What Do You Know About DSP?


I learned DSP with a mix of books, manufacturer’s literature and today’s Internet resources. I must have a dozen books on DSP. Some are just incomprehensible academic tomes while others do get to the practical side of DSP. It took this mix of books to make things clear. I still have not found the ideal book but some come close.

Digital signal processing is not one of those subjects that older engineers learned in college.  If you learned it at all, you did so with books, seminars and maybe a colleague.  So many of you never learned it, period.  But of course you all know what it is.  Awareness does not mean you can design with it, but at least you can appreciate how something so complex can become so commonplace.  DSP has become a common feature in most electronic products today.  In fact, it is difficult to think of a modern product that does not use DSP. It is in TV sets, smartphones, iPods, any digital radio, medical equipment, and others.  Help me out here.  Name one that does not use it.  Maybe a guitar amplifier, but then again I could be wrong about even that.

I first learned of DSP in the 1960’s when we processed geophysical seismic ground readings from geophones recorded on magnetic tape to find specific signal shapes and frequencies.  This was just after Cooley-Tukey put forth their famous fast Fourier transform (FFT) algorithm.  The processing was done on an IBM 7090 and the output was a long fan fold paper printout.  It was very time consuming but it worked.  It was fine for research but not ok for practical signal processing.  Today, geophysical signals are routinely processed in DSP on fast PCs.  No big deal.

I learned DSP with a mix of books, manufacturer’s literature and today’s Internet resources.  I must have a dozen books on DSP.  Some are just incomprehensible academic tomes while others do get to the practical side of DSP.  It took this mix of books to make things clear.  I still have not found the ideal book but some come close.

Just recently I came across a new book that is worth looking into if you are just learning DSP on your own.  It is The Essential Guide to Digital Signal Processing by Richard G. Lyons and D. Lee Fugal (Prentice Hall).  This book does a great job of detailing analog signals, sampling, spectra, and related topics.  These are the real basics of DSP and are often misunderstood.  Some DSP books just assume you know this already.  Many do not and I certainly was not that versed in those fundamentals when I started.

The book goes on the give several examples of how FFT works and digital filters.  What the book does not provide is details on algorithms for specific processes, but it does a fine job of giving you the concepts that prepares you for the more advanced books.  If you are starting from scratch, read this book first then go on to some more advanced books.

Author Richard Lyons also has a more advanced book called Understanding Digital Signal Processing (Prentice Hall, 2001) when you are ready for the gory details.  Other books that really helped me are Digital Signal Processing: A Hands-On Approach by Charles Schuler and Mahesh Chugani (McGraw Hill, 2005), Digital Signal Processing Technology by Doug Smith (American Radio Relay League, 2001), and Practical Digital Signal Processing for Engineers and Technicians by Edmund Lai (Newnes/Elsevier, 2004).

DSP was initially done on large fast computers.  Then in the 1980s we got DSP microcontrollers like those from Texas Instruments and Freescale (Motorola).  Today PC and smartphone micros are fast enough to run DSP.  Much of today’s DSP is done in an FPGA or in custom logic.

If you have not learned DSP yet, it is a great self education project with lots of resources.  What we really need is a cheap development kit so we can demo FFT spectrum analysis or try out different FIR or IIR filters.

Discuss this Blog Entry 8

on Jul 9, 2014

Lou - you're, like, so last century. A number of guitar amplifiers today use DSP to "model" other guitar amplifiers. So instead of owning a Fender amp, a Vox amp, and a Marshall amp, a guitarist can own a digital modeling amplifier and get the sound of several amps with one piece of gear. Nothing escapes the use of DSP anymore ;-)

on Jul 9, 2014

Another book that gets the basics across in a very readable manner is "The Scientist and Engineer's guide to Digital Signal Processing" by Steven W. Smith. And it's free!

on Jul 11, 2014

This is the best introductory book I have found for DSP. Plus it is free as Ben says. Just type dspguide in your browser's address bar. I bought a hard copy I loved this so much.

on Jul 9, 2014

Rick Lyons is an excellent author, both in terms of knowledge as well as ability to convey it clearly. I have his previous book, also cited, Understanding Digital Signal Processing and it is one of my most valuable references.

There are a number of ways to experiment with this. Several DSP vendors offer relatively low cost development boards with full or limited design tools. The difficult part about that is that the hardware (specifically ADC and DAC) requirements are so application specific that it may not be possible to do real work on a development board.

Another path is the FPGA (as mentioned). Even the lowly Spartan 6 can be acquired with enough MACs and block RAMS to do serious FIR filtering at RF frequencies. The biggest downside to this is that unlike DSPs you have to create the "processor" yourself. There isn't a simple path to writing code that can use the MACs in a generalized manner at operating speed. The intended path is MatLab/Simulink feeding the DSP version of Vivado or Plan Ahead. Sadly that tool chain path adds up to well over 10 grand--not really learning tool stuff.

MatLab/Simulink or even MathCAD can do a lot of modelling. I suspect Octave can, too (and it's free).

I find that I can do my modelling just fine with a C compiler. Yes it is orders of magnitude slower, but collect data with a suitable ADC into a file and then run it through some C code that is implementing the DSP constructs, placing the output in another file. It isn't real time, but very useful.

Another tool I've found invaluable is Scope FIR, which designs a variety of FIR filters. Excellent time saver at a very reasonable cost. It's been in my toolbox for almost 15 years.

on Jul 9, 2014

Hello whelm, this is Richard (Rick) Lyons. Thank you for your kind words. If you send me an e-mail, using r_dot_lyons_at_ieee_dot_org, I'll send you the errata to my "Understanding DSP" book.

on Jul 9, 2014

Lou, take a look at the TI TMDX5505EZDSP, available from DigiKey for $51+shipping. And no, I don't work for TI nor a distributor.
"What we really need is a cheap development kit so we can demo FFT spectrum analysis or try out different FIR or IIR filters."

on Jul 11, 2014

Check out SciLab for learning DSP concepts. It is an open source math package similar to Matlab. Of course the algorithms are really pretty simple when you see them so you could also use any good programming language, if you are so inclined.

on Jul 12, 2014

Our standard DSP textbook was written by Proakis/Manolakis. But before I took the course, I did some advanced reading over the summer using the book given by Mr. Ben Shamburger (Scientists and Engineers Guide), which worked out for me since it was very comprehensive and easy to understand for a beginner. Over time, you'd discover more DSP techniques cooked up in journals or when you do a written assignment (for me it was the short-time fourier transform which also included information on signal duration). Using Octave (mentioned by whelm) to do DSP is (for me) like trying to use a calculator in implementing a numerical method algorithm (compared to MATLAB's superiority and easy GUI)

Newsletter Signup

Please or Register to post comments.

What's Communiqué?

Blogs on topics such as wired and wireless networking.


Lou Frenzel

Lou Frenzel is the Communications Technology Editor for Electronic Design Magazine where he writes articles, columns, blogs, technology reports, and online material on the wireless, communications...
Blog Archive

Sponsored Introduction Continue on to (or wait seconds) ×