[time-nuts] World's most accurate PC clock!

Magnus Danielson cfmd at bredband.net
Mon Jul 4 07:31:35 EDT 2005

```From: "Poul-Henning Kamp" <phk at phk.freebsd.dk>
Subject: Re: [time-nuts] World's most accurate PC clock!
Date: Mon, 04 Jul 2005 09:49:14 +0200
Message-ID: <28486.1120463354 at phk.freebsd.dk>

> In message <20050703.152133.118722707.cfmd at bredband.net>, Magnus Danielson writ
> es:
>
> >> days and generate the 14.318MHz by PLL instead.
> >
> >Which should be an approximation to 14,31818 MHz I would assume, since that is
> > 2
> >2 *5*715909 Hz, but an approximation to that would be quite acceptable
> >considering it is usually +/- 100 ppm (for real TV uses it should be
> >+/- 3 ppm). Ever measured what is actually there in PLLed cases? (curiosity)
>
> The correct frequency is 315/88 MHz, but that is not the original derivation
> of it.  I found the real derivation on the web some time back, but now
> I can't seem to find it again.

The authorative reference is the ITU-R BT.470-6. In table 2, table item 2.11
you find

for NTSC:     for PAL (B, D, G, H, I, K, N) for PAL (M)  for PAL (M combi)

455            1135    1                    909           917    1
f   = --- f   f   = (---- + ---)f           f   = ---f   f   = (--- + ---)f
SC    2   H   SC     4     625  H           SC    4  H   SC     4    625  H

where f  is (in table 1)
H

15750000
f  = -------- f  = 15625                    f  = 15625   f  = 15625
H     1001    H                             H            H

I'm not covering SECAM here thought...

Notice that Note 1 in table 1 needs to be read in order to read the correct
value. The ratio-notation is not used in BT.470, but is used in other NTSC
related documents.

> >One should figure out which pin is the input to the CMOS inverter usually used
> >in such oscillators. This is where the signal shall be fed in.
>
> Not all of them use a cmos inverter any more, so the data sheet should be
> consulted.

Note taken!

Cheers,
Magnus

```