Synchronicity & Wurlitzer: a challenge for engineers

For the past couple of months I have been designing a system to copy
Wurlitzer style 165 rolls (and incidentally APP, Caliola, and other
Wurlitzer rolls) that would replace the service once provided by
Play-Rite Music Rolls, Inc.

Good as available recut rolls have been up to now, they do contain
slight errors in perforation which can be visually observed in the
uneven onset and termination of chords and in the ragged appearance
of the perforations in a tremolo.  Due to the forgiving nature of band
organ music, these errors do not come through in the music to the
extent they would if the instrument were a piano rather than an organ,
I think.  But errors in the length of percussion-instrument perfora-
tions do cause audible flaws in the form of a weak strike or total
omission of musical accent.

Discussions with Dave Saul, Wayne Stahnke, and others have convinced
me that a perforator control system that doesn't achieve synchronicity
between source -- whether a paper roll or a computer disk -- and the
perforator which the source drives is bound to introduce perforating
errors of the kind just described.

Synchronicity.  My original but naive understanding of synchronicity
in perforating was that if the speed of the paper moving under the
perforator punches exactly matched the speed of the roll being copied
-- so that a yard of original roll produced a yard of finished copy,
no more and no less -- then you were doing synchronous perforating.
If this were the definition of synchronicity, then Play-Rite's per-
forators operate synchronously.

But true synchronicity, as I now understand it, requires that the per-
forator cycling (stop - raise punches - advance paper - PUNCH - stop)
match exactly the cycling of the original perforator, which cycling
is coded in the form of discrete punches and advances in the original
roll.

I have played a bit with some web files provided by Wayne, using his
VIEW editing program, and have seen from the roll scans included with
the software the amazing ingenuity with which old-time roll makers
established the size of the punch-advance of their perforators, such
that the number of advances in a measure could accommodate notes of any
duration without compromising the precision required for proper attack
and note cut-off.  Wayne's web files faithfully mirror each punch and
advance as found in the original roll; therefore a perforator which
is truly synchronously driven using such files will make an exact and
error-free copy of the original roll.

Therefore, I intend to incorporate both Wayne's software (VIEW, and its
companion perforator-driving software, PUNCH)  and true synchronicity
into the system.  Now my problem and question -->

The perforators used by Wurlitzer were designed with a tempo compen-
sation feature built into their operation.  Because any roll playing
in a roll frame where the paper is drawn across the tracker bar by the
rotation of a take-up spool necessarily moves across the tracker bar
at an constantly increasing speed due to paper build-up on the take-up
spool and a consequent increase in the spool's circumference, Wurlitzer
found it necessary to correct for this acceleration.  The acceleration
factor couldn't be ignored in Wurlitzer's long ten-tune rolls, although
it could be -- and apparently was -- ignored in the production of piano
rolls.

Wurlitzer's tempo compensation feature operated to microscopically
increase the length of the paper advance through the perforator with
each perforator "stop-advance-PUNCH" cycle.  (In a few weeks I plan on
posting to MMD some images showing the design of this mechanism.)  The
result of this feature is that in Wurlitzer rolls, while the number of
punches and advances in a measure of music remains constant, the linear
distance in a measure is constantly increasing by a very slight amount,
not detectable by measuring two adjacent bars of music but detectable
if you compare the first measure of a tune with the last measure in the
same tune.

That's the problem.  My question is: how does this affect synchronicity
when a scan of such a roll is run under VIEW and PUNCH to operate a
synchronous perforating system?

Matthew Caulfield