Official Luthiers Forum!

An interesting article. Anyone familiar with it? Any thoughts on it?
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3827644/

Sent from my SM-T813 using Tapatalk

_________________
Under Compensated Nut!

It's interesting but didn't bring any new information to light.

It starts by assuming that luthiers use tap tones to direct brace shaving to sculpture the tone of the guitar. Then they go through some measurements and math to demonstrate how it might work. And then state that since top wood is very consistent in certain qualities, and once those qualities have been measured, they can predict the outcome of brace shaving and perhaps tone (more study is needed for tone of course).

I see as shortcomings that many variables that should be considered are omitted, such as top thickness, side and back wood type and construction, box shape and dimensions, top and back brace stiffness, material, size, shape, bracing pattern, soundhole size and location and many more that I have overlooked.

But overall it is a good attempt and I'm sure the big manufacturers will find some things to incorporate into their process and their guitars just might show an improvement in consistency and maybe tone and volume.

The way I see it manufactures build everything to the weakest component. That leads to an overbuilt guitar. The process may lead to less overbuilding and therefore better sounding instruments. But I wouldn't hold my breath.

_________________
Joe Beaver
Maker of Sawdust

I'm pretty familiar with the literature, so I didn't go through the whole article. A couple of thoughts:

Consistency.
The authors seem to assume that wood in more consistent than it, in fact, is. Lengthwise stiffness (young's modulus) of softwoods varies with density, and other things, and can range plus or minus 20% from the average for a given species from one piece to the next. Cross grain Young's modulus varies far more, with no obvious correlation except with the degree of 'quarter' of the wood. Would anybody manufacturing, say, an automobile, use steel with that range of variation? Note that, except for the degree of quarter, there are few obvoius visible correlates to the physical properties of wood; you've got to measure them, and that takes time. Time is the most expensive input in any manufacturing setting.

It's very difficult to make guitars with 'the same' tone, even when you start with wood that is 'the same' within reasonably tight limits. I've tried. The guitar is essentially complex, in that is consists of strongly coupled systems of nearly linear oscillators, such at the top, back, and the air in the box, which are, themselves compex. Once you get into the 'resonance continuum', where the individual resonances are spaced more closely in frequency than their band widths, there is no way, even in theory, to predict the exact outcome of any given change. Small perturbations can cause relatively large changes. Given that the deviations of most of the systems from linearity are 'small' these are not 'chaotic'.OTOH, give the nature of the 'audience', the ear-brain system which is set up to notice small changes near the limit of perception, they are important.

A practiced maker can, of course, fine tune an instrument after it's assembled to bring it more in line with a desired timbre. This calls for a lot of skill, which might, at some point, be amenable to automation, but in any event will also add to the time required to make it.

At which point you get to the question of how 'consistent' the manufacturers need to get? The better manufacturers, exercizing reasonable levels of quality control in terms of material, manufacturing tolerances, fit, and finish, do seem to sell everything they make. They do this in large part because customer tastes vary so much: everything they make is likey to be somebody's 'Holy Grail'. All they need to do is find that person and they have a happy customer. The design of the guitar is such that it is both reasonably well optimized and fairly fault tolerant. Once you;ve settled on a size and shape, and materials, the instruments will be 'pretty similar' despite the variation. All D-18s sound like D-18s, even though they don't sound 'the same', and they're all a bit different from D-28s, which are themselves pretty similar.

Of course, manufacturers who can't take the time to fine-tune things according to the variation in wood properties need to play it safe: they have to assume that the weakest top might end up with the weakest set of bracing simply by chance and design accordingly. On average they're going to be a bit over built, and less responsive or powerful than they 'could' be. This is the opening for the individual luthier who can take the time to fine-tune the structure to wring a little more sound out, or otherwise match specific needs/wants of a customer.

Joe:
Of course they left out a lot; the whole point was to look at one variable at a time.

One thing I caught in my skim of it was a statement that the role of 'scalloping' braces was not well understood. I can't give an exact citation, but I know that Evan Davis delivered a paper at an ASA meeting a few years ago on that subject, based on a computer model of the system. It increases the 'bellows' action of the top in the low 'bass reflex' range, and delivers 'punch', along with an increased risk of a 'wolf' note due to the lower impedance at the bridge.

I'm gonna jump into this without reading the whole thing too. I will later but... Having read the abstract and skimmed it a bit, I'm finishing up Trevor Gores book now and I find it interesting that he doesn't really 'tune' the braces in any manor as suggested in this paper. In fact the build book just uses a more or less standard J-45 bracing pattern because in a manor of words - it doesn't matter.

I have always thought that tap tuning was either voodoo or only possible with a tremendous amount of experience, a life time of experience. I use deflection which is imho a more concrete number then just guessing at what tap tones make sense. I guess in either case what you are trying to do is establish some sort of base line before you couple the parts all together. Because once you close the box then everything you just did has now changed. But having said that once the box is closed you can certainly make changes and changes that make a very substantial difference.

That's always been my approach... Just built the thing, then tune it later as necessary.

I wish I could find the article but cannot, perhaps someone has come across it... But anyway a violin maker had stumbled across a phenomenon when he was doing repairs on very old instruments that had cleats from former repairs all over them. He got to wondering about it and started doing some research and experimental builds. Now he incorporates cleats as part of the building process and claims that he can 'tune' a violin to a players preference simply by placing cleats in the appropriate spots.

I figure we can do the same thing.

that cleat thing's not a new trick. folklore has it that one puts a lump of clay in an area of interest to simulate wood mass, trim as needed, then replace with a cleat.

Arthur Benade in "Fundamentals of Musical Acoustics" talks about adding wax (mass) to soundboards to lower the frequency of different modes and then (removing the wax and) thinning the soundboard in those places to reduce stiffness - which then lowers the frequency of the mode. Key to this technique is identifying areas of maximum movement of the mode to use the least amount of weight and which will then require the least amount of thinning of the soundboard. Benade's book is a bargain ($5 -10 used amazon) for anyone interested in musical acoustics.

Alan Carruth said,
Joe:
Of course they left out a lot; the whole point was to look at one variable at a time.

That's my point. If you are looking for analytical process that will improve tone, more than just brace shape should be factored in.

_________________
Joe Beaver
Maker of Sawdust

Yeah, but if you change a bunch of stuff all at once you can't say which thing did what.

Most of the articles I've seen where they tried to evaluate the effect of various brace profiles have come to the conclusion that, as my friend Mark Blanchard says: "The sound is in the top". The best bracing setup in the world won't make a bad top produce great sound, but it's distressingly easy to mess up a good top with poorly thought out or executed bracing.

If you look at the ways an unbraced top vibrates when it's off the guitar, you'll find that it's pretty much like a properly braced one, just at lower frequencies. You could get the same result in most respects by simply leaving the top thicker, so that it would be stiff enough to withstand the static bridge torque over time. The problem with that is that the top would be too heavy to make much sound. So one way to look at it is that we add bracing to stiffen the top up structurally, and then try to minimize the damage it does acoustically by getting the profiles 'right'. Brace profiles can alter the tone some in the lower end in particular: as I mentioned, 'scalloped' profiles favor low end and 'punch', while 'tapered' bracing tends to favor higher end and, especially, sustain. But this is not a simple system, and there are lots of ways to get similar outcomes. The 'right' brace profile for one top or shape of guitar might be 'wrong' for another.

As I said, the designs we have are pretty robust. Lots of people have been making guitars for a long time, and everybody is always trying to improve them. Once in a great while somebody comes up with something that works a little better, and everybody else copies it right away. That feature becomes part of the 'standard'. That's why it's hard to improve on the 'standard' by very much, and why building carefully to the 'standard' usually results in a pretty good guitar. But there's a catch.

Whenever you look at a highly evolved system you'll find that there will be little variation in the features that matter. All cheetahs can run about the same speed for the same distance. If one came along that could run faster or further at no cost those traits would be spread throughout the gene pool in a hurry, and pretty soon they'd all meet the new standard (and the antelopes would get faster, too...). This puts a premium on small advantages: the cheetah that got a better night's sleep will probably be a much more successful hunter the next day. In guitar terms the equivalent would be taking a basically good guitar and giving it a proper setup, fussing with the intonation, and maybe a little bit of brace shaving or even just swapping out pins or tuners to make the guitar better.

To my thinking, 'tap tuning' or other such methods of brace shaping are efforts to 'make a better guitar', rather than 'making a guitar better'. It's an attempt to do 'up front' what an experienced luthier could do with the assembled guitar after the fact. With such a complicated thing it's hard to 'prove' that such methods work. It's hard, for example, to show systematic differences in the outcome, if only because it's hard get the measurements, and impossible to control some things you'd like to control. There is, however, an interesting way to think about it.

It can be argued that the differences between 'good' and 'great' guitars are mostly in the high frequency range. This is, however, the 'resonance continuum' where, as I said in my previous post, the resonances are closer together than their band widths. This makes them so tightly coupled that it's not possible to say exactly why you see a particular peak in the output, say, at a given frequency: you have no direct control over it in advance. We do know, however, that some makers can consistently produce 'better' guitars than others using the same woods and basic designs. How is this possible if they have no direct control over the things that make the guitars better?

There are two possible answers that I can see. One is that they are 'making their guitars better' after the fact by some sort of fine tuning. This is certainly common. However, there are some makers who don't seem to need to do much of that; they are 'making better guitars'. In that case, what they must be doing is using some sort of indirect control to establish the conditions that will make the guitar better once it's assembled. A technique such as 'tap tuning' could be such an indirect control.

Tim White, who published the 'Journal of Guitar Acoustics' some decades back has said that all systems of guitar acoustics can be seen as 'religions': they work for the people who believe in them, and make no sense to anybody else. Perhaps he's right. OTOH, vaccination for smallpox was once such a 'belief', that caught on because it worked (when it didn't kill the person vaccinated), and was only understood later. They figured it out through careful observation and by isolating variables to understand the causes. And that was a more complex system than we're dealing with.

"Once in a great while somebody comes up with something that works a little better, and everybody else copies it right away. That feature becomes part of the 'standard'."

Like Taylor V-Bracing.

In regard to this article, the author has made the supposition that Luthiers make good sounding guitars by refining the sound through brace shaving. To me the important things regarding tone are mass and stiffness of the top and braces, guitar design, materials used, brace layout and shape, and probably several I've overlooked.

If I were designing a program to voice a guitar, brace shaving would have a small or non-existent role in it. Most would be the mass and stiffness of the top and brace material. I feel those variables need to be corrected up front, not compensated for after the fact.
Just my pennies worth.

_________________
Joe Beaver
Maker of Sawdust

"brace shaving would have a small or non-existent role in it."

Benade might disagree with you on that. Where the braces are located on the areas of greatest excursion of the various modes (monopole, dipole, etc) then shaving them at those points (or not) could adjust the frequency of those modes (by adjusting the top stiffness). Scalloping the X brace might be an example of this.

Joe is talking about 'making a better guitar', rather than 'making a guitar better'.

Alan,
I´d like to ask you to elaborate a bit more on this, if you may. Are you talking about chladni patterns? if so, that doesn't´t seem to be my experience. There are modes that are pretty similar, specially on the lower range, but i usually get lots of modes on the braced plate that are absent from the bare plate and vice versa. Or so i think. Have i been fooling myself? (honest question)

I only build classical guitars, fwiw.

thanks,
Miguel

_________________
member of the guild of professional dilettantes

"Joe is talking about 'making a better guitar', rather than 'making a guitar better'."

So was I.

mqbernardo:
Yes, un-braced and braced flat plates do show many different modes, but they also show many that are similar. Bracing does, after all, change the ration of stiffness to mass of the structure as a whole, and also of different areas on the top. It was Mark Blanchard who pointed out to me that many of the modes that we look at in 'tuning' plates show up on both braced and un-braced ones. In particular, he noted that the 'ring+' mode, which seems to be one of the most important, shows up in both cases, and that when it is not well formed on the un-braced plate it will be difficult to correct it with bracing.

For most flat-top steel string or Classical guitars, you will need an upper transverse brace to resist neck torque and displacement. This brace, by itself, has a major effect on the modes of the plate, displacing some radically in pitch, distorting many in shape, and 'quenching' others entirely. So I guess you could say that in stating that un-braced and braced plates have many of the same modes, I was rather over simplifying things.

On the other hand, I will note that even tops that are similarly shaped and braced can show 'different' modes. I suspect this has to do with the way the braces and the top work together, with small differences in brace profile interacting with differences in stiffness and mass of the top from point to point.

I would also say that, in my experience, one of the hallmarks of a 'good' top is that it shows a lot of well-formed and highly active modes in the 'free' condition. This is also characteristic of un-braced plates, and differs from tops that have been over braced or otherwise badly braced.

Clay S:
Sorry if I misread you: I was assuming adding the mass to an already assembled guitar to determine where to remove material from the braces to correct it. The same thing can be done with a 'free' plate, of course.

Nobody denies that bracing, and particularly top bracing, affects the tone. I already mentioned Evan Davis' paper on scalloping braces. I guess I'd differentiate between the 'character' of the tone and it's 'quality'. A scalloped braced Dread will have a different character from a straight-braced one, but both could be excellent guitars for different users. Overall brace profile, such as that, affects the character. Quality comes from getting the bracing and the top to work well together, IMO, and I feel that can most easily be accomplished at the 'free' plate stage, rather than after the guitar is assembled. For one thing, it's easier to get at things!

"Quality comes from getting the bracing and the top to work well together, IMO, and I feel that can most easily be accomplished at the 'free' plate stage, rather than after the guitar is assembled"

Hi Alan,
When speaking of brace shaving those were things I was considering. Benade seemed to indicate that the places to change the stiffness and thus modify the frequency of a mode was at their areas of greatest movement and that the different modes could be individually modified by adjusting different places on the plate. That modifying the stiffness of one mode would not effect the other if where you modified it was on the node line of the other. These modifications would most easily be done to the top before it is assembled onto a guitar body ( is this what you are calling free plate?). I'm assuming the craftsman at Martin were tuning the plates before assembling the bodies rather than reaching in blindly through the soundhole.
Now as to what ratio or dispersion of frequencies of the modes would give the optimal result I would be curious to know.

Did I miss something? When were they doing this?

_________________
The name catgut is confusing. There are two explanations for the mix up.

Catgut is an abbreviation of the word cattle gut. Gut strings are made from sheep or goat intestines, in the past even from horse, mule or donkey intestines.

Otherwise it could be from the word kitgut or kitstring. Kit meant fiddle, not kitten.

It has been said that the workers at Martin used to scallop the braces after the guitars were strung up. I'm not sure what era that was; they don't do it that way now so far as I could see in the usual factory tour. Perhaps there was more leisure for that in the 'Golden' thirties when orders were off, but it would cost too much now.

Benade is correct as far as that goes. In some respects it doesn't go very far.

With enough experience building to a given pattern with 'standard' bracing a maker can get pretty good at controlling the pitches of the lower order top and back modes of the assembled guitar. It's difficult to get exact matches from one instrument to the next even in this range (up to , say, 350 Hz or so). We're saved to some extent by the fact that even fairly large changes in those pitches don't make much difference in the perceived timbre of the guitar over all (see Wright's 1996 thesis, given at U/Wales Cardiff). If a low order mode, such as the 'main top' resonance, is shifted off a played pitch to another, or to fall between notes, the timbre of the affected note will change noticeably, but not the overall sound of the guitar. In general, then, in this range 'close' is 'close enough'.

Note that the assembled mode frequencies can't be accurately predicted from 'free' plate information by any simple means. As I say, with enough experience you can get close, if you stay in familiar territory, but changing, say, the B&S wood, or the outline, will change the assembled mode frequencies, due the fact that all of these things are more or less strongly coupled. With a complete enough computer model you'd do a lot better. Gore stretched the limits of Excel with a model that he claims will get you pretty close, but it's limited to the lower order assembled mode frequencies, and those don't need to match exactly anyway in most cases, as I said.

The devil lives in the 'resonance continuum' range, say above 600-700Hz or so. Here a small change in some mode can alter the response of one or more others with which it is coupled, and the output can shift, both in the spectrum and also in the direction of the radiated sound. I'm pretty sure this is why it's so difficult to make guitars that sound 'identical'; very small local changes in top or brace stiffness produce proportionally large changes in spectrum and direction of radiation.

I will also say that my experience strongly suggests that 'free' plate mode shapes tell you more than the frequencies do. This is one reason I think that what we're really doing with these techniques is 'tuning' the higher frequency response, in some sense. Here I'm using the term 'tuning' as you would in saying 'tuning up' a car engine: you're not trying for a particular RPM or whatever, but rather striving to optimize the overall working of the machine. The usage originated with the magneto ignition system on the model T Ford. This included a make-and-break buzzer to get AC from the battery's DC, which could then be stepped up to high voltage. The gap in the buzzer was set using a tuning fork rather than a feeler gauge: too slow and you might not get a spark when you needed it, and too fast gave a weaker spark that might not work to fire the cylinder. Nobody except a few die-hard enthusiasts uses that ignition system any more, but the term 'tuning up' has stuck in the language.

As always, there are limits. As you go higher in frequency it gets harder to drive plates hard enough to form Chladni modes. Laser interferometry can show these, but most of us don't have that equipment. Small difference that can be hard to 'see' can still alter the timbre of the guitar. Also, these 'free' plate methods, whether tap tones or Chladni, don't tell as much as you'd like about the stiffness and mass of the edges, and, again, those can be important. At any rate, it's not just the braces or just the plate; it's how they work together that seems to count.

Hi Colin,
It was during the anthropocene.

I admit it is somewhat conjecture on my part. If we can assume humans were using tools during the early paleolithic period because were find a few sharp edged rocks, then I think it isn't too far a stretch to assume the scalloping of the braces by the Martin guitar factory workers was intentional, and was used to adjust the responsiveness of the top. But I could be wrong.

Alan, thanks for clearing that out. I was a bit puzzled by your comments, but at the same time it feels awkward sort of challenging you on this very topic since everything i´ve learnt about it has came from you, via the plethora of internet fora.
I suppose now is a good time to thank you for all the info you so graciously share and for the patience and time you put into your posts - they´re one of the reasons i keep coming to these boards. i think it qualifies as public service, really. i could go on...

_________________
member of the guild of professional dilettantes

OK: yes the braces on Martin guitars are scalloped intentionally. Scalloped bracing does produce a different timbre than straight or tapered profiles, with a stronger 'attack', less sustain, and more bass.

One of the things that seems to happen when the bracing and top are properly 'tuned' to work together is an improvement in the treble response. It my not be 'more' treble, but it does tend to be 'clearer' or more 'defined'. Sometimes when it's pointed out that scalloping gives 'more bass' people will come back and give examples of scalloped braced guitars that have good treble as well. As with anything, scalloping can be either be done well or not so well. Simply matching a set profile will sometimes result in bracing that works well with the top, and other times not. It's a crap shoot depending on the properties of the wood and the bracing. Doing better, getting more consistent results, requires further effort, in wood selection up front and in finding the correct brace shape for the wood used.

mqbernardo:
Don't worry about challenging me. If I can't explain it to you I probably either don't really understand it myself, or I've just plain got it wrong. Either way I want to know so that I can correct the problem.

So then maybe the measure twice cut once rule should apply to top tuning? IOW build a guitar with straight braces and hey, if it sounds wonderful then you win, leave it alone. But if not then use your skills as an experienced luthier to either taper or scallop or even both, the braces till you get what you are after.?

I've gone through Gores book now and while I have not actually tried building in this method yet I remain skeptical. I know all the math is there and believe it or not I understand [most] it too! But still there are a few things in it that bother me and in the end we are just monkeys assembling a box together and hoping it sounds good. THe one thing I really did appreciate from the books is the methods for tuning the guitar after the fact and also an understanding about how all the parts are coupled together. It's good knowledge to have when you are aiming for a particular type of guitar but I can't imagine having total control over it none the less.

" I can't imagine having total control over it none the less."

I think most honest people share this sentiment. Having some control over it I think is what most hope for. The more you know , the more confused you're allowed to be.