Restoration of a Knabe Ampico Reproducing upright for Carol Drummond
70) 
71) 
72) 
70) To the left are valve bodies which have had their glue
joints cleaned and dressed. While I do the dressing of the joints, I also use a
dowel rod that is perfectly flat at one end and covered with fine sand paper to flatten
any imperfections in the Bakelite valve seats. To the right are valve bodies which
haven't been cleaned up yet.
71) Switching to the primary valve chest now, I have cleaned off all the old leather
valve seats. Now I'm cleaning off the old shellac sealant and dirt in preparations
for making the surface flat and new again.
72) The valve board for the primary chest has now had a new layer of lacquer sealant
placed on the outer surface. This is really very different from the usual Amphion
primary chest as you will see in a moment.
73) 
74) 
75) 
73) While I wait for the primary valve chest to dry, I move
on to the secondary valves again. Now the pouch blocks have had their new pouches
put into place. Each pouch must be installed in such a way that it has a cup or
"dish" in it that will allow it to move up and down. It is not enough to
just glue the leather on flat. A special tool must be used (called a pouch setter)
to get the pouches installed correctly.
74) A dot of masking tape is placed at the center of each pouch. Then the
pouches are sealed with thinned rubber cement. The rubber seals the pouches totally
and protects them from future rot as well. The black tube is inserted into the
signal hole of each pouch block and the air tight condition of each pouch board is tested.
If any pouch is still seeping air, more sealant is applied until it is certain that
the pouch is air tight. Then the masking tape dots are removed. These dots of
tape prevented the rubber from getting into that area of the pouch. A dot of glue
will be put in that spot in order to attach the lifter disc to the pouch. The glue
will not soak into the leather and form a good bond unless that spot is free of sealant.
75) Now the lifter disks are in place. These discs are needed for two reasons.
First, the valve stems that are actuated by the pouches are not shaped in such a
way that they will interface well with the pouch. The disc gives the valve stem a
platform to ride on. In addition, the disc has the ability of transferring more of
the pouches lifting energy to the valve than would be transmitted without it. One of
the keys to good player piano design is getting the proper balance of valve size to pouch
size to lifter disc to pneumatic size set up correctly. The size of the pneumatic is
set based upon the amount of work that the pneumatic needs to do. Then the valve
size is set to be large enough to evacuate the pneumatic instantly. The pouch and
lifter disc are sized to ensure that the valve is able to move from one seat to the other
rapidly at all possible suction levels. If the pouch is too big, it will slow the
action of the valve. If the pouch is too small, the valve may not work at all.
It is a delicate balancing act...
76) 
77) 
78) 
76) The next step in the process on the primary valve chest
is underway. The lacquer that is on the wood where the leather valve seats need to
be glued is sanded away so that the glue will stick. Then the new leather valve
seats are glued into place and pressed down flat onto the wood with a piece of glass.
By doing this, any imperfections in the wood or leather will be corrected for.
The glass will press the leather down into the glue ensuring a flat leather surface
for the valve head to seat against.
77) The Bakelite valve heads have been flattened as described earlier. Now
they are placed onto the valve stems and a gapping "feeler gauge" is used to set
the gap between the valve head and the leather seat. The inside of the valves have
new leather as well. The overall thickness of the new leathers were ever so slightly
greater than with the originals. This made it necessary for me to work a little
harder getting the valve travels set properly since the valve stem to valve head
adjustment was not exactly the same as original. Once the gap is set exactly, a drop
of glue is placed onto the top of the valve stem. These stems have been thoroughly
cleaned of old glue. The new glue soaks into the wood valve stem and locks the stem
into the body of the Bakelite valve head. This is strong enough to hold the valves
in the correct position forever. However, it is just weak enough that the valves can
be knocked apart again in the future without doing any damage to the valve in any way.
78) Next to be restored is the pouch board to the primary chest. All the old
cork gasket material and old leather pouches have been scraped from the wood. Then
the surfaces were sanded clean with medium sand paper backed with a piece of glass to keep
the surfaces absolutely flat. The pouch wells were then resealed with fresh shellac
to ensure air tight operation. In this photo, half of the new pouches have been
installed.
79) 
80) 
81) 
79) In this photo, all the pouches are installed into the
pouch board and the leather has been sealed with rubber cement. You will note that a
number of the pouch wells have no pouch in them. The Amphion design allowed them to
mass produce mechanisms in advance. They were modular enough that the specific needs
of individual piano scales could be compensated for by simply skipping over any spots that
didn't align with a note that needed to be played. After sealing is complete and
perfect, the surfaces of all pouches are dusted with talc to prevent them from being
sticky from the sealant. Due to the shape of the primary valves, lifter discs are
not needed because the valves themselves are shaped in such a way as to ride very nicely
on the pouch. The reason for having this chest of valves in the system at all has to
do with that balance of sizes I mentioned earlier. If we ran a signal tube from the
tracker bar (where the paper roll runs) straight to the secondary valves, the operation of
those valves would be somewhat sluggish. The size of the valves is such that they
are too large to be able to operate quickly with just that tiny signal from the roll.
To overcome this problem, the signal from the roll is first sent to a very small
valve which is capable of operating at lightning speeds in response to the weak signal
coming from the roll. The only purpose of the primary valve is to send a very strong
suction and/or normal air pressure signal to the larger secondary valves. The result
is a system that operates at lightning speeds otherwise unobtainable.
80) The primary valve chest is finished. All new leather gaskets have been
installed, all the pouches and valve leathers have been replaced and all seating surfaces
have been freshly flattened. In addition, all finished surfaces have been restored
to a new looking condition. The chest, as it sits, is more air tight and more easily
serviced than it was the day it left the factory.
81) In this photo, I am in the process of gluing the secondary valve's pouch boards
back onto the main part of the valve blocks. To get this joint as tight as possible
requires considerable pressure. The design of the secondary valves makes it
necessary for the leather pouch to actually be in the way of the joints between the
boards. Therefore, significant pressure is used to more or less crush the leather
down to enable the surrounding wood surfaces to mate properly. In this photo, eight
spring clamps per valve block have been used to close up the joints. I stacked them
one on top of the other as I went along. As soon as the glue has gelled to the point
where it will not longer slide about, the spring clamps (which are easier to position) are
removed and an even stronger ratchet clamp is used to really compress the joints.
This is allowed to stand over night before the clamps are removed.
82) 
83) 
84) 
82) While the secondary valve blocks dried, I went back to
the primary chest. Now it is time to test the chest to see that there are no leaks.
The first test involves blocking off all the signal tubes in and out of the chest
with knotted rubber tubing. Then a suction supply is placed on the chest and air
pressure gauges are used to test the pressure inside the chest. Also, it must be
remembered, that because this is an expression piano, the mechanism is divided into a
treble and bass half. So the treble side is tested first. I put that side
under the highest pressure I can manage. Then I block off all air ducting to the
bass side and test it to see that no air is sneaking past the divider into the other side.
Once I'm convinced that the central divider is totally air tight, I turn the
suction down lower and listen to each valve head with a stethoscope to ensure no leaks are
occurring. Then I operate each valve one by one by signally the valve through its
signal tube. First I listen to the valve in the on and off positions to ensure that
no leaking is going on. Then I test the valve for rapid fire repetition. The
entire chest passed all tests with flying colors. However, there is one last test I
like to perform that is more archaic. With all the signal blocking tubes still in
place, I suck on the main suction supply nipple to each side to see if any leakage shows
up with this test. I just finished a more standard Amphion primary chest for a Mason
and Hamlin grand which had the standard primary chest. When I sucked on its main
supply nipples, the air pulled up hard under that test. In other words, it was tight
as a drum. Due to the design of this chest, however, it didn't pull up quite so hard
as with the other design. This means that this chest doesn't seal quite as well as
the other design. I suspected that this might be the case when I first saw the
chest. The trouble comes in the form of the leather valve seats. The size of
the seats are rather large. The Bakelite valve heads that rest on those seats are
equally large. Common sense would usually make you think that the larger the joint
between the valve and its seat, the more likely it will be to resist air seepage.
Actually, the opposite is true. The larger contact surface of these valves means
that the suction pressure holding the valve in place must be spread out over a larger
area. This means that the valve bears down onto the leather seat with less actual
pressure than with the more common design. In the future, I think that I will remake
the design of the leather valve seats to create a smaller surface area for the valve to
rest on. This might prove to make it more air tight. Even with this flaw, the
chest is sufficiently air tight to be very effective for many years to come. The
design flaw was original to the factory and yet it was made and sold in mass quantities
just as it is. The changes I have in mind will make future chests of this design a
little better but the improvement isn't enough to warrant pulling the chest apart and
reworking the leather seats on this unit.
83) The pneumatic boards have been held in clamps until they were totally dry and
flat again. Now all the old shellac, hide glue and dirt are being sanded off all the
surfaces of the boards. With this unit, the boards which are glued to the trunks
(called the stationary boards) are made from poplar while the movable boards are made of
maple because they must retain a screw that hold the lifter fingers in place. It
would be better if both boards were maple but other brands of pianos make both boards from
poplar so either one works fairly well.
84) This is the process of reattaching the two boards of the pneumatics to each
other by the installation of a cotton hinge. The hinge material is cut to size and
folded over and ironed into halves with a sharp crease. Plastic wrap is sandwiched
in between the halves of the hinge. Then the boards are coated with glue and the
hinge put into place on the first board. Then the second board is laid into place
over the hinge and the two boards are carefully aligned. Then a light clamping
pressure is applied just until the glue has a chance to gel up some. The pneumatics
are hinged in thirds. When the first third is hinged, they are taken out of the
light clamps and ganged together under the pressure of a single clamp which holds them
much tighter. Then they are allowed to sit over night to get hard. The plastic
wrap that was placed inside the folds of the hinge keeps any glue which might soak clear
through the cloth from causing the hinge to become glued to itself. Some makes of
pneumatics in piano and in organ work are occasionally built without any internal hinges
of this sort. Others have internal and external hinges. The reason for the
hinge is simply to make the pneumatic more strong at that point and also to make the job
of covering the pneumatics with rubber cloth (or in some cases - leather) more easy.
85) 
86) 
87) 
85) After the hinges had set up on the pneumatics, two more
tasks needed to be carried out before they were covered. First, the inside surfaces
of the pneumatics were coated with thickened orange shellac (usually referred to as burnt
shellac.) This is not usually done to pneumatics. However, this type of
pneumatic needs to be almost totally assembled before it can have its outside sealant
applied. This makes it very hard to test the pneumatics for air-tight-ness as I go
along. By sealing the pneumatic on the inside I am doing two things. I am
adding structural integrity to the wood and making the wood air tight without any need for
any sealant being applied to the outside. The second task is to glue a felt bumper
inside the pneumatic. This bumper (red felt in this case) keeps the pneumatic from
closing tightly onto the new cloth after it is covered. When the pneumatics are
clamped onto the rails, much pressure will be brought to bear. If there is nothing
to keep the boards apart, they will squeeze down onto the pneumatic cloth causing that
cloth to take on a hard crease. Rubber cloth which has a hard crease will only stay
air tight half as long as cloth which has only a gentle crease.
86) Strips of rubberized cotton cloth have been made and layout lines set up on the
work table. A pneumatic which is showing the internal sealant and felt bumper and
cotton hinge is ready to have the first strip of cloth glued to it. The layout lines
will make the process easier to get right. The boards are more likely to remain
parallel to each other during operation. Since the pneumatics are aligned onto the
trunks with the movable boards as the guides, they can look misaligned when viewed from
the fixed boards unless both boards are absolutely aligned with each other during the
gluing up process.
87) The first three sides of the pneumatics have had the cloth glued to them.
This must dry a little before trimming.
88) 
89) 
90) 
88) Now the first section has been trimmed and is ready to
have the last side glued up. The pneumatic is carefully and neatly closed and then
the two tail pieces of rubber cloth are glued down one by one. Then the excess is
trimmed off.
89) Finally the pneumatics are stacked under a weight to keep them closed.
They must not be allowed to open until the hide glue has dried. If you
attempt to glue the hinge ends of the pneumatics up while they are open, you will cause a
condition called "hinge binding" which prevents the pneumatics from closing
properly.
90) A layer of cheese cloth is glued to each trunk board after the index marks have
been transferred from the sides of the trunks to the tops using a square to get them
perfect. The cheese cloth will make it easier to remove the pneumatics in the future
while not interfering with the joint enough to cause any chance of pneumatics falling off
on their own some day. This is the same technique used at the factory to ensure that
the system is easily rebuilt in the future without compromising its function in the
present. The cheese cloth and thinned glue used to adhere it are not significant
enough to hide the layout marks that will guide me as I install the pneumatics back into
place.
91) 
92) 
93) 
91) Before the pneumatics can be glued to the trunks, each
one must be tested to ensure it is really air tight. The air holes are taped shut
with the pneumatics held in the closed position. I try to pull the pneumatics open.
If they refuse to open, I know they're air tight. If they open slowly or even
quickly, I know that something didn't get glued down well enough or I have a crack that
was missed in the boards. Only a couple of the pneumatics failed this test and were
repaired to make them totally air tight. The iron is pictured here because it is
often the only thing needed to correct a leaky pneumatic. If a corner or edge of the
pneumatic is not glued down properly, it is often enough to warm that spot with an iron
and rub it down again. This often fixes minor leaks.
92) A jig is set up to guide me in gluing on the pneumatics. Most rebuilders
don't use this which leaves the pneumatics looking a little misaligned and sloppy.
The straight edge in combination with a square that is used as I glue each pneumatic down
ensure that they are aligned properly. In the event that I find an original lay out
line which is flawed in the tidiness of the original glue up, I correct this so as to make
the finished product look very neat, tidy and well aligned.
93) These three valves show the process of reassembling the valves once the
rebuilding work is done. First, a restored valve is dropped down into the valve
block and allowed to rest on its inside (Bakelite) seat. While blocking the signal
hole to the pouch with my finger, I suck on the air suction hole (bottom left on the upper
left valve body in the photo.) If the seat is good, the valve body is good and all
aspects of the valve itself are good, I will not be able to pull any air through that
hole. Once I have the valve air tight at this stage, I insert the outside metal
valve seat. A gapping tool (left) is used to set the spacing between the outside
seat and the valve to the correct distance. Too much travel in the valve and it will
waste air and make a troublesome spitting noise while it also makes the action move
sluggishly. Too little travel and the valve will not permit air to pass through it
quickly enough to collapse the pneumatics promptly. Once I have the seat correctly
adjusted, I seal it into place with burnt orange shellac. The outside edges are
coated with the shellac as are the relief cuts that were needed to shape the metal when it
was original stamped out. Once the shellac is dry, I will be able to check that the
valve is also sealing well against the outside seat.
94) 
95) 
96) 
94) Two of the three tiers of this stack had wooden dowel
plugs in their centers. These were added in order to take a stack that was
originally set up to be a lower end manually pumped player piano and make it into a
reproducing unit. These plugs were not air tight after installation so Amphion
covered them with rubber cloth to seal them up. The problem with this method is that
the cloth gets old and brittle and must be replaced from time to time. Rather than
repeat this mistake, I covered these areas with wood veneer which will never need to be
replaced again and which will keep these areas air tight permanently.
95) All the valves have been completed.
96) Now it is time to install the gaskets which will enable them to mate with the
trunk boards in an air tight fashion. Normally, this is done with cork gaskets that
are available from my suppliers. However, I don't agree with the Amphion reliance on
cork for gasketing their mechanisms. Cork takes far too much pressure to make it
seal and often requires that it be glued down on both sides to ensure a permanent seal.
When gaskets are glued on both sides, you cannot take the item apart for servicing.
Leather seals much better than cork with much less pressure. This prevents
stripped screws and gives me an overall tight result. I have rebuilt Amphions using
the original cork gasket designs. They worked fine. However, I have come to
believe that the leather option is a better one. Since I cannot buy pre-made leather
gaskets for these valves, I must make them by hand. This photo shows some of that in
process. It takes about an hour to glue on pre-made cork gaskets to a set of Amphion
valves. It takes an entire day for me to make and install leather gaskets. I
do this extra work at no additional cost to the customer. The better result make it
well worth my while.
97) 
98) 
99) 
97) All the cork gaskets have been installed. The six
valves that are set differently than the others were valves which did not pass the air
tight tests I made before putting on the gaskets. For various reasons, these valves
had very tiny leaks that I was not willing to put up with even though they were very
slight. I heated and removed the outside metal valve seats on these and found and
repaired the causes of the leakages. The one valve that is shown still in pieces is
the only valve that gave me real trouble. No matter what I did, it seemed to insist
on having a small seepage. It has been sealed all around with extra burnt shellac.
After it dries over night, I will test it again to see if I have corrected the
problem. If no cause for the leak can be found at that time, I will replace the
valve entirely.
98) The three tiers of the pneumatic stack with the pneumatics glued into place.
The fixed boards and trunks have been coated with clear burnt shellac to make them
more attractive, new looking and protected from dust staining. This will also slow
moisture penetration into the wood.
99) The opposite sides of the pneumatics shown here cannot be shellaced until after
the lifter fingers are glued into place. Since the insides of these pneumatics have
all been sealed already, I can test each pneumatic for air-tight-ness again. I know
the pneumatics themselves are air tight but now I need to check that the glue joint
between them and the trunk is air tight. This is done by blocking the supply hole on
the trunk with my thumb as I try to pull the pneumatic open. If the pneumatic will
not open, I know the glue joint is air tight. If the pneumatic tries to open, I know
that the cause of the seepage will be in the glue joint itself.
© Copyright 2010 {David Rodgers' Piano Rebuilding}. All Rights Reserved.