Oscilloscope vs Spectrum analyzer, what can be done and not

Hi, What are the things that can't be done on a spectrum analyzer and can be done with the scope?

I am talking only for RF work, I do not care about inspecting the edges of a square wave for example.

These are the basics: An oscilloscope is of limited use for RF work. It can show a modulation envelope and also the modulation input, but it doesn't show you what you need to know for an RF signal. A spectrum analyzer gives you all of the frequencies contained within a RF signal. It is the frequencies within a signal that is the most important, since you need to know the amplitude of the non-fundamental frequencies.

Ok the SA can show you the distortion of the signal, which may be hard to see in the scope (if the distortion is small). It measures the harmonics levels, thus the distortion. It also shows the modulation signals, for example in AM, the sidebands, if the resolution of the SA is enough.
I have forgotten about the modulation envelope you mentioned and can be measured with the scope indeed.
Is there any way I can measure the modulation envelope and also the modulation input by just having a spectrum analyzer, so I do not need a scope at all?

Viewing a signal in the frequency domain provides a whole different view than attempting to view it in the time domain.

For instance, have you ever connected a high speed scope to a CATV signal? It appears like pure noise.

If that same signal is viewed by with a spectrum analyzer, one can clearly see the individual channels, the absolute and relative level of each of its carriers and subcarriers, their frequency separation, modulation index....etc.

In time domain different things can be shown in contrast to the frequency domain, of course.
I wonder what are these things related to RF work.

Post #2 mentions the modulation, which I do not know how can be shown on an SA (if it can).
What other things?

Oscilloscope displays amplitude in the time domain.
Spectrum analyser displays amplitude in the frequency domain.

My oscilloscope is good to several hundred MHz so I dispute the claim they are not suitable for RF. I can clearly see the waveform from my HF and VHF transmitters. I agree that for RF they are better for observing the amplitude envelope rather than the signal itself.

An SA will for example show you the main signal, harmonics and also the sidebands because although they are component parts of the signal, they are at different frequencies. To some extent you can mathematically determine the spectrum from the amplitude information by using FFT, it has limitations, especially at higher frequencies.

Basically, they are both essential tools if you are working with RF and because they measure in different ways, you can't substitute one for the other.

Brian.
(Tek MSO3032 scope, Anritsu 610 Spectrum Analyser, HP8648 signal generator)

I didn't say that they were not suitable for RF - I said that they had limited use for RF. I had a 250MHz scope on my bench that could see a VHF transmitter output. Although I could see it, the scope didn't tell me what I needed to know, particularly the harmonic levels. Therefore, my first reach wasn't for the oscilloscope, but rather the spectrum analyzer.

So, they both have their uses in RF. If I only had a choice of one, I would choose the spectrum analyzer, hands down.

A spectrum analyser is undoubtedly much more useful for pure RF work, but there are still quite a few applications where seeing actual analog waveforms and sometimes quickly varying dc can be quite helpful.

Trying to properly bias an amplifier, mixer, or modulator and diagnose actual causes of waveform distortion (rather than just knowing how much) can sometimes be more illuminating than a bunch of dbm numbers.

For instance, setting up and analysing and fault finding the dynamic performance of phase locked loops and VCOs might be rather difficult without an oscilloscope, and these infernal PLLs are now popping up in just about everything that uses RF.

The other factor is frequency limit. There are very few osciloscopes that will work beyond 500MHz. A really basic SA cover to at least 1.5Ghz and most will go much higher. You also have the ability to have a span starting at an arbatary frequency. Scopes need to start at DC.

I have used both for RF work, but I would say that at least 95% of the time the SA is the tool I reach for.

SDR radios are another useful tool that I think many RF engineers have not caught up to yet. You can do one helluva a lot with a cheap S30 dongle and free software.

There is a need for the SA to go much higher in frequency. It is the need to see higher harmonics. The scope instead does not have that need. The scope speed is more important when noticing the fast transients of the edges of a square waveform for example, which can be thought more like a much higher frequency sine.

The span feature you refer is a major limitation in my HP54520A FFT. I can measure the harmonics (limited dynamic range and possibly sensitivity) but I cannot zoom in on a modulated signal to measure it's sidebands, or at least I have not figured how to do it. In that sense FFT is of no use.

I am susceptible about the performance of such a cheap SDR equipment, it is very easy to perform false measurements and think that they are fine. Have you made any comparisons to a real spectrum analyzer in such things?

Ah, but if you only need to drive half a mile to the supermarket, you do not need a 200+ MPH Ferrari to do it in (although it would be really nice).

Any and all test equipment is useful, even the ultra cheap Chinese, or ancient boat anchor stuff on e-bay, but every piece of test equipment has its limitations.

Confucius say :
Being able to measure parameter rough as guts, better than having no clue at all and guessing.

My limited experience related to measurements, is that the measuring equipment, must be much more as good as possible than the circuit you try to emasure. I mean, the measurement equipment better be a ferrari, even though you need to measure a donkey performance. There are many characteristics in a signal, you would ignore that they exist, if cheaper measurement equipment was used. Of course we try to do our best with our limited budget.

I totally agree with you though, even the most crappy piece of equipment is better than having none at all. But for serious emasurements you need much more capable equipment than the thing you try to measure.

Suppose you wish to measure if your car battery is flat (it will not crank the engine).
A five dollar multimeter will work just fine.

Most often fairly close is all you need to tell you all you need to know.

Unless of course you are selling a professional calibration service with full traceability.

All test equipment can lie to you if you dont know what you are doing. Even the expensive stuff.

The SDRs are good because they can be used as a realtime spectrum analyser. That means you capture the whole bandwidth all the time and can garantee not to miss transients that a sweeping SA may not pick up on. You can also record and re-analyse the same signals with various FFT lengths etc. They do have problems with spurs, dynamic range and other effects. You can get round some of these by buying a better SDR radio such as ettus for instance, but i think anyone playing with RF should at least get a RTL dongle at the very least.

If you switch the SA to manual sweep, linear display, wide video bandwidth you can tune the display into your AM signal and measure the RF level at the peaks and troughs of modulation. So you can work out the mod percentage.
You can also work out the mod index of FM signals by looking at the carrier and adjusting the amplitude or frequency of the modulating frequency until the carrier disappears (Bessel zero).
Frank

Hi
I really liked the discussion it is very informative. But my issue is some what different.
I have joined a project and in lab we have 4GHz OS and Network Analyzer and still they want o buy 4GHz SA.
From the discussion above I can make clear picture of OS and SA but it would be great for me if someone also includes Network Analyzer in the discussion as it will complete the picture of Microwave Lab.

A network analyzer in it's simplest form is a spectrum analyzer and tracking signal generator. The are many other things it may do such as 'S' parameter measurement, phase shift, admittance and reflection measurement and so on.

The concept is that if you produce a fixed frequency and feed it directly to a SA, it will read it's amplitude. If you sweep the frequency and at the same time sweep the SA window, you would see a constant amplitude across the frequency spectrum. You would be selectively measuring the output as it changed and if the equipment is accurate you see a constant level.

Now, if you introduce something in the signal path (the network you are analyzing) it will have an influence on the passage of the signal from the signal generator to the SA. As the frequency sweeps, the SA will show the amplitude change according to how well the network let is pass through it. For example, it the network was a bandpass filter, and the sweep of frequencies was from below it cut-off to above it's cut-off, the SA would show the filters frequncy response. I doesn't have to be a filter, you could be looking at the frequency response of an amplifier for example.

So a basic network analyzer is a combination of signal generator and spectrum analyser. Unlike separate instruments, some of the internal functions can be combined to simplify design.

Brian.

Thanks a lot for a comprehensive answer.
So I can say safely that with the Network Analyzer available there is not need to go for Spectrum analyzer?

A network analyser does more, but invariably costs more for similar performance.

You may need to balance your requirements with how deep your pockets are.

77
Well we have network analyze and OS at 4GHz already in lab. Discussion was to buy a SA at 4GHz for which I needed information. As I believed that with NA at hand there is no need to buy SA.