Xref: utzoo rec.video:6889 sci.electronics:6734
Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!apple!ames!xanth!mcnc!unccvax!dya
From: dya@unccvax.UUCP (York David Anthony @ WKTD, Wilmington, NC)
Newsgroups: rec.video,sci.electronics
Subject: VTR colour recording (and how does SECAM do it?)
Message-ID: <1533@unccvax.UUCP>
Date: 26 Jun 89 22:38:04 GMT
References: <124@nisca.ircc.ohio-state.edu> <566@axis.fr> <367@ctycal.UUCP>
Organization: Univ. of NC at Charlotte, Charlotte, NC
Lines: 107

In article <367@ctycal.UUCP>, ingoldsb@ctycal.COM (Terry Ingoldsby) writes:
> In article <566@axis.fr>, philip@axis.fr (Philip Peake) writes:

> First of all, unless I'm very mistaken ordinary VCR's do not separate the
> chrominance and luminance info when they store it on the tape.  It is all
> stored together.  Somebody please correct me if I'm wrong about this

	Wronnnnggggg :-) :-)

	This point is not often covered here, but basically, you can't
record baseband video (with encoded chrominance) on a helical scan
video tape recorder.  The entire reason is a **mechanical** one, and
is basically the fact that you can't obtain a constant linear scanning
velocity where the rotating headwheel's head tip contacts the tape.

	Basically, the headwheel (and the head tip) suffers all kinds
of accelerations.  These are due to tape pack variations in tension
within a reel and as a function of reel position; power supply
variations; "stiction" (humidity, fingerprints, emulsion chemistry);
dirt build up, the list goes on and on.  Your VTR is doing quite
well to maintain a fully framed image.  Your camcorder is doing an
even more remarkable job, due to gyro errors (because you move the
camcorder, the headwheel speed to a fixed reference is the same, but
relative to the tape, it changed).

	NTSC and PAL depend on a constant phase sampling clock
to encode and extract their colour information. That sampling clock
is contained as a colour burst signal; the 8-10 cycles of 3.58
(or 4.43) mHz sinewave right before the start of the line. The
presumption is, basically, that the colour information following
that burst was encoded with a constant phase clock identical to
the signal which made the burst.

	However, velocity errors will cause frequency modulation of
any signals on the tape, and therein lies the problem.

	Soooo, any FM which occurs as a result of footsteps across
the floor, tape pack irregularities, wind velocity, humidity, and
not the least of which mechanical wackies from the headwheel servo
system itself (which is barely fighting to keep your picture framed
vertically...burst phase coherence, you must be joking !!!!!) will
cause the relationship between the burst (sampling clock) and the
colour (the data) to be lost.  A hue stable image probably has less
than 6 nS p-p of residual burst to chrominance jitter, if that.

	Your VCR, on the other hand, doesn't have phase jitter, it
has gross frequency jitter.  We're talking about microseconds gained
and lost during recording and playback!

	The way 'round this is recording colour in a different
band than the luminance on the same tape.  There are several
different implementations of this, they are all called
"colour under" recording.

	Luminance and chrominance are separated. The luminance
goes on its merry way to be recorded as a narrowband FM track
at about 4 mHz with a 1 mHz deviation. (Mr. Video will no doubt
correct me as to specifications).  Chroma, on the other hand, is
processed in such a way that it is **heterodyned down** below
the luminance track.  The signal which does the heterodyning could
come from two places:

	1) be harmonically related to the colour burst
	2) be any old reasonably coherent CW signal of
	  desirable frequency (chosen to minimize artifacts
          with the original signal)

	I'm not sure what consumer decks do; in fact, I'm not 
sure what even our type "C" machine does!!!  In either case, the
party of the first part (the "heterodyned" signal) and a replica
of the party of the second part (the "heterodyning" signal) are
both laid down for a total of three video signals on the tape.

	Now, velocity errors still occur with impunity, but they
proportionally affect the heterodyne signal and the heterodyned
signal in such a way that when everything is played back, there
is no difference in frequency or phase between the two! The 
receiver sees a burst (directly converted or otherwise reconstituted)
and a colour subcarrier which are properly related, even though
their absolute values are changing violently from line to line!
To the **naked eye** correct colour playback is obtained.

	There really isn't a better inexpensive way for recording
baseband colour signals on MAGNETIC TAPE (<--laserdisc flamers,
take note).  Now, though, we have component video formats, which
record basically Y-I-Q (or Y-U-V), where the chroma is broken down
into two signals whose vector component span the colour space.
We have even digital videotape recording.

	But, the mainstay is still colour under. These signals can
even be broadcast, with appropriate massaging using a time base
corrector (basically, a big fat FIFO which is trying to be kept
half full by prompting the VTR headwheel to "slow down" or 
"speed up" its filling, and whose output is being clocked by 
precise, stable signals).  

	A question: how is SECAM recorded on a VHS or Beta machine
of the appropriate scanning system?  It would seem that the two
FM subcarriers used could be directly recorded if there were an adequeate
C/N ratio on the tape; or they could be filtered out, digitally
divided down, recorded under, and digitially multiplied back.
Do SECAM consumer decks specifically enjoy the advantages of that
system?


York David Anthony
DataSpan, Inc