[RFI] New chipset vs. EMI in notebook computers.

Morse, Earl (E.A.) emorse at ford.com
Wed Jul 21 16:29:39 EDT 2004

This technology is old hat.  Almost every personal computer used a clock generator with the Spread Spectrum or "dithering" enabled when I was working on them 18 months ago.  CISPR uses receivers with 120 kHz bandwidth at the frequencies where this technique is effective. (above 30 MHz) The higher the harmonic the more it is spread.  It isn't too effective on 2nd or 3rd harmonics but at higher harmonics it can make a huge difference.

The only place we couldn't use it was on analog video because the dithering would make the lines of text swim on the screen.

In a nutshell,
Very effective on higher harmonic frequencies.
Don't expect much relief on emissions below 30 MHz where CISPR bandwidths are only 9 kHz.

If you truly want to see the effect of the dithering take a look at your modern PC and find out which clock generator you have ICS, IMI, Cypress etc.  Get the data sheet off the web and get their utility for flipping the register bits on the chip.  You will be able to toggle the dither off and on via the register and see the difference.

BTW, as manufacturers move closer to one chip solutions the clock generators are being incorporated into the main chipsets.  


>Message: 2
>Date: Tue, 20 Jul 2004 18:47:38 -0700
>From: "Kevin Rowett" <n6rce at arrl.net>
>Subject: RE: [RFI] New chipset vs. EMI in notebook computers.
>To: <rfi at contesting.com>
>Message-ID: <00c401c46ec4$b3ebffd0$b600000a at N6RCELaptop>
>Content-Type: text/plain;	charset="US-ASCII"
>The cypress part is from a company they bought a few years ago.  The
>part is a crystal OSC that outputs a "Spread Spectrum" clock, rather than
>a CW clock.
>The idea is the clock frequency moves around just a bit - not too fast, and
>not too much.  Not enough to create timing problems inside ICs, but enough
>that no narrowband CW signals show up on the PCB or chassis. 
>This is all targetted at passing CISPR 22 and FCC class B Part 15 emissions
>for unintentional radiators.  The net effect of the wandering clock is
>the total power in the passband of the test receiver.
>As for the units being less noisey into the radio sitting on the same table
>- YMMV.
>de -KR- N6RCE

More information about the RFI mailing list