A valve that switches its output between supply and atmosphere
basically has two valve disks and seats.  The positions of the disks
are set by a pouch (membrane) that senses a controlling input signal.

Simple so far, but this note comes from my frustration as a novice
organ designer with the question: why is every new valve idea so
different from the old ones, how many ways are there really to arrange
these components?

It appears we have three basic binary choices:

   * The system may be operated by pressurized air or by suction.
Fritz Gellerman once aptly pointed out to me the analogy of PNP and
NPN transistors, such that operate from negative or positive power
supplies.

   * The valves may be 'amplifiers' or 'inverters', either the outputs
follow the control inputs in terms of pressure, or they go the opposite
way.

   * The valves may be 'inside' or 'outside', referring to where the
disks are located with respect to the output chamber of the valve.

Three bits can be combined eight different ways, and it is easy enough
to find out working layouts for them all.  Notably they are all differ-
ent and this piece for meditation shows what they look like: [image]

All are shown in the state with zero pressure, atmosphere, as control
input to the upper left connection.  The chambers connected to the
supply suction/pressure are marked with S and P respectively.  Only
two of the designs, as indicated, are the ones conventionally used in
player pianos.  Note that the 'pouch disk' that carries the actuating
force from the pouch on to the valve stem must be put at the correct
side of the pouch.  In some cases this means the stem must penetrate
the pouch.  This is bad for leakage and good in that we get a stem
guide for free.

There are a host of critical requirements to make a valve work
properly, like air tightness in the valve seatings and pouch, and
smooth guiding of the stem.  This is covered in the literature [1,2]
and can be skipped here, but except for pouch leakage.

The input signal may come from the output of a similar valve.  Then its
pressure levels are well defined and it is relatively easy to make any
of the alternatives operate.  A more critical case is when the input
comes from a single open/close mechanism like a pallet valve, a cutout
valve, or a tracker bar.  Then the closed input state of pressure/
suction must be defined by use of a 'bleed', a narrow channel from the
supply to the valve input.  All the layouts above include the location
of this bleed.  In four of

the cases, those in a solid frame, the bleed is in parallel with the
pouch.  This means that the pouch tightness is not critical, you can
even implement the bleed as a small hole through the pouch although
this would be bad practice giving inferior control of how much bleed
flow there is.  In the other four cases any pouch leakage will reduce
the control pressure the bleed is trying to build up.  If for instance
the leak and the bleed flows are the same magnitude the pressure/
suction in the pouch well is only about half the supply and the valve
will likely cease to work.

For this reason the four alternatives illustrated inside solid frames
are the preferred ones, and we can feel content the old time player
designers probably knew what they were doing.

Johan Liljencrants

 - - -

References:

1.  Brougher, Craig : The orchestrion builder's manual and pneumatics
handbook. Automata Press, Independence MO, 1992.

2.  Reblitz, Arthur : Player piano servicing and rebuilding. Vestal
Press, Lanham MD, 1985.

[ See the drawings and this article at the MMD Tech site,
[ http://mmd.foxtail.com/Tech/    -- Robbie