Keyboard: Balance pin holes and rack slots

Keys must be guided in two places to keep them from shifting sideways as they are played. All keyboard frames have a balance rail slightly in front of the midpoint into which balance pins are set. These pass through holes in the keys. The second guide location varies among the various national styles of harpsichords, but a common arrangement for Italian harpsichords is to use a rack at the back of the key frame. A slotted strip of wood standing on edge receives metal pins or wooden slips driven horizontally into the key ends and restricts the keys to up-and-down motion only.

I'm using balance pins of 0.096" diameter and drilling a 3/32" hole to receive them. In order that the holes through the keys line up with the holes in the balance rail, the key panel is simply screwed onto the balance rail, after making sure that the panel is correctly positioned both back-to-front and side-to-side. Drilling holes this way is certain to align them properly, as any slight discrepancy is transferred from one part to the other:


The sharps have their balance pin holes offset to the rear, since their key levers are shorter.

Now for the rack itself. It must fit under the wrestplank so that the keyboard can be installed and removed freely. The key ends pivot in small arcs when played, so the rack can't be right up against back of the keyboard, nor can rack be much thicker than the upper belly rail or the back row of jacks might rub against it. These parameters determine the height and thickness of the board that will become the rack.

I'll be using 1.5 mm pins in the backs of the keys, so I want the rack slots to be 1.6 mm wide. This provides a minimal bit of clearance that avoids having the pins scrape up and down in slots that are too tight. The old makers made their rack slots with a dovetail profile so that the pins wouldn't jam inside the slots, the idea being to guide the pins with as small a contact surface as possible. Another way to accomplish this is to drill a bunch of oversize holes just shy of the front face of the rack:

These 3/16" holes were located by temporarily clamping the rack piece to the key frame and putting tick marks along the face at the midpoints of the key ends. The holes are drilled about 1 mm from the edge.

Next, the actual slots were routed with a 1.6 mm router bit and the table tilted at 8 degrees. No ordinary router table is capable of tilting, but the Shopsmith I'm using can do so, luckily for me. The slots must be angled because the back of the keyboard isn't square to the sides.

Once the slots were routed and residual fuzz cleaned up, the rack was glued to the back of the key frame:


Here it is with the glue dry and 2 screws (not seen) securing the left and right ends to the key frame:


Note that I've already glued the decorative maple arcades onto the key fronts. These are a little wider than necessary so that when the keys are cut apart the arcades will be trimmed flush with the sides of each key.

Lastly I have to transfer the exact locations of each rack slot back onto the rear edges of the keys so that the drill holes for the rack pins will let each pin enter the appropriate slot straight on.

Keyboard: Natural covers complete

All the natural key covers have now been glued on, and the left side of the keyboard has had a little wood grafted on to correct my miscutting of the key panel:


The last couple of keys in the treble were slotted at the backs of the tails to provide a little room for creating the balance pine mortises. You can see the two lines for these mortises marked in pencil and labelled with a natural and sharp:


Most of the mortises will not interfere with the key backs, so only a few slots were needed.

Keyboard: Natural backs

Now it's time to glue down the natural back portions, but this isn't quite as simple a task as gluing down the fronts was.

It all boils down to the fact that the typical keyboard octave is divided into two unequal parts of 3 naturals (C-E) and 4 naturals (F-B) in front, while the back is divided into 5 and 7 parts, respectively, when the accidentals and naturals are considered together (C-C♯-D-E♭-E and F-F♯-G-G♯-A-B♭-B). Thus the rear part of the keyboard has a width of 3/5 of a natural in the lower part of the octave and 4/7 in the upper part: a paradox, it would seem.

What ends up happening depends on how one approaches the process of gluing on the natural key covers. It's possible to glue down one-piece rectangles that are as wide as the key fronts, reaching all the way to the back of the key and overlapping where the accidentals will later go. One then draws out the keyboard design using a keyboard ruler, which is basically a pattern showing where the accidentals are. The backs of the naturals automatically appear in the spaces between the accidentals. The keyboard is then sawn apart, the wood covering the accidentals is removed—a reversible glue like hide glue must obviously be used—and the sharps are glued on.

The second option is to use separate key fronts and backs, and let the act of gluing down the backs automatically leave space for the accidentals. To do so, a few observations on the general layout of keyboards are worth noting:

• The front and back of C are flush along their left edges (the same is true for F)
• The front and back of E are flush along their right edges (the same is true for B)
• The middle of the back of D is aligned with the middle of the front of D
• The middle of G♯ is aligned with the middle of the gap between G and A

When these facts are digested, the layout of the keyboard using separate fronts and backs is largely reduced to finding the proper position of the backs of G and A such that the F♯, G♯ and B♭ keys aren't hemmed in too much by the adjacent naturals. If the key backs are all identically sized, the paradox described earlier is resolved by letting the gaps between neighbouring keys vary slightly in the two parts of the octave.

To be on the safe side, I took some of my sharp stock, which was purchased in the form of a stick of ebony already milled to the correct cross-section (narrower on top and wider at the base), and cut myself 5 sharps about 2.5" long. Then I temporarily attached natural backs against the fronts in various positions with double-sided tape and checked to see if the sharps fit correctly. The C-E portion of the octave was fine; F-B needed some juggling. With a ruler accurate to 0.5 mm, I made notes of how far each back was from the right-hand edge of the corresponding front for the notes D, G and A.

Here's the setup used to glue the backs:


This is largely the same as that used for the fronts, except the triangle used to align each piece now has the ruler taped to it to help take measurements:


The green piece of tape has the name of each note written on it, as a mistake in laying down the backs would be quite serious. I'm using a thick cyanoacrylate glue to attach the covers, and there's no turning back once the stuff cures.

The far ends of the keys must match the spacing of the jacks, which is of course determined by the registers. This is easily done by laying the register along the line showing the position of the wrestplank gap and ticking off each slot:


These tick marks will be connected to the rear edges of the key backs with pencil lines showing where each key is to be sawed out.

Note that the final few key backs haven't been glued on yet. This is because the key backs come very close to the balance pins in the extreme treble, so I'll have to notch the top couple of backs slightly to allow the key to move without interference.

Keyboard: Layout and natural fronts

The key frame, key panel and rack have been sitting around since last autumn, and I really need to complete the keyboard, as I need to know how high it sits in the case before I can cut the nameboard to size and apply moldings to it. So I've temporarily allowed myself to get sidetracked.

The key panel needs to stop a bit short of the rear of the key frame to leave room for the rack, a board standing on end with vertical slits in it that will guide metal pins sticking horizontally out of the back of each key. In front, some space for decorative arcades glued to the front edge of each key must be accounted for. The natural key covers overhang these arcades a bit, and the key fronts need to end short of any case moldings along the front bottom edge. After factoring in all these requirements, I cut the back edge of the panel to its final size at an 8-degree angle.

Next, various layout lines were drawn on the panel: a line showing where the front part of the natural key covers falls, a line showing where the nameboard will be, two lines for the balance pins that extend up from the balance rail through the key levers—the naturals have their pins slightly in front of the accidentals, hence the two lines—and a line showing the front edge of the gap behind the wrestplank.

The natural key covers are in two parts, front and back. I purchased these pre-cut and therefore had to spend some time sorting out the various pieces to try and match grain and colour. Here is one half of the total, properly matched:


Covering the key panel starts with the natural fronts. The gluing setup looks like this:


The ruler is exactly along the first layout line, and the yellow plastic triangle controls the side-to-side placement of the key front:


My natural fronts were made a little narrower than the octave span I'm using, so to make up the correct octave size I need to allow a controlled amount of space between each front. I used a feeler gauge to set a gap of 0.965 mm. By placing the feeler against a previously-glued key front, sliding a new front up to it and butting the plastic triangle against that, the correct placement for the next front is assured:


Here's the panel with all but one front glued on:


I left off the leftmost key front for the time being as I may need to extend the left key edge up a bit, having trimmed the panel to width a bit too closely. Oops! (but nothing that can't be fixed, fortunately).

Next up: the natural backs.

Moldings: Bentside hitchpin

After letting the molding stay in the form for a day, I unclamped it, mitered both ends and glued it in:


I didn't quite bend it enough to fit perfectly, despite trying to bend to a sharper curve this time around, but the additional curvature could be imposed easily enough with minimal resistance from the workpiece, so I'm satisfied.

Before gluing, I had to preserve the hitch pin marks made at the soundboard edge, since they will get covered up by the molding. Using a small square, I transferred them up to the top of the bentside, where they'll be covered by more moldings later on:


Later I will transfer these back onto the walnut molding and drill for the hitch pins.

This completes the walnut moldings, but lots more are to come, made of yellow cedar: on either side of the top edge of all the case sides, all along the outside bottom edge, and a cap molding glued onto the top of the case edges.

Moldings: Steaming the bentside hitchpin molding

I finally got around to working on the bentside molding at soundboard level, which I decided had to be steam-bent. Kerfing with a test piece of walnut looked rather ugly, even though it did allow the piece to bend. The problem is that this molding needs to be bent along its thick dimension, and doing so makes it twist out of flatness against the soundboard. My handscrews don't seem able to clamp the molding while it's trying to twist at the same time.

The steaming apparatus is a consolidation of information from numerous web pages discussing the art of bending wood. I used a 1.5"-diameter PVC pipe 7.5 feet long, capped on both ends with threaded covers. On one end I put a Y-junction into which steam flows from a boiling kettle. In several places I drilled through holes and installed a 2" bolt to act as a shelf on which the wood sits while inside the pipe. At the far end I drilled a small drain hole to allow condensed moisture to exit the pipe.

Here's the setup:


At the far end, the kettle spout nestles into the Y-junction and the connection is wrapped with aluminum foil to prevent the steam from escaping. Bricks keep the whole thing from tipping sideways onto the floor, while at the near end a tray collects condensed water from the drain hole.

After putting my molding into the pipe, it was simply a matter of filling up the kettle, plugging it in and letting it steam for about 20 minutes. The general rule of thumb for steaming is 1 hour per inch of wood thickness, regardless of width. My piece is not even 1/4" thick, but the bare minimum steaming time seems to be about 15 minutes for thin pieces. I gave it a little extra to be sure, then uncapped the near end of the pipe, pulled out the molding and quickly rushed it over to the clamping jig:


This is largely the same as that used to bend the bridge, except that the clamping blocks have notches marginally larger than the width and thickness of the molding. These contain the molding and keep it from twisting out of flatness as it is bent.

I'll let this sit in place for a day or so, then I'll see how it looks when it's unclamped. I screwed my clamping blocks along a tighter curve than when I bent the bridge. The bridge needed a little extra bending at gluing time and I want to minimize that with this piece of molding.

WARNING: in case anyone is thinking of trying to steam-bend wood at home, make sure your steam pipe isn't completely air-tight, or it might burst after it fills up thoroughly with steam. My drain hole provides some pressure relief, and I also have a few unused bolt-holes that I plugged with plumber's putty. In an over-pressurized situation the putty would get forced out and excess steam would safely escape.

The stand

Once again, several busy weeks got in the way of this project, and when work finally resumed last week, I wanted to do something that wouldn't take too long and would make a visible difference to the instrument, so I decided to make the stand. This also let me take the harpsichord off the saw horses it's been sitting on since last summer, which I've been tripping over from time to time in the crowded basement.


The legs come from Woodturners Incorporated, an outfit in eastern Ontario that sells turnings by Internet and mail order. I picked up 4 maple legs with a 2.25" cross-section, cut them down to 27" in length and joined them together in pairs with mortise-and-tenon joints, using two 3/4"-thick maple stretchers (the upper 3.5" wide, the lower 2.5" wide).

I connected the front and back pairs with two long stretchers, using 2" 1/4-20 bolts which thread into cross dowels embedded into the stretchers. I thought these would be safer than trying to put in threaded inserts, which might have split the thin ends of each stretcher.

The harpsichord is screwed to the stand with three 4"-long bolts that pass completely through the upper stretcher into threaded inserts set into the baseboard. This way I can pick up and move the whole instrument without it coming off the stand. Two dowels in the ends of the legs mate the instrument to the stand and align it with the screw holes for the bolts. It's amazing how many instruments I've encountered that aren't fastened to their stands. This makes repositioning the instrument an annoying task, if not a dangerous one: sometimes moving the instrument even a few inches is liable to tip the whole thing onto the floor!

I plan to peg the mortise-and-tenon joints to give them extra strength. Eventually the stand will be stained a dark walnut colour, which will match the bridge, nut and walnut moldings.