Electronic Design

Bob's Mailbox

Dear Bob:
I realize that you work in the analog realm, but you may have heard of the latest fad in digital circuit design—spread-spectrum clock generators. ( No, I hadn't. Wild! /rap) These are phase-locked-loop circuits that generate a clock signal that's frequency-modulated with a 100-kHz or so triangle wave. This trick is employed to transform the clock EMI spike seen on a spectrum analyzer into a flat-topped mesa shape, so that the peak EMI from the device appears to be 5- or 10-dB lower and the device will gain FCC approval. The ads for these clock generators typically claim that they "reduce EMI." (Yeah, sure! /rap)

Now, you and I both know that frequency modulation does not reduce signal amplitude. It has no discernable effect in the time domain. The clock signal will have the same wave shape and size as before when it's viewed on an oscilloscope or detected with an RF power meter.

This trick is only being used because the swept-frequency spectrum analyzer employed by EMI testing labs is a poor indicator of actual signal amplitude. This machine consists of a radio receiver whose receive frequency is swept through the frequency range of interest. The EMI signal from a test antenna is passed through the spectrum analyzer's IF amplifier and video detector. These circuits have low enough bandwidths that they don't "see" a 1-MHz-wide FM signal at its full amplitude. What's the result? Reduced EMI! (Or, "APPARENTLY reduced" EMI! /rap)

Now back to the real world. The practical EMI-reduction effect of modulating the clock frequency is negligible on a TV receiver "victim," as it has a 4-MHz-wide IF strip. And the resultant herringbone interference pattern will just get a bit wiggly if it's frequency-modulated. (Have you tried this? I haven't. /rap)

Because the amplitude of the interfering spread-spectrum clock signal is permitted to be higher than a single-frequency clock and still pass FCC muster, it's more likely to annoy the user. That's when the problems begin.

Sooner or later, the folks at the FCC will catch on. They will see more complaints of interference caused by approved devices with spread-spectrum clock generators, and change the rules so higher-fidelity EMI sensing equipment will be used. At that time, the spread-spectrum clock circuits will become poison. (Check. Ha! Cheaters do not ALWAYS get caught, but they should think about the rule-benders who will get caught! /rap) Is there any way to put a stop to this silliness before it takes over the world of PCs and set-top boxes?
David Forbes
via e-mail

Probably not. But thank you for griping!—RAP

Hi Bob:
I just finished reading Electronic Design and the thought struck me with some force: there must be a lot of analog design engineers out there who are faced with either becoming programmers (yuck), or moving their design skills to silicon. (This may be PARTLY true, but there are a LOT of people who can integrate a SYSTEM using purchased parts. For sure, that's a MAJOR part of the electronics business—often WITH a processor, but sometimes WITHOUT one. /rap)

I find that analog people have a keen desire to touch the reality of the stuff, whereas programmers are somehow fascinated by other areas, such as their own cleverness. (That's one way to put it. /rap) My point is that for a guy who loves analog, the decision to go software is probably fundamentally unacceptable. Opening a fish-and-chips shop would be more satisfying.

So, the action moves from circuit boards to chips. Obviously you made that jump some time ago. But, the subject that many of us might like to read about is how we should go about it ourselves. (Wow. There was still a need 25 years ago for small but tricky circuits, where good straightforward engineering gave you a pretty good chance to get an IC working on the second try. These days, most ICs are more complicated, involving perhaps hundreds or thousands of transistors, and two, three, or four levels of CAD—Spice; Layout; Logic; Timing. /rap)

A lot of us have large amounts of applications experience (mine is audio, for example) and could probably be an asset in the right structure. (I do know several "discrete" engineers who went into applications engineering. We who have bought and used ICs for YEARS have a lot of the necessary EMPATHY for other people who buy chips and design with chips. Not so many have gone into IC design. It's possible to learn in school some of the skills needed to design ICs. But, it's NOT so easy to learn the skills required to APPLY them. /rap) What do you think?
Dave Mate
via e-mail

I think that's a hell of a GOOD and TOUGH question, and I'll ask some guys. Give me a reminder every WEEK for three weeks, and we'll see if I get any answers.—RAP

Subject: Bob Pease
Electronic Design has a great asset in Bob. Long may his pen trail the ink.
Bernard Sharnbrook
via e-mail

Keep writing? I'll keep on trying. In fact, Bob Milne sometimes says I'm VERY TRYING! But, does my pen "trail ink?" Maybe it's more likely that my floppies keep dribbling out the bytes.—RAP

All for now. / Comments invited!
RAP / Robert A. Pease / Engineer
[email protected]–or:

Mail Stop D2597A
National Semiconductor
P.O. Box 58090
Santa Clara, CA 95052-8090

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.