Jack tongues

The tongue fits in the slot routed into the jack body and is kept in place by a small axle pin. A mortise is punched completely through to receive the plectrum.

Holly, a fine-grained wood, was historically one of the commonly used materials for the tongues. I'm sticking with tradition and will be using the sheet of holly seen in the photo that also shows the resawed walnut the jack bodies came from.

First I cross-cut strips 28 mm wide from the sheet. Each strip can be imagined as a bunch of tongues attached together side-by-side, hence the need to cut across the grain. My tongue slot is 30 mm, but I feel the tongues should stop slightly short of the top of the slot so that the tongue doesn't absorb any of the repeated impacts of the jack hitting the jackrail over and over. That might bend the axle or break the tongue. The tongues will be slightly inset from the front face of the jack, instead of being flush.

Once the strips were made, I chamfered one edge at 25 degrees with a chamfering router bit. This edge will rest against the angled bottom of the tongue slot; the difference in angles between the two (25 degrees versus about 40 degrees) gives me some clearance that will be explained shortly.

Next, I needed to mark where the plectrum mortise would be. On some historical jacks, a small groove across the back of the tongue shows the location, and also thins the tongue to facilitate punching the mortise successfully.

I'm placing my mortise 9.5 mm from the top. That location was marked by making a 1/32" groove with a special small router bit. The groove is about 1 mm deep, reducing the thickness that needs to be punched through from 3 mm to 2 mm:


Individual tongues were sawn from these strips at the bandsaw. With a few strokes of sandpaper, the sawn edges were cleaned up.

A finished tongue:


I punched a test mortise on this one, and luckily it worked without splitting. More on making the mortises in the next post.

Jack bodies, part 2

Once all the jacks were planed to the correct thickness, I took them over to the strip sander and sanded the top and bottom ends clean, shortening each jack at the same time to a final length of 9.7 mm.

Next, I measured the thickness of the top and bottom ends. The process of hand-planing something small and short tends to pull the object up into the blade slightly, creating a microscopic taper. I found that most jacks differed by approximately 0.03-0.04 mm from top to bottom. I marked the narrower end with a black marker dot: this end now becomes the bottom of the jack, since it's marginally easier to slip it into the register slot.

Because I plan to use end screws to provide a little adjustability in the jack heights, I drilled pilot holes for 1/2" long #2-56 steel screws in the jack bottoms. This was done using a horizontal boring setup and a wooden rail to keep the jack parallel to the #54 drill bit:


Before the screws go in (once the jacks are completely finished), I'll tap the upper portion of the hole to help the screw get started, but I won't tap it all the way. This means the screw will tap part of the hole itself, which will keep it tight enough that it won't unscrew as the instrument is played.

After drilling, I chamfered all four edges on the bottom of each jack at the strip sander. This makes it still easier to slip the jacks into their register slots.

Now it's time to cut a slot for the jack tongue. The slot goes all the way through the thickness of the jack and terminates with an angled bottom so that the tongue's angled base can stop against it. This will allow the tongue to tilt backward, but will prevent it from tilting forward past the vertical.

Harpsichord makers usually use some type of circular saw blade to make this slot. I'm using a 3-wing slot cutter in a horizontal router table setup to cut a slot 3/16" wide:


To keep the jack from getting chipped, I made a zero-clearance table surface and plunged the cutter up through it. This supports the face grain of the jack and minimizes tearout at the end of the cut:


The walnut strip is a stop block that establishes the 30 mm length of the slot.

Before plunging the cutter through the table, I had to decide exactly where to locate the slot within the width of the jack. Each jack is 13.4 mm wide and the slot is about 4.8 mm (3/16"). I needed to leave room for the damper that will mute the string as the jack settles back down. The damper will slide into a thin kerf parallel to the tongue slot, which means the tongue slot should be a bit off-centre to leave room for this kerf. The simplest thing to do, I decided, was to subtract the tongue slot width from the jack width and divide the remaining width in thirds, with 2/3 assigned to the damper kerf position and 1/3 left over. These jacks will have the damper on the left, so a width of 5.8 mm is reserved for that. Next is the tongue slot at 4.8 mm, and 2.9 mm remains on the right.

The angled base of the tongue slot is made by setting the cutter height to terminate the cut at an angle of about 45 degrees from the jack face. As you can imagine, the higher the cutter goes, the more the cut angle approaches 90 degrees, so it has to be set relatively low. The angle is produced on the underside of the jack, as this photo makes clear:


Once everything was up and running, I found it necessary to use a push block to press the jacks firmly against the table. This minimized the vibration and chattering that the jacks had experienced when fed freehand into the cutter:


The finished slot:

Jack bodies, part 1

Now for a very critical part of the instrument: the jacks.

Last autumn I resawed some walnut I bought back on my very first lumber buying expedition in August 2007. The walnut was planed to a 4.9 mm thickness, with a projected final jack thickness of about 4.6 mm. After resawing and planing, I stickered everything in layers, under bricks to keep it all flat:


The top item is a sheet of holly 3 mm thick, from which the jack tongues will be made. The sheets of walnut are underneath.

Having sat around for about a year, all this material is quite stable, which is an essential basis for producing jacks that are to be well-behaved.

The first order of business was to slice the walnut sheets up into long strips slightly wider than the finished jacks:


These were stacked together on edge, a dozen at a time, taped together on the underside, and planed to establish the final width of the jacks (13.4 mm):


The jack slots in the register are a bit over 14 mm wide, so there is a clearance of about 0.6 mm. This is fine; in fact it could be a little more and still be OK: Skowroneck's book suggests that even 1 mm of clearance isn't problematic.

These bundles, still taped together, were cut down into individual jack lengths on the bandsaw. I'm aiming for a final length of 9.7 cm, so I cut to 9.8 cm to give me a little room to sand the ends and eliminate the roughness left by the bandsaw. The required jack length is actually 10.4 cm: the extra length will be provided by an end screw that will allow the jack height to be adjusted. I know that historical harpsichords didn't have this little convenience; it's the one place where I feel a modern screw could possibly be useful. A generation ago, horrible modern jacks were made that had far too many screws all over the place: see this web page for photos.

Here's a box full of jack blanks:


The most critical part of the jack body is its thickness; the clearance in this dimension is about 0.2 mm at most. Too little and the jacks might rub in the register slots during the dry winter months; too much and the plucking of the strings will be inconsistent as the jacks wobble around.

I suppose one could thickness jacks by machine until the required dimension is reached, as I did with the edges. However, a machine-planed surface isn't completely smooth; under raking light a washboard-like series of ripples can easily be seen. It's best to hand-plane the jack faces, since the hand plane gives a completely smooth surface without ripples. An alternative might be to thickness-sand instead, but sanding tears wood fibres and mats them down instead of cutting them cleanly like the plane does, and these fibres might decide to stand up again sometime later, compromising the smooth surface. Since the jacks have a more generous clearance in the direction of their width, I don't think there will be any trouble leaving the edges machine-planed. They feel smooth, even if they aren't on a microscopic level.

Here's the setup for hand-planing jacks to a controlled final thickness:


Two hardwood rails are screwed to a plywood board. Each rail has a groove with its base exactly 5.0 mm above the plywood. The hand plane seen at left slides in these grooves. A jack is held in place between the rails as shown below:


Scrap wood pieces keep the jack from shifting sideways or backward as the plane rides over it. These scraps must obviously be thinner than the finished jack so as not to interfere with the plane. Note the white paper shims inserted under the jack: these are used to raise the blank up each time the plane cuts away the top surface.

The hand plane is a Veritas low-angle smooth plane with a 38-degree bevel-up blade. The blade is bedded at 12 degrees, yielding a cutting angle of 50 degrees (York pitch, for the plane experts out there). This yields a smoother surface than the usual 45 degree cutting angle, at the expense of more physical effort to push the plane.

A well-adjusted plane should be able to take off a shaving just one thousandth of an inch thick:


To thickness a jack with this setup, a jack blank is put in place and is planed until no more shavings come off. Then a paper shim 0.07 mm thick is put underneath and the jack is planed again. Next, the jack is turned end-over-end to keep the grain angle at the surface consistent, and the other face is planed and shimmed a few times until the correct thickness is reached:


The final jack thickness is about 4.6 mm. The register slots are about 4.76 mm, and the wiggle of the planed jacks within the registers seems right to me: there's just a little bit of play.

Two registers full of jacks:

Chipping up to pitch

Chipping refers to a rough tuning designed to get strings up to pitch without worrying about complete accuracy. Now that the instrument is strung and the nut is pinned, it can be tuned for the first time.

There's no completed action as of yet, so the instrument can't be played in the customary manner, but the strings can still be tuned to an electronic tuner by plucking them with a toothpick.

Many, many tunings will be required to stabilize the harpsichord at pitch. The wire stretches out a great deal in the beginning, which affects the tuning stability at first. Brass wire takes several weeks to develop its proper sound: at first it sounds quite dull, but eventually it acquires a kind of high-frequency sizzle that, to me, is the hallmark of a good-sounding string.

Nothing broke during the first tuning! So far so good...

Pinning the nut

At least one register pair needs to be in place to pin the nut. I've installed one and wedged it place so it doesn't shift from side to side:



Here are the tools needed to pin the nut:


From top left: 1.2 mm bridge/nut pins, Dremel Stylus with #57 drill bit installed, marking awl made from a nut pin installed into a dowel, and a marking jack with two pencil lines showing the proper string spacing for the wide string pairs (10.75 mm).

With the exception of the marking jack, the exact same tools were used to pin the bridge.

Pinning the nut starts by installing the first nut pin a known distance from the case edge. When pinning the bridge, I put the lowest bridge pin 37 mm rightward of the spine, so the first nut pin must match that position. Here it is:


The leftmost string is now in the correct position.

Next, the marking jack is dropped into the leftmost register slot and the wedge is adjusted until the side-to-side position of the register brings the left pencil line on the jack into alignment with the leftmost string. Now the register position and marking jack together will ensure the correct spacing of the remainder of the string band. Since the spacing of the register slots is 13.75 mm, and the pencil lines on the marking jack are 10.75 mm apart, the narrow string pairs end up 3 mm apart, as desired.

Each register slot can be used to pin two strings by matching them with the left and right pencil line positions, respectively. Pinning is as simple as catching the string with the marking awl and pushing it leftward until it matches up with the appropriate line, as shown in the next two photos:



When the string position is satisfactory, the awl is used to make a dimple in the nut:


Since the awl uses an actual nut pin, the thickness of the pin is automatically taken into account.

Next, a hole is drilled at the dimpled location and a nut pin is installed with the same pushing tool used when pinning the bridge. The tool automatically leaves a few millimetres exposed:


All that's left is to lift the string over the nut pin. Obviously the strings are pretty loose at this point to make them easy to manipulate: they've been tightened just enough to eliminate any visible slack.

Here is an overhead view of the process. Correctly pinned strings are to the left, unpinned strings and mess from drilling lots of holes are to the right:


Not only do the strings end up spaced correctly, but the string band is now parallel to the spine. Previously the strings all sloped slightly to the right as they reached the tuning pins. This was done to ensure they would gain some sidebearing once the nut was pinned.

A method such as this, done purely by eye, will yield some very slight inconsistencies in the string spacing, but this is accounted for when the plectra are cut to length and voiced.

When the nut pinning was complete, the position of the two gap spacers was rechecked. In order to keep out of the way of the jacks, each spacer must lie exactly below a close pair of strings. I moved one of them a little bit; the other appears to be correct. The spacers were made 3 mm wide to match the close pair string spacing.

Finishing the registers

The four registers need to be cut down in length and assembled together in pairs. Each pair is held together with spacer posts, which have a profile designed to fit into the grooved underside of the upper register:


These posts go not into the register slots, which are for the jacks, but in between them.

First, the registers are stacked in pairs, and to make sure the slots are vertically aligned between the two, blocks sized to fit snugly in a register slot are slipped through slots at opposite ends:


Spacer posts are installed into the bottom register at 5 places, and the entire assembly is drilled at these locations with a #54 drill bit. The hole depth is controlled so that the bit passes through both registers and makes a small dimple in the post. It is not possible to drill much further because the drill bit is pretty short.

Next, the assembly is taken apart. The dimple mark is used to align the bit correctly and drill the spacer posts vertically all the way through. The holes in the registers are then enlarged with a 5/64" bit.

This process creates holes for #2-56 machine screws in the spacer posts that are accurately aligned with clearance holes in both registers. The spacer posts are tapped with the appropriate tap from both ends, and all the components are temporarily screwed together with 3/4" brass machine screws to check the configuration.

Each register pair hangs from a pair of gap spacers which span the gap between the wrestplank and upper belly rail:


In other harpsichord designs, these little struts keep the gap from closing up due to the tension of the strings. Italian harpsichord designs don't need this kind of help; if they do, the design is fatally flawed. I'm using them just to hold up my registers. The upper register lies on top of the gap spacers and the remainder simply hangs down inside the gap, like this:


Each register pair is installed by screwing the spacer posts to the lower register and pushing it up into the gap from the back of the keywell. Then the upper register is slid sideways into the instrument through the spine window, pressed down onto the spacer posts, and screwed in place.

One final detail is this walnut cover plate for the spine window:

Stringing technique

And now for some actual hands-on stringing.

First, a generous amount of string is uncoiled and a hitch pin loop is made at the free end. The loop is a set of double helix twists, in which two strands of wire wind around an imaginary central axis and not one around the other (think of DNA to get an idea of what I mean). After a dozen or so twists are made, the free end of the wire is wrapped around the base of the emerging loop with a couple of straight turns.

There are numerous ways to make these twists. Professionals do so by hand. For fun, I've been using this gadget that fishing tackle makers use to make double helix twists on wire leaders. In the fishing equipment world, this loop is called the "haywire twist".

Whatever the method, the result looks like this:


The loop is slipped over a hitch pin and the free end of the wire is guided around the appropriate bridge pin and pulled towards the wrestplank. Since the wire is kind of springy, it wants to coil back up again and make a nuisance of itself. To keep it from slipping off the bridge pin, a hemostat acts as a handy little clamp:


Note that the hemostat isn't actually clamping the wire to the bridge pin: that would cause the wire to break once it is put under tension. Instead, the hemostat presses the wire down against the bridge, acting like an extra set of helping hands.

Next, the wire is drawn about 4 inches past its final position in the wrestplank and cut free from the coil. The free end is inserted into a little hole in the tuning pin shaft, and the pin is rolled forward to make the wire wind itself on. When the tuning pin is over the correct hole in the wrestplank, the pin is hammered down into the hole with a hammer and a tuning pin setter (basically like a glorified nail set, except the notch in the bottom end is as wide as the tip of a tuning pin).

The wire, as it leaves the tuning pin and heads out towards the soundboard, must be on the right-hand side of the pin. In addition, care must be taken to ensure the wire doesn't angle downwards too severely as it leaves the nut and approaches the pin. This downbearing, if excessive, will exert an upward force on the tuning pin and try to unseat it from the wrestplank. Ideally, the wire will meet the tuning pin at right angles to the pin, which means it exerts only a forward force. The tuning pin holes were drilled leaning back at about 5 degrees to provide some resistance to this forward pull.

The best way to control the downbearing is to ensure there isn't too much excess wire to wind onto the tuning pin. The wire coils on the pin can then be spaced widely or narrowly, as needed, to control how much downbearing there will be. If there are any problems when the pin is hammered into the wrestplank, the coil spacing can be adjusted with a screwdriver tip before the pin is twisted clockwise to put the wire under tension, which freezes the coils in place.

Here are all the tuning pins with their wire coils:


Notice that there is comparatively little wire wound on the pins. The wide spiral between the upper and lower sets of coils allows me to adjust the downbearing.

Here is a view of the instrument fully strung. Note the change in sidebearing for the lowest couple of strings:


Some hitch pin loops, up close:


The hitch pins lean back a bit to resist the pull of the wire. I did this by tapping each pin with a hammer and a short piece of dowel.

With stringing completed, I examined the spacing of the string pairs at the bridge. It should be identical to that of the register slots (13.75 mm). I found a few visible discrepancies in the spacing, which also affected the sidebearing when the strings came off the bridge and headed for the hitch pins. I don't think this has anything other than a cosmetic impact, but for the sake of consistency I figured out which bridge pins were wrongly placed by using a caliper set to 13.75 mm and checking successive pairs of strings to see which one was out. Then I pulled out each wrong pin with a small vise grip, redrilled the hole correctly and installed a new bridge pin.