At 04:44 AM 10/5/2004 -0400, you wrote:
>I've been doing some research for a project (that'll probably never happen)
>where the main constraint is to measure a time interval with nanosecond
>(preferably picosecond) resolution and repeatability. Obviously we're not
>talking about timer0 here :o)
>
>I've found some exceptionally high speed counters ( OnSemi mc10E016 -
>better than 1 GHz) That could reach 1 nanosecond, but finding a GHz clock
>precise enough adds another challenge (<20 PPB). Even if I had a suitable
>clock, calibrating it adds more challenges.
>
>There's also the option to measure it with a R/C time constant. Let a known
>capacitor charge (or discharge) through a known resistance for an unknown
>time and measure the resultant voltage. This option adds some complications
>of its own. I doubt that I could find (or measure) an adequately precise
>capacitance to derive the time completely mathematically. But I would
>expect that high quality components in a controlled environment should be
>able to achieve that level of repeatability - then it's just a (not so
>trivial) matter of comparing it to other instruments and deriving
>calibration data.

Maybe there's a way of comparing it to a reference time that's
conveniently long, perhaps in the hundreds of nanoseconds. Then you're
looking at relatively low resolution in the analog bit and the
errors cancel out to the accuracy of your clock. I was looking at this
a couple of years ago, and got to the point of doing some spice simulations
of my core design elements using a fast current source and good device
models, which confirmed the validity of the approach. The linearity was
quite good.

My friend John Larkin at Highland Technologies in San Francisco has a
company that does this sort of thing for a business. It's pretty
specialized, and the delay boards are $thousands, but I know that they
work. Just the equipment to design at picosecond speeds is pretty
pricey (even on eBay, many $thousands). If it's a one-off, you may be
better off just buying a VME board or whatever, and adding value in the
software rather than re-inventing the hardware. Just a thought.

>kind of capacitors have the best characteristics (ageing - tempco, etc),
>20+ bit A/D converters (about 10 samples per minute, max)?

Capacitors change a percent or more over temperature, depending on
the dielectric, but that's a good type. You're going to need to
compensate for short-term and long term capacitance variations and
for current source variations, output impedance, and of course
dielectric absorption.

Best regards,

Spehro Pefhany --"it's the network..."            "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com




_______________________________________________
http://www.piclist.com
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist