Laurent Coray was asking about the aperture effect and the sensitivity
of player valves (I suppose) to the opening and closing of holes in a
roll. I'm sure he is going to get some first-hand information from folks
more knowledgeable than I about this subject, but I wanted to add this
info to what might also be received.
When you are talking about vacuum valves versus the sensitivity of optic
readers and the like, there is a "hysteresis effect" in each. It's just
that the vacuum valves' hysteresis is a bit greater. For example, vacuum
valves take more pouch power to actuate than they do to remain actuated
by a factor of about 5. That means, they are much slower to return than
to actuate, even without much of a load. But when you load them with a
large pneumatic, like a pedal operator, that larger bellows acts like a
counter source for vacuum which keeps the poppet bouyed up at half-mast
in its well, preventing the stack source from "sucking" it down against
the inside seat again as quickly as it actuated. This is in addition to
the valve's own natural reluctance to return.
(Larger bellows also close more slowly as well (with a given intensity
and valve), requiring either the timing of the roll perf to lead, or the
trackerbar hole to be adjusted accordingly for simultaneous actuation.)
So to answer your question, Laurent, as far as vacuum valves are concerned,
you can probably figure that a valve will actuate cleanly at half a diameter
and drop out again at about a third or fourth a diameter-- and they vary all
over the place! But the speed of the roll enters in, too.
The faster the roll is going, the cleaner your "on/off" points are going
to be because of better resolution during play (but that doesn't apply to
reading or copying. The slower a roll is copied, the better the resolution,
in general. It is why a finely graduated stepper that also opens and
closes the reading circuits with each incremental movement of the roll
is a necessity).
As an aside, Simplex used a trapezoidal trackerbar hole which purposely
overlapped its neighbors (or at least removed the space between holes
completely and then the mistracking roll did the overlapping). The idea
was to improve the natural tracking of rolls without having to add a
tracking mechanism. You'd think perhaps that would play the notes to
each side of every hole, but it won't because of the hysteresis effect
One of the pneumatic readers (maybe someone has the answer to this) I
heard of had two trackerbar rows of holes spaced with shim brass or
something, but very close together. The idea was to use the first row
to unlatch the triggering of the punch and the second adjacent row "read"
and trigger the punch. The first row then could prevent overpunching by
sensing the end of a perforation a diameter or so before it appeared over
the second row. Such a mechanism will also have the capability of sensing
chain bridging and duplicating it.
When holes are read electronically, you have the same problem, by the
way. If you have made no provision to sense a perf stop, then you will
always extend certain note and expression perforations beyond their
original length. This may be done somewhat by giving yourself a
sensitivity adjustment for each reader, but I'd think it would be better
done with a double row reader.
My concern with these readers is this: Ampico rolls used 3 to 1 masters
out of necessity: The reason is that the very high speed of the Ampico
mechanism required that finely adjusted expression coding be reproduced
precisely. If you try to copy an Ampico roll on the existing readers
today and reperforate it from an ordinary roll, it sounds muddy in some
places, simply due to errors made by not stepping the roll and reader
through together at the same rate as the original. So you are wise in
asking about this and wanting to get it right the first time! I hope you
will get many replies.
[ Editor's Note:
[ I'm hoping that Wayne Stahnke might write to us in more detail about
[ his roll reader and the techniques he uses to "reconstruct the master
[ Both Wayne's optical scanner and the one I helped Mike Ames build use
[ CCD arrays and "see" an individual hole as __MANY__ pixles. As a result,
[ it is possible to compute (simulate) the exposed area of a tracker bar hole
[ and the associated pneumatic behavior. Wayne's system captures so many
[ pixles that it is possible for him to see the scalloping of the edges
[ of the holes and thus learn additional information about what was going
[ on with the perforator. Even without seeing the scalloping it is possible
[ to reconstruct the master for some of the perforating systems.
[ Optical systems which are based on one photosensor per hole position
[ must find some other way to reproduce the behavior of the tracker bar
[ and pneumatic, or accept the inaccuracies, which may include playing
[ chain perfs at staccato notes, or unnecessarily elongating notes.