Recently, I was wondering exactly how much signal propagation delay
there is in the tubing of a pneumatically played instrument.  I set up
the following experiment.

I made an electrically operated pneumatic valve out of a four-inch
loudspeaker roughly as described by Craig Brougher.  The speaker opens
a 1/8-inch hole into a 1/8-inch I.D. brass tube.

I initially connected a hot wire air flow detector directly to the
valve to measure the delay between application of power to the speaker
and the subsequent detection of air flow by the sensor.  The output
signal from the hot wire circuit as viewed on a digital oscilloscope
is a curved ramp.

It rises slowly for about the first 20 percent, then rises quickly to
about 80 percent, and finally rises at a decreasing rate until full
output.  The 80 percent of full flow point was somewhat arbitrarily
selected as being a suitable reference point for the time measurements.
The directly connected delay was about 2.5 milliseconds.

Then I inserted additional brass tubes of one foot, two feet, and
three feet and measured the additional delay.  The output signal with
the extra tubing in place has a progressively slower rate of increase.
This makes the estimate of time delay less accurate for the longer
tubes.

The measurements increase almost exactly one millisecond per foot,
indicating a propagation velocity of 1000 feet per second.  This is
about 88 percent of the speed of sound at 20 degrees C. (from my
Handbook of Chemistry and Physics).  These measurements were made with
a vacuum of about four inches water gage.

Richard Tonnesen
Richardson, TX

 [ The test pulse at the transmitting end of the tubing is a square
 [ wave, approximately, but the wave shape becomes rounded as more
 [ pipe is added.  This suggests that the sound wave in the thin
 [ pipe experiences phase distortion similar to the distortion in
 [ electric pulses transmitted through long cables.
 [
 [ The hot filament of a tiny "grain of wheat" lamp is cooled by the
 [ air flow; the flow rate is detected as the increase in the power
 [ needed to keep the filament at the same temperature.  Richard
 [ estimates that the sensor responds in less than one millisecond.
 [ See his article, "Hot Wire Airflow Sensor for Pneumatic Reader" at
 [ http://mail.foxtail.com:89/Archives/Digests/199901/1999.01.30.18.html
 [
 [ -- Robbie