Exclusive! Video Tektronix RSA306 USB Real-Time Spectrum Analyzer Review

[Hugoneus]

In this world exclusive episode, Shahriar takes a close look at the all new Tektronix Real-Time USB Spectrum Analyzer. The RSA306 has an RF bandwidth from 9kHz to 6.2GHz with a real-time analysis window of 40MHz. While being entirely powered form a single USB 3.0 interface, it provides a 100% Probably of Intercept for any signal event of at least 100us in duration. The RSA306 is compact, lightweight and shock resistant. At the same time Tektronix has made its Signal-Vu PC software with 17 measurement capabilities free and significantly reduced the price of other Signal-Vu advanced applications!

The SFDR, linearity, noise floor and phase noise of the instrument are all measured and presented. The unit is then used to capture and analyze an intermittent RF interference, analyze and demodulate an RF FM signal using an antenna as well as demodulation a 256-QAM signal with -65Bm of total power.

Watch it here [1 Hour]:
http://youtu.be/GDcuRTOCj_s

More videos at The Signal Path:
http://www.TheSignalPath.com

[jeremy]

I know I'm going to be disappointed, but what is the price? 

[dr.diesel]

Wow, in total, price disruptive might be an understatement!   

Excellent video once again.

[ronaldlijs]

RRP of £2,340.00 showing on Farnell's website. I think this is a great price for such a fantastic piece of kit. I've got a cheapo Rigol SA with tracking gen, but would rather prefer a small factor like this for the occasional work I do on FCC/EC compliance, of course would be double the price... but will think about it no doubt!

[fcb]

Just watched the first 25mins of The Signal Path's review - I don't think Tek have finished the unit yet and have released it prematurely, those -55dB spurious tones put me off - I don't think even the Rigol SA is that bad.

[Hugoneus]

Just watched the first 25mins of The Signal Path's review - I don't think Tek have finished the unit yet and have released it prematurely, those -55dB spurious tones put me off - I don't think even the Rigol SA is that bad.

The tones at -55dBm were certainly an issue. However, keep in mind that they scale down with the reference, so the worst case SFDR is -55dB. During most of the band, you can get even better than that.

I do wish they weren't there though.

[German_EE]

K5OBS also has a "World Exclusive" review of this device on his YouTube channel although the review on The Signal Path is more extensive. I would like a new spectrum Analyzer, I really would, but when you consider the cost of this device plus the cost of the new laptop that would be needed to drive it the stand-alone units from Rigol are still a viable option.

I would be interested to see a side by side review of this unit together with the Signal Hound and (say) a $3000 Rigol device.

[es]

Just watched the first 25mins of The Signal Path's review - I don't think Tek have finished the unit yet and have released it prematurely, those -55dB spurious tones put me off - I don't think even the Rigol SA is that bad.

The tones at -55dBm were certainly an issue. However, keep in mind that they scale down with the reference, so the worst case SFDR is -55dB. During most of the band, you can get even better than that.

I do wish they weren't there though.

Theoretically speaking, as a selectable option, would it be possible for the software to do one (or more and average) capture with the RF input off and use that to subtract the spurious tones from a capture with the RF input on? Would that technique be effective?

Thanks for the very nice video btw.

[fcb]

Just watched the first 25mins of The Signal Path's review - I don't think Tek have finished the unit yet and have released it prematurely, those -55dB spurious tones put me off - I don't think even the Rigol SA is that bad.

The tones at -55dBm were certainly an issue. However, keep in mind that they scale down with the reference, so the worst case SFDR is -55dB. During most of the band, you can get even better than that.

I do wish they weren't there though.

Great review as usual Shahriar, love to see a tear-down of that unit.

What worries me isn't so much the annoyance of having those tones there, but the mixed product of those tones with the signal under examination, even more so if you were to use that device for spectrum monitoring.  I would find it unnerving/annoying having to constantly check if what I was seeing was really there.

Also 914MHz is smack in the middle of the 902-928MHz ISM band...

[HighVoltage]

Thanks for a great review on this SA.

One of the reasons why Tektronix has to be aggressive in the price structure for a USB based SA, is probably the German based company AARONIA. They have USB based SP for many years and very aggressive pricing.
http://www.aaronia.de/

[Lukas]

I'm wondering how this compares to the midrange SDRs floating around. Take the USRP B210 for example: It's also up to 6GHz, has more capture bandwidth (although lower sampling rate) and costs around $1100. Does tek provide an API to stream the data out of their application? This would put their device into the SDR market.

Great video btw, can't wait for the teardown!

[es]

Does tek provide an API to stream the data out of their application? This would put their device into the SDR market.

Looks like they got that covered, search for "RSA306 API" on Tek site. By looking at the Dll and C headers files, there are functions for data streaming and audio demodulation.

[Rasz]

Theoretically speaking, as a selectable option, would it be possible for the software to do one (or more and average) capture with the RF input off and use that to subtract the spurious tones from a capture with the RF input on? Would that technique be effective?

you would still get mixing product between those ghost signals and your  real signal



as a total noob I dont see the value of this tek product, like Lukas said there are plenty of SDRs that are many times cheaper, as soon as you stop caring about accurate signal power measurement and can accept relative levels any SDR will deliver similar or superior experience at a fraction of a price.

[G0HZU]

I didn't watch all of the video but my impression is that the software is not really 'free' because the RF performance of the analyser front end is so poor.

Therefore, my interpretation of this product package is that you are paying for the very versatile software and getting a fairly ropey 6GHz USB signal analyser (with 40MHz RT BW) thrown in for free. The analyser is probably going to perform best as a learning tool when looking at modern digital transmissions and is better described as a signal analyser rather than a conventional spectrum analyser for this reason.

The phase noise of the analyser at 25:55 when set to a 1MHz span is amazingly poor. It looks like it is about -92dBc/Hz at 100kHz offset. That's about 35dB worse than my old japanese made analyser model that was first introduced back in 1981. The other spurious terms it produces look very poor indeed. 

Maybe they will improve all this with future hardware revisions of this analyser but it looks like this hardware platform is a cheap educational gateway to Tek's powerful DSP/software engine.

[Hugoneus]

Just to answer a few questions:

Yes, there is an API where you can grab data and do what you please with the data. It is part of the driver installation and is free.

The spurs that you see limit you to -55dB SFDR. The absolute power of the spurs shrinks with reference level as I have shown in the video.

The phase noise at 100kHz offset is -92dBc/Hz (at 1GHz), which of course is not amazing. But as you can see, you can demodulate 256QAM signals with it without any issues.

Whether all this fits your needs, is of course up to you. O0

[Hugoneus]

The phase noise of the analyser at 25:55 when set to a 1MHz span is amazingly poor. It looks like it is about -92dBc/Hz at 100kHz offset. That's about 35dB worse than my old japanese made analyser model that was first introduced back in 1981.

What analyzer is this?

[G0HZU]

It's an old Takeda Riken (Advantest)  TR4172. At 1GHz it measures -128dBc/Hz at 100kHz offset

Of course, it isn't a 6GHz analyser (only 1800MHz) but I also have an old (1982?) HP analyser here that goes to 22GHz and it shows a typical -120dBc/Hz at 100kHz offset at 1GHz input.

The point here is that some people on this thread will be looking to use this new Tek USB analyser as an all round spectrum analyser and the noise and spurious performance is really quite poor compared to other options. Even the modern and cheap Rigol analysers manage about -106dBc/Hz at 100kHz offset at a similar input frequency and they aren't dogged with all those -55dBc spurious terms either.

Maybe try sticking an antenna on it and see what happens. I think a lot of hobbyists will want to play with antennas on a spectrum analyser and I suspect that this new Tek analyser could perform very badly with a common task like this. That's why I think it is really better described a signal analyser where the range and qty of signal inputs can be controlled by the user.

[Hugoneus]

It's an old Takeda Riken (Advantest)  TR4172. At 1GHz it measures -128dBc/Hz at 100kHz offset

Of course, it isn't a 6GHz analyser (only 1800MHz) but I also have an old (1982?) HP analyser here that goes to 22GHz and it shows a typical -120dBc/Hz at 100kHz offset at 1GHz input.

You just can't beat a good YIG...

[w2aew]

It's an old Takeda Riken (Advantest)  TR4172. At 1GHz it measures -128dBc/Hz at 100kHz offset

Of course, it isn't a 6GHz analyser (only 1800MHz) but I also have an old (1982?) HP analyser here that goes to 22GHz and it shows a typical -120dBc/Hz at 100kHz offset at 1GHz input.

You just can't beat a good YIG...

Yeah, there are a lot of things you simply can't do when you've got a total power budget of 4.5W.

[G0HZU]

But as you can see, you can demodulate 256QAM signals with it without any issues.

That's why I think it would be better described as a signal analyser because it will perform very well when demodulating/displaying complex signals within its 40MHz RTBW. Especially under controlled conditions such as in a test lab. So I would see this as its main role. It looks like the key design driver was to be able to cover the wifi bands up to 5.6GHz.

But because of the poor spurious and phase noise performance it would struggle if used with an antenna on a busy band elsewhere. So it wouldn't perform well as a spectrum analyser when trying to display numerous signals over a wide frequency span and over a fairly wide spurious free range.

Is there a block diagram of the signal path? I'm guessing that it uses a simple up/down dual or triple conversion for the whole range (apart from the direct baseband range?). This would need a very wide tuning range VCO and mixer at the front end and would go some way to explain the high phase noise.

Have you checked to see if the phase noise is a lot better down at baseband? eg at 20MHz?

In recent years there have been various lowcost wideband VCOs and mixers available from the main manufacturers like Hittite. The specs on these wideband devices get better and better from the various manufacturers over the years so maybe it uses the latest generation of these wideband devices from someone like Hittite.

[Hugoneus]

I intend to do a teardown video of it, I am sure we will learn more about it. The phase noise is much better at 20MHz, I'll check what it is.

[wn1fju]

Personally, I would be a little hesitant about this new Tek product.  I buy most of the test equipment for our company and had the pleasure of spending $90,000 to buy the Tek RSA6106 spectrum analyzer some years ago.  The RSA6106 software has a similar look and feel to this new RSA306.  Let me say, in my opinion, the software was without question the worst implementation of a piece of test equipment I have ever used.  Even after a few software updates, the RSA6106 was extremely frustrating to use and had many things that did not work properly.  Sure, the 110 MHz instantaneous BW and the DPX spectrum (which has something like a 23 usec update interval) was nice, but once you were beyond the faulty software, the basic spectrum analyzer wasn't even anywhere near a mid-range Agilent or Rohde&Schwarz analyzer.  So I probably only used the thing a total of 10 hours in a couple of years (certainly got my ninety grand dollar's worth, didn't I).  Then I turned it on one day only to have it fail.  Tek wanted $11,000 for a one-incident fix.

[G0HZU]

At work we have several of the big Tek RSAs and they are quite popular with the engineers involved in custom RF waveform development. So for stuff like this they seem to be very capable and powerful. However, I don't think I've ever seen an RF engineer use one of our Tek RSAs for general RF design work as they are not really optimised for this. So I guess it's a case of horses for courses.

Note:
It would be interesting to see how well the little USB analyser performs at baseband. eg up to 40MHz? Because it would presumably connect direct to the ADC via buffering and a LPF the spurious free dynamic range and phase noise would hopefully be a lot better and it may prove to be a very useful tool up to 40MHz.

I wonder how it would perform if connected to the IF port of a conventional spectrum analyser? If it has a 14bit ADC it may prove to be a usefull addon to something like my old 1500MHz HP8568B or 22GHz HP8566B analysers at the 21.4MHz IF output.

I'd certainly like to try one here if it proved useful. Especially if the Tek SW allowed user offsets to be entered (eg 21.4MHz IF to true RF). If the controlling/display PC could also steer and control the analyser via a USB to GPIB interface it would be really neat. I keep meaning to make my own modern digital IF back end for these analysers but never get around to it.

[Mr Simpleton]

We had Tek showing this little marvel at the office during a training course for the bigger RSA6000 that we have. But unfortunately I was away on a field trial, bugger!
Pretty cool it uses basically the same s/w as the bigger brother, i.e. signalvu

Using real time spectrum analyzers, I prefer NOT to use the RSA6000 for anything that our "normal" SA's can do. I find the user interface not the best for just watching the spectrum, but that might just be that I doesn't feel comfortable with the UI.
 
And do not forget that the RSA306 is a lot slower than it's bigger brother in capturing signals, it's not a replacement for the RSA6000, but a compliment (but it IS ~25-50 times cheaper) Tek even got slapped on its fingers by governement agencies for selling too good instrument, forcing them to slow down exsisting instruments in a firmware upgrade and then the user had to apply for an option (free of charge) to bring it back to "normal" if they were worthy   

Now the RSA306 do have nice speed, size and portability, making it a really cool tool when out in mother nature doing measurements. I love running instruments on USB power!

Looking at phase noise data published it is clear that this is not a threat to normal bench-top instruments. Tek has cut some corners when designing this and it shows in the numbers.
Still I'll try sqeeze out some fundings for this little box, it sure will be handy when tesing equipment outdoors.

[macgyver0815]

Hey Shahriar,

all your Videos are awesome!

Maybe you could do a comparision of low cost USB Analyzers, that would be great.
(Tek RSA306, Signalhound BB60, Aaronia Spectran V4 and V5)
Does not have to be all in the same video, just for example measure the same signals or so and
compare saved results.

Cheers!