John, I'm glad you enjoyed the translation of the article:

        Juergen Hocker, Horst Mohr, Walter Tenten: "Computer Control and
        Synchronization of two Ampico Self-playing Grand-pianos", in
        "The Mechanical Music-instrument", Nr. 63 (August 1995) p. 42-48,
        published by the Society for Self-playing Music Instruments (GSM),
        Germany.

The writers say that many valve designs were studied, including yours,
but none met all their requirements. I think that they rejected your
solenoid valve design because of concerns about power consumption,
repetition rate, consistent response from valve to valve, and
adjustment stability.

Consider the difference in armature masses between (1) a solenoid,
(2) an ordinary relay, and (3) an organ "magnet" valve, such as the
Reisner design.  ("Armature" is the moving piece in any magnetic
device like these.)

In order to realize similar transit acceleration, the force of the
return spring must be proportional to the armature mass.  Recall the
familiar equation, "force = mass x acceleration".

The return-spring of a solenoid valve does not have very good
stability because of high force and large compression percentage --
after many cycles of operation, and heating and cooling, the spring
force is likely to change.  This will degrade the consistency of the
valve-to-valve response times.  Also, the sliding armature of a
solenoid is very susceptible to contamination, hence varying friction.
I can't recall solenoid valves being used in any industrial application
which requires long-term stability like in a reproducing piano.

The basic requirement of an electric valve for a reproducing piano is
only to open a tiny aperture to a hose, from 4 up to 40 (Ampico) or 70
(DuoArt) inches-of-water vacuum.  There is no need for the long stroke
and "brute force" of a  solenoid, nor for the large port area which a
relay armature can cover. Therefore, the organ-style "magnet valve"
(such as supplied for many decades by the Reisner company) is adequate.
It has other deficiencies, though.

The organ valve design sacrifices speed and magnetic path efficiency in
favor of simplicity.  The valve assembly is bulky, but the armature is
tiny, and the return-force is gravity.  The low efficiency results in
high power dissipation, but this is not a problem due to the great area
of the exposed coils, and in an organ there is plenty of space for
convection cooling. (And _inside_ the wind chest there's plenty of
moving air!).

The valve for a reproducing piano, as you correctly point out, must be
both  low-power and small. It must also be quiet!  Author Walter Tenten
told me that their valve assembly had to be contained in a solid, felt-
lined box, in order to reduce the noise.  Most of the air vents ultimately
were covered, too, because of the noise.  Fortunately, he said, the
temperature of the relay valves didn't rise so high that a fan was
necessary.

The magnet valve designed by Wayne Stahnke for the IMI Cassette Converter
improves the Reisner valve in these ways: (1) it's considerably smaller,
(2) power dissipation is reduced, because a strap-steel frame forms a
complete magnetic path with the armature disk, (2) the mass of the
armature is minimal, and (3) a rather "long" return spring is used, which
experiences only about 15% active compression.

The extremely rare DuoArt Concertola has a magnet valve with even
better efficiency, due to the steel cup enclosing the coil, but
without a return spring it is slow compared to the Stahnke design.
Also, all 98 valve  components are assembled into single wooden
board, much like a primary  valve set from an Ampico, and this
assembly is difficult to adjust.

Mike Ames inquired about the cost of duplicating the Concertola
valve, and was told it would be about $10 each in a small quantity
(1000 units, I believe).  A few years ago Ampico technician Randy
Cox and I got a similar quote for a short run of Cassette Converter
valves.

All I know about the "new" Reisner valve is that the company is
talking about modifying their old vacuum valve design to make it
smaller (and less expensive, we hope).  The "old" design costs only
about $4.00 in 100 quantity,  which is appealing, but otherwise it
fails to meet other important requirements.

All of the potential designs for a cheap reproducing piano valve
seem to cost the same, and $10 per valve is still too much for me.
The only apparent way to reduce the cost is through production in
large quantities, like 10,000 units.   We (that's Mike Ames, Randy
Cox and, with advice from Wayne Stahnke) are  studying all the
possibilities.

I agree with you, John, that a market is developing right now for
good quality valves.  You and I and many subscribers in this group
are definitely planning to someday connect the computer to the
piano, just as described in the German Society article.

I want to see all the options before I commit my meager funds!

-- Robbie Rhodes