[time-nuts] Interfacing a 8dBm sine output of an OCXO toadigital logic standard
cfmd at bredband.net
Fri Sep 16 20:29:39 EDT 2005
From: "Bill Hawkins" <bill at iaxs.net>
Subject: RE: [time-nuts] Interfacing a 8dBm sine output of an OCXO toadigital logic standard
Date: Fri, 16 Sep 2005 18:42:36 -0500
Message-ID: <006b01c5bb18$50e325c0$0500a8c0 at darius.domain.actdsltmp>
> Magnus Danielson wrote, in part,
> "I wonder if that chip isn't really a DLL after all. They usually are."
> When my brain parses DLL I get data linked library.
> Did you mean phase locked loop?
No, I mean Delay Locked Loop.
A Delay Locked Loop is a quite different animal than what we usually perceive
as a PLL, but have similar behaviours. Wonderfull for CMOS implementation, but
the drawback is the jitter.
> How do you measure jitter? What instrument?
I use a high BW undersampling scope with high resolution time base, such as a
Tek CSA 803, CSA 8000 or Agilent 86100. You need to use a power-splitter to
get a trigger signal (for all three you need a separate trigger) and the other
to the sampling head but only after being delayed through a cable to compensate
for the trigger delay in the scope as they are not compensated within, that is
only done in real-time sampling scopes which just samples and store before the
trigger kicks in. The delay-cable needs to be longer than the delay-time, so
that one clearly can view the trigger point. This is necessary in order to
establish the trigger jitter. Tek made a box called DL-11 which was a delay-
line intended for this purpose, but I use a standard cable cut into a suitable
delay. I can use the reading from the next rising edge directly, but I should
really compensate it by taking the root of the squared value minus the squared
trigger noise to get a more accurate measure.
There are other scopes that can be used too.
Notice that I use scopes to get the cycle-to-cycle jitter measure, which is not
the same as a normal phase-noise analysis. You can take a phase-noise plot and
convert it over, but it is so easy to measure it with a scope you have standing
around in the lab anyway, so that is what I choose to use. Wouln't mine a real
phase-noise system, but it really isn't needed.
More information about the time-nuts