William Cumpiano's 
String Instrument
Newsletter #17


I get letters...lots of letters.
William R. Cumpiano 1999, All Rights Reserved


Hi Bill,
Last spring I completed my first classical guitar following the method outlined in your book. I took the guitar to a professor (and composer) of classical guitar at Vassar College here in Poughkeepsie. Overall, he said that it was very good for a first guitar and was fine for use as a student guitar. As Terry was evaluating the guitar, he played the notes at the 12th fret and then played their corresponding harmonic at the 12th fret. He said that the harmonic was a little sharper than its note. Terry (not a luthier) thought that maybe if I shimmed the saddle so that it did not lean in the slot, it might improve.  The saddle is so tight to begin with, that the most I could do was get a piece of fine sandpaper in the slot. So, the saddle is sitting as straight as possible now. (It wasn't leaning horribly before, though) Do you have any ideas as to why this happened? What do you do at set-up time to check for this problem? How can I prevent this in my future guitars?

Congratulations on your new guitar!

I presume you meant that ALL the strings played sharp, not just one or two. The problem may be in the bridge placement.

The saddle location may be a little bit closer to the nut than it should be. Measure from the front face of the nut to the center-point of the saddle. It should be twice the twelfth fret measurement plus a tenth of an inch (about 5/32"). If all the strings intonate sharp, measuring across the string array will reveal all the lengths are shorter than they should be, that is, with insufficient compensation, causing them to play sharp. The saddle would have to be leaning more than a paper's thickness for the problem to be caused by a leaning saddle. Short of taking the bridge off and re-gluing it a bit further back (or making a new bridge with just the saddle slot moved back, placing it in precisely the same place as the original) the easiest remedy is to just move the crown of the saddle back. Just file the top of the saddle, instead of to a rounded high crown in the center, instead, to an edge that coincides with the back face of the saddle. There is another remedy. It requires a lot of patience and a very sharp chisel, and a X acto blade. The idea is to widen the slot so it accommodates a fatter saddle blank. You would fit the fatter saddle blank (say, 1/8th inch thick) by scoring, then shaving some material towards the BACK face of the saddle slot. Thus it will accommodate a thicker saddle with the slanted-back saddle crown in turn allowing even more compensation length to the strings.


Hello sir, how are you? I had a question about the steelstring guitar in your book. With all things being equal, at what thickness would you suggest a cedar top to be at? I'm making a cedar, quilted mahogony version of the guitar in your book and I have the top down to approx. .145" .150". I realize that there are a lot of factors to consider, but as a generalization what "wood" you suggest?

Figure about 15% thicker than spruce. So if you're going down to .10 inch in spruce, go to .115 in cedar. For a mid size steelstring, .120" in cedar is a good starting point.


How to proceed from here? Is there a way to avoid this? Reshaping the fretboard while it is on the guitar looks to be painful (if, for no other reason than holding the guitar while planing/sanding is more difficult).

Touching up the slope of the fingerboard with the plane (edge or sanding plane) is the only way to insure a precise action geometry before fretting. The only way to avoid having to do this is forcing a much higher level of precision in the assembly and parts-making procedures than a person working in their home basement/ kitchen table can reasonably hope to attain.

You could for example use an aluminum workboard which is not only perfectly flat, but will not sag in any way when the back is roped on. I have improved the accuracy of this process since writing the book by stiffening the workboard with the addition of a massive beam down the centerline on the back side. The use of more precise machinery has also aided me in avoiding the kind of random problems which you are experiencing. But the advice in the book is still good: before fretting you must sand or plane the
fingerboard to remedy the accumulated errors which crop up during parts making and assembly. For that you must have a well-sharpened and adjusted plane and a way to secure the guitar while you plane it. You can imprison the guitar with clamps and curved blocks, or acquire a leather-lined swivel-jaw vise, that will hold the guitar by the neck securely.

During construction, the neck was always clamped to the workboard so I am not sure how the neck and top could have crept out of alignment. Assuming that some amount of misalignment will occur, is there some set of measurements that one can do that will indicate how to shape the fretboard before it is glued? For example, lay the straight-edge on the neck/top before the fretboard is glued on and measure the clearance at the 12-th fret?

You can certainly devise a system like this that will help. But your accuracy may still be degraded in the subsequent steps of gluing the fingerboard (you inadvertently force a distortion in the neck by how you clamp your fingerboard on during gluing) and so on. Touching up the fingerboard slope with the plane is the only way to fly for people at this stage in the game, 'seems to me.


I would greatly appreciate some words of advice. I was in the middle of assembling the soundbox of a maple steel-string when I was unavoidably held up - work related - and consequently the guitar went on the back burner for six months. The soundboard is fine but the back - as yet to be glued to the ribs - looks a little like Schwarzenegger's abdominal muscles.

The braces have held well, but with the very considerable change in humidity from a hot dry summer to a cold wet winter, the wood between the braces has rippled. Do I

a) scrap the back and make a new one
b) Steam off the braces and let it all settle down (another six months on the backburner )
c) Leave the braces on and let it all settle down.
d) Try to dry the back out a bit, using a hygrometer to assess the moisture content
e). Something else.

I'm not surprised. Maple--especially flat-sawn maple--is among the most reactive (greatest dimensional changes in response to environmental humidity changes) of all hardwoods (by the way, mahogany and Indian rosewood are among the least reactive). My guess is if your back rippled and bulged, its not only flat-sawn, and highly figured, but fairly thin besides. Lesson 1: use only well-quartered maple. Lesson 2: never start to assemble a soundbox and come back to it six months later. Finish assembling it and THEN leave for six months if you have to. You were cruisin' for a bruisin' all the way around.

It will not magically return to flat by itself. It'll probably get worse, and if the back is braced, its likely to start cracking besides. I don't have any good news. If the back is indeed flat sawn, get another back that isn't. If I guessed wrong and it ISN'T flat sawn, it probably distorted because you waited six months to finish assembling it. If that's the case, carve (don't steam) off the back braces, iron (yes, iron) the bulged plate flat with a warm, not hot, iron. A little bit at a time and on both sides, and mist a little water on it, like you were doing a shirt. Rebrace and mount on the guitar. Quick. Don't wait another six months.


Hello, Mr. Cumpiano,

A quick question: I am working through your great book on guitar building (excellent book, BTW). I am considering diverging on the body design, however. I am considering building a fan-braced steel-string. I am curious to know your thoughts on this...I had a great (albeit inexpensive) guitar some years ago that used steel strings and fan braces, and it had an unusual "classical-steel" sound that I'd like to replicate.

I don't want to take a great deal of your time; if the answer is involved, just say so, and I'll research it further on my own. To date, I think I have sized the braces and bridge-plate properly (thanks to AutoCAD stress analysis based on sitka spruce's stiffness numbers from LMI catalog). Just want to know if you've tried it & found it wanting....or if you think it's a worthwhile experiment.

There's nothing inherently wrong with using a non-x brace pattern on a steel string, as long as whatever pattern you do use is adequate to the stresses. If you have indeed "have sized the braces and bridge-plate properly (thanks to AutoCAD stress analysis based on Sitka spruce's stiffness numbers from LMI catalog)" you should definitely have a worthwhile experiment.

That is a relief...I was afraid that it would be a dead-end.

I would add, however, that I'm interested in that stress analysis that you did. What were the sizes that you found adequate, and what sort of inputs did you put in and results that you got that reassured you that they were adequate?

OK -- here's the process: First, I modeled the top with braces (based on my design -- 5 fans, two cross-braces, a classical-style bridge plate (albeit oversized), and a smaller curved brace on the treble side near the bottom of the lower bout (stole the idea from John Mello, whose interior top design I saw last week). I also added the glue (or rather, designed in a .001" layer between each brace and the top). After doing that, I assigned the specific properties of each material (stiffness-to-weight ratio, etc) to each piece, added the properties of the glue (relatively insignificant for stiffness, but essential to ensure torsion was correctly entered). Then I added 155lbs of string tension pulling up a virtual neck....I added endpoints, but didn't account for the leverage of neck on body; it seems to me that a properly joined neck/body probably doesn't affect the tension on the top that much. I may be wrong, and I may add it in next time.

Please clarify...

I defined the neck and its material, and how it was attached (i.e., that the surfaces of the joint don't move against each other in normal operation). I did NOT specifically define and measure a set of parameters that accounted for the leverage of the neck against the body when under tension. The neck did show some distortion (in the .0005-0010" range) in tension, but the body itself didn't deform appreciatively from the rim of the upper bout (neck joint) to the butt joint -- the top and back appear to provide some stiffness there. I figured that a correctly built body wouldn't become "shorter" from the neck to the butt, so I didn't measure it this time...another thing to measure.

The 155lb. measurement came from a set of D'Addario strings (12-58) that I commonly use. Each string was adjusted for the appropriate tension under load (this is one of the processes that accounted for a great deal of time).

Once all of that (it doesn't sound like much, but it took about a week of spare time) was done, I took the DXF to work and ran a stress analysis on it on a fairly high-end Sun box (had to steal the cycles -- this was assuredly NOT an approved project). During the process, I increased the height of the braces to about 16-18mm and narrowed them to about 6mm;

Why did you find it necessary to do this?

Because I started w/ braces that were far too heavy and far too short -- they were between 9 and 12mm wide and only about that tall. I wanted to see what a square-section brace would do under tension versus the conventional wisdom of a tall narrow brace. The tall narrow brace won out -- it allows more overall deformity ("bellies") but also a lot more 'freedom' for the top. I suspect that graphite composite might enable me to change the shape. What I'd like to do (another time....) is define a top braced only by increasing thicknesses of some extraordinarily stiff material -- don't know what yet. Another experiment for my next significant downtime.

The bridge plate (assuming maple, assuming the stiffness in the LMI catalog) came out to about 4-5 mm deep.

In order for what to occur?

The purpose of the bridge plate (to my understanding) is to provide an area of additional stiffness so that the stress load of the bridge (a small, very stiff plate attached to a larger plate of a less stiff material) will distributed over a larger area of the top. I kept adjusting it until I got a bridge plate that kept the top reasonably flat under tension -- I measured from a 'no-tension' state to a 'tense' state and wanted to see only a couple of mms of deflection (don't really know what's reasonable -- I settled on about 2mm distortion, which may be too stiff).

The top is dished in a 30' radius (in the plan) and this seemed to strengthen the areas that I was afraid would fail -- the area between the bridge and the soundhole and the areas beside the soundhole.

What indeed did you think was the effect of dishing ("doming"?) the top? How did you
propose to effectuate that dome on a classic bracing system?

put the dome in the top because I have read that it can help alleviate seasonal fluctuations. I proposed to do it by gluing braces curved to a 30' radius (less the top's thickness -- this is where CAD is useful -- I can print out what my braces look like) into a form that I think I can make with a router/scorp/concave spokeshave/scraper.

The design is depressingly conventional -- I didn't come up with anything particularly new. It did seem to confirm the current design, but there are some interesting dead spots, particularly near the edges beside the soundhole -- I would have thought that area would have been more live.

What do you mean by "dead spots"? Presumably this was a stress analysis, not an acoustical analysis... What sort of information were you getting that revealed "dead spots" near the soundhole?

The stiffness (or relative lack thereof) of the wood from the soundhole to the rims indicates that it won't vibrate as much as I had thought. It is lightly braced and would seem to be a prime spot to deform under tension -- and it does that. But it wants to turn into a wave shape as the tension increases and the top materials are made thinner -- so it looks like the 'crush zone' in a car. If it's doing that, then I'm guessing that it's eating energy. It might need more bracing if I want to make it more lively. I don't know yet what I'm risking by doing that.

You were correct; it was a stress analysis. I'm extrapolating results based on my experience with acoustic guitars as a player -- not as a builder. I'll have building experience once I get this thing done....then I can fuss with it a bit more.

I think the next guitar is going to be more adventurous -- maybe a lattice bracing a la Greg Smallman (or actually, Joshia DeJonge -- I played one of her guitars at Steve Swan's shop in Berkeley and it was GREAT), but I want to get a first acoustic done so I have a body of experience. I've built solidbody electrics, but this is a new thing -- it's cabinetmaking with extraordinarily delicate wood. Plus, if it sounds good, it'll be a different-sounding counterpart to my 1990 Santa Cruz OM.

I'm convinced that the effectiveness of the lattice systems has to do with spreading the tension evenly and in an undifferentiated manner across the top rather than dumping it onto discreet and restrictive beams. In this way the entire surface becomes a tension-diffusing "field" rather than a weak membrane propped up (and interrupted) by a relatively massive girder structure.

I'm also convinced that the top becomes a far more effective radiator in this way. The ideal direction to go, is to really discover what the absolutely minimal adequate structure is. Success lies there. This is why I'm so curious about modeling the stresses

I think you're right. I am interested in your experiments with the graphite top, since it's possible, if one's inclined, to mold bracing or additional thickness into the top, making it a unitary structure.

The concept of top as "sound radiator" is not one that I'd thought of in those terms. I've tended to see it as a plate that is braced by trusses to resist stress enough to create sound. That's sort of backwards....it's a computer geek's way of looking at it, I guess.

One thought: a bridge of some as-yet-undiscovered material that allows the builder to define a stiff web that attaches to the top across a variety of points, something like a dobro cone but more three-dimensional and more adapted to the lopsided stress of the strings. That would take some doing.....

Thanks for your time. I'll mail you the scatterplots when I get back home if you'd like....

Yes, I would. What's a scatterplot?

It's a graphical distribution that shows particular data plotted on an x/y axis. I used it to graph the stress encountered at different points, measured by deflection. I got the idea from one of Edward Tufte's books (If you don't have them, get them -- they're worth reading no matter what you do. The Visual Display of Quantitative Information, Envisioning Information, and Visual Explanations).
Thanks for your time.

Thanks for yours.