
I'm not sure of the input or output. I have assumed that the Input is with the Female SMA with the Male pin on the left side w/respect to the labeling.
I just want to check the VSWR of my homemade 5.8GHZ antennas.
As a retiree, I just want to minimize expenditures.
Thanks for any help.
bob k
Reply | Reply with quote | Quote
# Webmaster 2016-08-01 20:15
Hi Bob, unfortunately is hard to know from the picture, it may not even be a directional coupler but a bias tee or some other device. Assuming is actually a DC, the suggestion is to follow our process documented at to check the performance of the DC in the frequency range of interest. If you find the right connection setup and the directivity is 20dB or better, you should be able to use this.
Reply | Reply with quote | Quote
# jimmy 2017-09-12 17:08
what is the calibration process for the device?
Reply | Reply with quote | Quote
# Webmaster 2017-09-13 00:09
Hi, not sure to understand which calibration do you mean. If it is the test calibration, that is done as part of the SNA process called "Normalization", where all the underlying losses and non-linear responses are mathematically neutralized.
If you mean the actual device calibration at factory, that is done with internal bench lab calibration tables for every unit.
Reply | Reply with quote | Quote
# Matthew 2018-05-04 16:51
How fast is the sweeping for a resolution of 10kHz? What is the processing time for storing a output data? Can the sweeping be programmed to automatic sweep with command line?
Reply | Reply with quote | Quote
# Webmaster 2018-05-06 12:30
Sweep speed depends on how many sweep points you select. As an example the typical 50 points sweep can take about 2secs sweep time.

Tracking SNA is only available through the RF Explorer for Windows application, currently no command line option for that.
Reply | Reply with quote | Quote