The reservoir recovering is similar to that in any player piano, save that it will contain pressure, not vacuum.  It is so big, almost six feet wide, that one must do the gluing in more stages than usual, and must work quickly, slopping on the glue without time to wipe up any spills.  Not doing this for a living, I can't run a thick brushful of glue freehand like the pros, without some getting where it shouldn't be.  So I masked everything, including the insides, since there are bungs that give access to the interior.  This also protects the already installed flap valves from any stray drops of glue.

    Put the assembly on the A-frame, open side up.  Using a piece of the old cloth as a gage, establish the gap, then wedge the boards open and fix them in the open position with sticks and nails.

    Starting at the center of the fixed side, glue the center bit of the cloth, and tack it down.  The original builder made the reservoir, with both fixed and moving boards as massive frames, separately from the base panel, player piano fashion, and the glued it to the panel.  But there was no way to brake such a huge joint, so the reservoir had to be covered like a surface mounted bellows.  The wooden stick and clamp holds the fold of the cloth tightly into the angle between the frame and the panel for a moment, until it sets.  Continue along the back edge in steps.  At each gluing step, apply glue to the board, put down the cloth, and lift it again quickly.  Apply more glue if necessary, until the white side of the cloth shows itself fully wetted with the glue.  At the end, put in a tack.  Then nail on the stick.

    Likewise, starting at the center, glue the other edge of the cloth to the moving board; pull it tight and tack it at the end.  Then lift the end piece, and apply more glue to the corner.  Quickly grab the loose end, pull it down very hard, and put two more tacks near the corners on the side.  It must be emphasized that tacks do not hold cloth on; hot glue is a hundred times stronger than tacks.  Their only function is to hold the corners of the cloth tight and square to the wood.  Otherwise, the elasticity of the cloth will pull the corner open into a little rounded shape, which will leak.

    The sides are now glued the rest of the way.  One must work quickly here, since the whole length of both sides must be done at one time.  Tack the bottom corners, and nail on the sticks.

    After the glue is well set but not dry, trim the upper half.  The masking tape, excess glue and excess cloth, peels off in one piece.

     Turn the whole thing over, hinge side up, in the A-frame.  Remove the gap sticks, and clamp the bellows closed. The hinge is covered with the bellows closed, so that tension in the lapped end and the hinge line cover will not bind the bellows open.  Cut the overlap end to the correct size, and glue and tack it around the corner.  Complete by gluing the rest of the lap piece, and then put in four more tacks at the inner end; two as close to the hinge line as possible, and two more at the outer corners of the piece.

    Finally apply the narrow strip of cloth that covers the hinge line.  Sometimes, the constant flexing of the bellows causes the lap piece to work itself loose along the hinge line.  Then air can sneak along this line from the corner, and work its way out through the little triangular gap where the long strip overlays over the lap piece.  A bit of pouch leather will prevent this; perhaps I overdid it here.

    Next day, when everything is fully dry, mask off the cloth, and refinish this side.  First apply thick shellac, allowing it to saturate and seal all the edges where the inner layer of the cloth shows.  Semi gloss spray paint completes the job.  Set up the finished assembly for the test.  Carefully tape over the normal openings, and temporarily unhitch the flap valves and tape their holes.  Install the springs.  I made a gasketed panel that fits over one of the supply openings, with a small pallet check valve covering the opening of the hose from the ShopVac.  The bellows can then be inflated, and when the ShopVac is turned off, the wind is trapped inside.  Note the odd appearance of the inflated cloth.  There are no gussets in the reservoir, and this is the same shape that the original cloth had.

    Time how long it takes for the reservoir to leak down from 4" WC, the maximum extent of the spring range.  (This will never happen in the finished organ, when the safety pallet is installed and set to open at the normal maximum 2-1/2" WC.)  At first I got a time of 72 seconds.  Givens says that 30 to 60 seconds is an acceptable goal for the amateur player rebuilder, but this reservoir is much larger and at lower pressure, and I was sure I could do better.  The leaks were chased down, first with a wet finger, and then with a squirt bottle filled with dishwashing liquid solution.  Any leak will blow beautiful Welkian bubbles. The first big problem was the access lids.  They were made from the same polar-oak-poplar lumber core plywood that has given me problems throughout.  The lids were so warped and partially delaminated that they wouldn't hold air, no matter what.  Also, the screw holes were worn loose by constant opening and closing, and some of them were oversized, from using replacement English screws, which are just a little bit bigger (or smaller) than American.  I thought I might get away with reusing them, but no.  I made new bungs from well shellacked Finnish birch plywood, larger than the originals to allow for a fresh screw line.  I gasketed them with thick soft blotting paper from Conservation By Design Ltd., in England, just like what was used throughout the organ.  I was wary of using this instead of leather, but it worked fine.  I got the time up to over 11 minutes just by doing this.  Now I had time to hunt for more leaks.  No bubbles were seen around the new cloth, but a couple of hairline cracks in the wood that had escaped my notice before were patched.  Most of the screw holes where various things are attached to the feeder side blew big streams of bubbles.  They did not go all the way through to the interior, but I suppose the drillings went close enough to the inner surface that air leaked through the soft surface layer of the plywood.  Perhaps these holes would seal themselves once the attached items where screwed on, but to complete the test, I sealed them anyway.  Once all the leaks big enough to make bubbles were chased down, the emptying time improved to 22 minutes and 50 seconds; not too shabby.

    Again, perhaps I overdid it, but every extra pump stroke that doesn't make music is lost effort.  But more important, the entire organ has hundreds of feet of packed joints, and hundreds of pouches, valves and pallets in parallel, all potential leakers.  It would be impossible to isolate any individual leak after the entire instrument was assembled.  The only way is to attach the components one at a time, and test each one in turn.  Of course, each item may leak a little, and each additional item will shorten the reservoir time by some amount.  But without a solid starting point, it will be impossible to judge whether that additional leakage is acceptable, or if that item has to be further checked for hidden leakage.

    [back] [forward] [index]