At 10:42 AM 5/31/00 -0500, you wrote:
>Some colleagues here took on a liquid-level sensor based on ultrasonic
>ranging, and were suprised at what a major project it turned into.
>
>For starters, cost was an important issue.  It turns out that, like with
>many things, there are two kinds of ultrasonic transducers -- cheap ones
>that aren't very good and very expensive ones that are still only pretty
>good.
>
>After spending a lot of time dealing with analog noise, they got down to
>difficult problems with ringing in the transducer, multiple reflections, and
>the fact that all the transducer's properties varied a whole lot with
>temperature.  Then they got to work compensating for the fact that, while
>the speed of sound in air and water vapor is fairly well-behaved, the speed
>of sound in diesel fuel vapor is not.  It varies a lot, and in a highly
>nonlinear fashion, with temperature (and to a lesser extent, from one batch
>of fuel to another).
>

For a slightly different viewpoint, I consulted with a company that
used std Polaroid ultrasonic transducers to determine the level of
ice in an icemaker machine. They had 10s of 1000s of unitsin the
field. This device is quite easy to interface, and fairly cheap,
and used just a couple of chips running off a 6805 uC, which provided
control and took measurements.

However, the transducers always had a heavy coating of water
condensate on them, so I was always amazed they didn't have
horrendous reliability problems. Eventually, they gave them up for
a simpler/cheaper solution involving a thermistor in a tube.

best regards,
- Dan Michaels
==============