[sci.electronics] Police radar and part per billion measurement.

fjs@nntp-server.caltech.edu (Fernando J. Selman) (02/18/91)

Running the risk of antagonizing the people following
this discussion I have to explain my claim to parts per
billion accuracy. Let us consider a different example, not
to different, but one that will explain my point more clearly:

Consider a Michelson interferometer where one of the arms
can change length (e.g. the passage of a gravitational wave).
The two waves are made to interfere via a beam splitter and then
projected into a detector. At the detector we observe the interference
fringes. We can measure theire displacement with a microscope and a
ruler. The measuring device can be good at most to a few microns
over a range of say a millimeter, but nevertheless it can detect length
changes of one arm with respect to the other of the order of a wavelength
of the light used in the interferometer. If the arm length is 500 mts and
the light used has a wavelength of 500 nm, then the accuracy of this device
can be characterized as 1 part per billion. The important point here is that
the measuring device is the interferometer itself, NOT the microscope or ruler
used to measure fringe displacement! Replace microscope and ruler by
frequency counter, laser by local oscilator, beam splitter by circulator,
and you have quite a good analogy. The measuring device is not the
frequency counter or the rest of the circuitry, but in a way the
whole radar unit. The counter is just used to extract the information,
as the microscope and ruler are used in the interferometer.
	I hope this clarify my position.

					- Fernando

P.S. This also show that even though we can nominally characterize
     the accuracy to a part per billion, there are so many factors
     that can affect measurements that before claiming that accuracy
     many tests would have to be conducted (e.g. the standard arm might
     change length, the beamsplitter could introduce systematics, 
     vibrations introduce lots of noise, etc). Most of these problems
     have been addressed and  the technology is way pass the one part
     per billion, by orders of magnitude.