Vector network analysers (VNA) come in all sort of price ranges. Most of them are out of reach for a typical hobbyist. Very few VNAs are available to be bought by someone like you and me. For example, the $500 miniVNA and the sub-hundred dollar nanoVNA are very much affordable in comparison to the thousands of dollar equipment available out there. In this post, we will be looking at the two EXTREMES available in the market today; the sub-hundred dollar nanoVNA and the Keysight fieldfox N9952A.
The Keysight Fieldfox N9952A can operate over 300kHz to whopping 50GHz. It offers a dynamic range of 70 to 100dB and comes with plenty of features such as Time domain reflectometry, adjustable output power, adjustable IF bandwidth, and so on. In short, the N9952A has everything a professional RF design engineer requires. Additionally, this device can also operate as a Spectrum Analyzer mode to observe the frequency spectrum. Let us get to the comparison real quick. In this article, we will be testing a a low pass filter and an antenna and then comparing the measurement results from both the instruments.
Calibration is the first step before you even begin to use your VNA. To do so, we will use respective calibration kits that come with the two devices. In the case of the Keysight VNA, we will be using the 85561A 2.92mm 4 in 1 calibration kit. Whereas, the individual open-short-load will calibrate the nanoVNA.
I made sure that the devices were warmed up for at least 15 minutes prior to calibration. Furthermore, the calibration process takes approximately five to ten minutes. Once calibrated, the ideal return loss response should be a flat line indicating 0dB across the frequency span of interest. Let us find out how our VNAs performed in this test.
After calibration, the Port 1 of the VNA was left open. Obviously, the Keysight VNA shows a flat line whereas, nanoVNA seems to be struggling with ringing in the return loss plot. It appears that the nanoVNA may not handle full span calibration very well. Therefore, I further try reducing the span and calibrating it again. Let us look at the calibration done for 10MHz to 500MHz span.
Even after reducing the span, we still see the ringing while keeping the port open-circuited. Furthermore, a constantly varying spike also appears at 300MHz. Looking at the circuit of the nanoVNA, we come to know that the stimulus generator works only up to 300MHz. Beyond 300MHz, it uses the harmonics for calculating S-parameters. As a result, we see reduced dynamic range above 300MHz. Nevertheless, the readings are still good which we will see further in this blog post.
Looking at the graph carefully, we see a return loss error in the range of 0.5dB to 0.7dB. For a hobbyist, this error is negligible. You can't complain about this kind of error for the price you are paying. In my view, nanoVNA PASSES the calibration test.
In order to obtain a very good calibration, I reduce the span further in the range of 100MHz to 200MHz and only then do I see a very flat graph.
As you can see the problem lies only in very large spans. The operation is flawless when we bring the device in its comfort zone.
Testing a filter
Observing filter response can be an interesting thing to do on a VNA. I have a Minicircuits SLP-200 Low pass filter having a 200MHz cut off frequency. Let us observe its response on both the VNAs at once.
Return loss performance of the low pass filter on both the network analyzer looks absolutely close. An error of approximately 2dB exists in certain frequency regions. Nevertheless, nanoVNA being a low-cost device is able to give us a pretty fair idea about the filter performance. Huge thumbs up to the nanoVNA!
Now that we have observed the return loss, let us have a look at the insertion loss or the S21 measurement.
Keysight VNA definitely has a huge dynamic range and we are able to observe that in the cut of the region of the filter. On the other hand, the nanoVNA isn't able to show true suppression characteristics of the filter. In my opinion, the nanoVNA works very well up to 300MHz and then takes a hit past that. Even so, the measurements are quite spot on.
1-port antenna reflection coefficient test
I connected a 400MHz antenna to both the devices and observed the performance.
My previous concern appears to be coming true. The device performance starts to degrade past 300MHz. We can see a major difference between the two readings. Although, both measurements clearly show the antenna resonance point at 400MHz, the -10dB bandwidth does differ by several MHz.
Apart from a few errors here and there, the nanoVNA appears to perform very well. As they say, "Something is better than nothing". The nanoVNA is something to give us a rough idea about our antennas, be it RF filters or something else. There are many things that could improve in the nanoVNA circuit. For example, the Si5351 could be replaced with a better performing wide band PLL. The RF circuitry could be better shielded to improve the dynamic range and reduce high-frequency errors. The list goes on and with the list, so does the cost. We can't have it all. So, we all have to settle somewhere and I prefer to settle with the $60 nanoVNA than a $50,000 FieldFox.