Picking up from what Bob Billings said yesterday.

When copying rolls, it isn't strictly necessary for the copy to match
the original step distance between rows.  It is clearly desirable in
general, but may not even be a good idea if the size of the punches is
significantly different, as the width of the bridges needs to be kept
reasonable (both upper and lower limits).

A general-purpose perforator will inevitably have to deal with a variety
of punch sizes and advances.  If the copy step distance is different,
accurate performance can be obtained by adjusting the copy roll speed by
the difference in the step distance between original and copy.

Obviously, the copy roll will not be the same length as the original, so
cannot be laid over it for comparison, but it is still an exact copy of
the punch pattern and should perform identically - except for the speed
pickup throughout the performance.

Recent 45-to-the-inch copies (0.022" steps) were mentioned on 8 Feb 97 by
Craig Brougher, and he felt that even at this degree of accuracy accents
in Ampico rolls were being missed from time to time.  The system needs
very accurate placement of control codes.  I am pretty sure the same
would be true of Duo-Art theme perforations.  You need a lot more accu-
racy than might be expected from empirical 'how short a distance can I
detect' arguments.

In general, having two cyclic operations with different frequencies will
always produce an interference beat pattern.  In roll copying this
manifests itself by the copy roll periodically inserting or losing a
complete row, jumping from being in advance to being behind the original
(or vice versa) -- this is on top of individual reading errors.  It comes
over as an intangibly rougher performance rather than a specific
mistake.

Such periodicity can never be completely avoided, no matter how well the
original roll's advance has been matched.  That's why making hole-for-
hole masters is so important.  Once you have that master, you don't need
an exact match for the original advance rate when making the copy.  In
essence, you synchronise the two cycles and so eliminate the beat
pattern.

The point I was trying to make is that if the technology to make hole-
for-hole masters can be made widely available, all copy rolls will be of
perfect musical accuracy and cost exactly the same as current rolls.  If
this is possible, why persist in pursuing a second-best solution?

In reality, pneumatic direct copying or low-tech light sensor techniques
are easy to grasp and lots of us could have a reasonable go at making
such a copier -- I guess that's why virtually all discussions about roll
copying end back there, even though it's second best.

Recognising a hole pattern and aligning it to a fixed advance step seems
to require computer optical processing techniques which are totally new
ground for most.  It's in making the jump to this new technology that we
need the boost from all you experts out there!

Julian Dyer