Christopher Felton (@cfelton)
Constrained Integer Behavior
Overflow and underflow are not always bugs, they can be useful in DSP when fixed-width integers wrap during processing. Christopher Felton demonstrates with moving-average (recursive-windowed-averager) and CIC filter examples how 2's complement wraparound in MyHDL's modbv cancels between an integrator and a comb via pole-zero cancellation. He also covers fixed-point resizing choices, saturation versus wrap, and how rounding error can accumulate.
Python scipy.signal IIR Filtering: An Example
Christopher Felton walks through using scipy.signal IIR filters to demodulate PWM signals, using spectrum and spectrogram analysis to show what works and what does not. He demonstrates using filtfilt to avoid phase delay, compares a single narrow IIR to a very high order FIR, and shows how staged IIR filtering and multirate ideas give much better attenuation. Includes an FPGA-ready MyHDL PWM model.
Python scipy.signal IIR Filter Design Cont.
Christopher Felton continues his practical tour of SciPy's iirdesign, moving beyond lowpass examples to show highpass, bandpass, and stopband designs with concise, code-focused explanations. He highlights how ellip and cheby2 let you tighten specifications for sharper transitions, and shows that the iirdesign workflow is consistent across filter types. Read for clear, reusable examples to produce IIR filter coefficients with scipy.signal.
Python scipy.signal IIR Filter Design
Christopher Felton walks through designing infinite impulse response filters using scipy.signal in Python, focusing on practical specs and functions rather than theoretical derivations. He explains normalized passband and stopband definitions, gpass and gstop, and shows how iirdesign and iirfilter differ. Plots compare elliptic, Chebyshev, Butterworth and Bessel responses, highlighting steep transitions versus near-linear phase tradeoffs.
Curse you, iPython Notebook!
Christopher Felton shares a cautionary tale about losing an ipython 0.12 notebook session after assuming the browser would save his interactive edits. He explains that notebooks at the time required clicking the top Save button to persist sessions, and autosave was not yet available. He recommends basing interactive work on scripts, saving often, and testing export behavior to avoid redoing text, LaTeX, and plots.
scipy.signal calling all developers
There has been some chatter on the scipy-dev mailing list lately about enhancing the scipy.signal package. Unfortunately, there seems to be a split. Some are going off and starting a new package scikit-signal. The original developer, Travis Oliphant, appears to have strong interest in seeing the scipy.signal evovle. If you are interested in signal processing you should check out the mailing lists (
[Book Review] Numpy 1.5 Beginner's Guide
Christopher Felton's review gives a pragmatic take on Ivan Idris's Numpy 1.5 Beginner's Guide, praising its hands-on, exercise-driven approach while flagging several shortcomings. He finds the book a useful starting point for newcomers to Python numerical computing thanks to practical examples and a chapter on testing, but warns the title, incomplete installation guidance, and some factual errors may mislead readers.
Python number crunching faster? Part I
Christopher Felton walks through simple benchmarks comparing raw Python, numpy, and PyPy for numeric workloads, and shares what surprised him about performance. He shows that idiomatic Python optimizations such as list comprehensions and built-ins plus the PyPy JIT can sometimes beat a numpy approach for small tests, and explains why native PyPy numpy progress matters for scientific users.
Impulse Response Approximation
A stepped-triangular impulse approximation represents an FIR low-pass using a cascade of recursive running-sum filters, offering big savings in computation. Christopher Felton outlines the quantization step that maps a true impulse into three stepped-triangular types and shows how the approximation is built from recursive running-sum and sparse-sum blocks. Inspect the frequency tradeoffs and decide if the efficiency gain is worth the approximation error.
A Fixed-Point Introduction by Example
Christopher Felton walks through binary fixed-point representation with clear examples and a simple W=(wl,iwl,fwl) notation. He argues for designing to range and resolution rather than bit counts, then shows how multiplication and addition affect bit growth and alignment. These concrete examples make it easy to see why rounding, resizing, and radix-point bookkeeping are essential in DSP implementations.
Matlab Programming Contest
Love puzzles or want to sharpen your MATLAB skills? Christopher Felton highlights MathWorks' biannual MATLAB programming contest, a week-long set of clever algorithm challenges that require only base MATLAB. Whether you're experienced or new, you can compete, compare solutions, or simply study others' code when later phases disclose submissions. No toolboxes or mex files allowed, so it's a pure programming playground for learning and bragging rights.
Use this form to contact cfelton
Before you can contact a member of the *Related Sites:
- You must be logged in (register here)
- You must confirm you email address
