Policy question - drop low-precision clocks?

[Eric S. Raymond]

Hal Murray <hmurray at megapathdsl.net>:
> 
> esr at thyrsus.com said:
> > In the wake of the dumbclock removal, I'm now wondering if it makes any
> > sense at all to retain refclocks that report only in-band time to a
> > precision of a second.  That is, without 1PPS. 
> 
> I think you are on a wild goose chase.
> 
> You can get very accurate timing from a serial port without PPS when the text 
> doesn't have fractions of a second.  All the hardware has to do is send the 
> packet at a fixed time relative to the second tick.

Yes, and it is clear that (for example) the Spectracom driver does
exactly this.  It "receives 1PPS from sats and issues precise time
in-band" is the way I put it.

truetime is weird because some versions of the device get 1PPS from sats
but the driver doesn't use it.  Here's what it says:

/* This should be an atom clock but those are very hard to build.
 *
 * The PCL720 from P C Labs has an Intel 8253 lookalike, as well as a bunch
 * of TTL input and output pins, all brought out to the back panel.  If you
 * wire a PPS signal (such as the TTL PPS coming out of a GOES or other
 * Kinemetrics/Truetime clock) to the 8253's GATE0, and then also wire the
 * 8253's OUT0 to the PCL720's INPUT3.BIT0, then we can read CTR0 to get the
 * number of uSecs since the last PPS upward swing, mediated by reading OUT0
 * to find out if the counter has wrapped around (this happens if more than
 * 65535us (65ms) elapses between the PPS event and our being called.)
 */

So, the only way to get accurate time out of this box is to pass its 1PPS
output through an outboard conter chip.
-- 
		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>