Part One

Lately there has been discussion in MMD about ways to pneumatically read
music rolls with sufficient precision to compete with the accuracy of
optical scanning, for making accurate soft-file or MIDI records of old
rolls.  Robbie suggested that strain-gauge transducers (electronic
devices which measure force without much movement) could be used as the
interface between the pneumatic system and the computer.  This could be
done in the following manner.

The basic problem:  Any player system consists of two orifices in
series; the hole in the paper over the tracker port, and the bleed, with
the space between these orifices being the pouch-well.  When the port is
covered, the flow through both is very small (only what leaks past the
paper), but most of the resistance to flow, and therefore most of the
pressure loss, occurs at the tracker bar. Therefore the pressure in the
pouch-well is very near the full stack vacuum.

Conversely, when the hole in the roll uncovers the tracker port, this
opening is large compared with the hole in the bleed.  Although the
total flow through both orifices now increases dramatically, most of the
resistance to flow, and the loss of pressure in that stream, now occurs
at the smaller bleed opening.  The pressure between the two openings
(the pouch-well) now rises dramatically.

The pouch, and the valve attached to it, 'measures' this pressure
change, and responds to it by inflating, and playing the note.  The
problem is, this process involves changing the volume of the system, and
moving objects (the pouch and whatever is connected to it) which have
mass, friction or varying pressures acting on them.  Time delays and
hysteresis are inevitable.  Since all these objects are hand made, and
thus vary from note to note in mass and dimension, errors which are not
constant, and therefore cannot be allowed for in the computer, are
introduced.  Thus any mechanical switching based on ordinary pouch-
action may not be able to compete with optical reading, to replicate
the hole-for-hole resolution (about 1/50-second) of reproducing rolls.

My idea is to dispense with the pouch altogether, and measure the
pressure between the tracker bar and the bleed electronically, using
miniature strain gage pressure transducers.  The only inherent time
dependent phenomenon would then be the volume change in the system as
the pressure in it changed.  The volume change in the detector is
negligible since the 1/4" diameter 'loadcell chip' deflects only
micro-inches during measurement.  The volume change in the system and
its connecting tube is small (1.03 times its total volume between
atmospheric and 24 inches vacuum) and that volume can be made small and
constant by making the tracker bar tubes short, and all the same length.

The following setup could be made quite compactly, using four such bars
drilled on 1/2" centers, as shown below:

                            MINIATURE STRAIN-GAUGE
                   ___      PRESSURE TRANSDUCER (1),
SHORT             |___|<----TO COMPUTER SCANNER (3)
TUBE           ____| |________
FROM          |    | |    _   |<----RAIL DRILLED
TRACKER_______|____| |_ _| |  |     AS SHOWN
BAR-->________ ________0_ <======D<-----BLEED AND
         _____|__________| |__|______   AJUSTING
        |    ____________| |_____    |  SCREW (2)
        |   |                    |   |
        |   |<--WINDCHEST OR---->|   |
        |   |  VACUUM CHANNEL    |   |


Per Robbie's request, to avoid MMD overload, more technical details
about this setup will follow tomorrow in Part Two.

Richard Vance