Flow rate measurement of Ampico-A secondary valves
  and strange marks on upper valve seals.

I will soon be reassembling a set of Ampico-A secondary valves from
a 5'-8" Chickering Ampico manufactured in 1926 (right-side-up valves).
The piano has been professionally restrung with new soundboard,
bridges, pinblock, hammers, voicing and regulation.  With the
touchweight of all keys being somewhat uniform, it will be important
to make sure each striker pneumatic can lift each key with the same
amount of force at a given stack pressure.

The amount of power delivered by each striker pneumatic is dependent
upon several factors, such as using the correct span when covering the
pneumatics, keeping the stroke length equal on each tier of the stack,
minimizing lost motion and friction, sealing all potential leaks and
replacing both upper and lower valve facings with a good grade of
leather.  Even with the very best workmanship and materials, there
can be some variation in leakage among a set of new pouches and the
sealing of valves against their seats.

In searching the MMD Archives and AMICA Technicalities, there is
an abundance of advice on how to rebuild and regulate these valves.
Most of the discussion concerning regulation centers around setting
the valve travel to between .038 and .040".  The subject of flow rate
is touched upon but no article I've found has actually published
a suggested flow rate (standard cubic feet per minute) for these
secondary valves.

Spencer Chase did do some testing with a rotameter connected to the
upper valve seat.  (A rotameter has a tapered bore inside a transparent
scale with a ball which floats up and down.) Spencer's test is very
easy to perform and would be helpful if you were having issues
achieving uniform repetition rates.

In order to really measure the valve's capacity to evacuate the striker
pneumatic, I believe you would have to make your measurements in a test
jig with the flow meter connected directly to the output port of the
secondary valve.  This measures the actual flow between the lower valve
face and the lower valve seat and factors in the tiny but significant
leakage of the upper valve face against the upper seat as well as the
tiny but significant variations in the lifting force exerted by the
pouch.  These leakage factors may prove to be insignificant most of the
time except when there is an actual failure of the valve to seat fully
or a puncture in the pouch.

At some point in time, flow test experimentation was claimed to have
been done in the Ampico laboratory as it is described in some of the
advertising.  (I need to find and reread this info as it may have
answers to some of my questions.)  Does anyone know if and when flow
testing was implemented on the assembly line?  Could anyone suggest
an appropriate flow rate for the stack and piano described above?
I suspect one would want to perform this test at a relatively low
vacuum level, say, 5" regulated water column.

In the absence of empirical data, the plan is to set one valve at
the recommended gap with a dial indicator and measure its flow rate.
All other valves will be set to match the flow rate of this reference
valve.  The test fixture will be constructed so that the output port
of the valve is tubed to the top port of the rotameter gauge (to suck
the ball upward off its seat).  The bottom port of the gauge will be
open to atmosphere simulating an infinitely large striker pneumatic.
Flow restrictors can be inserted here for a range of measurements which
will be recorded.

The rotameter I purchased yesterday is without the adjusting knob since
it is for indicating flow, not regulating it.  The range is 0.8 to 8.2
standard cubic feet per minute (SCFM).  (I hope I'm in the ballpark.)
The fixture will include an adjustable mandrel for forcing the upper
seat into position.  This mandrel will be relieved at its center to
allow the tiny but significant leakage of the upper valve face to be
factored into the net result.

I believe the dial indicator should still be used for reading the final
valve travel, but not used for setting it.  If the travel falls within
an acceptable range, you are assured there are no serious valve seating
or pouch defects and your vacuum loss will not be excessive.  If I do
a good job hinging the striker pneumatics and regulating the flange
fingers, repetition should be trouble-free.

Has anyone ever noticed strange chisel marks on the upper valve seats?
Each of the metal upper valve plates from this original unrestored
stack has a strange mark on it and the marks are all in the same
relative positions.  Holding the valve with the upper seat facing you
and the ports at 12 o'clock, the marks are at the 2 o'clock position,
a short distance from the center hole.  The marks are about 3/16" long
and look like they were made by a sharp chisel.  The mark is always
perpendicular to the radius of the metal seat.

On a few parts, the 'chisel' was dragged a short distance leaving a
scratch.  Some of the marks are deeper than others and the deepest ones
began to bend the metal very slightly.  The simplest explanation I can
think of is that the factory might have had a dial indicator fixture
and a way of inflating the pouch.  The upper seat would be placed into
its bore without any sealant and the valve assembly would be placed
under the probe of the dial indicator.  While repeatedly inflating the
pouch, the operator would press down on the valve plate (with the
'chisel' somehow) while watching the dial indicator until the desired
valve travel was reached.

This is a very fast way to set the valve travel if you are not too
concerned about maintaining the perpendicularity and flatness of the
upper valve seal plate.  I would prefer to see the upper plate pressed
into position using a vise or a drill press mandrel but you'd have to
repeatedly remove and replace the dial indicator.  Varying amounts of
force are required to push the seal plate into position which would
account for why some of the marks are deeper than others.

I guess the factory could have found it more efficient to press the
plates down with a hand tool or some kind of lever to avoid disturbing
the orientation of the valve assembly under the dial indicator.

It must have taken quite a lot of clamping force (or impact?) to begin
deforming the shape of the valve seals.  Maybe there was a chisel-tipped
clamping lever in contact with the valve seal which could be forced
down by turning a knob.  You can begin to see the advantages of using
flow testing.

I reviewed all the popular disassembly methods and settled on removing
the pouch board first with a sharp knife.  It's kind of like opening
a clam by applying constant force until it yields and then moving the
knife slowly clockwise around the perimeter until separation.  No
serious damage occurred to any of the wood parts.

A 10-second blast from the big red heat gun softens the sealant around
the metal valve seal.  Then, with the pouch board gone, a special dual
purpose mandrel mounted in a drill press will push down on the inverted
valve spool which pushes down on the seal plate, ejecting it.  Another
quick blast from the heat gun to the lower valve seat and the mandrel
pushes it out without damage.

85 valves were disassembled in one day.  I didn't break a single one
and I can prove it!  Since the lower shellac valve seats can have
invisible cracks or be loosely sealed I decided not to reuse them.
I ordered a bar of 303 stainless steel to make replacement lower seals
on the lathe.

Larry Doe