Rotola: The Nut

So far, I've talked about the rotola's construction fairly idiosyncratically, discussing what I'm doing as I'm doing it, but, apart from sharing pix of the original design, I haven't contextualized what I've done within the instrument overall or within the principles of chordophones generally.  We're coming to a point where a little of that background might be useful.  

On most stringed instruments, the string runs between two anchor points that keep the string in tension and across two contact points that define the length of the resonating section of string.  Thus:

Note that the  contact points are by definition between the anchor points.  This is true of all stringed instruments in one way or another, with variations in how their different jobs get done.  

There can be some further specialization, however, within these jobs.  One of the anchor points will usually have a mechanism for adjusting the string tension and one of the contact points usually has the additional purpose of transferring the vibrations of the string to a resonating body.  We name these different points according to their roles:  

The anchor point responsible for adjusting the string tension may be called a tuner, tuning pin, tuning peg, etc.  The anchor point that holds down the other end of the string, usually immobile, is sometimes called an end pin (in some instruments, instead of a pin, there is a hole into which a thickened part of the string catches).  The contact point that transfers string vibrations to the resonating body is usually called the bridge and the contact point that holds the other end steady is the nut.  Most string instruments have some version of all of these, although sometimes a piece might do more than one job, as in the bowed psaltery, in which the end pin also serves as a nut.  The most important job of the nut is to provide a solid point against which the string can vibrate, i.e., by being a rigid surface, it inhibits the string's movement as minimally as possible.  (This property is necessary for the bridge, too, but I would say it's first job is to transfer sound to the body.)

Whether the bridge and the end pin are located at the same end of the instrument or not is usually dictated by the structure of the instrument and how it gets played.  In violin, viol, and guitar family instruments, for example, the tuners and the nut are usually near each other and the bridge and end pins/holes are together at the other end.  But it can be the other way around, which is the case for the rotola:

More precisely, per my initial design:

So, as you can tell from both the end of the previous post as well as the title of this one, I'll be talking here about the design and fabrication of the nut.  

Despite the octagonal cross-section seen in the above initial design, as discussed elsewhere, I decided as I began to build the instrument to aim for a cylindrical shape instead; thus, the nut will need to be circular as well.  However, the grain of the wood comprising the nut, in this case walnut, will need to run perpendicular to the strings it supports.  As no trees grow with the grain in a circle (growth rings don't count, as the fibers are still longitudinal), the nut will have to be assembled similarly to the pinblock -- what I came affectionately to call pizza slices, as I struggled to get yet again perfect 22.5° cuts.  Using a bandsaw, this is tricky:

The one on the left, you see the cuts are too shallow and on the right, too steep.  These are just two of my many attempts to get the jig set right using scrap wood.  In the end, I finally got one close enough:

You can still see some small gaps, artifacts of having to use the oscillating belt sander to get the angles just right, but I had wasted a good deal of wood already, and my tools and my skill with them were not going to let me get more precise than this anyway, so I glued it up:

Next step was to round it off, which entailed first circularizing the center hole and then using that as a pivot to do the same with the circumference.  

I spent quite a bit of time laying out and fashioning a jig that would allow me to center and hold the nut on the drill press; unfortunately, I neglected to capture that work, but here is a shot of the completed center hole with the nut correctly placed on the jig:

The slots in the frame allow the jig to be bolted to the drill press table; the alignment marks have complements on the table and those on the nut then line up with the jig.  Using this system, I was able to place the 3/4" blade bit precisely in the center of the inner octagon and make a perfect pivot hole.  It's amazing what good marking up can yield.  

Next was to make a jig for circularizing the outside of the octagon.  Using a scrap bit from a practice axle, I set up a way to hold the nut firmly in place on the table of the oscillating belt sander sufficiently precisely at the radius I wanted:

The long edge allowed me to keep the whole thing square, which was the key to controlling the radius (from the center pin to the sander face).  Thus:

I was very pleased with the end result:

In working with the walnut thus far (this was the first part of the rotola that used walnut), I began to realize that, while it is a relatively hard wood, it's not as dense as a nut really needs to be.  My original plan was that the strings would be supported directly by the nut, draping across its edge (as you can see in the third figure, the one labeling the various string supports), but I began to think that the walnut was just soft enough that it could have a dampening effect.  Given that the rotola's strings should be highly resonant in order to create the sound I aurelize, this was a risk I was unwilling to take.  

I played with several ideas for solutions, including remaking the nut with ebony or maple on its edges or even fashioning a new one out of those materials.  However, in the end, I took inspiration from the bowed psaltery, which has a slim bar of brass on its bridge.  Brass is hard enough to support the steel strings rigidly, malleable enough to work with in a woodshop, and looks good, too.  

Having procured some 1/8" brass rod from my local Big Box Home Store, the next step was to mount it onto the nut's edge, most simply by cutting a groove into it.  As with much of this project -- indeed lutherie in general -- this would require making a jig.  Fortunately, I needed only modify the one I'd made to circularize the nut, standing it on its edge and mounting it to the router table.  After some trial-and-error, I came up with this:

A v-bit in the router could be raised (via the black and yellow handled adjustment shaft on the left) into the edge of the nut, then the nut could be rotated on its axle, creating a fairly straight and even groove.  Both the axle and the router bit are seen better from the below angle (seen from the other side of the fence): 

I was pretty happy with the result; it's not perfect, but more than good enough for a proof-of-concept:

Next, I wanted to taper the outside edge, giving a smoother transition for the strings on the end pin side (outside) of the nut.  After more trial-and-error, I came up with a setup that would allow some consistency:  

The rounded nut (here a practice scrap) rests on the bench, with a bench dog through its center hole (serendipitously the same size as the axle), then a guide keeps the plane at the same distance -- and thus the same angle -- relative to the nut; counting strokes and rotating the nut the same number of degrees as I worked my way around helped me rough in a taper:

These large facets (seen below in the nut itself) were smoothed by making smaller facets in between:

These were then blended and rounded over with sandpaper.  

Next step was bending the brass rod.  I expected that, in order to get a good fit, the rod should be wrapped around a cylinder that was somewhat smaller in radius than that nut itself, to account for springback.  I made a form, which turned out to be too large, then a smaller one -- 

-- which was still too large.  Rather than continue to make circles out of wood, I decided to look for ready-made cylinders, like cans of finish:

-- which turned out to be just right.  

I got the brass rod to a rough fit, cut it, and then taped it to the nut to see a) how well it would sit in the groove on the edge of the nut and b) how much pressure from the strings would be required to get it to sit straight and evenly in the groove.  

You can see some of the wobbliness, especially in the lower of these pix.  Masking tape was capable of holding the brass down fairly strongly, but I'm unsure if the strings (even considering that there are 16 of them) will generate enough power to keep it firmly in its slot.  That said, I decided those were problems to resolve another time and I needed to keep going to finish the nut.  

The last fabrication operation with the nut would be drilling the holes for the end pins.  This would require some accurate marking out:

Here, you can see I started by marking out the radii at 1/16th circumference (or 22.5°) arcs.  I'm using the seams between pizza slices as references, which, in turn, I will match up with the seams in the soundboard slats; this will then allow the end pins to line up perfectly with the tuning pins, which were also drilled in with an analogous method at the other end of the instrument.  Then, returning the nut to its jig, I used a compass to draw circles at 15mm and 20mm from center.  These intersections I then marked to indicate alternating drilling points, like so:

The places with the hash marks (or crosses) are to be drilled.  I drilled a few test holes in some scrap to see which of the candidate pins would work best (I had several ideas:  brads, screws, etc.) and settled on brass-plated twist brads (no pix yet; to be revealed later).  To keep the alignment consistent, I returned to the edge circularizing jig (the one I used on the oscillating belt sander and the router table), now mounting it onto the drill press and tipping the drill press bed 5° to give the pins a tilt away from the direction of the pull of the strings.  After drilling the holes, I ended with this: 

And that is as far as I'm taking this blog post.  I have since cleaned up the markup pencil and begun pre-finish sanding on the nut.  Still remaining for the nut is to be glued to the body, which I'm holding off on only because I have this nagging feeling that I'll regret it if I'm too hasty, and for the brass rod to get a final fitting (i.e., hand-bending to get as nearly perfectly circular as possible) and possibly to be glued (epoxy? cyanoacrylate?) in place.  

Next post I plan to be about the bridges, which I have just begun to experiment with.  After that, there will be very little left of the complex, inventive parts of this build:  I'll need to make a support for the instrument (basically a couple of Ys to hold the axle), glue up whatever isn't yet glued and apply a finish (very likely shellac), put the strings on and tune it, and make a bow (the design of which is still unclear).  

Rotola: An Instrument Takes Shape

In my last post, I focused on finalizing the basic frame of the instrument:  two blocks at each end of an axle.  With that complete, and having given up on further refinement of the soundboard slats' radiusing, it was now time to attach the latter to the former.  This post will cover those steps.  

Before I could glue the pieces of the soundboard to the frame, a few things needed to happen to them.  First, they needed to be fit, i.e., cut to final width (or more properly, degrees of arc) and jointed along their long edges so as ultimately to create as unified a cylinder as possible.  Once that was done, the soundholes needed to be cut.  

Jointing was done in several steps.  First, one edge was selected as most straight; this was then made as perfectly straight as possible by either sanding or planing or both.  Given the thinness of the soundboard the wood was brought to the cutter, rather than the usual other way around:  for sanding, I used sandpaper attached to glass and, for planing, my #7 Stanley jointing plane was placed upside down in a vise and the piece of soundboard was moved along it.  Care was taken that the edge being jointed was perpendicular to the tangent of the cross sectional arc (or coplanar with the radius) so as to join as tightly as possible with the adjacent slat.  Hopefully, the below diagram clarifies this statement:

That done, a wheel marking gage was used to mark a parallel edge just a bit wider than the final width, e.g., given there were eight slats, each slat would be 22.5 degrees of the cylinder's circumference, so the edge would be marked at 23 or 23.5 degrees; translated into millimeters at the outer radius of the pinblock, this was something like 45mm on the wheel gage -- I don't recall exactly.  The excess would be planed down, then, if necessary, sanded to fit precisely with the adjacent slat with the same method as the first side.  Here I'm test fitting two pieces of the soundboard, taped to the frame:

Next, the eight soundholes needed to be cut.  This would normally be a tricky operation, made more difficult by the curve of the soundboard sections.  A typical way of doing this would be to drill holes into the center of each strip and then connect them by cutting with a coping saw or knife, risking splitting the thin and delicate spruce, requiring much clean up, and making it difficult to get them all identical.  Alternatively, one could make plunge cuts with a router or router table; this would require yet another jig -- probably a fairly elaborate one -- and, again, the curve of the soundboard would make it difficult to keep the soundboard stable during the operation.  

After much deliberation, I decided instead to route out half-holes on each side, which would then create soundholes between slats.  There were several advantages to this:  it would be a very simple operation on the router table and require only stops on the fence, rather than a jig, it would be simple to keep the soundholes neat, straight, and centered, since they would register to the precisely jointed edges rather than a curved face, and it would reduce the surface of the edges being glued (and thus save time during a tricky glue-up).  

I spent a fair bit of time testing with scraps and getting the table and fence set up and then proceeded cutting sixteen half-soundholes, two into each of the eight slats.  Rough off the router table, they looked like this:

As you can see, the routing left some rough edges, especially at the ends of each hole, but these were easily cleaned up with sandpaper.  Too, there were a couple of holes I screwed up, one in which the router bit climbed out of the intended hole and bit a little too far into the spruce and one where the bit chipped out the edge of a hole.  I managed to repair these passably well by cutting tiny bits of spruce, matching the grain as best I could, gluing them carefully into place and the filing and sanding them to the correct outlines of the soundhole.  

In the end, I was pleased with the overall results and, if I go with this soundboard and soundhole design again, will probably use this method to cut the soundholes.  It was a bit fiddly, but I think less so and much easier to make consistent than the two methods I described above.  

So, now I was finally ready to begin gluing up the body of the instrument.  I had numbered the strips to keep the jointed edges together:

Six of the eight are shown here.  Number one is off camera, either taped or already glued to the frame; number eight is yet to be finalized, as I decided that would be easiest to do once the rest are on.  (You may notice in the first pic above that the pinblock sections are numbered as well; they correspond with the slats.)

Gluing arc-cross-sectioned strips of wood to cylindrical hunks of wood using square-faced clamps was not easy.  I tried several configurations, one of which was this: 

Note how the two clamps on the right cross each other, each taking a basically diametrical orientation, which is the only way a flat clamp can get any purchase on a cylindrical surface.  The one on the left is serving to bolster the rope holding the soundboard down, as I could not get enough radial pressure with the rope alone; that experiment was not repeated as subsequent strips were glued.  In the end, I solved these problems by making custom, cork-lined clamping cauls (excuse the accidental alliteration); although I don't have pix of these glue-ups, the cauls show up some of the next steps.  Here is the instrument with more slats attached:

You can see the clamping cauls being used as supports there.  Next is a view of I believe the same stage (six of the slats glued) showing the soundholes:

The end result was not what I had hoped:  most of the slats had radii tighter than the cylinder they made up, which meant that, rather than being perfectly round, the tube of the instrument was slightly lumpy.  So, instead of having a cross section like this:

It's more like this:

This effect was further exaggerated by the need to scrape and/or sand some of the glue joints.  It is, however, largely an esthetic concern at this point.  Unquestionably, the unroundness will affect the sound of the instrument, but my goal here is to see if it will simply hold together and produce something like the sound I have in mind; presuming it does, I can experiment with the effects of structure on timbre on later iterations.  Presuming I do so, I have some ideas as to how to increase the chance of ending up with a nice, smooth, round cylinder, if that seems like the shape to pursue.  

One other outcome of this stage was learning something about the acoustics of a stopped, cylindrical soundboard.  The tap tones for the individual slats were decent; they were clear and reasonably sustained.  As I added each strip to the frame and glued them into an ever larger resonating surface, they retained this quality and I was increasingly excited about how the whole would sound.  However, after gluing in the final piece, the soundboard deadened dramatically; it is now much stiffer than I expected.  It has some resonance, but nothing like I had hoped.  In retrospect, this makes sense, but I'll share my speculations on that elsewhere.  Ultimately, this design -- a capped, tubular resonating body (as opposed to an open or stopped-at-one-end chamber, like a marimba) -- may not work.  I expect to experiment further with it later on.  

With the body of the instrument assembled, it was time to drill the holes for the tuning pins.  My original design (along with being octagonal and having both the soundholes and the bow at the other end) had only eight strings.  As a reminder:

Back when I was assembling the pinblock, I realized I had grossly overestimated the space I needed for the pins.  My ultimate vision for this instrument is of something much larger, with, say, 30 or 40 strings; being able to put more strings on this prototype would allow for a better proof-of-concept, so I decided to set it up for 16 strings rather than the original eight.  

After doing the layout for drilling the holes (which can be seen in one of the pix below), I needed to create a drilling jig.  Not only do the pinholes need to be perfectly radial (i.e., run from the surface of the cylinder exactly toward its center), but they also need to be tilted a bit along the longitudinal axis to counter the pull of the string; I decided, more or less arbitrarily, on five degrees.  (You can see this tilt in the pic above of the original design.)  To accomplish this, I repurposed the clamping cauls and set up a jig on the drill press thus:

The carriage bolts secure the jig solidly to the drill press table and the extension helps stabilize it.  In the next pic, it's easier to see the five degree rake.  

Setting up and doing the actual drilling was quite nerve wracking:  if I screwed this up, I'd have to scrap what I'd built so far and start over and, having spent more than two years on this project already, that would be a bad thing.  So, following the maker's adage "measure twice, cut once" -- or in this case, measure many times -- I forged ahead with a trial hole, which, much to my relief, came out exactly as planned.  Testing it with a pin: 

You can see the layout marks for the pinholes here; "x" marks the spot!  The first of sixteen holes being successful, I plunged on (sic) and completed them.  

The next step is to create, assemble, and attach the nut:

The nut also has had some significant redesign, but I will cover that in the next post.  

Rotola: Compromises and Progress

My last post about the rotola focused on the production of the cross-wise bent strips of spruce that would ultimately comprise the cylindrical soundboard.  Of the 14 months since that post, 10 were spent in frustrating, off-again-on-again experiments trying to get the right radius in the strips without cracking them.  After gaining some confidence with scrap pine, the Sitka spruce that I had been husbanding for the "final" steaming ended up with more strips cracked than not.  Swallowing my pride and scratching my head, but with undimmed determination, I brainstormed with a luthier at the timber supply company I like to go to about new approaches.  We decided to try thinner stock (the original stock was 4mm; the new is 3mm), which helped with the cracking, but now my molds were too tightly radiused (the thinner stock rebounded less after drying).  My near obsession with getting the radius perfect became a running gag in my family and, eventually, I found that even identically milled and carefully steamed and molded strips varied too much from one to the next to produce a consistent radius using steam bending alone.  

In the end, I threw up my hands and used a heat gun to get each strip to something close to right radius.  As most of them were too tight, flattening them with heat and a little water worked to a degree -- although I snapped a few from too much confidence and too little heat and moisture.  When all was said and done, I could only get a few of them to exactly the right radius, so the cylinder I was ultimately able to make was far from perfect.  As I have said, though, this is a prototype and, as such, I don't expect it to be perfect (even if I want it to be).  

But I'm getting ahead of myself.  Let me pick up at the point where I decided to go with the strips I had.  Next step was to create an axle.  Just as a reminder:

(This original CAD workup was octagonal in cross-section; the current version is cylindrical, but the location and function of the axle is the same.)

Unfortunately, I did not photograph any of my initial work on this part (spaced it), but I will describe it.  The axle needs to extend far enough from the ends of the cylinder to allow for a support at each and a crank handle at one.  Obviously, it needs to be round (cylindrical) at the points of support, but it does not have to be so within the instrument.  In fact, I realized that the shoulders of a square cross section could help relieve some of the compression stress on the soundboard by allowing the pinblock and endblock to rest against them (if this is not clear, I do have pix of this aspect below).  Further, as the original mahogany dowel I had purchased to serve as the axle was not sufficiently straight, I decided I needed to make my own from scratch.  

After thinking about what type of wood would best suit as the axle, I settled on ash for its strength and straightness of grain and purchased a bit from a nearby timber supply company.  I started with some oak offcuts to practice my cutting and planing on and was able to produce some satisfactorily square cross sectioned blanks, then made the final from the ash.  Next was to cylinderize the ends.  Again, practicing on the oak blanks, I used a quarter round bit on the router table for this; once I had my setup right, I ran the ash through.  This process went surprisingly well, which was a great relief in the context of my struggles with the soundboard.  This was my end result:

Here is a shot of the major parts laid out together:

Note the endblock blank on the left still needs to be cut to a circle and both it and the pinblock need to have center holes cut to the correct diameter.  The soundboard strips have been taped together to get a rough sense of how they will join, but have not been properly fit.  

A close up of the endblock:

...and the pinblock:

Note the pencil marks aligned with the inside edge of each block; those represent the shoulders to be cut against which each block will eventually rest.  

Next step was to round off the endblock blank and to cut the center holes in it and the pinblock.  Again, I didn't get pix of cutting the endblock to a circle, but I used the same jig that I used for the pinblock (see the top of the last post).  That done, cutting the center holes had to be extremely precise, lest I build in a wobble.*  The process was very fiddly and I ended up getting absorbed in it and forgetting to photograph it, as well (I will, however, show in an upcoming post the very similar process of cutting the center hole for the nut).  Here are the results:

Endblock: 

...and pinblock: 

I was very happy with these results; despite my post hoc approach, the enlarged holes were concentric with the blocks' outer circumference. 

Next was to cut the shoulders of the axle.  Here you can see the slope left by the router cutter and the nicks I cut with the ryoba saw to mark the shoulders:  

I then chiseled in the shoulders roughly:

And:

Followed by more detailed chiseling:

I was then able to use the endblock and pinblock to finalize the shape for a nice, sufficiently tight fit:

Honestly, I'm pretty proud of the pinblock side.  

Next, I glued up these three parts to create the dumbbell structure to which the soundboard will be attached:

In order to keep these updates in bit-sized (or at least single-serving) pieces, I'll put the subsequent steps -- fitting, shaping, and gluing the soundboard, drilling the pin holes, and constructing and shaping the nut -- in the next post.  

* If I make more of these in the future, I will create the center holes first, then round of the circumference to guarantee concentricity, but the nature of this experiment meant that I ended up building the pinblock, endblock, and nut more or less from the outside in.