Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!usc!wuarchive!zaphod.mps.ohio-state.edu!rpi!batcomputer!gaarder
From: gaarder@batcomputer.tn.cornell.edu (Steve Gaarder)
Newsgroups: sci.electronics
Subject: Re: video chroma circuit
Message-ID: <1991Jan17.192816.23881@batcomputer.tn.cornell.edu>
Date: 17 Jan 91 19:28:16 GMT
References: <767@ssc.UUCP> <14501@milton.u.washington.edu> <38104@cup.portal.com>
Distribution: usa
Organization: Cornell Theory Center
Lines: 30

In article <38104@cup.portal.com> ISW@cup.portal.com (Isaac S Wingfield) writes:
>In article <767@ssc.UUCP> markz@ssc.UUCP (Mark Zenier) writes:
>>Does anyone know the reason they picked 3.579545... Mhz for the

>Most of the energy in a (B&W) television image lies at frequencies which are
>multiples of the horizontal rate (15,734.26 for NTSC color). The chroma signal
>carrier (actually a suppressed carrier) is at an odd multiple of half the 
>horizontal rate; this causes the chroma sidebands to interdigitate with the
>luma information and produce minimal interference with it.
>
A good way to think of this is that the subcarrier produces a very fine
pattern of alternating black and white dots on the screen, which toggle
between black and white with every frame.

>Another part of the
>compromise, and for which I can't remember the reason, involved changing
>the horizontal and vertical rates from the B&W values (15,750Hz & 60Hz)
>to slightly different ones (15,734.26Hz & 59.94Hz). This may also have to
>do with minimizing crosstalk with the sound carrier.

Actually, it was done to make deriving those frequencies from the 3.579545
master easier.  A the time, frequency division was done by syncronizing 
an oscillator, which was fine for small divisors but impractical for large
ones.  Thus a division ratio was chosen that could be factored completely
into small numbers. (Divide the burst by 13, 5, and 7, then double it, to
get the horizontal frequency)

Steve Gaarder
gaarder@theory.tn.cornell.edu
...!cornell!batcomputer!gaarder