I have been following this thread with great interest and have not
responded sooner as I wanted to see what others had to day on the
subject.  Tempo compensation of perforated music rolls is something
I have had to deal with for the last 27 years.  I sincerely believe
I have made just about every mistake that can be made on the subject,
but we all learn from our mistakes.

When mechanical music machines started to be operated by perforated
paper rolls, there was really no need to concern oneself with Tempo
variations.  Why?  Well, most of the instruments were operated by
endless rolls and Tempo compensation was not needed.  Witness the fact
that the Link Piano Company used endless rolls until the end of their
production.

When roll frames with take-up spools were initiated Wurlitzer, and
maybe others, had a unique solution.  A small metal roller rested
against the take-up spool and as the diameter of the spool increased,
the small roller, which was linked to the friction drive of the roll
frame, compensated for paper buildup.  Problem solved again.  Wurlitzer
would later use perforators that Tempo compensated literally every turn
of the machine's crankshaft.  Others took a different route.

Punching multi tune rolls with lengths of 170 to 230 feet in length
mandated some form of compensation for the increase of speed of the
music sheet over the tracker bar.  It is my belief that Clark, Capitol,
U. S. Music and the coin-op division of QRS all used the same length
master rolls.  The change in length of the completed roll came from
altering the paper advance on the perforating equipment.

Paper advance on original perforating machines I have studied consisted
of a ratchet moved by a pawl connected in some manner with the rotating
crankshaft.  Things have not changed much since except that now stepper
motors have replaced the mechanical linkage between the tractor feed
which incrementally advances the paper through the punch head.  I must
add here that I am speaking of a perforator with a full compliment of
punches and not the single punch machines that travel back and forth
across the paper.  They must also advance the paper when a _row_ is
completed and I assume that is taken care of by the software that
operates the punch.  I have no experience with such machines.

Wayne Stahnke ascertained years ago that coin operated nickelodeon rolls
of ten tunes or greater length changed the incremental advance, called
the step rate, between tunes three and four and six and seven.  Each
roll used three step rates.  If this is correct, and time has proven it
is, then the original perforators had some method for the operator to
change the step rate at those times in the production of a roll.

Stahnke expressed all of this in ratios, e.g., 25:25 basic step rate,
25:24 intermediate step rate and 25:23 final step rate.  I have used
this system for years with great success and comparison of newly punched
rolls with the originals bears this out.  The trick is to determine the
basic beginning step rate.

Naturally every roll manufacturer used a different beginning rate.
Clark used the most coarse rate of about 240 punches per foot.
Capitol's basic rate was 256 punches per foot.  I have not done enough
originally QRS produced coin piano rolls to determine just what they
used.

Using Wayne's formula, if we start with 256 punches per foot, the next
step rate changes to 245.76 punches per foot and the final step rate
would be 235.92 punches per foot.  I suppressed mathematics beginning in
the third grade so apologies if those calculations contain errors.

Wayne's software suite, which he wrote to control the PD-1 perforator
control board built by Octet Systems and later Ron Perry, dictates
that two separate files be used to produce a roll.  He names them "web"
files for the musical notes and "ann" files which tell the machine just
what to do with those note files.

In preparing files for punching an End of Selection (EOS) mark is
placed in the web file when you want the machine to change to a new
step rate.  When the machine encounters the EOS mark it reads the next
step rate specification in the ann file.  You really don't notice when
it happens.  Any step rate can be specified and now with super accurate
optical scanning those step rates can be ascertained to four decimal
places -- a bit of over-kill, in my estimation.

When I acquired the perforator built by Richard Tonnesen, it was a
whole new ball game.  There was no way to change the step rate -- the
machine chomped along at a steady 540 punches per foot.  The reasons
for such a high rate would take far too much bandwidth to explain, but
that was how it was.

A pesky feature of the machine, built in 1979, was the fact that once
a roll was started the machine could not be stopped without starting
all over again.  I cannot estimate how many pounds of punched partial
rolls I sent to the land fill.  I finally got smart and spooled those
mistakes and used them for dunnage in the packages I mailed out.

Then Rick Crandall commissioned me to make composite rolls for his
Coinola SO.  At nine holes to the inch, the rolls had to be produced
on the Tonnesen machine.  Tempo compensation became a whole new ball
game.  At this point enter Richard Brandle's exceptional "Wind"
editing program.  An extremely versatile program designed for the 
perforation of music rolls, "Wind" contains all the tools you need
to manipulate files to suit your needs.

In the case of the style "O" rolls I simply changed the Tempo of
each musical selection depending on where it fell in the tune lineup.
I started with Tempo 70 for the first selection, changed tune #2 to
71, #3 to 72 and so forth.  By the time I got to selection #10, the
roll was playing at Tempo 80.  In effect, I had _stretched_ the roll
longer, thus compensating for paper build up and paper speed.

A few test rolls winged their way between Rick and myself as I had no
"O" roll machine to test my efforts on.  Rick sent me video recordings
of his magnificent Coinola SO playing the rolls produced.  It worked
and production proceeded.

Three days ago I punched a Capitol "A" roll that changed step rates
no less than seven times over a ten-selection roll.  The final length
was 232 feet.  Why did this happen with the original?  Dunno -- you'd
have to ask the folks that produced the roll.  I don't believe they
are talking anymore!

My announcement of retirement was premature.  The 9-to-the-inch machine
has found a new home and the 6-to-the-inch will take its location.

Regards from South Texas,

Ed Gaida - Still preserving music by punching holes in paper.
San Antonio, Texas
edgaida@sbcglobal.net.geentroep [delete ".geentroep" to reply]