Instrument Basics: Spectrum Analyzer -- Workbench Wednesdays 15

While one can use an SDR as a "low cost" spectrum analyser, there are a few disadvantages to be aware of.

Some SDRs are perhaps not built as rugged, and are designed to receive on-the-air low-level signals. As a result, they can be designed with LNAs on the input which can easily saturate with higher level signals. The linearity of these LNAs is perhaps not guaranteed, with higher signal levels possibly causing intermodulation images due to non-linearities. As a result, I wouldn't trust any absolute measurements made with SDRs as I would with a calibrated spectrum analyser. Sometimes SDRs don't do a good job of letting you know when you have a signal over-range causing ADC clipping which can cause the spectrum to "bounce" due to the broad-band noise that happens when it overloads.

The other issue is that sometimes you do get images and stray signals which may be internally generated by the SDR or picked up "over the air" especially with cheaper, unshielded units. These can take a little head-scratching to figure out sometimes, and can come down to I/Q path gain and DC offset imbalances resulting in a "peak" in the centre and mirror image signals appearing on the waterfall. A proper spectrum analyser should have this "taken care of".

But with some care, many of the modern SDRs offer a good amount of real-time bandwidth and can function as a basic spectrum analyzer to understand signal emissions. Just don't expect them to do EMI pre-qualification testing.

- Gough

Thank you for this video, James.

I have never used a SA, but we just bought one at work, so I will try the steps to identify frequency of some HW laying around.

shabaz  wrote:

The software I tried was called QSpectrumAnalyzer. It's ok-ish, and will work at a pinch but still needed a lot of work the last time I tried it (about a year ago).

Interesting, I think I looked at that about a year ago as well. I forgot about it. I want to see how it works with the HackRF. I didn't realize the it has a sweep mode.

I did have a vision of building a Pi + SDR based spectrum analyzer. Probably wouldn't be very fast, but then again, it would not need to be.

Hi James,

I've found the same, the software for SDRs leaves a lot to be desired. Even flagships like GNU Radio need a lot of work. For some reason the radio-focussed software that is out there tends to throws all UI and usability concepts out of the window.. it would be easier to control a spaceship, than navigate the UI that some of the radio or ham apps have* : )

There is the occasional piece of 'simple' software that can display a view a bit like a real spectrum analyzer, but with no real concept of actual power levels, so that would be guess-work. The software I tried was called QSpectrumAnalyzer. It's ok-ish, and will work at a pinch but still needed a lot of work the last time I tried it (about a year ago).

* Horrendous UIs like this.. (source: google images)

From YouTube:

It looks alot like SDR (software defined radio). wondering if sdr will work similar?, I'm currently trying to make a NFC aerial for a phone.

Software defined radios (SDR) use a similar heterodyning technique. (Many radios do.) The RF portion is a mixer that downconverters the RF into a lower frequency, but wide bandwidth, baseband. Then an analog-to-digital conversion happens. The difference between a Spec An and an SDR is how that information gets used.

Functionally the difference between a swept spectrum analyzer and an SDR is that the SDR typically tunes to a specific center (or carrier) frequency. It does not sweep; it just digitizes a specific bandwidth around that single frequency. So when you tune in 100 MHz, the SDR digitizes its full baseband bandwidth, but then a filter like 16 kHz (Narrow FM) is applied to the digitized data. This wide-band digitizing is what makes SDRs so cheap. Instead of having a bunch of analog filters, all of the baseband filtering happens in software.

The difference for a spectrum analyzer is that if you set the center frequency to 100 MHz and the Span to 16 kHz, the analyzer is repeatedly tuning its local oscillator from 99.984 MHz to 100.016 MHz in steps of 100, 1000, etc. At each step, it applies a filter (the resolution bandwidth) before digitizing the data.

Visually, the screens may look similar. But the there will be more detail and a higher signal-to-noise ratio (SNR) with the spectrum analyzer's display. The SDR is more similar to an oscilloscope plotting an FFT than a spectrum analyzer doing a sweep.

Two notes to the above explanation:

1. It would be possible to program an SDR to work similarly to a spectrum analyzer. I looked around for some software to do just that, but most of what I found was really "radio" focused.
2. "Real-time spectrum analyzers" effectively do the sweeping SDR technique. They are both sweeping and digitizing wide-band. The analyzer I showed in this video the traditional sweep type.

neilk  wrote:

 

James, I loved that video - I learned so much!! I've never seen a spectrum analyzer in the flesh, let alone used one!

Glad you liked it Neil. Spec Ans are a very misunderstood and underutilized tool.

baldengineer I think you nailed it on both those points. It was a great video.

shabaz  wrote:

 

Especially for a lot of engineers who may never have used a spectrum analyzer

Awesome, I am so happy to see you call this point out, because this is the audience I targeted!

When I got my FPC1500 almost two years ago, I searched around for basic usage videos. Other than a few random times, I had not used one for 15+ years. Most tutorials I found were either difficult to watch or were manuals on a specific model. My goal for this one was to show what a spectrum analyzer could measure and to highlight the 3 or 4 controls needed to get something reasonable on screen.

shabaz  wrote:

Another cool part of the video was the automated features highlights

I almost cut the "advanced" measurements shown at the end, since there is not much explanation about any of them. However, I wanted to leave viewers with the impression that modern spectrum analyzers can do a lot of useful (and cool) stuff.

James, I loved that video - I learned so much!! I've never seen a spectrum analyzer in the flesh, let alone used one!

It's a fascinating instrument.

Thanks

Neil

Great video! It's nice to see modern content like this, especially for a lot of engineers who may never have used a spectrum analyzer (at least, when I studied, I went through all 4 years of labs without seeing one - only to have to use one in my first job). What's nice is that nowadays those in non-radio jobs can make use of them too, whereas in the past they were so expensive they were definitely not on a lot of test benches.

Another cool part of the video was the automated features highlights (because often they really are hard to do without the automation - without that, it is very easy to get a measurement  procedure misunderstood or a calculation wrong, and be out by orders of magnitude. Particularly with an SA, the automated features are like having an expert sitting next to you, rather than just a convenience/productivity thing.