S&S D Duplex

[Sarah Fox]

Hi Don and list,

OK, serious suggestions from someone who did her dissertation on acoustical
type stuff -- not an amateur to be ignored on this one... (Besides, I would
like to think of technical exchanges being a two-way street.  I wouldn't ask
for information if I didn't feel I had some to trade.)

> I wish someone would take two nearly identical pianos. Tune the duplex on
> one by lowering the pitch as you recommend. On the other merely lower the
> pitch and then *not* move the duplex. Then compare before and after times
> on both instruments, as well as real time analyser data.

I would appreciate this as well, and I will probably do it myself, once I
have time and am able.  I will add that in order to avoid bias on the part
of the experimenter/technician, this experiment *must* be done completely
quantitatively.

GENERAL:  I would highly recommend using an SPL meter (or better, a
microphone connected to an oscilloscope.  If available, the experiment would
be improved by feeding the microphone signal through a bandpass filter (or
even a graphic equalizer with the gain on the relevant channel turned up and
the gain on the other channels turned down.  This measure would eliminate
noise.  If at all possible, eliminate the lower frequency noises, since they
are most likely to mess up your measurements.   The experiment would best be
done on the lower-pitch end of the duplex range, since dwell time would be
greater.  Also, it should be obvious that this should be done with
damping/muting of all notes except the one note being measured.  I would
suggest doing the measurements on the *same* piano on several test notes (1)
with the duplex in tune and muted, (2) with the duplex in tune and unmuted,
(3) with the duplex out of tune and muted, and (4) witht he duplex out of
tune and unmuted. Tuning and detuning can be accomplished from the tuning
pin without moving the aliquot plate/bar.  Detuning should probably be by
1/4 step.

WITH SPL METER:  Use A weighting if you have it.  If not A, then C.  Flat,
band limited, would be next.  Flat unlimited would be worst.  The reason?
Better attenuation of extranious noises.  The A scale mimics human hearing,
BTW.  As for the response time, fast (125 ms halftime) is better than slow
(1000 or more ms halftime, depending on manufacturer and adherance to ANSI
standards).  If you've effectively filtered your noise, you should have a
steady reading and will not need a slower setting anyway.  Finally, I
suspect you'll need an old-fashioned meter with a needle movement, rather
than a digital display.  Although a B&K or even a GenRad would be great, the
humble, garden-variety $35 Radio Shack model should be fine.  These
measurements are relative and therefore will not require the exacting
accuracy of a B&K.

Set the meter to as high a dB SPL range as possible so that a note can be
played mf/f, but not with obvious distortion, just pegging the meter's
needle.  Watch as the needle falls from full scale to some arbitrary
reference point.  The +4 dB mark would be good.  As the needle sweeps past
the +4 dB mark, start a stopwatch.  As the needle then sweeps past the -2 dB
mark, stop the stopwatch and record the time.  This would be the 6 dB decay
time, which is the half-time with respect to the amplitude of the note. You
could also do a 3 dB decay, which would be the half-time with respect to
power (amplitude^2).  I don't think anyone would throw stones at either
measurement, but the 6 dB measurement would be easier to measure with some
degree of accuracy, IMO.

WITH AN OSCILLOSCOPE:  Run the microphone through a bandpass, preamp, and
into the scope.  AC couple.  Trigger however you wish.  If possible,
calibrate the oscilloscope readings with an SPL meter.  Obviously the only
reading that will be of beenfit to you on an O-scope would be peak-to-peak
amplitude in the waveform.  Use a slow sweep so as to get a nice line of
peaks.  Do your measurements as described above.  A half-amplitude decay
will correspond to 6 dB (e.g. from 6 divisions to 3).

SCIENTIFIC PERSNICKITINESS:  Repeat the measure several times and take an
average.  Then repeat the measurement in a lower dB SPL / amplitude range,
playing a softer note.  I suspect the half-time would come out longer.
Here's where you get to do something scientists have to do:  come up with an
operational definition.  You decide how to define your decay time for the
sake of your study, and you stick with that definition.  You must decide how
loudly a note is to be played and precisely which dB limits define your
decay limits.  These should be the same numbers applied to all notes
measured.

HELP/COLLABORATION:  Yes, I'm willing.  I'm in Columbus, OH and frequently
travel to Newport News, VA.  If there are any RPTs in these areas or
inbetween, I'll do the measurements if you do the piano.

ACKNOWLEDGEMENTS: ... if you publish, yes, I would appreciate being
acknowledged

Peace,
Sarah