[ Editor's Note:
 [
 [ The following message, which is attributed to Mike Ames
 [ is actually an experiment.  Mike is very active in the
 [ rebuilding and "MIDI-fication of instruments.  He's been
 [ involved in several serious archival projects, and is
 [ currently putting the finishing touches on a Mills Violano
 [ playing system.  He's very busy and is not a particularly
 [ fast typist.  He does, however, read the digest every day
 [ and often comments to me over the phone about things.  The
 [ experiment here is that I asked him to dictate a reply to Jim
 [ Heyworth's thread on Concertolas" onto my voice-mail and I
 [ then interviewed him to clear up some confusion I had
 [ when I listened to the dictation.  Because of the way the
 [ voice-mail system works, Mike was unable to review and edit
 [ what he said.
 [
 [ Typographical or interpretive errors are mine, not Mike's!
 [
 [ Jody

Message for Jim Heyworth in response on his discussion on Concertolas.

Concertolas, "remote multi-roll reading devices for a player pianos",
come in two breeds.  One, the most common, is designed for a "so called"
Duo-Art organ roll which is very wide (14 inches).  The other is designed
for a standard piano Duo-Art roll which is the normal player piano width.
The Aeolean Concertola system, shortly after the tracker-bar, has a series
of pouches, typically in glass topped chests with the pouches underneath
that flexes a limber contact switch on the end of which are the typical
organ crossed silver chucks of wire.

I'll digress slightly now into contact bounce.  Limber flexible members in
particular, as well as most switches, when they close, bounce.  If you don't
believe that all you have to do is get an oscilloscope out or a bigger
magnifying glass.  Its very difficult to damp the billiard ball effect of
two pieces of metal coming together.  They inevitably bounce in an unpleasant
fashion electrically.  If you are trying to record this signal you can imagine
the difficulties you get into, the MIDI encoder throws up its hands and
puts its feat up in the air.  The electromagnets, of course, don't respond
quickly enough to care about the contact bounce.

 [ Editor's Note:  A well behaved "contact-to-MIDI" encoder will record the
 [ contact bouncing as closely spaced sequence of MIDI On and Off events.  At
 [ best the extraneous "off-on" sequences make editing a pain.  However there
 [ are two "pathologies" we've observed which can make things worse.  The first
 [ pathology is that the "extraneous events" consume time in the MIDI output
 [ stream, distorting the apparent time of other events which happen "nearby".
 [ The second pathology is that some  "contact-to-MIDI" encoders may actually
 [ get confused and drop events if there is too much contact bounce.  This can
 [ result stuck notes on playback and "uneditable" notes w/r/t the MIDI editing
 [ program.  This second pathology afflicted a homemade (built under contract)
 [ MIDI encoder Mike had built before he switched to the Devtronix units mentioned
 [ below, which I don't think exhibit this second problem.

Back to the system.  The system for playing the piano is kind of fascinating.  The
piano has no roll frame.  The pneumatic input to the system is a series of
wooden members which clamp small, approximately 1/2" diameter magnets with ferrous
disks with leather and felt glued to either side (as Jim described in his
message).  There weren't very many of these pianos made.

There is a multi-conductor cable that comes from the strip containing the
myriad of electromagnets that runs off to the Concertola.  Yes, it would be
very easy to connect this to a MIDI driver system such as the
Devtronix 293 system.  Now I'll get in to the Devtronix nomenclature here.
The playback system, as we envision it, is quantity 1 model 293-1B MIDI output
processor board ($250) and from our point of view quantity 2 model 293-3A MIDI
output board ($125 each).  So for $500 you get a MIDI playback system.  Each
output board has 64 outputs (contact closure "equivalents").  Each output board
is "channel adjustable" and we've used them on many different systems here.

To record you buy a very similar system only there are input boards instead of
output boards and an input processor instead of an output processor.  But it
won't work with the kind of contact bounce you are going to get.  You'll have
to go through each recording and do a lot of cleanup work.  It can be done,
I have done it, but its a lot of work.

 [ Editor's Note:  This cleanup can be done with a program if the system
 [ doesn't drop events and doesn't induce too much time distortion as a result
 [ of the extra events.  I've written such a program for Mike.

It would be better to do the recordings another way (but thats another subject).

Returning to the  original message:  Jim talks about putting diodes across the
coils.  This is a fine thing to do. It will cut down the inductive kick, but
the price one may pay is that if the magnets are high Q,  it will delay the
armature falling of the electromagnet.  This is because the diode continues
to keep the current circulating in magnet.  If the coils are low Q than its
not an issue.  I've seen and know of examples of both ways, so you'll have
to try it out.

 [ Editor's note:  The delay characteristics can be improved by putting a
 [ small resistance value in series with the diode.  An alternative is to
 [ use a metal oxide varistor (MOV) with appropriate ratings.  Some bench
 [ testing will be required.

For future reference, I have one of these system so I'm familiar with what
he is describing.  We consider the system very reliable.  We don't see anything
the matter with it.  It represents something to achieve;  that is, the smallness
of the magnet (1/2" x 1/2") is pretty impressive.  We don't have anything that
small right now.  We've costed them out right now and they're going to cost
more than existing larger valves to do a similar job so we have not pursued
that.

For a record/playback computer, I think that his XT is fine, and CakeWalk is
fine. CakeLive would also be fine if he has it (it support a "juke box mode").
We're unfamiliar with the MPU-IPCT card.  We use the MPU-401 as our standard
one.  Not all MPU-401 "clones" work correctly.

Mike Ames   (transcribed and edited by Jody Kravitz)