Hi to Larry Smith from DEC-world.  I'm a hardware type myself, but I
can write just enough "self-defense" software (i.e., hardware
diagnostics) to show you software guys where the bug _really_ lies!

But, like Jody Kravitz, I wouldn't be surprised if you bear soldering
iron scars as witness to hardware experiences.  So we're even!   And
like lots of enigeersxxx engineers (sheesh!) I like to dabble in
design in related engineering fields.  Mechanical musical instrument
design is therefore most satisfying.

I narrowed your music box problem to the task of creating a new
pinned cylinder for a worthy old music box, and talked to a few
mechaniker friends this evening.  Got a good education, too.

#1.  Al Frankel, retired manufacturing engineer at Hughes Aerospace,
but he served a full apprenticeship as tool & die maker before
earning the engineering degree.  He built the first magnetic drum
memories at Hughes, and now serves as the gear design guru for the
Dusenberg auto restoration crowd.  Al said to be wary of
thermoplastics for your task:  they continue to have dimension
problems after fabrication because the material goes through a phase
change.  He told me about "instantaneous temperature coefficient of
expansion" driving them nuts:  the material goes through a glass
[molten] transition point.  Most epoxies exhibit this behavior, and
thus it is tricky dealing with two different kinds, in the same
"weldment", for example.  Al recommends using a NC (numerical
control) machine to drill the metal cylinder, then insert previously
sheared pins by hand and do a finish grind to size.

#2.  Karl Petersen, designer & manufacturing engineer in Boise ID,
will hopefully join this discussion group this weekend with his
views.  Karl and I both enjoy funky Citroen cars and own late Ampico
B pianos: his is a Chickering, mine is a Weber.  (Yeah, Robin Pratt,
I hear ya -- it's probably a Knabe design.)

#3.  My brother, John Rhodes, designer and manufacturing engineer at
Hewlett-Packard, just sent me this e-mail note:

> ... Modern NC equipment can place holes to 0.5
> thousandths of an inch.  Indexing the cylinder can be with an
> auxiliary rotary table which is also NC controlled.  I assume
> the pins are ground to length after being placed in the
> cylinder.  Dig out the tolerances required; I can get data on
> the machines.

> Seems to me the cylinders take about 30 seconds/revolution for a
> 50mm cylinder.   I assume the notes need to be plucked with an
> accuracy of about 10 msec.  That'll tell us something about the
> placement accuracy.

Hokay, Larry, put on your mechanical hat:  what are representative
dimensions and the machine tolerances to consider?

Related subject:  I want to make a precision test "roll" for
calibrating a music roll transcribing machine.  The test pattern is
only a few feet of high-density "music", but should use Mylar instead
of paper or vellum for stability.  I could make it in a jig borer/
vertical mill using an end mill cutter, and running under NC control
from a computer, of course.  Big question:  where in the dickins can
I get the format specification for one of the standard NC disk files
(e.g., AEGIS)?  I sure don't want to make a drawing of a music roll
by hand!

-- Robbie Rhodes < rhodes@foxtail.com >  4 August 1995