I am experimenting with some ideas about making a single copy piano
roll perforator for home use. People will probably encourage me to
complete the designs and then have no interest when it comes to
spending a few dollars on the software.

Due to my current health issues I am trying to not waste time on
hopeless causes but I am having fun and might have some useful ideas
for others.

I made a prototype using a tubular punch with a pneumatic (high
pressure, not player levels) pneumatic and solenoid valve control.
The tubular punches do a great job of cutting the paper but alignment
of the punch and the surrounding guide collar is absolutely critical.
I have not been able to punch a single sheet reliably but I plan to
make a more precise version that might be able to punch a single sheet.
The punching is done against an iron sheet so it is important that
the punch not bash into the iron. If not possible to do a single sheet
while maintaining punch sharpness for a long run,it will be necessary
to punch two copies and throw the bottom one away.

The punch moves just a couple of hundredths of an inch, so it is
very fast. I put it on a cycle timer but the on and off minimums were
one-tenth second and I am sure the punch will operate at probably five
to ten times that speed. I have a much faster cycle timer that I just
made that runs a normal relay as fast as it can go and once the weather
clears up a little I will be testing with this and a more carefully
made punch head prototype.

Since the punch head is tiny compared to the armature needed for
a punch and die setup, it should be possible to run the X-axis very
fast as well. If there are multiple punches instead of just one it
will be even faster having fewer acceleration periods but I will
probably not do even the single punch version. I just want to prove
a point and maybe provide software to drive it. Any software I write
will produce G-code from punch MIDI files. I already wrote this a long
time ago so finishing it should not be impossible.

The other thing I am playing with is using a little consumer type
laser engraver. So far I have proven that holes can be cut with very
minimal charing and smoke damage. The smoke damage is only on the back
side and the roll looks really good from the front. Initially I thought
this would be impossible without inert gas or whatever but it looks
very acceptable.

My initial testing suggests that with a punch advance of 1 mm and
with the relatively slow G0 speed command (including acceleration and
deceleration) of the laser engraver, it looks like a 30-foot roll would
take 12 hours to cut. I have no idea if the machine is running at the
maximum speed nor how much of the move time is due to acceleration and
whether that can be increased as well. It takes about 5 seconds to cut
10 widely spaced holes across 12 inches. Edge trimming would add maybe
6 seconds. This is cutting each hole separately.

Also I have a 30 watt (output) laser which is close to the top end of
consumer diode laser engravers. The highest power is 40 watts currently.
However, since the cutting is really a small part of the total line
time, I think that a cheaper laser will be almost as fast.

Cutting slots where appropriate instead of overlapping holes, and also
not bothering to cut over air, could also speed it up a bit but the
programming might get complicated. I know that the 'optimization' that
the engraver does on its own is really pretty bad -- jumping all over
the place.

The engraver software lets you output the G-code used in cutting and
it is very simplistic. It is all line moves, no arcs. I am hoping that
the software does support arcs as that would decrease the size of the
file tremendously. I will make some G-code using arcs (G2 and G3)and
see what the engraver thinks of it. :) Writing the hole cutting macros
should not be difficult and only needs to be done once but I really
hope the engraver G-code parsing supports G2 and or G3.

It should be possible to modify a standard $300 engraver and $100
cylinder attachment pretty easily to do the job, so intense machining
would not be required.

Spencer Chase
Garberville, California
http://www.spencerserolls.com/ 
spencer@poodlex.com.geentroep [delete ".geentroep" to reply]