Electronicdesign 4928 Xl 63921 Tab

The Front End: Caveat Lector

Jan. 25, 2017
Kendall Castor-Perry, a.k.a. "The Filter Wizard," joins Electronic Design's Contributing Tech Expert community, offering his unique and insightful takes on electronics technology.

Electronic Design welcomes Kendall Castor-Perry, a.k.a. "The Filter Wizard," to our community of Contributing Technology Experts. Kendall is well-known in industry circles for his expertise in all things electronics, particularly in the world of filters, as well as for his myriad published/syndicated columns in prominent publications. He is currently Senior MTS Architect for the Programmable Systems Division at Cypress Semiconductor.

Front end (n): The entry point of signals into a system. Usually determines the performance of the entire signal chain.

Envisage the scene: A typical chip company office floor, the cloth-shrouded cube partitions sucking up stray sound, excessively bright colours, and any hope of getting away early.

A greying engineer, staring into the beige nothingness, blinks in astonishment.

“A column?  In Electronic Design?  Me?”  I look around the office, embarrassed at what my local colleagues might think. Fiddle with my headset, its puffy faux-leather cushion squeezed against my left ear as a thin, disembodied voice fights against the poor call quality.

“That’s [crackles]. They are interested. They’ve seen the things you’ve written in the past, and are [crackles].”

I stretch my fingers. Look at the keyboard that, should I go down this path, would be their almost-constant companion, the print on its key-caps already fading—worn away completely, in the case of the i, a key that I tend to overuse, rather like comma-delimited run-on sentences. And sentences that begin with conjunctions.

“I’ll do it.”

“[crackles] let them know what you intend to write about. Maybe make a list of [crackles][crackles]”

And with that, he’s gone. Another victim of inadequate RF channel SNR messing up information transfer, raising the packet error rate until the poor radio chip in the headset can stand no more. And that’s a great topic to add to that list of things to write about in this brand-new column. But first, a little about me.

I’m a physicist by training, a silicon chip architect by current profession and a sci-fi fan by spare-time inclination. For close to four decades, both outside and inside the semiconductor industry, I’ve been chasing analog signals through electronic circuits and systems, wringing out the information they are trying to conceal, and battling with monsters lurking within murky forests of transistors.

During this time I’ve written a lot of technical stuff to mentor and encourage others to follow in my tracks, often under the guise of “The Filter Wizard.” You could think of me as the Van Helsing of analog system design.

But man cannot live by fact alone. In what precious fragments of spare time may come, I also write fiction—short stories, novels, and feature screenplays in the sci-fi, technothriller, and horror spaces. You’ll find me somewhere on the what-if to what-the-Hell continuum…

Back to this column. Do we need another stream of thoughts on electronic design?  Isn’t there enough out there on the web already?  Well, caveat lector, my friends. Rather like Duff beer in The Simpsons, today’s richly stocked web seems to me to be both the solution to, and the cause of, many engineering problems. Misinformation is rife, and the old ways are being lost. If I can’t beat this, I can at least try to join it, to pass on some hard-won insights, experience, and comfort, from the trenches and from ivory towers (I’ve spent time in both).

So. What are the things that people ask for advice on the most? What are the beliefs that people hold about this discipline that get in the way, rather than help?  What has got me into trouble in the past?  Here are a few sketches of things I’m thinking of writing about here:

Jumping to confusions
Every circuit tells a story—but that circuit may be what fiction writers call an “unreliable narrator.” Don’t make up your mind too early.

Climbing the not-highest mountain
A tale of local versus global optimization. Don’t start fine-tuning your design until you really have done all of the coarse tuning. Otherwise, you’re neatening arrays of deckchairs on the Titanic rather than steering the ship.

Where does the wrong come from?
Governments may not be able to figure out a budget, but you can—an error budget, in which the wrong gets shared out fairly. Applies particularly to analog-to-digital converter resolution (hobby-horse alert…).

Nah, mate, that’s a filter
A useful rule-of-thumb difference between filtering and bandlimiting: If you leave out bandlimiting, performance will suffer; if you leave out filtering, function will suffer.

Beware the one-leaded voltmeter…
… and the person who declares that the voltage *at* that node is 1 V. Also starring the mythical GND beast, a monster that, despite what some seem to believe, cannot be everywhere at once.

Schrödinger’s node
Or maybe it was George “Bishop” Berkley’s. A node that you don’t bother to monitor in your system, but that might—or might not—be crying out for help.

One slew over the cuckoo’s nest
When is an op amp not op ampy?  When it’s slewy.

Tell me what happened, and when it happened.
Save the data and the date. You can’t filter or transform a set of readings unless you know when they were taken. Time noise is just as bad as amplitude noise.

Are you deceiving me? Over.
Detection errors can have very nonlinear consequences. And what is “three-and-fourpence” anyway?

Is there something that is particularly bugging you about analog system design?  Email me at [email protected] to let me know. I’ll see if there’s something buried deep in the compressed layers of experience that might help!

Oh, and don’t forget to [crackles].

Sponsored Recommendations


To join the conversation, and become an exclusive member of Electronic Design, create an account today!