[ Editor's note:
 [
 [ Interest in transcribing music rolls to computer files seems to
 [ be growing now, as several people are developing the hardware to
 [ play reproducing pianos from disk files.  In this letter Wayne
 [ Stahnke shares his thoughts about pneumatic interfaces, including
 [ a novel design of his for an air-operated switch.
 [
 [ I hope that other MMDers will tell us of their experiences.
 [
 [ -- Robbie

I thought I would try to contribute to the ongoing discussion regarding
pneumatic reading insofar as I can.  Most of the correspondence has
been private, among interested individuals (including me), but it seems
to me that we should properly discuss these matters in the MMD, where
everyone can participate.  This might help relieve the total dearth of
technical discussion in the MMD these past few months.

Pneumatic readers can be very good.  Toward the end of the time that
I worked with them I was able to reconstruct master rolls from the
readers disk files, which gives an idea of the accuracy that is possible.
Following are a few points that may be of interest, in no particular
order:

  (1) When I first started out to build a pneumatic reader in the early
1970s, I wanted to use gold-contact micro-switches.  It turned out not
to be possible to do so for purely logistical reasons, so I settled for
silver contacts.  In the process I learned that gold is required for
what are called "dry" circuits (i.e., circuits with low voltages, in
the range of 12 volts or less) because gold does not form oxides in the
atmosphere (at least in theory).

I remember learning that the oxide layer that forms on silver is so
thin that it can be punctured by a voltage as low as about 24 volts,
and that using a lower voltage is asking for trouble.  If I were doing
it today, I would use about 48 volts for reliability.

  (2) Contrary to prevailing opinion, a modest amount of hysteresis is
a desirable thing.  It increases the switching speed and provides
sharply defined trigger points for both on and off.  Note that a single
"inside" or "outside" valve (an Ampico primary valve is an example of
the latter type) has inherent hysteresis, so if primary valves are used
between the tracker bar and the pouches the hysteresis is determined by
the primary valves, not by the pouch board itself.

  (3) Contact bounce is a real problem, and must be dealt with.  Do not
rule out using optical switches, with a small "flag" attached to a
pouch (or to the stem of a primary valve) to interrupt the light path.
Such switches are available at surplus outlets at very affordable
prices.

  (4) Toward the end of my work with pneumatic readers, I designed and
built a new pneumatic-electric transducer that consists of a cylinder
of Teflon-filled Delrin, about 4 mm in diameter and about the same
length, that fits closely within a glass tube with an outside diameter
of 5 mm and about 30 mm long.

Delrin is very light and easily machined, and the Teflon filling
provides lubricity.  (Precision glass tubes are available and should
be used in a production design, but if you want to build one to try
the idea, simply use the glass envelope of a standard "3AG" electrical
fuse, and make the cylinder of wood.)

The tube is placed between the arms of an optical switch, positioned
such that when the cylinder moves within the glass tube it interrupts
the optical path.  The total displacement of the cylinder is con-
strained by keepers in both ends of the tube to be very small, around
1 mm or less.

On testing this device I discovered that it would repeat very quickly;
I remember that the transit time was on the order of a few milli-
seconds.  Consider using this, rather than pouches, for a new design.

(5) For the transducer described above, the output of the phototransis-
tor was connected to the input of a Schmitt trigger device (actually
a 74C14) that provided both buffering and hysteresis.  The output
transition times were on the order of 5-10 nanoseconds, and the
hysteresis in the Schmitt trigger provided well-defined trigger points.

A single 5-volt power supply provided current for the LED, the neces-
sary pull-up voltage for the phototransistor, and power for the Schmitt
trigger device.  The whole thing was truly wonderful, and I am sorry
that I never built a complete reader using it.  Instead, I changed my
approach to optical scanning, using a CCD array.

I hope this note is of use.  With best regards,

Wayne Stahnke