On Tue, 04 Mar 2008 23:37:34 -0500 Michael A. Terrell
| email@example.com wrote:
|> YOU designed this equipment? I think not. Well, maybe one day you might
|> have been able to. But you clearly do not understand enough about math to
|> accomplish that today. You couldn't even come close on figuring out the
|> phase shift over time of 2 very close frequencies (2 Hz apart during the
|> time of 1 TV scan line in the NTSC system).
| A project that size is designed by a team, or it would never make it
| to market in time. Of course, no one expects you to know this simple
| fact of engineering. No one cares about you, or your anal retentive
I know _you_
don't care. But why is it _you_
need to keep making that
| As far as the chroma phase shift, I suppose you know EVERYTHING
| possible about Quartz crystals? If you did, you would know that they
| never lock, exactly. The further apart thery are, the more phase noise
| and chirp occurs. I'll bet you're going to go into great detail about
| them, but I've seen and used some you'll likely never see, like the 125
| MHz FUNDAMENTAL CUT crystals we used in some of our tuner modules. They
| were in gold based TO-5 cans, as well.
You seem to have some problem understanding locked vs. not locked.
If it is locked, it will have the same number of cycles as the source
in the long term. In that case, there is no accumulated phase shift.
The phase shift may jitter or wobble around. But it will go back the
other way just as much over that long term.
If it is NOT locked, then over the long term, there will be fewer or
more cycles, and the phase shift will accumulate.
If the local oscillator is NOT locked, your color will be distorted or
constantly changing. For example, if it gains or loses exactly one
cycle every field, you'll see the rainbow effect go completely full
circle from top to bottom of the screen. If it gains or loses that
one cycle every line (or more radical difference), that color will
vary full cycle from left to right. If it gains or loses exactly one
cycle every 10 seconds (rather close, but still not in sync), you'll
see the color cycle around slowly over those 10 seconds back to the
same as it was 10 seconds ago.
You better have it locked, or your color will suck.
If your crystal oscillator can't lock, then your color will suck.
If you have an oscillator that shifts in phase by N degrees leading
then shifts back to N degrees lagging, and back again, over some
period of time, but never accumulates or loses any cycles over the
long term, then it *IS*
locked. It may be locked poorly, but it is
locked. The quality can be measured by how far the phase shift goes
and the modulation of the phase shift.
| The burst is only availible for seven cycles per horizontal line.
| The early color burst circuts had to be set up by disabling the burst
| input, and trimming the oscillator as close as possible to 3.579545
| MHz. When the burst was turned back on, you had to fine tune it, to
| center the tint range to the center of the control's range. If this
| wasn't followed, the color could gothrough a full 360 degree change in
| one line, as the set drifted.
That is a case where it loses or gains a full cycle. That is NOT locked,
at least not to the proper subcarrier. If it happens like this over the
course of just one line, as you say, then it is locked TO A SIDEBAND of
the subcarrier, plus or minus the horizontal frequency from it. That
would be about 3563811 Hz or 3595280 Hz for NTSC. The crystal would have
to be quite a ways off frequency (at least half the horizontal frequency
from the subcarrier) to get locked out there. A good RC circuit could
do better than that. The reason to use a crystal is so that there is
very little phase noise across the line. Ideally, it will be very close
in phase when it gets to the next burst, preferrably within a degree or
two (any more than that and you'd see the phase drift as color shift).
Then the next burst can pull it back to being in phase, again.
How many degrees of phase drift do you think is OK?
| Phil Howard KA9WGN (ka9wgn.ham.org) / Do not send to the address below |
Click to see the full signature.