Minimum Shift Keying (MSK) - A Tutorial
How does MSK achieve both excellent spectral efficiency and a constant-envelope signal suitable for nonlinear amplifiers? This tutorial builds MSK step‑by‑step from binary FSK, shows the minimum frequency spacing and continuous‑phase construction, and then recasts MSK as an OQPSK (pseudo‑symbol) representation. It finishes by generalizing MSK into CP‑FSK and the wider CPM family so you can connect practical pulse shapes and modulation indices to performance.
New Video: Parametric Oscillations
Tim Wescott just posted a short new video titled "Parametric Oscillations." It’s a little off-topic for the channel, but he used the project as an excuse to break a months-long posting drought. If you follow his work, this quick update shows how small builds can rekindle momentum and prompt informal explorations of oscillation behavior.
Round Round Get Around: Why Fixed-Point Right-Shifts Are Just Fine
Jason Sachs explains why, in most embedded systems, simple bitwise right-shifts are an acceptable way to do fixed-point division rather than paying the runtime cost to round. He shows the cheap trick of adding 2^(N-1) to implement round-to-nearest, explains unbiased "round-to-even" issues, and compares arithmetic error to much larger ADC and sensor errors. The takeaway: save cycles unless your algorithm or inputs require extra precision.
Some Thoughts on Sampling
Sampling's 1/Ts amplitude factor is not a paradox but a consequence of axis scaling and impulse density, once you view the units correctly. This post walks through impulse trains in continuous and discrete time, uses DFT examples and Parseval's relation, and shows how downsampling and time scaling produce the familiar spectral replicas and their amplitudes. The geometry of the axes resolves the confusion.
Matlab Code to Synthesize Multiplierless FIR Filters
Learn how to build multiplierless FIR lowpass filters in Matlab using Canonic Signed-Digit coefficients. The post explains converting Parks-McClellan floating-point taps to scaled integers, then to exact CSD digits, and includes two m-files that search maintap scaling to minimize signed digits while preserving the filter response. Practical notes cover external gain compensation, the 2/3 full-scale CSD limit, and sensitivity to pass/stop edges.
Wavelets II - Vanishing Moments and Spectral Factorization
This post walks through how vanishing moments turn into concrete algebraic constraints on wavelet filter coefficients, and why that leads to Daubechies filters. It explains how a wavelet with A vanishing moments is orthogonal to all polynomials up to degree A minus one, and it shows how those continuous conditions become discrete sums like sum_k k^n h1(k)=0. Expect clear links between approximation power and filter length.
Fibonacci trick
Tim Wescott shares a compact, surprising trick linking Fibonacci numbers and difference equations. Start with any two consecutive Fibonacci numbers, negate the larger-magnitude one, and iterate the usual recurrence; after a few steps you'll arrive at the standard Fibonacci sequence or its negative. This behavior is specific to the Fibonacci recurrence and makes a great illustrative example for teaching linear recurrences.
The Power Spectrum
You can get absolute power from a DFT, not just relative spectra. In this post Neil Robertson shows how to convert FFT outputs into watts per bin using Parseval's theorem, how to form one-sided spectra, and how to normalize windows so power is preserved. Matlab examples demonstrate bin-centered and between-bin sinusoids, leakage, scalloping, and how to recover component power by summing bins.
New Comments System (please help me test it)
DSPRelated just got a practical upgrade, Stephane Boucher has released a new comments system built from his earlier forum work. It supports drag-and-drop or Insert Image uploads, MathML, TeX and ASCIImath rendered by MathJax, syntax-highlighted code via highlight.js, and in-place editing and deletion of comments. Improved email notifications alert authors and commenters to replies, and readers are invited to post test comments and report problems.
Wavelets I - From Filter Banks to the Dilation Equation
Starting from a practical cascaded FIR filter bank, this post derives the key equations behind the Fast Wavelet Transform. It shows how conjugate-quadrature analysis and synthesis filters give perfect reconstruction and how iterating the cascade produces the scaling function, leading to the dilation equation. DB4 coefficients are used as a concrete example and a linear-system trick yields exact integer-sample values of the scaling function.
Multiplierless Exponential Averaging
Rick Lyons shows how to implement exponential averaging without multiplies by exploiting a rearranged leaky-integrator form and binary shifts. He demonstrates reducing the standard two-multiply averager to a single-multiply form, then eliminating the multiply entirely when the weighting α equals reciprocals or differences of reciprocals of powers of two. The post catalogs practical α choices for fixed-point filters and flags quantization as an open issue.
What to See at Embedded World 2019
Skip the overwhelm at Embedded World 2019, Stephane Boucher lays out a practical preview of what to see and how to prioritize your time. The post helps embedded engineers focus on demos, vendor booths, and sessions that matter without getting lost on the show floor. Read it to plan a short, efficient visit that maximizes technical takeaways and networking opportunities.
Feedback Controllers - Making Hardware with Firmware. Part 2. Ideal Model Examples
An engineer's guide to building ideal continuous-time models for hardware emulation, using TINA Spice, MATLAB and Simulink to validate controller and circuit behavior. The article shows how a passive R-C network can be emulated by an amplifier, a current measurement and a summer, with Spice, MATLAB and Simulink producing coincident Bode responses. Small phase differences between MATLAB and Simulink are noted, and sampled-data issues are slated for the next installment.
Embedded TMS320C6000 Information Kick Off
Mike Dunn kicks off a series on the TMS320C6000 family, drawing on decades of hardware-first experience. He plans pragmatic posts that focus on device differences, odd quirks, and real-world debugging rather than DSP theory. Expect concise troubleshooting stories driven by his Yahoo c6x group activity and an emphasis on explaining why solutions work, not just what to do.
OpenCV for DSP/GPU, MSDN equivalent for CCS, and more
Porting OpenCV to DSPs could be a real business opportunity, but it is far from trivial, writes Shehrzad Qureshi. He highlights major obstacles: the engineering scale, mixed open-source licenses, and hard-to-parallelize primitives like connected components. He also criticizes Code Composer Studio's help system compared with MSDN, notes an ATI Stream talk, and announces a CUDA walkthrough on FFT-based image filtering.
Data Types for Control & DSP
Control engineers often default to double precision, but Tim Wescott shows that choice can waste CPU cycles on embedded targets. He separates numeric representation into floating point, integer, and fixed-point, then walks through the tradeoffs, including quantization, overflow, and performance. A concrete PID example highlights why integrator precision and ADC scaling should drive your choice of data type rather than habit.
DSP Papers, Articles, Theses, etc
Stephane Boucher invites the DSP community to help expand DSPRelated's Papers and Theses repository, which currently lists just over 100 documents. He asks contributors to find and submit recent DSP PDFs, ideally from the last ten years, and notes that each approved submission enters the submitter into a draw for Michael Parker's Digital Signal Processing 101; the draw is planned for early April.
Computing Translated Frequencies in Digitizing and Downsampling Analog Bandpass Signals
Textbooks rarely give ready formulas for tracking where individual spectral lines land after bandpass sampling or decimation. Rick Lyons provides three concise equations, with Matlab code, that compute translated frequencies for analog bandpass sampling, real digital downsampling, and complex downsampling. Practical examples show how to place the sampled image at fs/4 and how to translate a complex bandpass to baseband for efficient demodulation.
Simple Discrete-Time Modeling of Lossy LC Filters
Converting a lossy LC filter into a discrete-time impulse response lets you analyze mixed analog and DSP systems in one time domain. This post walks through computing the LC frequency response via chain (ABCD) parameters including resistive losses, enforcing the Hermitian symmetry required for a real IDFT, and using the IDFT to produce an asymmetrical FIR impulse response. A 5th-order Butterworth example illustrates insertion loss and impulse-shape effects.
Are DSPs Dead ?
Jeff Brower argues that the science of digital signal processing is far from dead, but commercial DSP chips lost momentum when Texas Instruments refused to embrace server-centric AI and 5G markets. He traces how TI's embedded-only culture, halted multicore CPU roadmaps, and lack of server-class products pushed customers to GPUs and FPGAs. A comeback would demand PCIe cards, VM and container support, open-source engagement, and bold leadership.
A multiuser waterfilling algorithm
Markus Nentwig shares a compact, heuristic multiuser waterfilling algorithm with ready-to-run C code, designed for practical radio resource allocation. The approach uses round-robin user handling, per-user power budgets and a mode switch between fixed-power and waterfilling distributions, and it is easy to extend for constraints or QoS tweaks. The implementation is suboptimal by design, fast, and requires verification before production use.
Online DSP Classes: Why Such a High Dropout Rate?
Rick Lyons digs into a startling statistic: online DSP courses reported a 97% dropout rate. He argues the main culprits are math-heavy curricula that overwhelm beginners and rigid, non-self-paced schedules that demand sustained 8-10+ hours per week. Rick urges course creators to rethink pacing and mathematical depth to improve completion rates and student engagement.
The Swiss Army Knife of Digital Networks
A single discrete-signal network can masquerade as a comb filter, a recursive section, or something much more versatile. Rick Lyons shows how this seven-coefficient structure can be reconfigured to realize a wide range of DSP functions, with tables of impulse responses, pole-zero plots, and frequency responses to illustrate each case. The full explanations live in the downloadable PDF, but the post gives a strong feel for why this is such a handy building block.
Feedback Controllers - Making Hardware with Firmware. Part I. Introduction
This first post kicks off a series on using DSP and feedback control with mixed-signal electronics and FPGAs to emulate two-terminal circuits and create low latency controllers. It frames circuit emulation as a feedback problem, highlights latency as the key practical constraint, and outlines the planned evaluation hardware, target devices, and software tools that will be used in later MATLAB/Simulink and FPGA work.
Add the Hilbert Transformer to Your DSP Toolkit, Part 2
This post shows a simple practical route to a Hilbert transformer by starting from a half-band FIR filter and tweaking its symmetry. It walks through a 19-tap example synthesized with Matlab's firpm (Parks-McClellan), explains the required frequency scaling, and shows how even-numbered taps become (or can be forced) zero through symmetry and coefficient quantization. Useful design rules are summarized for choosing ntaps.
How Not to Reduce DFT Leakage
Rick Lyons debunks a proposed 'data-flipping' fix for DFT spectral leakage, demonstrating with MATLAB that it can produce higher sidelobes and a troubling mainlobe dip for some input frequencies. He explains that windowing's goal is to reduce amplitude discontinuities in a periodic extension, not merely to force end samples to zero, and concludes the method is frequency-dependent and not recommended.
Harmonic Notch Filter
A practical, DSP-friendly recipe for scrubbing 60 Hz power-line hum and its harmonics from noisy ECG and EEG recordings is presented, using IIR notch filters built from second-order all-pass sections. The post derives how to set all-pass phase to place notches and compute biquad coefficients by solving a simple 2x2 system, then shows C code and precomputed coefficients for cascading the first eight odd harmonics at a 2 kHz sample rate. Engineers get a compact, editable implementation with explicit control over notch bandwidth.
Somewhat Off Topic: Deciphering Transistor Terminology
Rick Lyons unpacks a small linguistic mystery in electronics, revealing why the transistor's middle terminal is called the "base". He traces the name to the 1949 Bell Labs "semiconductor triode", where the device sat on a metal base plate described as a large-area low-resistance contact, and notes that later transistor sandwich designs kept the name for historical reasons. The post includes original references and a few trivia nuggets.
What to See at Embedded World 2019
Skip the overwhelm at Embedded World 2019, Stephane Boucher lays out a practical preview of what to see and how to prioritize your time. The post helps embedded engineers focus on demos, vendor booths, and sessions that matter without getting lost on the show floor. Read it to plan a short, efficient visit that maximizes technical takeaways and networking opportunities.
Decimators Using Cascaded Multiplierless Half-band Filters
In my last post, I provided coefficients for several multiplierless half-band FIR filters. In the comment section, Rick Lyons mentioned that such filters would be useful in a multi-stage decimator. For such an application, any subsequent multipliers save on resources, since they operate at a fraction of the maximum sample frequency. We’ll examine the frequency response and aliasing of a multiplierless decimate-by-8 cascade in this article, and we’ll also discuss an interpolator cascade using the same half-band filters.
