Hi all,  I used a different approach when transcribing tunes from the
barrel of the Clay's organ clock; I think it could be useful for someone
(and perhaps for Spencer Chase also) to know how I realized transcrip-
tion.  The system consists of a mechanical part, an electronic part,
and software.

The mechanical part is a wooden frame on which the barrel is mounted.
The original key frame is mounted on a bar having lateral sweeping
capability (the barrel contains ten tunes), arranged in a position
(relative to the barrel) similar to the original one.  On the same bar,
at the edge of the keys, there are sensors, consisting of simple
microswitches.

The electronic part is a specially developed card, containing buffers for
the sensors, and an up-down counter; it is connected to the serial port
of a Macintosh computer through a serial-to-parallel interface.  The
counter of the card is driven by an optical encoder, fixed on the rim
of the barrel.

The software manages the whole, operating as follows.  The counter is
reset at beginning of the action, manually, by a push-button, then
reading begins.  The program recursively reads the state of the sensors;
if one or more changed their state, then the program records data in the
memory, along with the position, and completes recording when the switch
goes in an off-state.

The position of data items (the pins, in this case) is determined by the
optical shaft encoder; this is a device that is similar to the roller/
encoder of Spencer's scanner; it appears as a potentiometer, and is
capable of a certain number of ticks per turn.  I used a low cost encoder
with 256 ticks per turn, but there are encoders with up to 5000 t.p.t.
(at proportionally higher cost).

The use of the encoder offers many advantages; since it is a position
sensor, it is independent of the speed, so the barrel can be moved with
any variable speed, although it is preferable to maintain a convenient
low speed.  I move the barrel by hand; the system can be improved by
providing a conventional or stepping motor.  You can even stop or move
backward -- the counter will follow counting backward, and still the
position is exactly determined.

The resolution of the encoder is very high: the circumference of the
shaft is about 25 mm., comprising a rubber ring for friction improvement,
so the resolution is 25/256 = 0.1 mm. (The version with 5000 t.p.t. can
achieve 0.005 mm.)  When reading finishes, then the data, stored in
memory in an internal format, are converted to a standard MIDI file, that
is written on the hard disk, and can be utilized by one of the common
music programs, such as Cubase or Finale and so on.

I think that this system could be used to scan whatever kind of medium,
by just providing adequate sensors and frames.  I'm studying a system
based on a laser light beam to scan punched supports (it's just an idea
for now, I'm going to experiment).

I don't know if this subject has been discussed on MMD before Andy La
Torre's optical fiber proposal.  (I read MMD since 10 January, and I have
not read previous MMD issues yet, because of lacking time).  As Spencer,
I am interested in knowing new ideas, so that a fast and efficient
scanning system can be developed.

Regards

Leonardo Perretti¶
perretti@mbox.vol.it

 [ That's a fine system you describe -- everything is commanded by the
 [ pulses from the measuring roller.  Our member David Wasson built a
 [ perforator in which the driving capstan is also the measuring roller,
 [ and which transmits pulses to the Midi Clock port on the computer.
 [ His perforator, therefore, has the same command method as your
 [ barrel transcriber.  -- Robbie