Spectrum Analyzers are very useful if you
work with radios. Unfortunately, they are very expensive. Until recently, when I saw this small device
for around 50$. Unbelievable. But is it any good? 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. Electrical signals are not visible to us. This is why we need instruments to see them. Simple devices like multimeters to see constant
signals or more sophisticated ones like oscilloscopes which show us how signals change over time. If we work with radios, typical oscilloscopes
are not capable of displaying very fast signals. This is why we use the so-called spectrum
analyzers to visualize what happens. In this video, we will:
- See, what such a spectrum analyzer does and how it works
- Have a close look at this cheap device and compare its readings with a more expensive
one - See the difference between spectrum analyzers,
Vector Network analyzers, and SDR radios Let’s get started with the first topic:
What does it do and how does it work? A spectrum analyzer shows the power of a signal
across frequencies. The x-axis shows the frequency, here, for
example, between 0 and 3 GHz. On the y-axis, we see the power of the signal
in dBm. Keep in mind: 3 dBm more means double the
power. 0 dBm is one milliwatt, and -70dBm is 0.00000001
milliwatt. A very small signal. This will be important to understand why such
devices are quite expensive. I use my Baofeng radio for a short demonstration:
If I press the transmit button, we see that the radio emits a signal on the intended frequency
of 145.550 MHz. But it also emits signals on other frequencies
called harmonics. And we see signals popping up which are not
emitted by the Baofeng. Typically, we see signals in the 2.4GHz Wi-Fi
band. The strength of the different signals is also
different and change if I change the distance from the antenna. This is one application for spectrum analyzers:
To see how our transmitter behaves if it emits on the intended frequency, and how much power. But how is this done? The principle is quite simple if we understand
one thing: Frequency mixing. Frequent viewers remember this principle if
they watched the “introduction to SDR” in video #286. If we mix two signals, the output of such
a “mixer” is two signals: Frequency1 plus frequency2 and frequency1 – frequency2. Frequency one, in this case, is called “local
oscillator, or LO. If frequency2 is similar to the LO frequency,
the difference signal has a frequency of nearly zero hertz. The sum is on double the frequency. The latter is not important for us, and we
have to filter it out. If we measure the power in the signal around
zero hertz and show it in a graph, we have the y-axis of the spectrum. Of course, we have to filter this signal to
cover only a very narrow band. The width of this filter is called “resolution
bandwidth.” On my Siglent SSA3032X, it is currently 1
MHz. So, all signals inside a bucket of 1 MHz are
added to the displayed power—also, the noise in this bucket. Therefore, the noise level is around -70dBm. To show you how spectrum analyzers work, I
change this bandwidth. Now we see that it starts with a very low
frequency and move up to the maximum frequency. The smaller the bandwidth, the slower the
scan speed, because the analyzer has to probe more points to cover the whole spectrum. In such a mode, short signals are no more
detected. This is why we can use “peak hold” and
hope for the chance of detecting also short signals. Like that, we see our 2.4GHz signal even if
it is not always on. Next, I show you another important function. Again, I use the Baofeng on 145.550 and display
the frequencies from 140 MHz to 150MHz. The scan is now much faster because it only
has to cover a small band. The signal seems to be about 4 MHz wide. Which is hopefully not the case. Only if I reduce the RBW, we see more details,
and we see that the signal is very narrow-band. If I modulate the signal with my “natural
sine wave generator,” we do not see a difference. Only if I zoom in to 100kHz span, we see the
modulation of the signal. The RBW is now 1 kHz. And the noise floor is down to -100dBm. Very sensitive. I show you all these details because later,
you will see the differences between this expensive and big device and this small and
cheap one. The next important application of spectrum
analyzers is to check out filters. To do that, we have to input a signal into
the filter and display its output. The output of a filter, therefore, can be
connected to the analyzer. To produce an input signal, we have to add
a component called “tracking generator.” If the frequency of the tracking generator
and the scan frequency of the spectrum analyzer are synchronized, the analyzer always sees
the full power of the tracking generator. BTW, this is why it is called a “tracking”
generator. It “tracks” the frequency of the analyzer. Let’s connect the output of the tracking
generator to the input of the spectrum analyzer. We still see the noise floor. As soon as we switch the TG on, we see a curve. The output of the TG is set to -20dBm, but
the spectrum analyzer does not show a constant curve. This is because both the tracking generator
and the spectrum analyzer are not ideal across all 3 GHz, and we have cables. Fortunately, this fact does not matter too
much if we press this button: Normalize ON. Now, the response is completely flat at 0
dBm. The Spectrum analyzer stored the differences
and adjusted its display automatically. Cool. Now I exchange this adapter between the tracking
generator and the analyzer with a filter. I do not tell you what its purpose is. You tell me. It has a very flat response across the whole
3GHz, very similar to the plain connector from before. Only if we have a closer look we see a difference:
Here, it has a deep dip of around -40dB. Let’s zoom in: The notch is from around
80 to 110MHz. Filters like that are used if you live nearby
of an FM radio station. Such stations can block your receiver, also
if you intend to listen to other frequencies. In those cases, such filters make sure your
receiver is much more sensitive to signals like LoRa, which are much weaker. The next filter shows this behavior. It blocks high frequencies and, therefore,
is called a low pass. It has a -3dB point at around470MHz. And really, it is sold as a lowpass with the
cut-off frequency of 450 MHz. The third application of spectrum analyzers
is the measuring of antennas. If you attach a reflection bridge like this
one, you can check your antennas. I will not do that because I have a nice and
cheap device that is better suited for this purpose. And the whole device costs less than just
this reflection bridge. Let’s now head over to the cheap spectrum
analyzer. Ah, here I see that I have to remind you to
like or subscribe if you want to see other videos like that! You find a few versions of the same design,
some with and others without case. I got this one from Bangood, and it is called
LTDZ 35-4400M. Its coverage is from 35MHz to 4.4GHz. The top frequency is even higher than the
one of the Siglent. But it costs only around 50 dollars, about
60 times less than the big one. How come? It has no display or buttons, and the PC controls
it. A first cost reduction. Let’s go into the schematic. Here we see how simple it is. Let’s try to find the relevant parts. Here we have the mixer chip. It gets the input signal directly from the
SMA connector and from this ADF4351 wideband synthesizer, which acts as the local oscillator. Here you see that this component limits the
range of the overall device: Its range is from 35 to 4400 MHz. Its output is connected to the second input
of the mixer. As expected. The output of the mixer is connected to a
logarithmic amplifier, which creates a signal for the ADC of the microprocessor. A low pass filter removes the high frequencies. This setup acts as a “power meter.” Between the mixer and the Amplifier, we find
a filter. This filter defines the Resolution Bandwidth. It is fixed, and we will later see what this
means. An STM32 microprocessor controls all. This 25 MHz oscillator generates the clock. These components form a working spectrum Analyzer
that can detect RF signals. Missing is only the tracking generator. We find it here. It uses the same ADF4351 as the local oscillator. Its output is connected to the second SMA
connector, and you have to switch it on with this button on the back. The Blue LED shows that it is enabled, and
the two yellow LEDs show which oscillator is working: The local oscillator alone or
with the tracking generator. The PCB looks like that, and the case is really
small. To compare: Here you see only one PCB of the
Siglent. If you are interested in a “hardcore”
teardown, you find a link to a video from the signal path. Excellent, but hardcore. Let us now do the same tests as we did with
the “big box.” First, we have to download software through
a link provided by Bangood. It is made in Germany by Ham operator, Andreas
Lindenau. I doubt that he gets anything from this deal. It also seems that he abandoned this project
and created a new one. At least, I did not find it on his homepage
where I found another version. I assume it is newer, and I use this one. Unfortunately, I did not find any documentation,
and also the help does not provide anything. Software for cliffhangers, as we will later
see. This spectrum analyzer project, BTW, seems
also have its roots in Germany many years ago and found its way to China like the transistor
tester story shown in video #290. I connect the device to a USB connector and
start a scan between 35 and 3000 MHz, comparable with the tests we did before with the Siglent. I select 1000 points, and the scan speed is
quite fast. Unfortunately, I do not see a signal from
my Baofeng. Does this device really work? Let’s quickly check what happens. One step is nearly 3 MHz. It starts at 35MHz. Every 3 MHz means that it measures at around
143 and 146 MHz. So it does not “see” my signal at 145.550
MHz. Why did the Siglent see it? Because it uses a wider Resolution bandwidth
for fast scan speeds. So it sees signals which are not exactly on
the scanned frequencies. It also does not see it if we increase the
samples to the maximum of 2000. On the other hand, if we decrease the samples
to 1450, it sees the signal but not the harmonics. Just by chance, because one of the scanning
points is close to 145.55MHz. If we zoom in to 140 to 150 MHz, we see an
interesting signal. Very different from the one seen before on
the Siglent. It consists of a peak with a notch in the
middle. It is obvious that this notch does not exist
in reality. The device itself produces it. Just ignore it, and you are ok. Now we focus even closer: 145 to 146MHz. Here we see a similar curve as before. Unfortunately, no more details are visible. So we can reduce to 100 samples and get a
decent scanning speed. Even now, we do not see a difference when
I modulate the carrier. Here we see the limits of the cheap device
because of its fixed RBW. Another limit is also its sensitivity. It is not as sensitive as the Siglent and
also has no built-in pre-amplifier. The noise floor is always at around -70dBm. Which means: No signal. Maybe this fact is not so important for makers. You find information to enhance this device. Three areas of enhancements are proposed:
The first is here in the input stage, where it is suggested to remove this resistor. It seems to reduce the input power and is
not necessary. Another area is this filter. And a third one is to add a small capacitor
for a cleaner power supply. I leave you links to the enhancements in the
video comments. One important thing: There is no input protection
or warning if your signal is too high. So pay attention! Some owners of such devices wrote that they
had to replace the mixer chip because they did not pay attention. Maybe you order a few with your device. They are not expensive. Always use attenuators if you work with powerful
transmitters. You get them cheap on Aliexpress, and they
usually are good up to the 2.4GHz Wi-Fi or Bluetooth band. Now we leave the power measurement of transmitters
and use the tracking generator to test filters. As said before, you have to enable the tracking
generator with this button. We also connect it directly to the input of
the analyzer to get the normalization curve. And really: we also get a curve which is not
flat and below 0dBm. Not an issue as we saw with the Siglent. Unfortunately, I did not find out how to calibrate
the device or normalize this curve. Maybe you can help me? I also tried with the old software provided
by Bangood. There is a calibration function, but I was
not able to use it. Here, I would have loved a manual to read. At least nobody can tell me: RTFM! But we can use the device also without normalization. We save the curve created with the direct
connection and display it together with the measured filter curve. Like that, we get a feeling about the filter
performance. Here we see the FM radio station filter, and
here is the 450 MHz lowpass filter. In this curve, we also see that the low sensitivity
of the cheap analyzer. It shows -70dBm. In reality, it is around -40dBm. Just to show you how expensive filters can
be: This one costs 75 dollars. Used. And here, you find another difference between
the two devices. The Siglent has filters everywhere, this small
box only a very few and simple ones. BTW: Don’t worry: I got this filter for
ten bucks at a HAM festival. To check the linearity of the device, I measure
a 20 and a 40dB attenuator. The 20dB shows around 24dB and the 40dB attenuator
reduces the signal to below the sensitivity of the device. If I find out how to calibrate this little
bugger, I am sure its linearity could be improved. But for sure not as good as the Siglent. Summarized: Is this small tool useful? - If you want to know on which frequency your
transmitter works and have an idea about its output power and harmonics, you can use it. - But signal or bandwidth analysis cannot
be done for narrow signals - For filter analysis, it is usable if you
are not interested in exact values like the dampening. But you should get indications if the filter
is ok or not - If you work with radio frequencies above
35MHz, it is a nice addition to a lab. Not the first and not the second priority. But maybe a birthday or Christmas gift. Because it can be fun to play with
- Definitively a higher priority has a network analyzer like the one I showed in video #191. Or, if you need lower frequencies, also this
NanoVNA. Which, BTW, uses similar chips from Analog
devices as the device tested today. Because it also measures the phase of the
signal, it is much better for antenna analysis. A must for all LoRa enthusiasts. And the nanoVNA can also show filter curves
- An alternative to this spectrum analyzer could be an SDR receiver. It, for sure, can show on which frequency
you transmit. And with the right software, you can also
get diagrams which are quite broadband. I never tried it, but would assume does not
have the same dynamic range - You cannot measure filters with an SDR receiver
only. You have to add a white noise source that
emulates a tracking generator. Maybe stuff for another video? I hope, this video was useful or at least
interesting for you. If true, please consider supporting the channel
to secure its future existence. You find the links in the description. Thank you! Bye
