Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site unccvax.UUCP
Path: utzoo!watmath!clyde!bonnie!akgua!whuxlm!harpo!decvax!mcnc!unccvax!dsi
From: dsi@unccvax.UUCP (Dataspan Inc)
Newsgroups: net.video,net.analog,net.micro.pc
Subject: Re: Composite=>RBG conversion (reposting of query)
Message-ID: <176@unccvax.UUCP>
Date: Mon, 1-Apr-85 09:28:24 EST
Article-I.D.: unccvax.176
Posted: Mon Apr  1 09:28:24 1985
Date-Received: Thu, 4-Apr-85 06:23:18 EST
References: <191@neurad.UUCP>
Organization: UNC-Charlotte
Lines: 48
Xref: watmath net.video:1084 net.analog:251 net.micro.pc:3636


    This is not terribly easy, particularly if you want to get the benefits
of I/Q axis wide bandwidth demodulation....
    What you need is:
       1) A chrominance/luminance separator, which could be either
          a comb filter (preferred) or simple bandpass filter from 3.08
          to 4.08 mc;
       2) A subcarrier regenerator to provide a 3.579545 mc sine wave 
          locked in phase to the colour burst signal in the composite
          waveform;
       3) Quadrature synchronous detection of the chromanance, to recover
          the original I and Q sidebands;
       4) Output filtering (DC-1.6 mc I, DC - 600 kc Q or to taste) as
          well as a fixed delay to match I to Q channel because of the
          lower bandpass of the Q channel;
       5) Basis transformation matrix to convert I and Q into R, G, B
 
     Note that you could perform demodulation among any arbitrary axes, 
provided that the basis transformation matrix was correct. However, if one
selects R-Y, B-Y; that narrow bandwidth is only possible (both channels 
limited to 500 kc/s. I think this is due to quadrature distortion caused
by VSB transmission of the I-channel and subsequent truncation of the
nonsymmetrical component contaminating the Q-channel. Another way to say
this is that high frequency crosstalk is minimized with I-Q axis demodulation.

     Also, you must add back in the Y (monochrome) component unless your
monitor has a separate Y-input (which is linear, etc.)

     There are devices which exist to do this (commercially black boxes)
but the ones which recover the full I-channel bandwidth are VERY VERY 
expensive. The typical mid-line video store that caters to the industrial
and basement-Eyewitness News market probably has one. 

     If you had to build one, get the data sheet for the RCA CA 3217E. This
is narrow band but has potential; i.e. a quoted bandwidth of 900 kc for
the chroma demodulator and the basis transformation matrix built right in.
It also doesn't use flaky, hard to wind coils; just the Dale IR-2 types.
Be sure to ground pin 16 through a 470 ohm resistor if you like accurate
colour reproduction (the production receivers leave it open, for orangy
fleshtones!) 

     One question: why go through the effort for your colour under VCR?
Unless you have a type "C" or quadruplex in your living room, I think you
will be ** VERY ** disappointed in the spurious junk brought to light
from the video output jack.

David Anthony
DataSpan, Inc.