Picoseconds Or PPM

or a synthesizer (see Figure 2 for an example of how a
synthesizer works). The same crystal oscillator feeds
another synthesizer that puts out 10MHz. Alternatively
the crystal oscillator can put out 10MHz directly and a
more complex synthesizer/multiplier feeds the rubidium
gas cell.
You may have been told that rubidium standards neces-
sarily deliver 10MHz. That’s not quite the case. When
you buy one of those tin cans from a lab equipment sup-
plier it puts out 10MHz but that’s only because laborato-
ries have standardized on that number. After all,
6834682610.904324Hz isn’t a very useful frequency. So,
off-the-shelf lab units synthesize 10MHz.
cêÉèìÉåÅó
jìäíáéäáÉê
mii
póåíÜÉëáòÉê
mÜçíç
aÉíÉÅíçê
jáÅêç
mêçÅÉëëçê
sçäí~ÖÉ
`çåíêçääÉÇ
`êóëí~ä
lëÅáää~íçê
oìÄáÇáìã
d~ë=`Éää
oìÄáÇáìã
i~ãé
~åíÉåå~
NMjeò
Figure 1: Block diagramme of a typical rubidium frequency
standard
Unfortunately, a 10MHz frequency standard doesn’t
improve things. None of the usual audio rates are simply
related to 10MHz. It would have been infinitely more
sensible to derive audio rates, or multiples, straight
from the rubidium gas cell instead of generating 10MHz
first. When a frequency standard sold by an audio com-
pany to the audio market puts out 10MHz this is simply
because it’s a rebadged OEM rubidium standard. This
requires yet another synthesizer to get to a standard
audio rate:
mÜ~ëÉ
`çãé~ê~íçê
ÑêÉèK
ÇáîáÇÉê
LPNOR
NMjeò
ÇáîáÇÉê
LPUQM
çê=LPROU
LORS
LNOU
LSQ
s`EuFl
QQKN
UUKO
NTSKQ
QU
VS
NVO
Figure 2: Audio word clock generator with 10MHz reference
input. This whole structure is called a PLL synthesizer. Some-
thing similar is used in the output stage of Figure 1.
Because of the fact that the two frequencies are not sim-
ply related there is a lot of scope for intermodulation.
The greatest common denominator between 44.1kHz
and 10MHz for instance is 100Hz. It is exceedingly diffi-
cult to make a synthesizer with such a ratio that doesn’t
have significant jitter products at 100Hz intervals. If the
aim of the undertaking is to minimize jitter, bringing in
a 10MHz laboratory standard is a surefire way of mak-
ing things much more complicated and much more ex-
pensive.
Do atomic clocks have low jit-
ter?
The crystal oscillator used to probe the rubidium cell is
top rate. It has to be because if it had significant jitter
the photo cell wouldn’t get a stable reading. So if such an
oscillator were used alone, without the rubidium and
without an extra synthesizer to get to 10MHz, jitter
would be superb. That’s not the case and the output
spectrum is rarely clean. It’s only centered at a very pre-
cise frequency, that’s all. That’s their only purpose.
I like my CC1. Will attaching an
atomic standard make it better?
No. The lowest jitter clock you can ever get is a really
good crystal oscillator that puts out the frequency you
want (or an integer multiple) directly. This is what the
CC1 does. Besides low jitter, the CC1’s great strength is
the slave mode. The CC1’s jitter performance does not
change when locked to a jittery source.
I run a broadcast facility so I
have an atomic standard to sync
the house. Will a CC1 improve
the quality of the clock?
Absolutely. That is what the CC1 is designed to do: to
generate a low jitter clock, either stand-alone or syn-
chronized to an external source.