Can VNA be used as SA?

[knick]

Can a VNA be used as a spectrum analyzer?
I understand the VNA has a transmitter and a receiver. Can I just use to receiver alone to check the frequency spectrum (magnitude only) of a signal?

Background: I don't have either piece of equipment at the moment. We have an immediate need for a SA to do some EMC pre-compliance investigations. We'll also probably need a VNA later down the track for another project. We're considering whether we could just purchase a VNA now to do both tasks.

[nctnico]

Can a VNA be used as a spectrum analyzer?
I understand the VNA has a transmitter and a receiver. Can I just use to receiver alone to check the frequency spectrum (magnitude only) of a signal?

Usually yes but the noise floor of a VNA is typically higher (AFAIK).

[hendorog]

I'm going to beg to differ and say you won't be happy with a VNA for SA work.

It's possible but it sucks.

[knick]

I'm going to beg to differ and say you won't be happy with a VNA for SA work.

It's possible but it sucks.

Hi Hendorog. I would be eager to hear where you think the VNA would be lacking. Noise floor issues, missing features, user interface unsuitable or something else?

[tautech]

knick, there are a range of instruments that do both VNA and SA, each in a different mode:
the SVA1000X series.

For pre-compliance and lower frequency SA work a SVA1015X with the current EMI promo package is one of the cheaper options.
https://int.siglent.com/info/detail-40.html

[rf-messkopf]

Can a VNA be used as a spectrum analyzer?
I understand the VNA has a transmitter and a receiver. Can I just use to receiver alone to check the frequency spectrum (magnitude only) of a signal?

Yes, in theory. But there are a number of problems with that:

1. A VNA typically has no image rejection. That is, for a single tone at the port, you get multiple lines. I once demonstrated this with my VNA after a question in a German forum, see this posting and the attached pictures: https://www.mikrocontroller.net/topic/506633#6472702. There a single tone of 100 MHz and 0 dBm is fed to its port, while it is set up to measure the incident wave at that port (with the source at the other port).

2. Amplitude measurement is not very accurate. After calibration for a S_21 measurement, the indicated amplitude for an externally injected signal depends on the reflection tracking correction, and hence on the amplitude of the VNAs own source. And even if the VNA is set up to indicate the incident and reflected waves (a/b-waves) instead of S_21, it may not be too accurate. In the linked picture, the indicated amplitude is shown for two different IF bandwidth settings, and it shows more than 20 dB difference. That is probably because the incident tone is not exactly on one frequency point of the VNA. A VNA is just not made for this application.

3. The IF filters have a different shape than in a spectrum analyzer, i.e., are not Gaussian. They are optimized for fast settling to facilitate fast sweeps, but their stopband attenuation is just good enough to yield the specified dynamic range. See the massive stopband ripples in the pictures linked above.

Unless the VNA is explicitly designed and specified as a spectrum analyzer as well (the high end ones aren't), this misuse does not seem to be too recommendable.

[hendorog]

I'm going to beg to differ and say you won't be happy with a VNA for SA work.

It's possible but it sucks.

Hi Hendorog. I would be eager to hear where you think the VNA would be lacking. Noise floor issues, missing features, user interface unsuitable or something else?

A VNA lacks the hardware and design required to make a good SA, the issues are as described already by rf-messkopf's post. I would add potentially poor phase noise and maybe spurs to that list and also the coupler/bridges reduce dynamic range further unless the unit has direct receiver access.

Another potential consideration, is that old VNAs, such as my 8753, used a pulse generator to downconvert the external signal into the receivers. Which means you would need to provide a front end set of mixers as well and drive it from the source. The mixers would need to produce the IF at the same freq as the source.

There is a receiver mode in the 8753, but it is a long time since I've looked at that. I thought it maybe could be a poor man's SA too, but after some playing around I remember deciding that it wasn't worth further effort due to the coherent source issue.

The software is a problem. Someone might have tried it before on your device, but it's a far cry from a purpose built SA.

As tautech points out, combined VNA/SA devices exist, work well, and I have one. But using a pure VNA as an SA would be painful.

There is an old thread here which asked the same question, and some dual SA/VNA devices were mentioned in that.

A reverse example, the Signalhound SA range can be used as a VNA, but only for a single frequency at a time. A new calibration must be done after each frequency change.

I recall that some of the very low cost vnas like librevna and nanovna do have an SA mode which work with limitations. If so then a nanovna in particular doesn't cost much money, so you could  try see for yourself if it will work for what you need.

[cdev]

@hendorog.

Nanovna2 doesnt have a spectrum analyzer "mode".

sometimes a little signal leaks into the receiver side when its not being used.

 For that reason, sometimes I keep another 50 ohm termination on the unterminated input. It seems like it does pick up some signals otherwise, sometimes, but instead of where and as it should be it looks more random. Its probably my local RF noise.

It would be great if it was possible to use it for finding signals, EMI, etc. with a software change.

[hendorog]

Yes thanks for correcting me. I did think someone had written firmware for it, but now I look I just people trying to use port 2 as an SA.
It does work - with limitations as described.

[knick]

Thanks for the detailed answers 
What I'm taking away from this is that we either want a VNA which specifically has a SA mode or separate devices.

[tautech]

Thanks for the detailed answers 
What I'm taking away from this is that we either want a VNA which specifically has a SA mode or separate devices.

Yes but you said immediate use is for EMI pre-compliance so a SA with an EMI test feature set is particularly what you need then decide if you want a single unit that has VNA and SA modes.

Skilled operators can decipher SA results from near field testing however having an EMI feature can help immensely like in this UI which is how all Siglent's SSA3000X Plus and SVA1000X look now updated from an earlier EMI UI:



More info here showing the older EMI feature set and a little bit of how to use it:
https://www.eevblog.com/forum/testgear/cannot-find-temporary-unlocked-additional-packages-on-siglent-sva1015x/

[Keysight DanielBogdanoff]

I talked with one of our VNA pros a little more about this because I was curious, and here's what they said.

Generally speaking, a VNA shouldn't be used as a SA.  Among other things, VNA architectures don't have a preselector for image rejection.  Even the VNA models we offer with a SA option are only intended for limited use applications such as harmonic / spur searches on amplifiers, for example.  VNA SA options that Keysight offers don't have demod capabilities.

One thought is to go with FieldFox, so that the user can get both VNA as well as SA with selection of options.  For example, N9913B provides a 4 GHz Combination Analyzer that measures RL and DTF out of the box.  Adding options 210 and 211 provide a two-port S-parameter test set for NA mode.  Adding option 233 provides a spectrum analyzer.  Adding option 350 on top of option 233 gives the user RTSA with 10 MHz analysis bandwidth standard.  Adding options B01 or B10 bolster RTSA with either 40 or 100 MHz analysis BW, respectively.

I think there's a future state where VNAs do more SA stuff, but for now it's targeted at specific applications.

[hagster]

Copper mountain released a tech neote about this very question.

https://coppermountaintech.com/spectral-analysis-with-a-vector-network-analyzer/

Conclusion is in line with other comments here.

The SignalHound is a tracking generator as far as I am aware.

I use a fieldfox combo at work. It's very good. I think it does sacrifice a bit of performance for portability. For instance our Copper Mountain VNA and SignalHound SA are orders of magnitude faster and better in their swimlanes. Its probably cheaper to buy those separately than one fieldfox with enough options to make it a fair fight. But having the all-in-one and built in battery does mean it gets used more often.

I haven't used the Siglent combi units. They only do S11 and S21 so not a full 2 port device.

[thinkfat]

As the nanovna2 was mentioned, I seem to remember that there's a fairly recent (as in, August last year) youtube video on the topic:

[Keysight DanielBogdanoff]

I use a fieldfox combo at work. It's very good. I think it does sacrifice a bit of performance for portability.

...

But having the all-in-one and built in battery does mean it gets used more often.

It's definitely designed to e used in the field and isn't spec'd as well as many benchtop options, but dang is it convenient. I got to play with the RTSA version and it was pretty nice.

[hagster]

I use a fieldfox combo at work. It's very good. I think it does sacrifice a bit of performance for portability.

...

But having the all-in-one and built in battery does mean it gets used more often.

It's definitely designed to e used in the field and isn't spec'd as well as many benchtop options, but dang is it convenient. I got to play with the RTSA version and it was pretty nice.

Portability is an underrated quality. Even in a lab.

[Joel_Dunsmore]

Why yes, and in many ways it can be better than a traditional spectrum analyzer, if you have the right VNA. I spent the last 5 years of my life working on the SA mode of the PNA and PNA-X (which then merged over to the new ENA, and the Streamline VNA, as well as PXI VNA).  In terms of power accuracy, it is maybe 2-10 times better than a traditional SA.  It has similar dynamic range although then noise floor appears higher because there is a 13 dB input coupler (which if you bypass will result in similar Displayed Average Noise Level).  Some of our best SA's are a bit better than the VNA with SA application, but in general either will work. Image rejection is accomplished though digital means that works remarkably well.  It always uses an FFT based method (traditional SA's have swept tuned or stepped FFT) which results in fast speeds than swept-tuned SA for low bandwidths; at 100 Hz RBW the VNA might be 100-1000 times fast than a traditional SA.

Newest versions have high performance source and LO synthesizers that allow the VNA to make phase noise measurements nearly equivalent to the best spectrum analyzers.  Details of how this all really works is the topic of chapter 8 in the 2nd edition of the Handbook of Microwave Component Measurements. www.tinyurl.com/JoelsMicrowaveBook 

Downside: OK, PNA is a bit dear, so not for a hobbyist.  But that's what it takes to get it all in the software and new DSP data-acq engines in Modern VNAs; they're a lot different than the ones from 20 years ago.  However, I will say the traditional SA's probably still lead in the role of specific applications (such as EM receivers, demod, and other specialty tests).

[2N3055]

Why yes, and in many ways it can be better than a traditional spectrum analyzer, if you have the right VNA. I spent the last 5 years of my life working on the SA mode of the PNA and PNA-X (which then merged over to the new ENA, and the Streamline VNA, as well as PXI VNA).  In terms of power accuracy, it is maybe 2-10 times better than a traditional SA.  It has similar dynamic range although then noise floor appears higher because there is a 13 dB input coupler (which if you bypass will result in similar Displayed Average Noise Level).  Some of our best SA's are a bit better than the VNA with SA application, but in general either will work. Image rejection is accomplished though digital means that works remarkably well.  It always uses an FFT based method (traditional SA's have swept tuned or stepped FFT) which results in fast speeds than swept-tuned SA for low bandwidths; at 100 Hz RBW the VNA might be 100-1000 times fast than a traditional SA.

Newest versions have high performance source and LO synthesizers that allow the VNA to make phase noise measurements nearly equivalent to the best spectrum analyzers.  Details of how this all really works is the topic of chapter 8 in the 2nd edition of the Handbook of Microwave Component Measurements. www.tinyurl.com/JoelsMicrowaveBook 

Downside: OK, PNA is a bit dear, so not for a hobbyist.  But that's what it takes to get it all in the software and new DSP data-acq engines in Modern VNAs; they're a lot different than the ones from 20 years ago.  However, I will say the traditional SA's probably still lead in the role of specific applications (such as EM receivers, demod, and other specialty tests).

Of course it works well because it is NOT an VNA.
It's a microwave lab put in single box, that costs more than 250000 USD...

Question was can a simple classic architecture low cost VNA can be used to replace SA. Answer is,  partially it can, with mediocre results...

There are other, combined instruments, that have both SA and VNA building blocks, that work well for both roles, but those are not VNA in sense of question asked..

[Joel_Dunsmore]

Of course it works well because it is NOT an VNA.
It's a microwave lab put in single box, that costs more than 250000 USD...

RF version is quite bit less, and USB versions are less than a new car.

Question was can a simple classic architecture low cost VNA can be used to replace SA. Answer is,  partially it can, with mediocre results...

Actually, PNA is nearly the classic architecture, as is Streamline USB.  Some difference is the IF structure (whether you can process the data as an FFT or you have to use a streaming digital filter) and mostly it is the software implementation.  The older "IM Spectrum" mode in the PNA used a very classic structure.  One really big difference is using Gausian filters instead of VNA filters (VNA filters tend to be nearly rectangular with very high sidelobes).  The next issue is image rejection, and to do image rejection it requires you to change the nominal IF frequency, if you don't have sweeping preselector like a classic SA.. 

But the real roadblock on older VNAs (HP models 8754, 8753, 8409, 8510) was they used a sampling receiver that caused hundreds of spurious images. No way to use those.  By 2002 or so, many VNAs moved to fundamental mixing which is the chief requirement for a reasonable SA response, so there are only a few images (at the IF offset, at the 3rd and 5th harmonics of the LO, for example).  The normal way to deal with the images, even if you have a fixed IF, is to flip the LO from low-side mixing to high-side mixing; images move and real signals don't and you can sort them out in software. But, as you say, most VNAs lack this and so no good way to use them to measure unknown signal.  And the final issue, VNAs skip frequencies, if you have a 1 GHz span, 201 points, you have a measurement every 5 MHz; with a 3 kHz IF BW filter, you're going to miss a lot of the spectrum, so you need to really increase the IF BW and the number of points to avoid jumping over signals.

So, to use a classic VNA, you have to have a lot of points, a wide filter and you will see two (at least) of every signal.