Rigol DSA815 Spectrum Analyzer

April 25, 2016

Jack Ganssle-April 25, 2016

Rigol has a new low-cost spectrum analyzer (SA). Saelig, their US distributor, loaned me one for evaluation, and the bottom line is that this baby packs a ton of capability into one small and inexpensive package.

They sent me a model DSA815-TG which spans to 1.5 GHz, but other models in the series go to 7.5 GHz. The 1.5GHz unit is just $1495; that climbs to $9500 for the wide-bandwidth model. If you can live without the TG (tracking generator) option the DSA815 is $1295.

But I’d never buy an SA without the TG. Ever. The tracking generator output is the SA’s local oscillator; as the SA scans from the displayed low frequency to the upper one the TG outputs the frequency being displayed. That’s critical for experiments with RF components. For instance, you could feed the TG into a filter and the filter’s output to the SA’s input to display the filter’s frequency response.

Rigol DSA815. Picture swiped from Rigol’s web site.

As is apparent from the picture, the front panel is well-laid out with individual buttons for each function, plus soft-buttons used to select sub-modes. One large wheel can adjust pretty much anything.

Perhaps a little background is in order. Everyone knows that an oscilloscope displays voltage on the vertical axis, and time on the horizontal. An SA shows amplitude on the vertical and frequency on the horizontal. The following picture shows the Rigol displaying the FM radio band. The center frequency is 100 MHz with a span of 20 MHz (that is, the SA is sweeping from 90 to 110 MHz), RBW and VBW (more on these later) of 3 KHz. The marker is barely visible as a “1” towards the left side of the screen. It’s positioned at 91.5 MHz, the Baltimore classical station. The peaks are various radio stations’ carrier frequencies.

Rigol DSA815TG showing FM spectrum. Photo by author.

In this case the SA’s input is one meter of wire acting as an antenna.

A couple of things to note: first, while the DSA815 and all other SA’s will display amplitude in a variety of units, the most useful is dBm, or dB referenced to 1 milliwatt. So 0 dBm is 1 mW, -30 dBm is 1 uW. WBJC, at 91.5 MHz in the photo, comes in at -75 dBm, or around 50 picowatts.

And, though the instrument can display four markers and do delta math between them, they sure are hard to see. When in the noise I found them just about impossible to recognize.

All of the important settings are clearly shown on the screen.

Every modern digital scope can show the FFT of an input signal, which is exactly the same display. So why spend precious bucks on another instrument?

A scope is not particularly sensitive. You’re lucky to be able to see tens of mV. In contrast, the SA is just like a superhet radio – a local oscillator sweeps from a user-selected low frequency up to a desired high frequency, with phenomenal gain since a mixer stage isolates just the frequency being swept at any time. If you’re doing any sort of wireless work an SA is an essential bit of test equipment.

The DSA815’s frequency display is spot on, measured with a high-end HP signal generator.

RBW is resolution bandwidth. It’s the width of the filter after the SA’s mixer stage. So setting this to a big number means a perfect sine wave, which should show as a infinitely-narrow spike in a perfect world, will be broad and mushy. A narrow RBW will display that sine wave as a spike, or nearly so. VBW is video bandwidth, a smoothing filter applied to the displayed data. Obviously you’d want to set RBW as narrow as possible and VBW just where the data is reasonably noise-free. However, there’s a cost. The sweep time of any SA is:

Sweep time = (k * frequency span)/(VBW*RBW)

So setting the bandwidths narrow hugely increases the sweep time.

My fancy Advantest SA can go to a RBW of 1 KHz. The DSA815 is much better, with a low-end setting of 100 Hz. Pretty slick! “k” in the above formula is an SA-dependent constant. I found that with a 1 MHz span, and RBW = VBW = 1 KHz, the sweep took 3.3 seconds, compared with one second for the much more expensive Advantest.

The displayed average noise level is basically the instrument’s noise floor with the lowest RBW. I measured it at -105 dBm, or around 0.05 picowatt. That’s pretty good, though the big test equipment vendors have much more sensitive gear, for much more money. The Rigol is more than adequate for wireless comm work.

One thing I don’t like: you can save an image to a .CSV file, but not to a picture file (e.g., .jpg). Sure, it’s easy enough to convert a .CSV to a picture in Excel, but that’s an extra step.

But it does come with a demodulator for AM and FM (which my Advantest doesn’t have). And there are a ton of other features, like the ability to do n sweeps and stop, where n is any number from 0 to 9999. And there’s a pass/fail mode as found on some scopes: if a waveform falls outside some boundaries a fail signal is generated. Lots of measurements are included like total power, power on adjacent channels, carrier to noise ratio, power of each harmonic, THD and VSWR. The manual, which is generally of barely adequate quality, is very good in the section on the measurements. USB and Ethernet connectivity are standard.

Rigol and other Chinese vendors have changed the test equipment market with a stream of very good gear for astonishing prices. The DSA815-TG has all of the features I deem important in a spectrum analyzer, and you can’t beat the price. Highly recommended.

Jack G. Ganssle is a lecturer and consultant on embedded development issues. He conducts seminars on embedded systems and helps companies with their embedded challenges, and works as an expert witness on embedded issues. Contact him at His website is


Loading comments...