How to avoid mistakes with a VNA
Is this antenna good or bad, and for which frequency is it useful? A question I am often asked. Because a lousy antenna reduces the range
of a device considerably. Or another question: Did the supplier cheat
on this filter? At the end of this video, you will be able
to answer these questions with confidence. And you know everything you always wanted
to know about the Smith Chart. For about 50 dollars. Is this a good deal? Grüezi YouTubers. Here is the guy with the Swiss accent. With a new episode and fresh ideas around
sensors and microcontrollers. Remember: If you subscribe, you will always
sit in the first row. If you use antennas or filters, your most
important tool is a Vector Network Analyzer or VNA. It is essentially like a multimeter or an
oscilloscope. But until recently, nearly nobody had one
of those because its price range started at around 10’000 dollars. With the appearance of these nanoVNAs, the
price dropped to way below 50 dollars. The first generation had a limited frequency
range. With the new V2 generation’s range of 3GHz,
such a device is a must for every advanced Maker—time to learn how to use it and avoid
the most common mistakes. In this tutorial, we will cover:
- What is the difference between a VNA, a Spectrum Analyzer, and a VSWR meter? - Which questions can be answered with a VNA? - How does the VNA display its results and
how to read them - Why is it important to know the limitation
of the cheap VNAs and how can we overcome them? - Why and where do we need to calibrate our
instruments, and what happens if we do not do it? - How is the quality of those devices? Do they provide correct information? - Which additional software packages are available? First question: What is a VNA? It generates a signal and measures the amplitude
and the phase of the signal coming from the Device Under Test or DUT. Simple. Other than the ohm meter, which only can measure
resistance, it also can measure capacitance and inductance because it also measures the
phase between voltage and current. This is the main difference between a VNA
and a Spectrum analyzer with a tracking generator. Spectrum analyzers only measure amplitudes. But the Spectrum analyzer can characterize
external signals from transmitters. What problems can be solved with a VNA? Let’s assume a simple system consisting
of a transmitter, a cable, and an antenna that emits the power into the air. Of course, we want that all power is emitted. This happens if the transmitter’s output
impedance is well adapted to the cable’s input impedance and the cable's output impedance
to the antenna. Matched means they have the same impedance. Generally, this impedance is standardized
to 50 or 75 ohms. For this video, I will concentrate on 50 ohms
systems. If these impedances do not match, signals
are reflected, do not reach the air, and heat the transmitter. In the old days, we had VSWR or short SWR
meters for that purpose. They measured the forward and reflected power,
as we see here. If there is no reflected power, the SWR is
1. If we want to know more, we use a VNA to measure
the different parts' impedances and decide if the match is good or bad and what we have
to change it. Cool. Unfortunately, a few errors can happen during
this process. And we need a little theory to avoid them. How does a VNA work? It has one or two connectors, called ports
or, in the case of the nanoVNAs, channels. Let’s first focus on port one. This port emits a signal and, at the same
time, measures the amplitude and the phase shift, short phase, of the reflected signal. If you know the amplitude and the phase of
the signal, you can calculate the impedance of the device under test and show it on a
screen. It is essential to know that these are the
only measurements a VNA does on port one. The rest is just displaying these results. For example, this VNA shows the the impedance
as well as the amplitude for one particular frequency. Nice. But most of the time, we want to know the
course of these values across a frequency range. This is why VNAs draw amplitude and phase
curves across frequency ranges. On the nanoVNAs, the amplitude is called logarithmic
magnitude or LOGMAG and the phase is called “PHASE.” The amplitude curve is more famous because
it is used to determine the quality of a match. As we saw before with the SWR meter: The higher
the amplitude of the reflected power compared to the forward power, the worse the match. LOGMAG and SWR show the very same thing, just
for a different community, as we will later see. The main differences between an SWR meter
and a VNA are: - The VNA can measure and display also the
phase of the signal - Because the VNA has a built-in transmitter,
it can sweep frequencies and show curves - The SWR meter often is built for much higher
forward power So, a one-port VNA like this N1201A is sufficient
to characterize antennas. Why does this nanoVNA have two ports? The second port is a receiving only port. For filters, we are not only interested in
the reflected power on port one; we are more interested in the curve between the input
and the output of the filter. And this is why we need this receiving port. The rest is the same. We can display amplitude and phase on port
two the same way as on port 1. But here, we see the measurements of the signal
coming off the filter, not the reflected signal. BTW: The reflected signal is called S11 because
the VNA transmits and receives the signal on port one. The signal measured by port two is called
S21 because it is transmitted by port one and measured by port2. There are also S22 and S12, which can be measured
by exchanging port 1 and 2 on the nanoVNA. So far, we only dealt with simple curves across
a frequency range. How does the VNA display impedances? It uses the Smith Chart. This is a Smith Chart: Scary! But it is handy and vital. Each point on the diagram represents an impedance. We go through the three most important points
first: A pure 50-ohm resistance shows up right in the middle. This is the holy grail where we want to be. All shooters know precisely what I am talking
about. The next point is here: Nothing connected
to port1 or “open.” And a short-circuit or “short” shows up
here. You will use these three points over and over
if you work with VNAs. All points on this line from the “short”
via the 50-ohm to the open are purely resistive. I show you that in an example. I have a 500 ohms potentiometer connected
to a nanoVNA. You see what happens if I change its resistance
from 500 to 0 ohms: The point moves along the centerline. Only in the end, it turns away from the line. Why? Because the potentiometer also behaves like
a coil and therefore has an inductance, which is added to the resistance. I did not know it, but the VNA detected it
as a small phase shift between voltage and current. A sign that an inductance is in the play. Why do I know that this is an inductance? Because I know that above the centerline,
we find all inductances. Below the line, BTW, we find all capacitances. You do not believe it? Here I connect this beautiful variable capacitor
I bought for my tube radio from video #355 It starts nearly as an open, which is quite
clear if we look at it: The plates are not connected and are far from each other. If I turn the capacitor, its capacitance gets
bigger, and we see the point moving along the outer circle below the centerline. Unfortunately, I do not have a moving coil,
but the same would happen with it, just on the opposite side. Again, the measured values are the same as
before: Amplitude and phase. The VNA only shows more information. If we add a capacitor in parallel to the potentiometer,
we can pull the point back to the centerline. You see, we can use a capacitor to “neutralize”
the inductance of the potentiometer. I do not need to care about the values; the
VNA shows me what happens and when I have to stop. Cool! But, unfortunately, I cheated on you. I only used a frequency of around 10 MHz. If I change the range from 1 to 20 MHz, things
become different. Suppose I leave the two parts where they were
before I get now a curve. Why is that? Capacitance and inductance change with frequency
even if the part does not change. And this is what the VNA shows us without
any effort. For example, we see that the curve crosses
the middle line a second time at 1 MHz in addition to the 10 MHz we saw before. And it nearly hits the “short” position
at 20 MHz. So our potentiometer and capacitor behave
like a piece of copper at 20 MHz. Strange but true. Now we can try to answer the question: Would
this be a good match if it were an antenna? Ask the shooters how much points they get
if they hit at those positions. The Smith Chart is precisely the same: The
distance to the sweet spot shows us the quality of the match. Here we would have the best match at the 10
MHz with the closest distance to the 50 ohms sweet spot. As seen before, we can also check the quality
of the match by looking at the SWR curve. It shows the distance to the sweet spot. And really, it is very flat up to about 15MHz. Then it increases rapidly because the distance
gets bigger and bigger. At the sweet spot, the SWR would be one. Now it is around three—a bad match. You see, we do not need a Smith Chart just
to know if an antenna is good or bad. The SWR is sufficient for that purpose. When do we use the smith chart? When we want to know how we can compensate
for a bad match, for example. And for calibration, as we will later see. BTW: If you talk VSWR to an RF engineer, he
immediately knows you are a Radio Amateur Operator. He would never use this word, he uses “return
loss” in dB, which is the same, just different. And of course, it sounds more professional. You can also display the return loss on a
nanoVNA. It is the LOGMAG from before. And it is around -6dB at an SWR of 3. Now we have the basics to start working with
a VNA. The most important thing you have to know
is “calibration.” Never use an uncalibrated VNA. Never! It will show wrong values and fool you. There are two reasons for calibration:
1. The VNA itself has lots of non-linearities
2. The measurements heavily depend on where you
attach the DUT You probably asked yourself: Why is it possible
to drop prices from 10’000 to 50 dollars. Is the quality of these devices ok or do they
display crap? And here comes the calibration into play. Even if those cheap instruments are not accurate
at all, you still can get good results. How is this possible? If we use well-known DUTs and tell the instrument
their exact value, it can remember this fact. For example, if I attach an exact 50 ohms
resistor to port one and tell the instrument that it is 50 ohms, it will later recognize
a well-matched antenna and also displays 50 ohms. So far, so good. But do we need to have a ton of resistors,
capacitors, and inductances to calibrate the instrument on all points of the Smith Chart? Fortunately, not. We only need three: An open, a short, and
a 50-ohm load. If we calibrate even a cheap instrument at
these 3 points, it will display all other points with acceptable precision. But do not forget: If we omit this calibration,
these cheap instruments can show very wrong measurements! Also, the professional ones, BTW. The calibration process is relatively easy. First, choose the frequency range you want
for your calibration. This is very important because the nanoVNA
invalidates a calibration as soon as you change the range. Then we go to the calibration menu and start
with connecting the “short” to the VNA. After waiting for a short time, we can go
on with the open and the load. This is called SOL calibration. If we connect both ports and tip “trough,”
we get a SOLT calibration. Now we can save it in one of the seven memories. Here the instrument shows us if it is calibrated. Without this “C,” it is uncalibrated. As we saw before, the impedance, as well as
the sensitivity of the instrument, changes with frequency. This is why we have to calibrate our instrument
at all frequencies used by our measurement. You probably remember: I said you would see
these three points over and over. Now you know why. We have to calibrate the instrument every
time we change the frequency range. You might say: Why not do a wide-band calibration
at the beginning? Then the instrument should know its calibrations
at each frequency. Theoretically, this is true, and the N1201SA
does it like that. But those small devices only measure 101 or
201 points per scan. If you start at 1 MHz and end at 3GHz, you
get a calibration point each 30 MHz. Which is more or less useless. Most antennas, for example, have a much smaller
usable range than 30 MHz. BTW: If we switch the calibration off, we
get very different and wrong results. We increased the precision of our device by
factors using this simple calibration method. This is why you also have to order your VNA
with a calibration set. According to Kurt Poulsen, a calibration specialist,
the SMA calibration sets coming with the nanoVNA V2 are quite good up to 3 GHz. Which is very astonishing. A professional calibration set can easily
set you back 2500 dollars and more. This was the first reason for calibration. The second is the calibration plane. If we calibrate our instrument with the standard
set, we establish a calibration plane right at the SMA connector. Maybe you asked yourself why we have a special
“open” calibrator? Because there is a difference of a few millimeters
between “nothing” and “open,” especially for higher frequencies. As soon as you add a cable or an adapter to
your VNA, the calibration is no more accurate. Here is a simple example: This antenna only
has a female connector, which does not match the female connector on the nanoVNA. I can solve this problem by adding this simple
adapter. Look what happens if I add the adapter: The
open point is shifted considerably because the signal travels to the newly established
plane and back. Our instrument is no more calibrated. I have to calibrate my instrument at the newly
established plane. But if I have no female calibration set? Fortunately, the nanoVNA can correct this
error by using a “port extension,” called “electrical delay” on the nanoVNAs. If I enter 165 picoseconds, the open is back
where it should be. Good enough for us Makers. If you do not trust me, you can visit Allan’s
W2AEW channel. He covers that and many other specialties
of the nanoVNAs. The same but much worse applies if you want
to measure on this Demo kit because you need a pigtail to connect the VNA to these tiny
U.FL connectors. Now we have two possibilities: Because the
board has an open, a short, and a load, you can calibrate on the board or use the port
extension as before. Of course, with a much longer delay of 1.95ns. Both establish a calibration plane at the
end of this pigtail. Here are the results: With calibration on
the board as well as with the port extension, the measurements are very similar. The result with the calibration plane at the
SMA connector of the VNA is very different and wrong. With all this knowledge, we can answer our
question from the beginning: Which antenna is better? They look very similar: Same manufacturer
and same type. They should work on 144-148 and 430-440MHz. As shown before, I established the calibration
plane on the adapter for frequencies between 100 and 500MHz. For an overview, this is ok. I use 201 measuring points and get a space
between two measurements of 2MHz. Enough to check if the antenna is terrible
but not enough to see if the antenna is good. Both antennas have two “dips.” Unfortunately, the dips of the fake one are
not deep enough. So it is not usable for our purpose. If we look at the Smith Chart, we see that
its curve always is in the capacitive area. Maybe adding some inductivity would get a
better response. The real one is far from an optimal antenna,
but its dips at least are in the right place. We could now calibrate our VNA around 144
MHz or around 430MHz to get better measurements for the two ranges. But I think you got the point. As said in the beginning, we can also measure
filters if we have a two-port VNA. We connect port one to the input of the filter
and its output to port two and measure S21. We have some filters on the Demo kit to check
the effect. But before we can do so, we have to calibrate
the “through. This is either done with such an adapter or,
in the case of the demo kit, with this through. After establishing a SOLT with a calibration
plane at the kit and for a frequency range of 5-15MHz, we see the expected dip at 6.5MHz
as well as the peak at 10.7MHz for the other filter. In my lab, the highest frequency I use is
2.45 GHz for Wi-Fi. And this will stay for the next years. I do not think that our gadgets will move
to 5.8GHz Wi-Fi soon. So now it is the time to invest in this new
technology if not already done. A few words about the available software:
I used the VNA-QT software for this video. You need it also for firmware updates. It is simple and does more-or-less the same
thing as the firmware on the device. The nanoVNA saver software is much more elaborated
but also has a steeper learning curve. You find the links in the description. Summarized:
- VNAs create a signal and measure impedances of DUTs like antennas or filters. They measure amplitude and phase to determine
impedances - They display amplitudes, phases, and impedances
over a frequency range. Either in XY charts or in a Smith Chart
- The Smith Chart is made for system analyses. SWR or return loss curves help for quick checks
- To get the needed accuracy, VNAs have to be calibrated
- Use a SOL calibration for one port antenna characterization and SOLT calibration for
two-port filter measurements - Always check that you work with a calibrated
instrument. Otherwise, it will display completely wrong
values. For the NanoVNAs, the calibration range always
is the same as the measuring range - Always establish the calibration plane where
the measurement is planned. Either by connecting the calibration set at
this plane or by using port extension - The nanoVNA V2 is an ideal device because
it covers all ham bands, the 433 and 868/915MHz ISM bands, and 2.4GHz Wi-Fi. All we need. For 50 dollars
- It has two ports and therefore can characterize antennas and filters
- It is crucial to know that it only measures a maximum of 201 points in the range. Only with the PC software, you can extend
this number - So it is necessary to adapt the frequency
range to the questions you want to answer. For a quick antenna check, the full range
up to 2.5GHz is good enough. If you want to characterize a filter, you
have to use smaller frequency ranges - The most important limitation is the small
number of measuring and calibration points with the effect that we have to calibrate
the instrument way too often - Its limited dynamic range does not show
the quality of powerful filters As always, you find all the relevant links
in the description. I hope this video was useful or at least interesting
for you. If true, please consider supporting the channel
to secure its future existence. Thank you! Bye
