[time-nuts] Rcvrs/xmtrs that can slave to frequency standards

[Magnus Danielson]

From: John Ackermann N8UR <jra at febo.com>
Subject: Re: [time-nuts] Rcvrs/xmtrs that can slave to frequency standards
Date: Sun, 05 Feb 2006 17:21:50 -0500
Message-ID: <43E67A7E.6050505 at febo.com>

> Tim Shoppa said the following on 02/05/2006 05:05 PM:
> > I discover that I am in posession of a WJ-8716 which happens to have a
> > BNC on the back for an external 1MHz reference. I also have a Z3801A
> > that makes 10MHz. Is it as simple as turning the 10MHz PECL into TTL,
> > dividing by ten (e.g. 7490), and feeding this into the WJ-8716?
> 
> Hi Tim --
> 
> The thing you may need to watch for is whether the receiver is happy
> with a square wave input.  Depending on the design, it may need a clean
> sine wave.  My recollection may be wrong, but I think I once tried to
> feed a square wave into my HP 3586C selective level meter, which is
> essentially a receiver, and the results weren't pretty.
> 
> If that's the case, a divider followed by a low pass filter should do
> the trick.

I would consider a LC-tank tuned to 10 MHx not too high-Q to reduce temperature
dependence phase-variation. As an alternative could a crystal based filter be
used.

A 50% PWM gives the strongest fundamental, så divide-by-five followed by a
divide-by-two gives the best result there. Even if a square is tolerated, I
would recommend that.

As a twist, if you re-clock the output of the counter using a D-flip-flop which
is clocked with the original clock, then you can lower the jitter, as the
jitter will be only thru that flip-flopp. Spend the money on that flip-flop
being low-jitter and the rest just provide the correct logic.

Cheers,
Magnus