The evolution of spectrum analysis

Cutting-edge, real-time spectrum analysis is leading a revolution in the ability to identify sources of interference.

It is said that necessity is the mother of invention. Nowhere is that more true than in the field of radiocommunications and, in particular, spectrum analysis for diagnosing interference problems. With intentional and unintentional radiating devices proliferating in more and higher bands, and with 5G just around the corner, the need for spectrum analysis has never been more acute (see ‘Diagnosis: interference’ in this issue).

Intentional radiators that have active transmitters include: broadcast radio and television, cellular, satellite, radar, mobile radio, WLAN and cordless phones. Unintentional radiators that use RF but not for radio transmission include: microwave ovens, radio receivers, industrial heaters and MRI equipment. And incidental radiators that do not use RF include: switching power supplies, clock and control signals, ignition motors and fluorescent and LED lighting.

Field spectrum analysers have been available for well over a decade now and have helped in the vast majority of situations where interference has been experienced. But initially they were large, cumbersome beasts.

“In the early 1990s when I was doing fieldwork, you took a full-sized bench instrument out into the field and you had a power inverter in the car or van,” said Steve Karandais, general manager of Keysight Technologies Australia. “You had to lug it all around or you would try to do some recording techniques.

“When you’re out there in the real world and they’re saying, ‘Look, we’ve got interference out here and we need to solve this problem quickly,’ it was always problematic having a full-size spectrum analyser.”

Portable or handheld spectrum analysers have made life much easier, and over the last five years they’ve become a lot more capable as a huge increase in processing power has seen the capabilities of high-end, research-grade, benchtop spectrum analysers make their way into handhelds.

A new breed

According to Karandais, standard swept-tuned and fast Fourier transform (FFT) spectrum analysers have Achilles heels that make them miss many kinds of interference.

“What’s happened in more recent years is that the overcrowding of the spectrum and the proliferation of electronic transmitters that are unintentional transmitters... has made the space very, very crowded,” added Karandais. “Sometimes the interference that’s occurring is sitting underneath the bandwidth of the signal that would normally be there, [so] we can’t detect it. Or, the signals that are there are for such short spans that as we try to use a normal spectrum analyser... we don’t see the interferer because they’re there for shorter periods of time than our sweep time allows us to capture.

“With a swept tuned analyser, unless you happen to coincide with your interferer being on at the time that you’re sweeping, you’ll lose it,” added Karandais. “And with a normal FFT analyser, there’s lost information while it’s calculating and then displaying data, [before] it starts amassing data again.”

Keysight's FieldFox spectrum analyser.

The new breed of real-time spectrum analysers, such as Keysight’s FieldFox, are still FFT based but the bandwidths are much larger — 10 MHz as opposed to the 100 KHz that was normal in the past — and they have multiple FFT engines running at the same time to provide overlapping measurements.

“What we do is we run three FFT engines built into 11 ASICs inside the FieldFox,” said Karandais. “This is processing power and memory speed. We can get a [measurement] in and out of memory, process it and display it really, really fast.”

Solving the unsolvable

According to Karandais, until recently many interference problems hadn’t been solvable. Either the source of the interference could not be found or the interference itself could not be isolated within the bands. “All we could do was fiddle with enough things until we’d either fixed it or it became less of an issue,” he said.

“Sometimes they’ve had to move and abandon whole bits of spectrum or they’ve had to increase amplitudes to levels they would prefer not to have to use because of power requirements and heat,” he added. “They’ve had to overcome the interference in some way without actually being able to always identify what’s going on.”

But with real-time spectrum analysis, “you can say, ‘All right, I have a signal at this frequency, at this bandwidth, and now I can look around and say, Who else is in this spectrum? Who else has licensing here? What else could this possibly be?’

“In the old days with a normal spectrum analyser, people would just walk away and say ‘I can’t see anything useful’, whereas now, within a minute of setting up, you’ll be able to say, ‘this area is clear’, or ‘there’s my problem’,” he said.

A growing need

What has changed everything is the ability to put vast processing power and fast memory into smaller packages.

“With products like these, you can take the technology of a very large, really well specified major spectrum analyser, take some of those algorithms and technology and press it into an ASIC,” he said. “With exactly the same methodology, the same measurement sciences behind it, but you’re putting it much, much closer to the coalface. And this is what’s really changing the way that we can track down these interferers.

“And we need to stay on top of this because of the level of complexity of electronics and the environment. Especially when 5G comes along, we’re going to have picocells all over the place,” he added. “They’ll be operating at a frequency that’s very, very high, which means that the attenuation will be extreme, [so] they’ll have to have one every 50 to 100 metres.”

When LTE and 5G sites are spread all across town and cities, plus Wi-Fi, other wireless bands, and automotive radars and electronics, “People will have a really difficult time figuring out who’s interfering with whom. So we’re going to need these spectrum analysis tools,” he said.

Follow us and share on Twitter and Facebook