[time-nuts] GPS disciplined oscillators - how not to do it.

Dr Bruce Griffiths bruce.griffiths at xtra.co.nz
Tue Apr 3 21:48:53 EDT 2007


An Australian Electronics magazine recently published a circuit for a 
GPS disciplined crystal oscillator.
This particular implementation is the worst I've ever seen. A partial 
schematic of a frequency divider chain they used is attached.
Spot the design errors.
They even went as far as to design their own crystal oven with a 
bang-bang temperature controller.
The VCXO used a logic gate as the amplifier and a varicap was added to 
the circuit so that the oscillator frequency could be adjusted.
A counter was used to measure the phase error between the GPS (Garmin 
GPS15L) receiver PPS output and a 50kHz frequency divided down from the 
reference.
The phase error count was then used as the input to an 8 bit DAC built 
using a handful of resistors and a CMOS octal flipflop. The DAC output 
was then connected to the varicap after minimal analog filtering. If one 
used a sufficiently long averaging time the fractional frequency error 
was claimed to be better than 1E-7.
Meanwhile a PIC was used to communicate with the GPS receiver and 
control a display.
Since one can obtain similar or better performance by using the GPS 
receiver PPS output directly to calibrate a counter timebase, why bother 
with such an inept design?

Bruce


-------------- next part --------------
A non-text attachment was scrubbed...
Name: GPS-FreqRef2divider.jpg
Type: image/jpeg
Size: 44206 bytes
Desc: not available
Url : http://www.febo.com/pipermail/time-nuts/attachments/20070404/23347a02/attachment-0001.jpg 


More information about the time-nuts mailing list