Restoration of a Knabe Ampico Reproducing Upright for Carol Drummond
145) 
146) 
147) 
145) Now it is time to rebuild the expression system for
this piano. Amphion made stacks very similar to what we have already seen so far on
this project that were never intended to be reproducing pianos. If we were
rebuilding one of those systems, the next item to be rebuilt would be the manually pumped
bellows. Since this isn't just a simple Amphion but a full blown Ampico Reproducer,
the reproducing system is next to be rebuilt. The Ampico reproducing mechanism is
one of the best designed suction level control systems ever developed. If it has any
flaws, they would first be the shear number of parts that go into each component.
See the next two photos to get an idea of just how many separate components go into each
expression mechanism. Secondly, these are designed in such a way that any leak
whatsoever, even if it is very small, will interfere with the function of the mechanism.
Everything must be perfectly air tight and also be rebuilt and assembled to operate
with total freedom or the system will not work. This mechanism has the ability to
take one individual note out of an entire series of notes that are to be played at the
same time all across the keyboard and make just one of the notes much louder than all the
others. It is able to do this by a combination of precise timing and suction control
that is made to operate literally on split second timing. If anything operates
slower than it is supposed to operate, the wrong note will get accented or the accent may
never take place at all.
146) Here is the first of two expression regulators that has been taken apart for
rebuilding. Some components I clean up and/or repaint as I take the unit apart.
Therefore, this photo will show some parts already looking better.
147) In this photo, all the various parts have been rebuilt and many of the
sub-assemblies have been put back together. The expression can now be reassembled.
If you will look to the center right of this photo, the control valve blocks are shown resting on their sides to reveal the new leather gaskets. Originally, these were gasketed with cork. However, cork takes additional screw pressure to make it air tight. Even though it is more expensive, I use leather because it seals better and is easier on the screw threads since it takes less pressure to make it air tight. Also, you will notice yellow circles of leather with holes in their centers which are glued to three of the four valve blocks. These are valve seats for the lock and cancel valves which are hidden deep inside the device. The lock and cancel system is used in order to make the construction of the paper rolls more practical. (That's a long story we won't get into right here.) Amphion originally used felt punchings for these valve seats. In fact, you can see the original red felt lock and cancel valve seats sitting in front of the valves in the photo. I replace the felt punchings (which are always reusable if you wanted to do so.) I replace them with buckskin. The buckskin is about the same thickness as the felt was so that the valve travel is unaffected. However, the buckskin is totally air tight whereas the felt always leaked badly. I have never had any luck getting an Amphion expression to set my regulating suction levels correctly using that original felt washer design. Air is allowed to seep through the felt causing the expression to operate inaccurately. The use of buckskin here is just one of a number of improvements I make on every Amphion I rebuild. I am always more interested in good function over originality. Some rebuilders reuse the felt seats even though they are awful because that is the "authentic" way to go. I say that, if Amphion got something wrong, let's learn from their mistakes and do it better.
148) 
149) 
150) 
148) Here is the expression after it is reassembled. This photo may help in an explanation of how the Ampico expression device works. To fully understand the function, you must also follow the work of rebuilding the crescendo mechanisms which will take place next. Inside of the main trunk box into (or onto) which all the other components fit is the main passage for air being evacuated from the pneumatic stack to the pump and then out into the room. Inside of this box are two large valves which are attached to wires that hold them in position. The main valve is the one that adjusts the suction levels inside the pneumatic stack. The job of the expression device is to micro-manage how far that valve is away from its seat. If the valve touches its seat, all air will be blocked completely. If the valve is moved far enough away from its seat, air will flow through the system totally unchecked.
To manage the placement of this valve, the Ampico expression has controlling pneumatics attached to the wire on which the valve is mounted. In fact, there are controlling pneumatics mounted at each end of this wire with the valve itself positioned between them. The blond board with the three square controlling pneumatics is constantly pushing that valve down towards its seat. At the other end is the wine colored pneumatic (which we call the "spring pneumatic.") This pneumatic is constantly attempting to push the expression control valve in the opposite direction. The two are something like the "Push-me-pull-you" of Doctor Dolittle fame. The three upper pneumatics (the square ones) are located at different points along the length of the blonde colored hinged board to which they all mount. The pneumatic closest to the hinge end of this board has the least ability to effect how hard that board pushes down on the expression valve. The pneumatic furthest from the hinge is the strongest of the three. You can see that this creates a situation where we have seven possible combinations of these three pneumatics which can effect how hard they push down against the pressure of the spring pneumatic below. Those combinations are: 1; 2; 3; 1+2; 1+3; 2+3; 1+2+3.
These seven settings create seven different possible suction levels in theory. In practice, they create six since 1+2 has the same strength as 3. This might be considered enough all by itself but there is more. The spring pneumatic at the other end of the valve wire is connected to a device called a crescendo. The crescendo's job is to increase and decrease the suction level inside the spring pneumatic. It can be set to raise or lower the suction levels at either slow or fast rates of speed. The suction levels that can be applied to the spring pneumatic are infinite.
This is, of course, a very simplistic explanation of a complex system but I thought it would be of interest to understand the workings of these mechanisms to some degree.
But wait, there's more! There is a second wine colored
pneumatic mounted at the bottom of the expression mechanism along side the spring
pneumatic. This pneumatic is called the "re-regulator." During
normal play, this pneumatic does nothing. However, when you select
"Subdued" on the selector switches inside the spool box, this pneumatic gets
turned on. It is fed whatever suction levels are being set by the primary expression
controls of this device. The re-regulator is spring loaded. At low suction
levels, the spring overcomes all the suction and does little to the pneumatic's position.
However, when the roll calls for increased suction levels in order to make the
music louder, gradually the spring on the re-regulator is overcome and the pneumatic
begins to close. As it closes, it moves a valve that is identical to the expression
control valve closer to its seat. The higher the suction in the system, the closer
this valve is moved to its seat. The purpose of this device is to enable all the
expression coded on the roll to be carried out as before. However, the louder
settings which are called for are toned down considerably without actually making the
variations disappear completely.
149) The same expression from the other side.
150) Both expressions after they were rebuilt sitting side by side.
151) 
152) 
153) 
151) This photo shows all the components of the player mechanism that are yet to be
rebuilt except for the pump. Starting at the far left and moving clockwise, you see:
the rewind pump control valve, the amplifier (which signals the pump to raise the
system pressures when you select "Brilliant" inside the spool box selector
switches, a pedal pneumatic and valve assembly, the motor governor which sets the tempo
and also keeps the motor running smoothly, another pedal pneumatic and control valve
assembly, the bass and treble crescendos, and the equalizers.
152) The two crescendos have been taken apart and laid out to reveal all the
components which make them up.
153) All the separate components have been restored. The main pneumatics can
now be recovered and the devices reassembled.
154) 
155) 
156) 
154) The restored crescendos. The job of these crescendos is to work in tandem
with the expression regulators. These devices govern how much suction is supplied to
the spring pneumatic and, therefore, govern how far away from its seat the expression
control valve is pushed. Although these devices are complex in their construction,
their functions are simple. They either increase the amount of suction slowly or
they do it quickly. Also, they decrease the suction inside the spring pneumatic
either slowly or quickly. Finally, these devices are also what we use to set the
lowest suction level that the system will ever drop to (which we call the "minimum
intensity.") The most important part about these devices and the spring
pneumatics to which they are tubed is that they must be air tight. There cannot even
be so much as a pin hole leak in these devices or in their spring pneumatics. The
valve and air channel that causes the crescendos to increase suction levels slowly do so
by drawing air through a pin hole sized orifice. If there is even the slightest
leak, the leak will match or exceed the size of the orifice that governs slow crescendo.
If this were to happen, the crescendos would be unable to raise suction slowly as
the leak would counteract the function of the slow valve. This is one of the most
common mistakes made by amateur player piano rebuilders. They fail to appreciate how
critical it is to ensure that all the components of the system are totally air tight.
It is not all that difficult to learn how to recover a player piano pneumatic.
However, it can be very difficult to learn how to do it in such a way that the
pneumatic is left absolutely air tight.
155) The bass and treble equalizers. These are like mini reservoirs of air.
They are spring loaded. The suction in the system draws them closed and
builds up stored energy in their springs. When the roll has the piano playing at a
specific volume level using only a few notes at a time and then calls for a large quantity
of notes at a similar level, these pneumatics will open. This draws extra air from
the system for just that split second needed to ensure that no notes are dropped during
the sudden increase in demand. In manually pumped player pianos, the equalizer is
called a reservoir and is vastly larger than these. Due to the much more even supply
of suction provided by the pump in an Ampico system, these equalizers need not be that
big. In fact, if they were too big, the more subtle effects created by the system
would get absorbed by them and would, therefore, be lost on the listener. In fact,
some designs of reproducing player pianos don't have any equalizers of any kind in them.
Weltes often have no equalizers, for instance.
156) The restored equalizers. These, too, must be totally air tight; no
exceptions.
157) 
158) 
159) 
157) The damper and hammer lift pneumatics before being rebuilt. These
pneumatics have an added advantage over the design used by Amphion in their Grand Ampico
installations. In the grands, a double valve arrangement like that on the left
device in the photo is used to operate both pneumatics. Once valve triggers one
pneumatic and the other valve triggers the other. With this design, the two valve
operate the one pneumatic. This lets that pneumatic close more crisply. Also,
the extra large valve body inside the control valve box of the right mechanism in the
photo permits air to pass rapidly from it as well. Care must be taken, however, to
keep the speed of the hammer lift pneumatic more stately rather than rapid. The
system expects the hammer lift rail to take 1 to 1 1/2 seconds to perform its duty.
Anything quicker could adversely effect the accuracy of the reproduction of the
music. For this reason, getting the valve travel exactly correct on this unit is
vital.
158) The two pedal pneumatics and valve have now been torn down into their separate
components. It is amazing that even the simplest of devices breaks down into a large
number of parts.
159) Here, you see the separated parts well on their way to being restored.
All metal is polished, all painted wood is repainted, all glue joints are cleaned and
trued up, all leather pouches and valve facings are replaced, all metal and/or wood valve
seats are cleaned and lapped to make the totally flat; all cork gaskets are replaced with
leather. Finally, all air channels including the inner surfaces of the pneumatics
themselves are freshly sealed with shellac.
160) 
161) 
162) 
160) The restored pedal pneumatics.
161) This control valve did get taken apart and examined. I found that the
pouch was in perfect, perfect condition. I resealed it and left it in place since no
new leather would be any better. I replaced the valve leather and repainted the
exterior. I will also have to, eventually, build a new valve cover since the old one
is missing. It serves to keep dust out of the valve itself. Unlike most
components in player pianos, this mechanism only got partial replacement work on the
leather. If I have even the slightest doubt about the condition and longevity of a
piece of leather I replace it. Often times I replace them even if I think they will
be fine. But this pouch impressed me since no matter how hard I pressed with my
finger nail, I was unable to damage the pouch in any way. So I permitted an
exception to the rule on this one control valve.
162) This is the amplifier valve. It is designed in such a way that it gets
suction from both halves of the pneumatic stack. Whichever half of the stack is
getting higher suction at any given time is the side which will supply this valve
assembly. Its function, when activated, is to close a pneumatic which is placed on
the pump and is attached to the pump's spill valve spring. As the pneumatic closes,
the spill valve spring is stretched causing the pump to generate a higher level of overall
suction to the system in general. This does not effect the dynamic effects of the
music but it does take everything evenly up a couple notches.
163) 
164) 
165) 
163) This is the amplifier after being torn apart but before restoration began.
164) The various components have been replaced. The cork gaskets have been
replaced with leather. The flap valves that ensure that this device is always being
driven by the half of the stack which is under the highest suction levels. I prefer
to use buckskin for these flap valves since it is very supply, thick enough to last many
decades without any signs of rot developing and because it is naturally quite air tight.
The pouch has been replaced, valve travel set, and new pouch leather has been glued
to the outside valve seat. At the factory, Amphion had a leathered valve head
sealing against this leathered valve plate. Leather never seals well against
leather. To correct this, I eliminated the leather on the outside surface of the
control valve. The wood has been lapped to ensure an exactly flat surface. The
bare wood of the valve face seals against the white pouch leather of the outside valve
plate far more tightly than the leather to leather joint did when it was first built.
165) All the last details have been cleaned up, repainted, put back together and
tested.
166) 
167) 
168) 
166) This is the wind motor governor. The action of the governor is meant to
compensate for minor variations in the suction supply caused by the constant suction
changes going on in the expression regulators. Also, any minor variations in motor
demand caused by slight variations in how it runs are all compensated for by this unit.
Basically, it evens out the suction levels and keeps the wind motor turning the
roll over the tracker bar at a consistent and predictable rate of speed. There is a
slight variation in this governor over those used in Amphion grand piano installations.
The tempo plates are a bit larger on uprights, permitting more air through the
system. Also a pouch is attached to the spring on the top of the governor that is
connected to the treble end suction supply. As variations in the treble suction
occur during play, these effect the suction level under this pouch and therefore micro
adjust the spring tension caused by the long black spring arm. Personally, I think
this slight change rather useless especially considering that Amphion eventually abandoned
its use. Other minor compensation adjusters similar to this were added to their
early expressions. It was thought that a very tiny pneumatic operating against the
spring pneumatic in the expression would make its function more accurate. Eventually
they discovered that this was of little use and dropped it from the design. However,
part of the history of this piano is that it has a governor with the old pouch
compensation attachment. Although it could be done without, I add it back in and
make it as air tight and strong as possible in order to honor the original design and to
improve upon it slightly. The original design called for simple pouch leather here.
However, thin pouch leather isn't really up to the challenge of drawing on such a
heavy spring with any real success. Therefore, I replaced that pouch with a triple
sealed pouch of sheepskin. This material is very air tight when sealed, as well as
being very flexible and strong. A much more perfect type of leather for this
application.
167) Much of the restoration of individual components is complete and I am ready to
start putting it back together. Certain design features of Amphion wind regulators
are such that the square pouch that operates the rewind valve along with the regulation
plate must be installed before recovering of the pneumatic. This is a touch
pneumatic to work with because it is impossible to check if you got the thing air tight
until you have it all back together again.
168) The completed regulator after it was totally reassembled and checked for air
tight operation.
169) 
170) 
171) 
169) All the components that are placed beneath the keybed
are completely restored, tested to be totally air tight and are ready for installation as
soon as the piano is rebuilt.
170) Now we rebuild the pump. The last pneumatic component in this piano to be
restored. Here is a frontal picture of the pump before work began.
171) The same pump from behind. From the color and condition of the cloth that
is covering the bellows and equalizer in this pump, I have determined that it has been
rebuilt once before. However, the conditions it has been subjected to since then
have been so damp that every screw has rusted and mold is growing on the cloth. Also
the cloth has taken on a stiff texture that indicates that it has been under adverse
conditions.
172) 
173) 
174) 
172) After taking the pump apart, I tackle the nastiest job first. Pulling all
the metal bearing components apart, cleaning out all the old grease and checking the
bearings to see if they are in good shape or in need of replacement. In the case of
this pump, I feel that the bearings are in good order and so I repacked them with fresh
grease. In this photo, I'm ready to replace the retaining rings and put this part of
the pump back together.
173) A shot of just some of the parts of the pump after it was taken apart.
Amphion is nothing if it isn't complex. Every component of the mechanism is loaded
with gobs of complex components; each of which must be in perfect operating condition or
the system will not work properly.
174) Here is the rest of the pump in the form of the four bellows. Each of
these four bellows actually has 50 different parts to them; all of which must be operating
properly for the bellows to function as designed. And I'm not even including the
complex connecting arms that join the bellows to the crank shaft in that count!
175) 
176) 
177) 
175) In the process of recovering the equalizer. This
is a heavy spring loaded bellows that helps keep the pump's suction level constant.
If you look at the pump suction separated from the rest of the system while the pump is
running, you will see the suction needle vibrating. Without this equalizer, it
wouldn't just vibrate, it would jump drastically. There are four suction bellows in
the pump which is not enough to create a seamless air supply. With the equalizer in
place, the air supply becomes sufficiently even that the system can operate perfectly.
176) The restored equalizer. The bellows cloth I am using is the same color as
what was used during the last restoration. Don't confuse the recovered parts with
the ones which have not yet been attended to. The last person to do this, skipped a
lot of important steps. Even though they recovered the bellows and equalizer, they
didn't replace any flap valves, they didn't replace the hinges, they didn't seal the wood
on the interior or repaint the wood on the exterior. In addition, there are signs
that they attempted to lubricate the bearings with plain oil instead of grease.
Fortunately, the pump didn't work for long after this botched attempt at
restoration. If it had been run long with that oil in there, the bearings would have
been ruined.
177) This is the cast iron cover plate which holds all the bearings. It has
been cleaned of old grease, dirt and rust, lubricated, repainted and reassembled.
178) 
179) 
180) 
178) This is the spill valve plate and the brilliant
pneumatic. When the louder setting is actuated from the spool box, this pneumatic is
pulled closed. At the hinge end of the pneumatic is a spring attached to a valve.
This valve has been recovered in new leather and everything has been carefully
sealed and recovered. When the pneumatic closes, it increases the spring tension
that is pulling on the spill valve. The spill valve lets a certain amount of air
leak into the pump to prevent it from bogging down. When the spill pneumatic is
closed, the tension on the valve increases which raises the suction pressure inside the
system so that the piano will play louder while retaining all of its expression.
179) This is a photo of the manifold which divides the suction from the pump into
the various lines that need it. The wooden block at the bottom of the photo is a
pouch block which fits into this manifold. When the roll is rewinding, the pouch
expands onto the valve seat seen inside the manifold at the top of the photo. This
blocks suction from getting to the pneumatic stack so that the piano will be silent during
rewind. All these parts must be sealed and all new leather installed. As
usual, Amphion used cork for the gaskets on this manifold. I will replace it with
leather which is more air tight and requires less screw tension to keep it secure.
The screens that you see in this photo are there to prevent any debris which gets sucked
out of the action toward the pump from becoming lodged under the cut out pouch or in any
of the pump's flap valves.
180) The restored cut out manifold.
© Copyright 2010 {David Rodgers' Piano Rebuilding}. All Rights Reserved.