Valve throw is so important to the proper action of a pneumatic
player, for the reasons already discussed by other MMDers.  When
I developed the interface valve unit for my Ampico, so it could
play from my computer "eroll" files, I soon recognised the problems
associated with valve throw.  That is, how much throw, and how to
ensure a consistent throw in all valves? In this case, the valves
were metal discs attracted by electromagnets that were positioned
the "throw" distance above the discs.

To determine the optimum throw, I set up a prototype valve and
applied all sorts of tests for different values of throw.  Fine,
I now have one valve set correctly, only 97 to go!  However, I could
now use this valve as my reference.  It was impossible to accurately
measure the valve throw, which would also not necessarily be an
accurate indicator of air flow through the valve.

My solution to the problem is perhaps the point of this posting.
Rather than use a static test, I applied a pulsing signal to the
electromagnets of about 10 hertz.  I then rigged up a set of 16
manometers, one for each valve.  Each manometer is a short length
(5 to 6 inches) of glass tube, one end immersed in a tray filled
with water.  The top of each tube has two connections, via a T-piece.
One goes to a common vacuum source of about 4 inches water column,
the other to the input nipple of the valve being tested.  These valves
simply open a hole to atmosphere.

The pulsing action of the valves causes the manometer readings to vary
between zero and whatever the vacuum supply is.  Because the valves
are operating rapidly enough, the height of the water in each manometer
will settle somewhere between the limits, giving an accurate indication
of the air flow through the valve.

It took me a while to establish the best settings, and I soon learnt to
be sure the vacuum level was not high enough to drain the manometers.
It was easy to make electromagnets pulse, perhaps not easy in a
pneumatic setup.  But I offer the idea anyway.

Peter Phillips
Sydney
http://www.petersmidi.com/