[time-nuts] questions on uncompensated crystal oscillators

[SAIDJACK at aol.com]

In a message dated 7/5/2006 13:00:41 Pacific Daylight Time,  
cfmd at bredband.net writes:

Yes, it  is hard.

PAL is 4433618.75 +/- 5 Hz or +/- 1.12 ppm.
NTSC is 3579545  +/- 10 Hz or +/- 2.79 ppm.

That is straight out of ITU-R Rec.  BT.470-6.

The MPEG transport stream is clocked at 27000000 +/- 810 Hz  or +/- 30 ppm.

That is straight out of ITU-T Rec.  H.222.0

:P

You know, you are really demorilizing me now. Stop  it! :P



Hi Magnus,
 
yup, that +-810Hz is the number I had in mind on the receiver side. The  
tighter <3ppm numbers you mentioned are for the broadcaster side I  think.
 
Although it sounds like a huge deviation from what we are  used to discuss 
here (0.001ppb versus 30ppm) it presents a very challenging  engineering problem 
since the factories don't want to pay more than about $0.25  for this 27MHz 
oscillator.
 
Most TV's can achieve lock with larger errors (I've seen up  to +-3KHz in 
some cases) and interestingly enough, the better the TV, the  smalller the 
lock-range. Sony professional Studio Monitors cut back on the range  to improve 
picture quality, and won't do much more than the 810Hz...
 
Don't be demoralized, the fix is easy - feed a good 27MHz into your AV  
equipment (preferrably from a GPS Disciplined frequency reference. You could use  
our FireFox synthesizer for example - but that would be total overkill :)
 
Also, to reduce the problem to a skipped-frame/repeated-frame issue you can  
use non NTSC/PAL baseband standards such as Component-out or HDMI out  etc.
 
bye,
Said