[time-nuts] Symmetricom CSAC is Cs acting like a Rb unit

Brooke Clarke brooke at pacific.net
Wed Jan 19 15:17:58 UTC 2011


Here's a response I got from Symmetricom:

· The SA.45s actually uses cesium, not rubidium. But it is a vapor-cell 
clock just like our Rb clocks.
· The Tx and Rx pins are for an RS-232 interface. Using the interface, 
the clock can be programmed, and the clock’s functions can be monitored. 
I have attached a User’s Guide that describes all the commands in detail.
· The SA.45s does not attempt to correct the 1 PPS input, nor does it 
use a sawtooth to correct itself to the 1 PPS input. Instead, it can be 
commanded (see User’s Guide) to discipline itself to the 1 PPS input. 
How fast it does this is a function of user-selected bandwidth inputs.

In the manual (CSACUsersGuide rev 004.pdf) they talk about the 
evaluation board and it's associated software. The main screen shot 
shows the following parameters for the physics package:
Laser Current: 0.86 ma
Heater Power 11.01 mW
DC Signal 0.996 V
TCXO Tuning 1.518 V
Contrast 4415 (no units)

A block diagram shows a 10 MHz TCXO driving a 4596 MHz Synthesizer which 
drives the physics package.
The error signal from the physics package drives the loop filter which 
tunes the TCXO.

The BITE output is a lock indicator.

I get the feeling that a vapor cell is not an absolute standard but 
rather works like a Rb vapor cell, i.e. the frequency has some drift and 
the starting frequency depends on how the cell was made.

The 1 PPS output can be synchronized (using the 1 PPS input) to the 
nearest 10 MHz zero crossing (i.e. to within 50 ns). With a continuously 
applied 1 PPS input after some time it will adjust both the frequency 
and 1 PPS output to about 5E-13.

"5.7.2 The "Art" of Disciplining
Implemented correctly, disciplining can be utilized to calibrate the 
CSAC frequency in the field, even if a reference source is only 
occasionally or sporadically available, thereby improving the long-term 
performance (phase and frequency drift) of the CSAC. Alternatively, the 
disciplined CSAC may be used to "clean-up" the short-term stability of 
an accurate, but noisy,, reference source, such as GPS.

Implemented incorrectly, however, disciplining may degrade the 
performance of the CSAC if, for example, the CSAC is disciplined with a 
short time constant to a source which is itself noisier than the CSAC, 
such as GPS.

Implementing a successful discipling strategy involves understanding the 
noise properties of the CSAC, the reference source, and the phase meter 
itself, and selecting the appropriate time constant that makes the best 
use of the available timing information."

Fig 16 is an Allan plot showing: CSAC, Phase Meter noise floor, Lab 
grade Cesium standard & GPS
At 1 second:
CSAC is 1E-10, Phase Meter is 4E-10, GPS is 2E-8, lab Cs is 1E-11
GPS gets better than the CSAC at about 5,000 seconds
The Phase meter gets better than the CSAC in about 20 seconds

So, for example, when using GPS to disipline the CSAC the time constant 
should be set to >= 5,000 seconds. Or when using a lab grade Cs standard 
the time constant should be set to >= 20 seconds.

R. Lutwak et. al., "The Chip-Scale atomic Clock - Low-Power Physics 
Package", Proceedings of the 36th Annual PTTI Dec 7-9, 2004

R. Lutwak, et. al., "The MAC - A Minature Atomic Clock", Proceedings of 
the 2005 joint IEE Freq Ctrl Sym & PTTI

Have Fun,

Brooke Clarke

More information about the time-nuts mailing list