For novel ideas about building embedded systems (both hardware and firmware), join the 35,000 engineers who subscribe to The Embedded Muse, a free biweekly newsletter. The Muse has no hype and no vendor PR. Click here to subscribe.
Episode 17: Rigol's DSA 815-TG Spectrum Analyzer
May 6, 2016
(Go to the complete list of videos)
I'll present my Better Firmware Faster seminar in Melbourne, Australia February 20. All are invited. More info here.
Thanks to Saelig for loaning me the unit.
I'm Jack Ganssle and welcome to the Embedded Muse video blog. This is a companion to my free Embedded Muse newsletter which is available online. Today we're going to do a short take about Rigol's new DSA815 Spectrum Analyzer. But first you might wonder, what is spectrum analyzer?
Perhaps the easiest way to answer that is by demonstrating the difference between a spectrum analyzer and an oscilloscope. So here is my test set up. This is the new Rigol spectrum analyzer. That is nice Agilent oscilloscope, and this is my Siglent generator which is currently putting out a 40 megahertz sine wave. As you can see on the oscilloscope, it looks like a sine wave, exactly as you would expect. The vertical axis of course is voltage, the horizontal axis is time, and here is the same signal being displayed on the Rigol spectrum analyzer. In this case, the vertical axis is not volts but is the amplitude that is measured in dBm, or decibels referenced to one milliwatt. More on that later. And on the horizontal axis, this is now frequency. So our 40 megahertz peak is right there, 40 megahertz, and everything else is noise.
What the spectrum analyzer is showing us is the distribution of frequencies of an input signal. This Rigol, this particular unit goes from 9 kilohertz all the way up to one and a half gigahertz and costs only $1,500, which is really a bargain for a spectrum analyzer. Other members of the family go up to seven Gigahertz and are correspondingly much more expensive.
Okay, so let's have some fun with the spectrum analyzer which will really demonstrate its capabilities. I've changed the input. No longer is the signal generator driving it. Now, it's connected to a clip lead, basically a meter of wire, and that of course is an antenna. I've tuned to the FM radio band, a centre frequency here is 100 megahertz, and the span which is starting from here going to there is 20 megahertz. In other words, it's displaying all the frequencies it picks up from 90 to 110 megahertz.
All spectrum analyzers have markers which let you identify signals. This particular unit supports four different markers and this is one of my very few beefs with this particular unit. You're going to have a hard time seeing it but here is marker number one. You can see it jumping around right there, or maybe you can't. The number is so small, the number one, indicator on that marker is so small I find it very difficult to see. I can ask it to search for the next peak to the left, and here it is right on top of this signal here which is 91.5 megahertz. That's WBJC, the Baltimore area classical station, and the signal is about minus 50, minus 55 or so dBm. That's the amplitude.
One of the cool things about this spectrum analyzer is that it has a demodulator. You can actually plug headphones in here and this thing will demodulate AM of FM signals. In other words, you can listen to the radio station. That'd be a very expensive radio, but there are other reasons why doing wireless development for example, it might be useful to be able to listen to signals. You can see it shows all the important parameters here, for example the centre frequency 100 megahertz, the span 20, resolution bandwidth right now is set to 30 kilohertz of video bandwidth to 30 kilohertz, and the sweep time is 22 milliseconds, and that's a nice feature. A lot of spectrum analyzers don't tell you what the sweep time is and I'm going to talk more about that in a little bit.
But what's this resolution bandwidth thing? That's kind of confusing. The resolution bandwidth is essentially the width of a filter that's sweeping across the range of frequencies that the analyzer is sucking in. So for example, suppose we had a signal that in the real world looks like this. Again, this is frequency and that is amplitude. The resolution bandwidth is a filter that looks like, maybe it's that wide and sweeps across the signal in that direction. When it gets to the signal of interest here, that's the filter that's supplied to it. So the resolution bandwidth is extremely narrow. For example, you'll see a signal looks just like the original signal. If it's very wide, suppose it's going from here to here, it's going to look much blurrier.
So why wouldn't you have a resolution bandwidth as narrow as possible all the time? The problem is the narrower the bandwidth is, the slower the analyzer sweeps across the frequencies of interest. In fact, we have an equation for it like most things. The sweep time equals some constant, times the frequency span that is being displayed on the analyzer, divided by the product of the resolution bandwidth, times the video bandwidth. So, the narrower these terms are, the slower it sweeps, which of course no one wants to wait a long time to see a display.
So let's demonstrate the resolution bandwidth and press this button here in order to set it. You can see it's currently set to 30 kilohertz. If I broaden it, now it's at 300 kilohertz, you can see the signal's getting rather mushy. They appear to occupy more bandwidth than they really do. If I sharpen it, you can see the sweep has slowed down as I've described, but the signals look much closer to their actual values. Just to illustrate the resolution bandwidth with a cleaner signal, here it's at 40 megahertz sine wave from my signal generator again. We have a resolution bandwidth of one megahertz, and if everything were perfect, the spectrum analyzer would show a sharp spike for a perfect sine wave. But because the resolution bandwidth is so broad, it's a very mushy, very spaced out signal.
As I reduce that, here is 100 kilohertz, 10 kilohertz, now going down here, I'm all the way down 100 hertz. Well, you're not going to see anything because the sweep time is now 1,500 seconds and I'm not going to sit around here and wait for 1,500 seconds. But you would see it as being a very very sharp peak. You'll go to three kilohertz with a sweep time of two and a quarter seconds, and you can see that effect. The video bandwidth, which I can also select here, is a filter that's applied. So all that does is smooth things out and reduces the noise level.
The Rigol, like most spectrum analyzers, can display the amplitude of the signal using a variety of the units, but in the RF world, the one that's most useful to us most of the time, is called dBm, and everyone's heard of course, of decibels. dBm is decibels referenced to 1 milliwatt. So, zero dBm is equal to one milliwatt. Minus 30 dBm is equal to one microwatt and so on and so on.
Another thing I don't care for with this spectrum analyzer, and that gives us a grand total of two little nits I don't particularly care for, is that you can't save the displayed image to a picture file. You can save it to a, say a .CSV on a USB stick for example, and you could of course convert that into a picture file in Microsoft Excel, but I prefer to get a JPG or a GIF or something of that nature.
Now, this unit comes either with or without a tracking generator. The $1,500 price I noted earlier includes a tracking generator which in my opinion is an absolutely necessary thing to get. I would never buy a spectrum analyzer without one, and all a tracking generator is is the local oscillator of the spectrum analyzer. So, as it scans across the frequencies, whatever frequency it's actually on at that second in the sweep or that millisecond in the sweep, that frequency, a sine wave, is generated at the tracking generator output and then you can do things with it.
For example, if you're designing a filter, you could feed the tracking generator into the input of the filter and then the spectrum analyzer could then read the output of the filter and you'd get a beautiful shape of waveform describing how effective the filter is with the cut offs, bandwidth and all that. The easiest way to understand the tracking generator is to feed that output into the input of another spectrum analyzer. And here the Rigol's tracking generator is going to my Advantest spectrum analyzer and you can see the tracking generator output, again the local oscillator of the device is marching across the screen as the Rigol sweeps from 0 to 40 megahertz.
Rigol's DSA815 has, in my opinion, pretty much any feature you would want for a spectrum analyzer, and some of the features are really surprising. For example, you can get a resolution bandwidth down to 100 hertz, which is a pretty spectacularly low number. My very expensive Advantest can only go down to one kilohertz.
All in all, I think it's a terrific value for a spectrum analyzer and I highly recommend it. So, there you are. There, you've seen the Rigol DSA815. I have to return it now to fine folks at Saelig who loaned it to me, unfortunately. I'd love to keep it here. But if you're looking for one, I highly recommend checking out this unit.
Thanks for watching and stay tuned to the Embedded Muse e-newletter for more videos and more information on building embedded systems.