spectrum analyzers (was Re: sidebands revisited (was: ben-wpan around the world))

Werner Almesberger werner at almesberger.net
Fri Mar 25 02:33:11 EDT 2011

Richard Sharpe wrote:
> This might be a useful starting point:
> http://www.wireless.org.au/~jhecker/specan/

By the way, I use an USRP2 with an XCVR2450 board for my measurements.

There are several classes of spectrum analyzers:

- high-end lab grade: wide frequency range (from a few kHz to >> 10 GHz),
  high frequency resolution (kHz or less), calibrated, very very expensive

- medium lab grade: moderate frequency range (up to a few GHz), high
  resolution, calibrated, expensive

- USRP-based: narrow frequency range (XCVR2450 just covers the 2.4 GHz and
  5 GHz ISM bands), high resolution, not calibrated, ~2 kUSD for a complete
  system, can also demodulate the signal (it's basically a vector signal

- low lab grade: like medium lab grade, but usually with some constraints,
  price a bit below a USRP solution

- WLAN analyzers: narrow frequency range, low resolution, cheap

- WLAN finders: narrow frequency range, extremely low resolution, very

Different devices are good for different purposes. E.g., a WLAN analyzer
can be used by a network admin to find holes in the coverage, detect
failing access points, and identify the place where some prankster has
wrapped tin foil around the antenna.

The medium lab stuff can be used to debug a device or a design. E.g.,
find spectrum leakage near and far (but not too far) from the desired
frequency, see how component changes affect the spectrum, etc.

The high-end lab stuff can also be used for certification. For
certification, equipment needs to be calibrated and it must be able to
detect even extreme outliers (e.g., to properly certify correct
operation of a 2.4 GHz device, one has to be sure that there's no
excessive spectrum leakage up to 25 GHz.)

The USRP is a compromise between getting something that can see many
(but not all) things, that can generate signals, that's flexible and
completely open, and not spending too much money.

- Werner

More information about the discussion mailing list