The Essential Guide to Digital Signal Processing - Embedded.com

# The Essential Guide to Digital Signal Processing

Rick Lyons sent me a copy of his latest book about DSPs back in June, and I finally had a chance to read it. It’s titled The Essential Guide to Digital Signal Processing , by Richard Lyons and D. Lee Fugal.

Most embedded.com readers probably will not want to read this book as the material is elementary. It’s aimed at those with really no understanding about DSP. However, I could be wrong, as the DSP segment of the market is small compared to general MCU usage, and perhaps many readers haven’t had a chance to delve into this topic. Signal processing is very important and interesting subject and I do think every embedded person should be familiar with it.

How elementary is the material? Chapter 2 describes analog signals in general. That’s followed by a chapter about frequency. Now, what could be simpler than that? However, the authors wisely talk in terms of degrees and radians, and the latter may be unfamiliar to non-EEs. You cannot understand any of the literature about signals without a solid foundation in radians. Every EE remembers the cos(x*pi+phase) notation, but perhaps CS graduates don’t. The chapter covers spectra and shows how a spectrum analyzer differs from an oscilloscope, in really clear prose.

Chapter 4 is about digital signals. Yawn – that’s what we’re all mired in. But are you familiar with decimation? That’s another critical part of signal processing.

Later chapters are meatier for techies. I have never seen a better explanation of aliasing, which is depicted with simple but very clear diagrams. Anyone using a digital scope needs to understand this. I once walked in on an EE who couldn’t understand why his 32 MHz clock looked like 32 KHz on the scope; a quick spin of the time base knob cured that woe, and an explanation of aliasing lifted the fog from his brain.

The two most demanding chapters cover, unsurprisingly, transforms. The FFT is described in general and a good example shows how it is computed. Wavelet transformations are increasingly-important and the books does a good job describing them. Wavelet transforms weren’t known when I went to college, yet now my son uses them extensively in processing seismic signals. The ten pages devoted to the subject gives a sense of what they are, when they should be used, and how they differ from Fourier transforms.

Filters are covered superficially. More detail would improve the book as they are so critical in many applications.

The rest of the volume is old stuff to old hands. Scientific notation and binary numbers are part of your DNA and there’s nothing new about them. An appendix on dBs is quite complete and worthwhile if you’re not comfortable thinking in logs. I wish the person on the street would read it and realize that “100 dB” is meaningless since dBs are always referenced to some value. Of course, these are the same folks who are giving 110% in football.

Interestingly, there’s nothing on DSP processors. This is a book about signals, not implementations.

The book is very well written with a quick, breezy style. Most engineers will get through it in an hour. My only complaint is that the \$39.95 list price is an awful lot for a 188 page tome. Amazon lists it at \$25, or \$17 for the Kindle edition.

This is a book for the vaguely-techie who needs just a bit more than a little familiarity with the subject. DSPers working for a non-technical boss should slip a copy under the bigwig’s door. It’s also probably the best work for a practicing engineer who wants a passing familiarity with the subject, as it is such an easy read. Early chapters will seem like CS101; skip the stuff you know but do look at every page as there are gems buried even in that material that might surprise.

## 2 thoughts on “The Essential Guide to Digital Signal Processing”

1. cdhmanning says:

“Signal processing isn't just limited to things EE would consider signals. nnAll values can be treated as signals and at least some degree of DSP can be applied. For example, a device reporting pulse rate probably needs some sort of filtering so that the