Official Luthiers Forum!

I know this is a huge question so any answers at all will be appreciated.

I've been looking at several plans for classical guitars and I have found that the size of top braces used in fairly similar fan-braced configurations vary a lot. Much more, relatively speaking, than what I've found for steel string guitars (well, except perhaps for the amount of scalloping). For example, on the plan I have for a Torres guitar, the braces are only 2mm tall. Other plans I have show them as being 6 mm tall. And there's everything in between - a few plans show them varying in height from the "bass side" to the "treble side".

So what's a fellow to do - i.e. what size bracing do you use for the tops on your classical guitars and what is your basis for using this size? Again, I realize this is not a simple to answer question, I'm just looking for anything that will help me decide what to try. My intent is to go with a basic 7 brace fan bracing.

Thanks,
Pat

_________________
There are three kinds of people:

Those that make things happen,
those that watch things happen,
and those that wondered what happened.

Copying a particular guitar's dimensions probably won't produce a good tonal copy, since you're unlikely to get exactly the same wood. A big piece of becoming a good Classical maker is coming up with some way to 'read' the wood, and figure out how to vary the top thickness and graduation, and the brace sizes and profiles, to get decent results with the wood you have. I think it's telling that many top performers on Steel string guitars play production instruments, while almost no good Classical players do. Use the plans as general guidelines, thickness the top based on your best guess about it's stiffness and mass, over build the bracing a little, and trim it down to get the top 'right'. Which only leaves the question of how you know when it's right.

Most makers seem to rely on either 'tap tones' or flexing and feel. Both of these systems can be made more rational with a bit of technology. Deflection testing, taking measurements with known weights deflecting the top, has been used successfully. David Hurd documented this in his 'Left Brain Lutherie', with a genteel sufficiency of math for those who like it. I use a 'tech' version of 'tap tones' called Chladni testing', which looks at resonant patterns on the 'free' top and attempts to control things by changing those. Any such system has to be built on an empirical basis: you compile data on what you did and what seemed to work, and try to draw correlations. Having numbers, either deflection maps such as Hurd uses, or mode shapes and frequencies in my case, make it easier to communicate these results, as compared with 'remove wood until the fundamental goes away' or it 'feels right'. In the end, though, there's no substitute for experience. No matter how you go about it, you'll probably have to make a few guitars before you start to get the feel for what works for you.

Just a quick note about the performers playing production instruments - these folks aren't typically playing an off the shelf guitar. Those guitars are more likely than not made in the custom shop and setup specifically for those performers. The best guitar advertising out there is to have a renowned player use your instrument...!

Trev

_________________
http://www.PeakeGuitars.com

I don't know that Al is speaking specifically of "big names."
I know an awful lot of musicians, pro and serious amateurs who play plenty of shows, and relatively few of those who play steel string play custom made instruments. (They do have their instruments professionally set up, though.)
On the other hand, very few of the classical guitarists that I've known who were even relatively serious students (much less pros) play factory made guitars.

You are right, this is a huge question. I have built about 25 classical guitars to date and I know what you mean. You look at various plans and they all seem to have a different approach. I have found that plans tend to be over built. So my general rule of thumb is to build lighter than the given plan suggests. I've had some very accomplished players tell me that my guitars sound nice. I also know that if you ever have to deliver a public speech and you know in your heart you completely blew it the audience none the less will tell you how great it was

It's not an easy question to answer.

The last classical guitar I built was a double top. It was an experiment. I just wanted to see for myself. I got a lot of good advice and helpful tips. When I laminated the top it felt to me that it was way too stiff. The Nomex core really stiffened this top but yet it was recommended that I still brace it 'normally' so I did. Was just playing this guitar tonight and am now considering furthering the experiment by first shaving the braces and after that re-topping!

So experience does help. Intuition is good. I knew this guitar would sound tight and it did!

The same goes when you are building a regular design.

You noted that the differences in brace thickness.
Also note the top thickness.

I have found that there are two (yes there are more than two, but they can be broken down to two) schools. First is the thin-top-big-brace-school. Second is thick-top--thin-brace.
Thick and thin are tough to discern because you are talking about, from the plan standpoint, the stiffest euro spruce possible. When thin it is 1.5-2mm thick. The thick stuff is 2.5-3.0mm


A half mm makes a vig difference on guitars built for nylon strings.

So, when analyzing those plans pay special attention to what role the originator put on the function of the top and braces.

dl

Add to this the amount of doming to further complicate the question. Some guitars, such as the Torres, had doming as high as 5mm. Others will have flatter tops, about 2-3 mm. Higher domes will be stiffer than flatter.

_________________
Expectation is the source of all misery; comparison the thief of joy.
http://redrivercanoe.ca/

Also note the angles the fans cross the grain of the soundboard. Changing the angle of the fans can give more, or less, cross grain stiffness.

My personal experience with bracing is that the 2mm high 7mm wide braces tended to favor bass response. The tops were less stiff relatively. and had very low arching, perhaps 1mm. For a while I was using braces 3mm wide by 6mm tall and I kept the braces full height for the whole brace except to taper the ends down. These tops came out much stiffer. The guitars sounded more even but perhaps a bit "thin". I was also making these tops pretty thin (1.8-2mm)with high aching around 5mm. My last several builds with one exception are really sounding "right" for me. The exception is a build using Douglas fir for the top. It sounds terrible. I am using braces 6mm x 6mm when gluing them on and fitting into an approximate 28' radius. I am also fitting the bottom of the braces to the top with a 28' radius meaning it is not a press fit when gluing but a fitted curve. The effect of the 28' radius is a top doming under the bridge of about 1.5-2mm. I work these 6x6 braces down a bit so that they are at full height from just behind the bridge to just in front of the bridge and straight taper them down to .5mm high at the ends of the braces. I am convinced that a bridge patch helps to even out the sound of the guitar so I am using one now. I like this way of brace shaping and there is a rich flavor now to my sound. I am also making the tops a bit thicker so that they do not deflect as much. They end up 2.4-2.8mm. My deflection test is to have a full size top not cut to shape but still a rectangle with a rosette installed. As I bring the thickness down with the plane incrementally I stop and wave the top up and down like a "wobble board" holding the top on both sides. At 3.5 mm there will be a lot of stiffness in the top. 3mm will show a bit more flex and this is where I take my time. Just about when I start to hear the "wobble board" sound is when I stop. I used to do deflection testing with weight and a dial indicator and found from top to top I was very consistent in my readings when checking my "wobble board" results. I just do it by feel now.

Things I would do differently if I was starting over
1. Use a bridge patch. It is a pain to fit braces over it but the sound is more focused for me using one.
2. Keep the transverse braces full height all the way across the guitar and anchor them securely to the sides. I have shaped transverse braces all different kind of ways and I think that it is a huge variable that I have not been controlling.
3. Use one plan! I am using the Hauser '43 from the Guild with some modifications from Jeff Elliott. I jumped around from plan to plan in the beginning. It really is not important which plan you use or if you make your own, just try to reduce the variables.
4. Probably 15 other things I cannot think of right now.


My steel string players generally prefer the sound of the low flat braces that give me a bass heavy response. If I were building a "folk" guitar or a nylon guitar for a steel string player I would lean in that direction. My classical players are really liking my Hauser/Elliott guitars I am making now and for me it is the way to go. The nice thing about all of this is that all of the guitars I have made have at least sounded pretty good if not great and that none were awful except the fir top. Every player that comes in will generally have a favorite but it is never the same guitar for everyone. Whatever you build will sound good.

_________________
http://www.booneguitars.com
https://www.facebook.com/pages/Stephen-Boone-guitar-builder/488208541257210

Excellent info, Stephen! Thanks a lot for writing that up. I've never really seen the point of a classical bridge patch, when any stiffness it adds could be added to the bridge instead. But I've been notching fan braces over the bridge patch in the lower quadrant of X braced steel strings lately, and it's really not that much trouble, so I guess I'll give it a try whenever I get to building a proper classical

Do you profile the fan braces to triangles or anything, or just leave them rectangular?

And did you measure the final weight of that Doug fir top compared to spruce? I'm always curious to add more data points to my theory that denser woods should be made thinner... in a sense, thicknessing the plate by weight, and carving the braces by stiffness. Although it doesn't work in the case of standard classical bracing, where the cross grain stiffness is almost entirely determined by the plate itself (which I imagine is why your wobble test works well). At least from the one hardwood topped steel string I've built, the higher relative mass seems to counteract the "thin" sound you get from thin softwood tops.

Dennis FWIW . I/ve got a d . fir top braced for a ss .The fir is not tight grained but vy wide so must be from cali. It seemed vy hvy and stiff , soo I thinned it to 2mm.It will have e. hard maple for sides and back . But will not start asembly till mar or apr, as I can only get to 35% humidity with 2 humidiers running now.

35% is right where I like it. I've been running 20% for the past few weeks, and what little moisture I had left is now frozen on the windows... no gluing braces for a while.

Let me know how that D.fir turns out 2mm sounds about right to me for steel string. Do you have an accurate scale to weigh it?

I have moved into the "thicker" top school, and I vary my braces in width and height. My center 3 fans are thinner and taller, while the get wider and lower as I move to the wings. Outside braces are roughly 6 mm wide by about 2 mm high. Center fan is in the 3.5 +/- mm wide by 5 +/- mm high. I do taper them a lot toward the tail block end, and have a much steeper slope at the sound hole end. My early guitars had thin tops and thin and tall fans. I had good sound, but head-room issues. The guitars just didn't have the required loudness to play in the market. Using thicker tops with the varied brace pattern, has solved that problem. Early tops were in the 2 mm +/- range in the center and 1.5 to 1.8 at the edges of the lower bout. Now I stick with 2.5 to 3 mm in the center and taper in the lower bout to 1.9 to 2.5 depending on the specific top. If the lateral stiffness is lacking I leave the wings thicker. I do not use a bridge patch, and see no need for one.

_________________
Waddy

Photobucket Build Album Library

Sound Clips of most of my guitars

I think Waddy has put his finger on the choice to be made here.
That is, to make a slightly more robust guitar that is ultimately capable of more volume (but requires a greater degree of player input and skill to realize) or one where a modest input from the player yields the optimal response. Let's set aside the tone question for now so that it isn't an issue.
In the past year or so I've "completed" (really just buffed, cleaned and set up) a number of 40 year old, never strung, N.O.S. Manouk Papazian guitars. These are rather robust with slightly thicker "Swiss spruce" tops (all about 2.6-3.0mm with a symmetrical 7 fan brace with no cut off bars).
These guitars have wonderful tone but many of the amateur players have passed on them because they "weren't loud enough".....professionals however are a different story and a number of these folks have owned and used these guitars to excellent effect........because they have the chops sufficient to get the most out of them.

One of the reasons that I only build on commission these days is that without having an idea of exactly who I'm building the guitar for, and they way they play, I feel that I'm just stumbling around in the dark and guessing at what is going to work for them.

DennisK wrote:
"I'm always curious to add more data points to my theory that denser woods should be made thinner... in a sense, thicknessing the plate by weight, and carving the braces by stiffness."

Not exactly, but close...

All the softwoods (but one: Eastern Hemlock) I've tested follow the same rule for relating long-grain Young's modulus to density. That is: if you know the density, you can usually (about 2/3 of the time) predict the Young's modulus along the grain within 10%. The main exceptions to this are samples that come from near the butt of a really big tree (redwood, commonly, for example), where the latewood lines are noiceably thicker than usual as a way to beef up the wood under the compression stress. This adds weight faster than stiffness. If make a plot of Young's modulus vs density, you get a pretty tight scatter around what is for all intents a straight line: in the range we look at, the relationship is nearly linear. Young's modulus is a predictor of stiffness: two pieces of material that have the same Young's modulus will have the same resistance to bending at a given thickness. For guitar tops, to a pretty good approximation, it's the long grain Young's modulus that limits how thin you can make the top for whatever scheme you're using: cross grain stiffness seems to be less of a help in resisting deformation than it 'ought' to be.

You can think of the top as a weird shaped beam, or, even better, a bunch of beams side by side. The stiffness of a beam varies as the Young's modulus, the width, and the cube of the height. We all know that a 2x4 on edge is a lot stiffer than the same 2x4 laid flat: same amount of wood, but depth helps a lot more than width. What this means is that making a given top, say, 25% thicker, will about double the stiffness, for 25% more weight.

So, we've got a linear relationship between density and Young's modulus, and a cubic relationship between stiffness and top thickness. If you do the math this means that for a given stiffness you should end up with a lighter top using lower density wood, and leaving it thicker. An extreme example of this would be substituting Indian rosewood for spruce. Generally speaking, EIR is about twice as dense as spruce, but has a very similar Young's modulus along the grain. Thus, to get the same stiffness from a rosewood top you'd leave it the same thickness a you would use for spruce, but it would weigh twice as much. Taking it down to the same weight would mean making it half as thick, and the stiffness would only be 1/8 what it should be.

The softwoods I've tested have densities that run from about 340 kg/cubic meter to a bit over 500 kg/m^3. The young's moduli I've measured average about 8000 MegaPascals at the low density and 16000 MPa at 500 kg.m^3. If you run the numbers what you'll find is that tops 'on the line' will trend a little heavier at a given stiffness as the density goes up, but the difference will only be about 15% over the range of densities. In terms of mass it's more important to avoid wood that is denser than it 'should' be, due to runout or compression wood, and find wood that's especially stiff for it's weight. You can often see runout and compression wood, but there's no way I've found to spot stiff low density wood other than measuring it. Some people are really good at feeling this, but I've been told that tests find most people aren't nearly as good as they think they are. I like to get actual measurements.

Most of the weight of the assembled top is the top itself: probably 2/3 or more. Bracing adds relatively a lot of stiffness for it's weight. That's what double tops are about: you make an 'artificial wood' that's about 30-40% lower in density than it 'should' be for it's stiffness, and then brace it normally. It's not as strong as a solid top in some ways, but most tops are more than strong enough.

Anyway, if you thickness by weight, you'll end up leaving low density tops too thick, and high density ones too thin, but most of the time the difference won't be too great. If you stick with a wood that's in the middle of the density range, as European spruce tends to be, the penalties should be small.

Yes, that is the idea (to an extent). A spruce top can be left quite stiff before bracing, and it won't be too heavy, but for rosewood you need to go thinner/less stiff to get the weight down.

Let's take the thickness of my experimental rosewood top, about 1.75mm. If that's about the same stiffness as a 1.75mm spruce top, then it should be about one third the stiffness of 2.5mm spruce.

So say you get used to building with 2.5mm spruce. You can build with 1.75mm, but you need to use a much stiffer bracing system, and even then it sounds "thin".

Making rosewood equal stiffness to 2.5mm spruce would be too heavy. So make it equal stiffness to the 1.75mm spruce, use stiff interconnected bracing, and how does it sound? Not like the 1.75mm spruce, because it's stiffer across the grain, and as heavy as 3.5mm spruce would be. That's pretty heavy, but probably in the same ballpark as most mahogany/koa tops built to the same specs as spruce, which seem to sound pretty good (at least with steel strings).

Sorry to derail the thread, Pat But hopefully this discussion of weight is somewhat relevant as well.

Yep, I can agree with that. As a rank amateur, I find much more enjoyment in a highly sensitive guitar with super low action, even though the volume isn't as much as you can get by playing hard on a stiff top with high action.

Both styles have their place.

Al - shouldn't the variation from the ideal thickness be within the 10% variation shown between the relationship between Young's modulus and the density? Not too bad for an inconsistent material like wood...!

Given that creep is a well known adversary to the long life of an instrument your approach makes a lot of sense to me in that it sounds like your goal is to build tops with a consistent stiffness. I think for structural integrity of the soundboard this approach makes sense, but I assume that to really get 100% out of a top you are still doing some voicing a la sanding of the outside edge of the lower bout (in addition to voicing via shaving braces earlier in the process)? If I recall my engineering lessons correctly, the boundary stiffness plays a big role in determining the vibration response of a vibrating membrane like a soundboard....I know I can really hear a box open up when I thin this area down..!

I'm admittedly more experienced as an engineer than a guitar builder & always appreciate your take on things like this since you seem to have a good understanding of both..!

Best,
Trev

_________________
http://www.PeakeGuitars.com

Thank you all for the thoughtful replies. Tremendously useful to me. My current thoughts are to start with 5 mm wide, 6 mm tall braces with a top thickness around 2.75 mm. I may reduce the height on the outside braces - that sounds reasonable to me. I probably won't put in a bridge plate because any I've seen in a classical guitar are narrower than the bridge and, as such, I can't understand why they should do anything.

Cheers,
Pat

_________________
There are three kinds of people:

Those that make things happen,
those that watch things happen,
and those that wondered what happened.

Here is a view of my current fan brace shaping. This has been working very well for me. Of course everything depends on everything else and merely shaping braces this way means nothing without all of the other decisions that I make. My workboard is domed with apex of the dome under the bridge. The lower transverse brace is slightly arched. The two upper braces are flat. Right now they are invisible. My cedar tops are just a bit thicker than Euro spruce usually. I use the same feel method as I described above to determine plate thickness. The electronic testing of plates is fascinating. The only thing I had a problem with is that I could never establish consistent base line data to make a determination about any thing. I could never get any tap tones to be the same from time to time on the same plate. I had a jig which I put a top in to measure tap tone with the computer. I would remove the top from the jig and then re install it with no changes. The measurements were wildly different. There was no baseline that I could create. If I could overcome this then electronic testing would open up many doors I am sure.

My argument for a bridge patch is tonal and not structural. My guitars with a bridge patch are more focused and have a more penetrating sound than the ones without a bridge patch.

_________________
http://www.booneguitars.com
https://www.facebook.com/pages/Stephen-Boone-guitar-builder/488208541257210

parser asked:
"...shouldn't the variation from the ideal thickness be within the 10% variation shown between the relationship between Young's modulus and the density?"

There's a lot of uncertainty in the whole procedure. As you know, the result is no more accurate than your least accurate measurement. In the case of a top, that would be thickness, most likely; how many of us really hold it to within .1mm, especially when you think about final sanding?

I use a vibration test to determine the Young's modulus and damping of tops that I get in. I find the fundamental bending mode frequencies, and the -3dB down points on either side, along and across the grain in rectangular top halves, and use these, along with the mass and dimensions, to find the Young's modulus. Several years ago McIntyre and Woodhouse did a series of articles on wood properties in the old Catgut Acoustical Society 'Newsletter' in which they discussed this. They pointed out that, aside from measurement errors, there's a fundamental limit to the accuracy of this whole procedure. The equation that is used to calculate the Young's modulus is necessarily simplified, to make the math tractable. It leaves out 'small' effects like the shear moduli, for instance. Even in the best case you can't count on being much closer to the 'real' value than plus or minus 10%.

On top of that, if the bending modes you're looking at are not 'pure'; if there is crosswise bending in the lengthwise mode, for instance, you're getting a mix of properties. You can tell when that happens because the node lines curve, and it may be possible to trim the piece in some cases to eliminate or reduce that.

On the flip side, this is not an aerospace application, where we really need to sweat out every last gram. Nobody builds a guitar that's right on the edge of failure, since, however nice it would sound, it would not hold up well. With all the errors, I think the testing helps to keep one in the right range, and then, as always, you do some fine tuning later.

From an engineering standpoint, I think you would be able to more accurately determine the young's modulus by measuring it directly via deflection tests. Inferring it from vibration results is bound to have more error. For damping, I don't think there is any other way to measure it...but I'm not sure what you do with the damping coefficients that result? Do you voice the guitar based on your damping results? Or do you voice it based on Chladni patters and/or tap tuning?

I certainly understand if there's a point at which you don't want to respond...everyone is entitled to maintain a few trade secrets. (c:
Trev

_________________
http://www.PeakeGuitars.com

Interesting point of view, Trev.

I've found that static and dynamic results match very well and if anything I'd argue that dynamic testing is more accurate, as in a dynamic test you automatically get a bend both ways, whereas in static testing, unless you flip the sample and take an average, you only get a one way test, which can be misleading. Either way, you still need good sample prep to avoid a whole host of potential problems (as Alan has inferred).

_________________
Trevor Gore, Luthier. Australian hand made acoustic guitars, classical guitars; custom guitar design and build; guitar design instruction.

http://www.goreguitars.com.au

FWIW I used to use a bridge patch and after some discussions here on OLF I now don't use one. I don't notice a difference except that it's a lot easier to brace the top

Also the guitars I have built with the thinnest tops seem to be the loudest. So now I'm confused

I am very happy with my bridge patches. A bridge patch is a successful part of my process. Other builders will find other ways.

_________________
http://www.booneguitars.com
https://www.facebook.com/pages/Stephen-Boone-guitar-builder/488208541257210