Perhaps one of my more silly Ebay purchases of late. I wasn't sure what I
wanted to do with a Rubidium frequency standard but it looked like an
interesting gadget so I stumped up the £50 or so and awaited delivery.
From the LPRO user manual, The LPRO makes use of the atomic resonance property of rubidium
(87Rb) to control the frequency of an unheated quartz crystal
oscillator via a frequency-locked loop.
In other words, this gadget uses the radioactive breakdown of
rubidium87 to generate a 10MHz sine wave. Once the device
has been switched on and warmed up, the accuracy of the output waveform is
nominally ±5 x 10-8 Hz or 5Hz in 1GHz.
That's quite accurate :-)
What does a computer fellow do with a 10MHz sine-wave?
In recent times, there seems to have become some competition in certain geek
circles to see who can build the most accurate clock - somewhat akin to the
schoolyard conversations of the 1980s:
"My watch is set by the talking clock."
"Well, mine is set by Big Ben."
"Shame... Mine is set by the Rugby signal."
etc. etc.
In order to have a clock to compete with my equally childish friends, the
plan is to build an interface board for the LPRO to convert the 10MHz to 1Hz, or
1 pulse per second (PPS). This PPS signal can be used by the Linux NTP Daemon to
discipline a time source, such as the computer's internal clock (or the time
signal from GPS) to produce a nanosecond-accurate clock.
Hardware
For clarity, the interface is divided into two logical sections (both built
on the same board) - one handling the power to and diagnostics from the LPRO
unit and one handling the conversion of the 10MHz signal to PPS.
LPRO Unit Pinouts
Pin
Function
Pin
Function
1
RF Out - 10MHz Sine
6
BITE
2
Chassis Ground
7
Ext. C-Field
3
DC Isol. RF Rtn
8
+24V Rtn
4
Chassis Ground
9
Xtal V Mon
5
Lamp Voltage
10
+24V
Ext C-Field is for calibration to an external, more accurate time
standard and is not used by this interface.
Power & Diagnostics
As well as taking power, ground and signal connections, the LPRO unit provides
several status outputs, which can be monitored for debugging purposes. Opamps in
this section of the interface compare the diagnostic outputs of the LPRO to
fixed voltages from potentiometers.
BITE or "Built In Test Equipment" provides information as to
whether the crystal oscillator is locked to the atomic transition. Once the BITE
output is low, (less than 4.2V) lock has been acquired and accuracy should be
within ±5 x 10-8.
The BITE LED on this
interface will light when lock is acquired.
Lamp Voltage is monitored for inherent degredation of the
photodiode. If Lamp Voltage drops below 3VDC, the LAMP V LED will light,
indicating that maintainence is required.
Xtal V Mon is used to indicate if the crystal is drifting out of
the available trim range once the unit has warmed up. If this output falls
outside the range of 0.55 to 12.6VDC, the XTAL V MON LED will light, indicating
that service is required.
Signal Processing
74HC390 "Dual Decade Ripple Counter" chips each contain two "divide-by-two" and
two "divide-by-five" sections. Using four such devices, it is possible to
"divide-by-one-million", producing a 1Hz TTL signal.
A MAX232 chip is used to convert the TTL signal to RS232 level for input to the
serial port of a PC.
A 7805 voltage regulator is included so that a single 24VDC supply can be used
to provide power to both sections of the interface and also to the LPRO unit.
The novely component in the circuit is the LED, which indicates the PPS. An
oscilloscope should also be used to verify signal correctness!