Spread-spectrum clocks

Thread Starter

sbkenn

Joined Apr 15, 2017
23
Just read a tech brief on a new clock generator which states "spread spectrum to minimise EMI". IMO, these don't actually reduce EMI, they just fool the measurement systems into indicating that the UUT has lower EMI, and bit like car manufacturers fooling emissions testing. Comments please.
 

nsaspook

Joined Aug 27, 2009
8,770
It's not fooling (that implies at least one party is being fooled, all parties here know exactly what's happening) anything but it might be a clever cheat for passing the buck. Spread Spectrum does minimize the coherent signal at the frequencies of testing. Interference is governed by the detectable level (spectral density) of the undesired frequencies in the desired range. If the specs say X and you meet that spec, it's working. That said, excessive EMI is a sign of inefficient design so it should be reduced at the source.
https://www.eetimes.com/author.asp?section_id=75&doc_id=1285083#
 
Last edited:

Lo_volt

Joined Apr 3, 2014
229
Good article. Explains Spread Spectrum well.

I agree, it's a cheat. You'll still have instantaneous spikes at frequencies within the dither range that may violate EMI emission criteria. Measured over time these won't appear as large.
 

Thread Starter

sbkenn

Joined Apr 15, 2017
23
It's not fooling (that implies at least one party is being fooled, all parties here know exactly what's happening) anything but it might be a clever cheat for passing the buck. Spread Spectrum does minimize the coherent signal at the frequencies of testing. Interference is governed by the detectable level (spectral density) of the undesired frequencies in the desired range. If the specs say X and you meet that spec, it's working. That said, excessive EMI is a sign of inefficient design so it should be reduced at the source.
https://www.eetimes.com/author.asp?section_id=75&doc_id=1285083#
The intent of the standards is to prevent one device from interfering with another. If a particular narrow-band frequency is measured while the device is working at a slightly different frequency, you will detect nothing. If the device is spread-spectrum, you will only measure a small part of the radiated (or conducted) power, because only a small proportion of that power is in the particular frequency being measured. If a broadband analyser is used, it may well catch the total power. Otherwise, the standards are being .... circumvented.
 

crutschow

Joined Mar 14, 2008
27,902
Depends upon how you look at it
If the device being interfered with is sensitive to only a narrow range of frequency than spread-spectrum of the interfering source should help.
If the device being interfered with is wideband than a spread-spectrum source may not help.
 
Top