survey of the "prior art"

Here are some pictures to illustrate how other people have built clavicytheria in the past.  I learn much from studying these examples, but in some ways much of what I learn is "negative", as in, here's what not to do!

This first image is the prototypical small clavicytherium, designed to sit on a tabletop.  This type probably pre-dates the more-common horizontal harpsichord.  Notice the archaic "partial" soundboard, and the bridge carved to suggest a bent tree branch.  Is that how bridges were originally made?  These small clavicytheria are graceful and attractive; since ceiling-height is not an issue, the tail doesn't need to be truncated, and so they often continue up to a dramatic sharp point, as in this example.  It's only when they try to get bigger and more complex, that the problems start...

Here is a fascinating one-off instrument, significant mainly because it is the only two-manual clavicytherium I have yet found (thanks to harpsichord-maker Peter Bavington for bringing this strange beast to my awareness).  It seems to be an assemblage of two smaller zithers, with keyboards added.  It's not clear to me what exactly the two keyboards do, or what the string disposition is.  Possibly, it is just single-strung, and the keyboards activate different types of plectra, such as quill and leather -- that's just my own guess.  Notice that the strings run horizontally, and the keys apparently lift organ-style trackers (the vertical rods in front of the strings) which operate some kind of jacks, mostly hidden under the "lightning bolt" jack rail.  Not exactly how I'm planning to do things... however, this instrument layout does have the advantage that it *could* have extremely long bass strings, going off to the side: it is not limited by ceiling height.

This is a Delin clavicytherium.  Albertus Delin was the best-known maker of clavicytheria, and his were noted both for good tone, and for light action.  The latter, in particular, is hard to obtain with an upright instrument, due to the more-complex action mechanism and less direct help from gravity (although Delin's action does use gravity, not springs).  These are typically 2x8 in disposition.  Delin used the old-fashioned method of simply having the registers protrude through the cheekpiece, to control the stops; modern copies often provide more-convenient stop levers above the keyboard.  Notice how tall it is; it fits well enough into the lavish surroundings of the Brussels museum where it's located, but for the average modern house, you'd either need a cathedral living room, or perhaps a Sawzall.  And in any case, if your area is at all prone to earthquakes, you'd probably better connect a guy-wire to the top: these large clavicytheria inevitably look alarmingly top-heavy on their spindly little legs.  I plan to do things differently...

The Delin above has no  lid (actually, looks like there may be hinges and the lid has been removed and stored separately).  When the lid is attached, as in the modern Delin copy below, the ungainly and top-heavy appearance is exacerbated even further, as you can see.  Notice that the center panel of the lid can be removed on this instrument, so that the silly-looking lid can be closed, but still let the sound out.  Just removing it altogether seems even better to me.  The sound is crucial, looks are secondary, but I find that when a design is really well conceived, it tends to look elegant as well as sound good.  A nine-foot-tall spindly giraffe of an instrument, which still has foreshortened bass strings of only around six feet long, is not quite hitting the mark in my book, even if it does have good tone and action.

A number of clavicytheria have been constructed in this symmetrical "pyramid" form.  It does perhaps look a little better than the off-center look of the Delins, and splitting the lid up thus alleviates some of that "barn door" appearance (while also trading off the effectiveness of the lid in directing the sound, for use on stage -- arguably not the usual purpose of these instruments).  However, it is necessary to either use organ-type rollers to re-distribute the notes left and right, with the lowest ones in the center, or, as in this one it appears, the strings must be placed with a slight angle to the right, so that the lowest bass string goes from the left end of the keyboard, up to the apex of the top, and then the rest fill in down and to the right.  Either way, the basic form-follows-function nature of the harpsichord is disturbed by these complications, for no obvious benefit other than appearance.  And yet, the appearance is still flawed in my opinion, because the instrument remains spindly and too tall.

Here is the only example I am aware of, of a clavicytherium which "reaches to the ground" (this was built by John Paul in the 1960s).  However, this was achieved only by having the jacks pluck the strings in the center.  That is, perhaps, fine for a lautenwerk (gut-string) instrument, but it will never do for the kind of bright, harmonically-rich tone I am seeking from my metal strings.  It is necessary for the jacks to be located closer to the floor, connected by "some kind of mechanism" to the keyboards above.  Of course, all the art lies in finding the right "some kind of mechanism"...  (It seems that John Paul was intentionally going for plucking right in the center; if he had been satisfied with a pluck more to one side, which could still substantially have that "plucked in the middle" sound, he could have extended the bass by a good 50% higher, still staying within the height of the average ceiling.  Clearly, long strings and Pythagorean length progression were not the priority here.)

Here's another lautenwerk clavicytherium, by Steven Sorli (that "o" should have the Nordic slash through it, whatever that character is called).  Again, we're back to the spindly strings-start-above-keyboard layout, though the shorter gut strings and the open-frame "harp" design mitigate this appearance somewhat.  If a single-strung lautenwerk instrument was what was sought, I wouldn't have any big changes to make, this design appears entirely adequate -- if not truly graceful.  (Not sure about the Sorli action design, however: I've been told that it involves short lengths of bicycle brake cable, to transfer the vertical motion of the keylevers to the horizontal motion of the jacks.  While certainly innovative and in a sense, clever, I find it hard to believe that such a mechanism can really operate with low-enough friction to compete with other more-conventional action designs.  It seems more like a quick way to "get the job done" using off-the-shelf parts and materials.  However, it works well enough to be quite playable, judging by videos, and it seems that Mr. Sorli has sold a number of these.)

---

To recap, the instrument I am designing will "reach to the floor", thus fitting good long bass strings within the approximate 8-foot limit of modern ceilings; it will (probably) have two manuals, controlling a string disposition of (possibly) 2x8,2x4; it will be festooned with multiple mutation stops, controlled by my flexible new machine-pedal design; and it will have a sustain pedal, along with the necessary unconventional action to accommodate this.  I am pretty sure there has been nothing quite like this, yet built!

more jack experiments...

Continuing experiments with different construction methods for the jacks.  I calculate that these will need to occupy a total width, each, of about 3/8", including the clearance to the next jack in the column.  The basswood I'm using is about 1/8" wide, so there's room for a bit of widening on each side, but *not* enough to use another segment of the same wood on each side to hold the axle for the tongue (i.e., as if the jack were made of 3/8" rectangular-section wood, with a slot cut in the middle, like conventional jacks).  I could potentially work with some wood (which I don't have on-hand at present) that was 1/4" wide, cutting out the (slightly over) 1/8" slot in the middle for the tongue.  In one of the experiments, I simulated this by whittling down two side pieces and gluing them on.  Other alternatives are having only one side piece (I have yet to get this design to work very well, for assorted reasons), or using aluminum as I did in the original too-large one.  This last alternative seems to be the best so far, in terms of ease-of-manufacture and in terms of reasonable stability of the plectrum.  However, none of these really have extremely confidence-inspiring levels of stability.  There's a lot of side-to-side play for the plectrum -- which I hope is a relatively less-critical direction, but any play has the potential to be problematic.  Rotation about the long axis (of the tongue relative to the jack stem; rotation of the whole jack in its guide is another potential source of, effectively, the same problem) is reasonably minimal.  And most critically, it seems that the front-to-back motion of the plectrum is well-defined and not compromised by too much play: the tongue swings back easily against the nylon "bristle" (return-spring), but when it comes forward it seats in a definitive position which seems quite stable and repeatable.  And when pressed downward by the string, the plectrum and tongue "dig in" and hold all the more firmly in the forward position, which will give a strong and consistent pluck -- as opposed to the string force tending to make the plectrum retreat backwards and the tongue begin to swivel, which would be exactly the motion we don't want.

It's hard to see the details, a diagram would probably help.  The aluminum moves with the tongue; a steel wire passes through for the axle; its ends are bent over and they cling in place to the sides of the aluminum piece, through friction (the axle moves with the tongue and turns in the stem piece, opposite from the larger jack where the axle was fixed in the stem).  A tab coming off the back of one side of the aluminum, is bent around the back to form a back-stop, limiting how far the tongue can swivel (to prevent it touching other strings or jacks during its operation).  And because there isn't really room "up inside" the bent aluminum for a spring, the way it was done in the larger jack, the spring is now located down below on the stem, angling out and upwards from the back face and hooking under the backstop tab (you can see the black duck tape that holds the ends of the spring, on the front face of this jack, below the aluminum).  This has the benefit that the nylon spring thus cushions what would otherwise be the impact of the tab against the wood of the stem, helping to reduce noise.  On the other hand, since the arcs of motion of the spring and the aluminum are opposite, there is a sliding motion of the tab against a small length of the nylon: which presumably causes friction though it's not a show-stopper, but more worryingly, may cause the nylon to abrade away.  If I persist with this design, I'll probably have to take steps to ensure that a smooth portion of aluminum contacts the nylon, not a sharp edge.  There may also be a way to extend the guide-wire, not present on this prototype yet (but visible on the large jack), so that the guide-wire and not the aluminum tab is what the spring pushes against.

the jacks

The first crucial element in my unusual action design, is the jack.  I have iterated through a number of radical designs already, all on paper, as I have gradually refined my ideas about this instrument, over about the past 5 years.  The points of commonality are that the design must be uncommonly cheap and easy to fabricate, *without* the well-equipped specialty woodshop (which is a non-negotiable pre-requisite for building almost any other serious musical instrument).  And, it must accommodate a sustain (damper-lift) pedal.

I sure have taken a lot of flack (well, good-natured skepticism, at least) from the ancient-instruments community, for my insistence on the sustain pedal!  But this is exemplary of how my philosophy and aims differ critically, from those of the majority of builders today, who are only concerned with making increasingly accurate copies of historical instruments.  I am glad for those builders and their painstaking research and efforts.  As well as making good instruments in their own right, they have contributed massively to the knowledge in the field, and although I do not hold myself constrained by the boundaries and practices of the past, I have liberally borrowed from the knowledge and ideas that I have found there, through the help of the aforementioned historically-accurate builders.  Lack of knowledge of (or respect for) the past seems to be the main flaw which led to the dead-end of the "revival" harpsichords of the early 20th century.  The return to historical designs was a welcome relief from those awful, piano-inspired revival instruments, but unfortunately, the lesson learned seems to have been "don't try anything new, it's bound to fail".  Even if this is the statistically-likely outcome, I still see value in persisting to try to find those rare and valuable new things which actually improve the instrument in a lasting way.  Of course, this view presumes that the instrument is not already perfect, and that there may be new sounds and new purposes for it in the future, which go beyond anything heard in the past, *without* destroying or debasing the underlying nature and beautiful tone of the instrument.

So yes.  We must have a sustain pedal.  This, in turn, means that the normal operation of the harpsichord jack, must be changed somewhat.  Normally, as the jack comes back down after plucking the string, the plectrum touches the string a second time.  It's not as hard a touch as the first time: instead of plucking the string, the spring on the tongue allows the plectrum to be pushed aside by the string as it falls past.  And right after (or even simultaneous with) this second touch, the damper mutes the string.  So, normally, the second-touch is no problem.  If it's audible at all, it's not objectionable, it just adds to that "harpsichord sound", a slight bit of noise upon damping.  However, if there's a sustain pedal, then the dampers are no longer directly attached to the same jack as the plectrum.  If the pedal is pressed down, then the string is not damped when the key is released, so noise from the second touch becomes more prominent.  Also, the second touch is likely to at least partially disrupt the vibrations of the string, set in motion at the first touch (the pluck).

(Many small lautenwerk harpsichords skirt past these problems, although having no dampers, because the undamped sustain time of the strings is so short that they might as well have dampers: the second touch doesn't disturb the vibrating string, because it's hardly vibrating any more anyway.)

In order to prevent second-touch, I attach a guide-wire to the tongue.  As the jack returns, the guide-wire encounters a blade-like sloped surface which causes it to pull the tongue back, just enough that the plectrum misses the string.  The guide-wires and the associated battens which hold the blades, are located in the space above the jacks.  Thus, there is no conventional jack-rail above the jacks; their upward motion will be stopped by other means, down below.

I built a prototype jack, to test some of these ideas.  My design has already changed, I will not build any more like this, but still the photos are probably helpful in describing the basic operation.

I used a bent piece of sheet aluminum for the tongue, instead of wood.  A nail (point clipped off) passes through horizontally, to form the axle.  The "plectrum" is the end of a nylon wire-tie (in actuality, I will probably have to use "real" plectrum material, i.e., the plastic called Delrin, because nylon wears out and bends and changes shape too much).  A hole is drilled at an angle, through the sloped front face of the wooden slip, and coming out the back.  A doubled-over piece of thin nylon monofilament passes through this hole, to form the spring.  It's visible, faintly, to the left of the plectrum in the photo above.  The two ends of the monofilament come out the back and are taped down with duck tape.

Up to this point, despite the unusual construction (considerably easier for me to fabricate than the traditional wooden jack), it still *behaves* just like a regular jack.  The spring is soft enough that the weight of the wooden slip is easily enough to make the tongue swivel and the plectrum return past the string, in the usual way: no hanging-jack problem seems likely.  Yet, in the upward direction, the angled plectrum catches the string and forces the tongue firmly forward to its stopping point, independent of the smaller force from the spring.  So I guessed right on the geometry.  This jack would work in a regular, horizontal harpsichord with no sustain pedal, just fine (except for lacking a damper).

But in the following photos, you can see that I've added the guide-wire to the top.  It is superglued into the corner of the tongue, on the other side from where the monofilament return-spring rides.

The guide-wire has an L-shaped bend with a horizontal segment at the top, folded over double at the end to keep the sharp end of the wire out of the contact area.  This horizontal stub is what encounters the 45-degree sloped blade.  On the way up, the sloped blade pushes the guide-wire forward towards the string, causing it to flex because the tongue is already against its forward stopping point.  On the way down, the other side of the blade pulls the guide-wire back, causing the tongue to swivel backwards on its axle, and thus pulling the plectrum back just enough to miss the string.

This is an exaggerated degree of bending, in situ it will just be a few degrees, enough to pull the plectrum back perhaps 2 mm or so.

As I mentioned, however, my design has changed since this one.  In order to fit all the jacks and dampers needed into the available space (the width of the gap between the wrestplank and the soundboard, about 2 inches, necessarily limited because the highest and therefore shortest strings are just barely longer than this), I need to make the jacks even narrower.  Also, the metal tongue is not desirable, because it doesn't give the right kind of seating for the plectrum (for this prototype, I just superglued it in there).

So in my next design, yet to be prototyped, the jack body will be square-section basswood, 1/8"x1/8".  A diagonal cut will separate the body from the tongue, which is just a continuation of the same wood.  A wire, firmly fixed to the side of the jack body, will go straight up along the side of the tongue, turn 90 degrees and pass through a hole in the tongue (this section of wire comprising the axle), then 90 degrees and back down; one end of the same wire will loop back up and form a backstop for the motion of the tongue, by standing in front of the lower, beveled portion which comes forward as the main part swings backward.

There will be a nylon spring, as in the prototype above, but the tongue will need to have an angled slot cut in its lower, beveled face, to accommodate the spring.

The upper end of the tongue piece will be drilled, front-to-back for the plectrum (I may use a specially-shaped hand-tool as a mortice-cutter to make these holes rectangular, after drilling the round pilot holes), and with a hole in the top for the guide-wire.

Thus, this jack will be like the "middle third" of a conventional jack: if you shaved off both sides until the whole thing was no wider than the tongue, and then added a wire to replace the axle whose fixed end supports you just shaved away.

(The bottom ends of my jacks will widen out into about the same width as a normal jack, to allow them to pass through rectangular guide-slots which will keep the rotational error to a minimum; the wider bottoms will also add weight, which will help them return when used in a horizontal context; the clavicytherium will of course have to address the issue of returning the jacks, more explicitly.  However, the wide dimension of the jacks, in the horizontal context, will go across the width of the keylever, so that the jacks still pack closely together on each keylever, separated only by their narrow dimensions; as opposed to normal jacks, where the long dimensions line up along the length of the keylever.)

An example of the useful knowledge I've gained from examining the methods of the past, is that I can make my own "drills" for drilling these small-diameter holes, by sharpening thin nails and pieces of steel music wire, or by using clipped-head pins and needles, depending on the diameter.  Initially I tried to use tiny storebought "hobby" drills, in an adapter in my drill press, to make holes like this; but it's entirely unnecessary to use spiral-cut drills *in wood* at such small diameters, and it only makes the drills frustratingly fragile and flexible.  A handheld electric drill, or even completely-manual means such as a bow-drill or crank-handle, are quite sufficient, given the right special purpose jigs to guide the drill direction.  (Very different terrain from drilling fiberglass PC boards, my other main area of experience with small drill bits.)

clavicytherium: an upright harpsichord

Here I will begin to document the design and eventually, the building, of a clavicytherium (upright harpsichord).  I am currently, mainly, involved with building my (horizontal) pandalon, which is a different blog (pandalon.blogspot.com).  But I'm designing the clavicytherium in parallel, and doing small amounts of prototyping to support the design process.  Hopefully, by the time the pandalon is finished, I'll be able to start work on the clavicytherium.

The entire design is of course subject to change, but at present, my thinking is to provide the maximum amount of harpsichord capability, in a small space.  Thus, the disposition must be at least 2x8,1x4.  But in fact, as of recently, I am considering a more-complicated design than that.  Two manuals, with 2x8,2x4.  Each manual capable of 1x8,1x4 (different strings and plucked in different positions), with a coupler.  Obviously, I am interested in providing *more* capability than has existed in the past, not in any way trying to exactly duplicate any instrument of the past.

In that same philosophy, I plan to include a sustain (damper-lift) pedal.  In fact, if there are two manuals, there will be a sustain pedal for each, arranged so they can be operated separately or together by one foot (the right).  There will also be a "machine" pedal, to allow control of the various handstops of the instrument by the other foot.  This machine pedal will be of basically the same design as what I am currently building into my pandalon (or an evolution thereof, more likely).

Other than the stops to turn on and off the choirs of strings, and the coupler, there will also be a number of "mutation" stops, i.e., things to change the sound.  My philosophy is, these are cheap and easy to add-in, so every possible option should be included.  At present, there are three which apply to harpsichord: buff, sitar, and bassoon.  Buff, also called the "harp" stop by some, is pads of leather or felt pressing against the strings near the ends.  It produces a muffled and shorter-duration sound, like pizzicato or hand-muting on other string instruments.  What I call, descriptively, sitar, is better-known to harpsichord people as the "arpichordum" stop, but this is not only opaque but simply incorrect by today's tonal standards (it means "harp" sound, but I've never seen a harp with this effect; apparently they commonly used to have it, which is strange to imagine).  Shaped metal wires are brought into contact with the strings, again close to their ends (in order for both the buff and sitar to have optimal string placement, one comes from beneath and one from above), giving the hypnotic "flangey" metallic buzzing of the sitar (and certain Japanese instruments).  Bassoon brings a curved surface of paper (or the right kind of "noisy" plastic, such as acetate) into contact with the strings, a little further out.  It creates a kazoo-like buzzing tone, similarly bright but different in sound from the sitar effect.

The bassoon stop tends to be constructed so that it hits both strings in a pair, such as both 8' strings in this instrument.  The sitar hits only one of a pair, so the other can be used to maintain some "normal" sound along with the sitar, or it can be turned off to emphasize the sitar more starkly.  The buff is normally constructed to hit only one string of a pair.  In my pandalon, the buff is like the bassoon and hits both strings, but this is a single-manual; instrument so the possibility to contrast buff and non-buff on two keyboards is not there anyway, and I wanted to ensure the ability to use the buff stop, controlled by the machine pedal, similar to the pedal-controlled dampers on concert hammered dulcimers, which were an important model for my pandalon.  In the clavicytherium, the buff will probably be more like a conventional harpsichord buff, with individual pads that hit only one string from each pair.  Probably all mutation stops which affect only one string will be tied to the upper manual.  However, if I find that there are two different mutation effects which I really want the ability to contrast against each other, one played from each keyboard, then I may make one or more of the stops such that it can be "assigned" to either keyboard (perhaps through some behind-the-scenes adjustment, not necessarily a quick change from a knob).  And finally, if I find after completing the pandalon and playing it for a while, that the hammered-dulcimer idiom of rapid buff-stop expression with pedal is quite important to me (even independent of the hammer-action dynamics of the pandalon), then I might conceivably fit two buff stops, to allow for one to be used expressively while the other (on the other keyboard) is constantly-on, via a handstop.

The overall structure of the clavicytherium, as well as many aspects of the action, will be rather unlike any others I've seen.  The most significant change, even more than the sustain pedal, is that this instrument will be designed to "reach the ground".  Meaning, the longest strings will have the full span available, from the floor to the height of the instrument.  This will allow good long bass strings, without making the instrument ridiculously tall.  Traditionally, clavicytheria have been build with the keyboards passing underneath the wrestplanks; thus, the entire height of the strings had to start above the level of the keyboard, going up from there.  Even with rather severe foreshortening of the bass strings, a full-compass clavicytherium can easily reach 9 feet tall.  More of a curiousity to stand in a corner of Versailles Palace, than anything that would be convenient for a regular musician with limited space.  With the strings not limited by the location of the keyboard, a significantly better instrument than those 9-foot-tall ones can fit in under 8 feet, which is about what I'm taking to be the maximum practical height for today's average ceilings.

(I am not the first to design a clavicytherium that "reaches the ground"; I have seen one example, from the 1960s, of a single-strung lautenwerk (i.e., having gut, or more likely, nylon strings) clavicytherium.  The strings reached "to the ground", but only by virtue of having the plucking point almost in the middle; i.e., the plucking point was still constrained to be near the keyboard, a constraint which I will do away with.  Plucking near the middle is fine and perhaps appropriate for a lautenwerk instrument, but it's not at all the type of tone that I am seeking from my metal strings.)