wrestplank and soundboard

Sprayed the wrestplank with polyu, keeping the areas masked where it will glue into the frame.

Hammered in the pins; no splitting detected...

Taped up the soundboard to mask the gluing regions.  Notice that I forgot to mask the landing spot for the diagonal beam which is situated between the two sound holes: I'll have to sand away some finish later, so that the glue will stick there.

Now for the decorations.  Lots of very precise cutting.  I realized that I could help myself at layout time, by leaving the segments of the circles connected together until the end.

The metallic gold looks way better than the copper I tried first.  I'm glad I waited and obtained the right paint.

I used Titebond II to paste the figures to the black circles, and also to paste the circles to the soundboard.  It seemed to stick well enough to the polyu-finished plywood; hopefully the final polyu overcoat will seal everything reliably.

After placing the first two by eye, I realized that I'd be better off cutting all around the figures except for a small connecting region, so that I could preserve the orientation of the figure elements relative to their wedges of the circle.  I carefully painted glue onto each figure, sparingly so that it wouldn't squeeze out too much and glue the surrounding paper down.  Then I positioned the whole gold-paper wedge over its corresponding wedge of the black circle, pressed the figure into place to establish the glue bond, and then carefully cut away the surrounding gold paper and removed it, leaving the figure against the black.

I taped these laser-printed circles around the sound holes, to give me a guide for positioning the decorations.  Even so, my placement was a little sloppy, but close enough for rock n roll we hope!  If I get picky later, I might try a technique or two to ream out the soundholes and make them more perfectly circular: if so, that'll give me a chance to also sneak the holes sideways a bit, to end up centered within the space delineated by the decorations.  Or I might just leave it...

After the decoration is securely pasted down, but before the glue is fully dry in case some cleanup is needed, I cut away the connecting isthmae.

After smoothing out all the air bubbles and wrinkles, and gluing down some corners which were not staying down using careful spot-applications of glue, it's time for another coat of polyu -- and another 5 days of drying (the polyu is dry to the touch within an hour or two, yet it takes many days to stop outgassing solvent fumes enough to be brought indoors).

Next up, additional structural work on the case and frame, preparatory to applying the paint and finish.

the bridge, and my "mistake bead"

Here are the bridge-blocks, all glued into place.  Took parts of three days, two more blocks every half hour...

How it was done.  As mentioned, I first drilled pilot holes in the oak moldings, using the #63 "wire guage" abrasive drill bit (so I don't know if final diam would be same with a #63 regular bit, I'm guessing not).

Coated with 3-4 coats of polyurethane, top and back surfaces only (not bottom)...

Bridge pins hammered into place.  Brass-plated steel "escutcheon pins", #18 x 3/4".

Dicing 'em up...

Now, crucially, right after dicing them, I painted the side faces with Titebond; I hope this prevents gradual splitting.

I rigged up two different clamping systems, ultimately, which made the work go at a reasonable pace.  Each block I allowed to dry 1/2 hour before removing the clamp.  This is what the Titebond II label suggests, and I found it to be about right, conservatively, under these pretty-good drying conditions.  I *probably* could have taken the clamps off at 10 minutes.  So 30 was perfect.

One clamp I did with bent wire, a spring.  One arm of the wire presses down on the middle of the curve of the molding, then it bends around and the other arm of wire presses forward against the back sides of the nails, thus giving a torque as well as downward pressure.

These molding pieces are particularly hard to clamp down, because the nails form the most-plausible attachment point on the slippery (polyu-finished) curved slopes, but the axis of the nails is *almost* at the tipping point: not only are the nails close to the edge of the quarter-round, but even worse, the molding has a small bevel on its corner edge, presumably to help it fit flush when used in its intended purpose.  But that bevel means that pressing down on the nails, or in their vicinity, is very prone to make the darn thing flip out from under you.  Hence the need for slight torque.

The other clamp used a weight, pressing a wire guided by the straw.  The straw snugly "held onto" the two nails, and kept the downbearing force from the steel wire corralled close to the top of the molding.

Neither clamping system was optimal, both required quite a bit of delicate fussing to get just the right downward pressure, slight forward torque, but no sideways or twisting torques, and no translational pressure, at each new block position.  Some of my positioning is a little ragged, I think a couple creeped slightly due to the "bad" forces I mentioned, but hopefully it's close enough for Baroque n Roll.

The bigger issue, and you can just see it in the top of the two photos above, is that, well in short, I screwed up!  As I was marking out the bridge positions, I realized that I didn't have room for the lowest, longest note, E2.  Oops!  Heh heh.  This is the "fog of war", when it comes to designing and prototyping.  Yes, I had originally ensured that there would be space, but many of the exact proportions have been fluid as I worked out the exact bracing of the frame, etc.; and crucially, I had not double-checked my plans, given myself a "sanity check", in a while.  It would have helped if I had constructed the big "ruler" with all the string positions marked, long ago.  The problem would have been obvious, as I moved the ruler around to estimate where strings would go.  Lacking the ruler, I measured key note positions individually, but clearly not that longest one!  (Also a factor might be my little pocket tape measure, which can only do about 1 meter.)

Anyway, once I realized that I had experienced some "measurement slip", I had to make a decision.  With the wrestplank not glued in yet, and much of "that end" of the instrument still very much unfinished and undefined, I could have "easily" moved the wrestplank back (well, forward); I also considered doing this to make the "gap" wider, in designing the action.  But the widening of the gap was the problem, here.  The amount of translation forward to actually fit the correct E2 length, would have made the distance the high notes would have to span, too wide.  I don't want to actually lose notes, from top or bottom.  So, the way to keep the E2 is to do what every other stringed keyboard instrument does, allow foreshortening in the bass.  For me, it will just be the one note.  I could widen the gap some, and have slightly less foreshortening on the E2; alternatively, I could begin the foreshortening a few notes above, and gradually "blend" the tone down to the E2.  But I chose neither of these mitigating strategies.  If the E2 can't be "perfect", then I won't disturb the rest of the design to make it only-halfway.  There may be a rather-noticeable contrast in tone colour, between this bottom note and all the others.  If so, that'll be an interesting experiment in its own right.  But I'll still probably build future instruments with the bottom note fitting better!

Rather than taking drastic actions or simply starting over, I am choosing to embrace this mistake, as a learning experience for myself and hopefully for those who read this.  Some of the Native American traditional artists insert intentional "mistake beads" into their work, as a sign to God that they are humble and know they are not perfect: not trying to upstage God.  I like the idea of conversing with God through artwork, but I have never thought it necessary to intentionally put any "mistake beads" into my work: they happen often enough naturally.  I'm not pulling any punches, I'm trying to do it as well as I can, and any mistakes are "the real McCoy"; if you confuse me with God then you'd better have your vision checked!  So I proudly preserve this big little f-up, right there at the peak of my instrument, always reminding me to measure early, measure often.

soundboard decorations

My spraypainted-paper technique is going to work, but I think I will go with metallic gold instead of the "hammertone copper": the latter just looks too brown, I think it won't have enough contrast with the soundboard wood (which turns much darker and redder when finished -- which is going to look nice!).

And here are my bridge blocks, drying: still in a long strip, not diced-up.  I've covered the top surfaces with several coats of polyurethane.  Next, I will hammer in the nails, then cut up and sand the individual blocks, and glue them to the soundboard.  Having the tops pre-finished will allow me to tape them up to keep the finish away from the nails, as I spray-coat the soundboard itself.  Originally I was going to hammer in the nails after all the finishing, but my concerns over splitting have changed this sequence.  Chances of splitting should be reduced by doing all the hammering while the blocks are still in a contiguous strip; and if splitting does still turn out to be a problem, I'll be able to change course without having already glued the blocks down.

the bridge blocks

So, I am planning to use a very unusual design for the bridge on this instrument.  I am now very glad that I've built this monochord, because it is allowing me to uncover and correct problems with my design, before I commit changes to the real instrument.

Most harpsichords and similar instruments use a single curved piece of hardwood for the bridge.  This works pretty well.  Usually, the builder just calculates the correct lengths for all the C notes on the soundboard, forces the bridge to pass through those points during the shaping process, and then the rest of the points on the exponential curve just kind of end up more or less where they should be, using the "curve fitting" of the bent wood itself.  It's not deadly accurate of course, but probably close enough, I've never heard of anyone asserting that the notes in between the measured points ended up noticeably wrong, on any real instrument.  (The only obvious problem would be if they were too long, because then the strings would break when brought up to pitch; other than that, errors in position would only lead to very subtle changes in tone quality.)  Another probably-insignificant-in-practice issue with the curved bridge, is that when there are pairs of strings per-note, one of the two ends up being slightly longer than the other.  Again, as long as nothing wanders over the safe maximum length, there is no real problem here -- though in this case, I have heard some people assert that they can hear the difference caused by the different lengths.  This would be very hard to pin down, because another factor probably has a larger effect on the tone in most instruments, which is the plucking position on each string: this is nearly always different for the two strings in a pair.  Having a difference in tone is not necessarily a bad thing, of course, as long as all the tones blend together well.  Many times, the two strings of a pair can be turned off and on separately, allowing the player to get different tones for different parts of the music.  Thus, some might argue that maximum tonal contrast, not minimum, is the right aim.

In addition to setting the correct lengths for the strings and coupling their vibrations into the soundboard, the conventional curved bridge has some other secondary effects, which may or may not be important.  For one thing, it leads to a relatively large amount of coupling between strings of different notes.  On traditional harpsichords, most of the notes will be damped (forced silent) at any given time; but on my pandalon, it will be possible to lift all the dampers at once, thus allowing strings to vibrate sympathetically with each other.  A one-piece bridge would probably enhance this effect.  Also, it's important to recognize that the one-piece curved bridge provides a tremendous amount of straightening and stiffening guidance, to the soundboard.

But I'm not using a curved bridge.  Instead, I am using small individual blocks of wood, one for each pair of strings.  Thus, there will not be as much coupling between notes.  And the stiffness of the soundboard will have to be maintained entirely through other means (I have fitted some braces, we'll see if it's enough).

Each block carries two nails, which provide the string terminations.  Well, actually (I'll have to post a link here when I find it), someone did a study and it turns out that the slight difference in length between the horizontal-plane termination of the string, provided by the nail, and the vertical-plane termination, provided by the point behind the nail where the string rests on the wood of the bridge (the "peak"), is quite critical to the tone.  Larger separations between the bridge peak and the nail, lead to softer attack, longer sustain, and less fundamental in the harmonic structure.  Of course there's a sweet spot, certainly less than 1 cm, which varies due to innumerable factors.  Out of a hat, I picked a distance of around 0.25 cm, as much because that's where I could best drill the holes as out of any tonal concerns; but I could have made the distance smaller, close to zero, and after reading this paper I chose not to do that.  One thing I could do with my monochord, is experiment with other distances, but I am so far satisfied with the tone quality I'm getting.

Of course, my bridge design looked really good on paper, in theory; but once I started building the monochord, I could see just how narrow the bridge pieces would really be.  1/2", to be exact.  1/2" wide sections of 3/4" quarter-round oak molding.  So, they look like tall quarter-slices of lemons, standing on their cut edges.  The tallness has always been a concern to me: would the sideways force of the strings, which make a 15-degree turn at the nails and then proceed to the hitching point, put too much torque on these pieces?  So far, answer seems to be no, the Titebond is amazingly strong when a good glue joint is attained.  However, the other factor which the narrowness brings in, which I somehow didn't forsee despite its obviousness in retrospect, is that I've pretty well done everything I can to make these blocks as likely to split as possible.  Two long nails, side by side, penetrating almost all the way through a narrow slice of molding, right near one of its faces.  If it were't hardwood, there wouldn't be a chance in the world of this working.  But now I see how poor oak is, as a choice for this.  I should be using something with a very dense and "closed" grain: I think maple might qualify better.  Of course I drilled pilot holes, but they are significantly smaller than the nail diameters, both because that's the available choice I had, but also because I want the nails to be very tightly gripped by the wood, I don't want them coming loose with vibrations and humidity changes, etc.; so I am expecting to significantly compress the wood fibers as I tap in the nails.  But how not to split the wood?  Well, I have revised my order of events, in processing these bridge pieces.  I was originally going to cut them up first, and then hammer in the nails.  Now, I plan to hammer in all the nails while it's still a single strip of molding, and then dice-up the pieces.  Even so, they might split after some time, being under constant strain from the fiber compression near the nails.  So, I am planning to "size" the side faces of the blocks with glue, during the operation where I also glue the blocks to the soundboard.  If this fails, well, I guess I'll try maple or some other variety of wood, next.  I might also try impregnating the wood with polyethylene glycol ("PEG"), which my woodworking reference suggests is a good treatment for making small wood parts like this more stable and easy to machine.

first sound! (in a way...)

Here, I am marking out the bridge positions on the soundboard, using a "ruler" I marked off on these steel rods taped together (material from the pedalboard project).

The string lengths are all based on the scaling (maximum tension before breaking) suggested by the string manufacturer.  I don't have a reason to doubt the value, but it would feel a little funny to start stringing up the final instrument without having any experience with this new iron wire.  It'd be a drag to have made some error of a few percent way back when, which I then propagated through all my calculations and stringing charts: I might see each string snap in turn, as I tried to tune it up to pitch.  This wire is notoriously fragile anyway; I'm used to strong steel guitar strings (and they also break!).

To check out the tension situation, and also to test several of my design ideas which have never been tested, I built a "monochord".  Actually, it's a duochord, replicating the two-strings-per-note disposition of the real instrument.  I sized the length for E4 with this iron wire, which is the high E of a guitar, and will be the middle note on this 49-note keyboard.  54-some-odd cm, i.e., only about 10 cm shorter than a guitar E4, which is strung in steel.  The iron is supposed to have a much lower breaking tension than steel, about half, and so the fact that the optimal length for E4 is so close to the guitar scale, shows how far guitars are tuned below their breaking tension.  I.e., I'm going to be operating these fragile iron strings much closer to their breaking point, than guitars...  (Actually, to estimate steel max tension relative to the "harpsichord" string materials, I use the high G4 of a 12-string guitar, which is the highest pitch I know of being strung at 65 cm.  And they sound pretty good!)

I also wanted to test out my bridge and nut design, based on quarter-round oak molding.  More and more, I see that oak is not the right wood for anything on an instrument, even though it has been used in the past.  It is strong and yet machines easily, but the grain is too coarse and uneven.  On my bridge pieces, the molding will be diced up into narrow chunks, 1/2" long each.  Thus, they will be taller than they are wide, so stability and the torque of the string tension, pulling sideways as it does, are of concern.  Also, I think, splitting is so much of a concern that drilling the pilot holes for the bridge pins may not be enough.  Maybe I should "size" the bridge pieces with glue, before finishing.  Or maybe I should give up on oak and find some other hardwood in a convenient shape, such as maple.  I don't have a definite answer, but now I can play with the monochord and ponder these matters...

Glad I tested this out!  I tried to use the same narrow-guage brass-plated nails for my hitch-pins, as I am using for the bridge and nut pins.  But the hitch-pin takes a lot more force, and as you can see, the poor thing bent right over as I tried to bring it up to pitch.  I'm glad to correct this once, rather than 49 times.  A slightly heftier nail seems to do the trick; although the ones I have in that size are not brass-plated, so I may look for some which are to keep a consistent look to everything.

You can see that I am using "double stringing", which refers to the string passing around one hitch pin and folding over to provide both strings of a pair.  The tradition with harpsichords has been to tie a loop in the end of the wire (a minor art in itself), and give each string its own hitch-pin.  But I have seen double-stringing in many other places, where it seems to work fine, including in pianos under way more tension, and in my little 15-note zither, where I can verify the somewhat-amazing fact that the two halves of the string can be tuned to entirely different pitches, without interacting or "slipping".  In the pandalon, the two halves will be at the same pitch, so I think I'm OK.  And stringing the instrument will be significantly easier.

With the monochord tuned up to E4, I can press it against the soundboard of the real instrument, and get a reasonable semblance of how its own E4 may sound.  All the signs are good.  The sound is reasonably loud, louder than I might have hoped.  It is delicate and harmonically-rich, similar to a harpsichord or a guitar, *not* similar to a piano (which is a good thing).  And when the monochord is pressed firmly against the soundboard, the sound no longer seems to emanate from the monochord itself: instead, the locus of sound seems to migrate to the larger sound hole.  So the soundboard is active, and it seems to be doing what it is designed to do -- perhaps even more effectively than my carefully-managed expectations might have expected.

drilling the wrestplank and nut pilot holes

I drilled pilot holes in the wrestplank: about 1/8", where final holes will be 3/16".  I hope that the pilot holes, done on the drill press, will suffice to guide the hand electric drill, so that I don't have to use the drill press for the final holes, once the wrestplank is glued into place and coated with finish.

I marked the holes with Sharpie, through a paper template.

I plan to use these little brass nails I found, called "escutcheon pins" (if I remember the spelling right), for the pins on the crowns of the nut and the bridges, which will form the crucial termination points of the "speaking lengths" of the strings.  The plan is to drill pilot holes in the nut and the bridge, using the drill press for nice vertical holes, and then to tap the nails in with a hammer; hopefully the fit will be just the right tightness, to not require glue and also to not split the wood.

The problem was, the nails were smaller in diameter than any of my drill bits: like with most regular sets, I can go down to 1/16".  Below that, they usually start using other measurement units for the drill diameters, and, it seems, they become a rare specialty item.  What, you people never had to drill a small hole?  All the "usual suspects" had nothing below 1/16" -- even places with Dremel tools and small hobby-type drills and such.  Fortunately, someone reminded me to check Hardwick's in the University District of Seattle.  It was a little obscure even there, but I found a set of small drill bits, and then the staff very helpfully were able to locate a tiny chuck in a different part of the store.  The drill bits are weird, they are coated with abrasives, for drilling through tile and stone, but they seem to go through wood OK without clogging up; indeed, the abrasive coating seems to let me "keep on trucking" and drill straight through even though the spiral grooves are fully clogged with wood shavings.  (They had a more-expensive set of bits that were regular steel, but I thought I'd try these first.)

The drill press has pretty poor mechanical stability, and between that and the two chucks, there was significant oscillation at the tip of the drill bit: it was not properly centered, even after much fussing and adjusting.  However, the drill bit was so flexible, relatively speaking, that it had a strong tendency to wander, which I suspect would have been a problem even with perfect centering.  The oak molding I'm using for the nut and the bridges, is quarter-round in shape, so at the line of the pins, I am drilling on a slight slope.  Wander-city for drill bits.  So, it was necessary to make pilot holes (for the pilot holes), which I did with the corner of a small screwdriver.  Once guided by the starter holes, the drill bit flexed enough to go right where I wanted it.  Still, for the many similar and more-demanding drilling tasks coming up in the construction of the action, I think I will need a smaller and more-precise drill press of some sort, for these little bits.  Perhaps just a hand-cranked drill built into the right frame.

Now that it's drilled, I can glue the nut to the wrestplank.  The whole thing will get sprayed with finish, and then I will tap in the crown pins and drill out the wrest-pin holes to their final diameters.  (I hope that the finish doesn't clog the tiny pin pilot holes so thoroughly that I can't hammer the pins into place; at worst, I might have to re-open the holes using the hand-drill: a whole lot of tedium but not the end of the world.)

The pandalon will have 49 notes, meaning 98 strings and 98 tuning pins.  The tuning pins come in sets of 100, so handily enough, there are just enough extra left over (if I don't mess some up!), so that I can build a "monochord", to test out all the aspects of my bridge and nut designs, test out whether I can get away with double stringing without tuning problems, as I believe I can, etc..  This "monochord" will have two strings (a "duochord"?), tuned in unison, just like the real instrument will have.  I'll be able to verify the actual breaking tension of the iron wire I've got, and I'll be able to hold it next to an electric guitar and get some sense for how well this wire will activate a magnetic pickup.