How to predict the outcome note by lightening the soundboard structure.

[Trevor Gore]

Hello,
for myself first guitar I'm using cedar for the soundboard and madagascar rw for back and sides. The bracing is Torres a bit creative, 7 fans with open harmonic bar. All the struts and fans are cedar except for hole patch that is laminated mahogany. The finished sb weighs 157 grams. Before bracing the tone was D and after braced and scalloped it raised to E. As an apprentice I'm very intrigued by the debates about voicing, but just as insecure where changes on structure could bring. My first thought is that lightening up again the struts, scalloping the harmonic bars , the pitch could move back to an intermediate note. Or could it move to E# o F?

Inevitably to E#.

Sometimes the satire doesn't transmit.

It seems to me that there are at least three ways of voicing a soundboard - tap tones, Chaldni patterns, and measurements of the board's flexibility. Do people use a combination of these methods, or do they gravitate to one favourite method? And if so, which one!?

Most of the sound that you hear that comes from a guitar is radiated by the low order resonant modes (about 85% according to research published in the 1980's (iirc)). Up to about 4kHz, it is only the resonant modes that radiate sound. The resonant modes are the ones you see with Chladni patterns on a completed guitar, the low order ones being quite efficient radiators quite a way from their centre resonant frequencies. If you want to alter the sound of a guitar, you therefore need to alter the amplitude, bandwidth and centre frequencies of the resonant modes, because very little else produces sound.

However, the free top mode resonant frequencies tell you very little about the resonant frequencies of the finished guitar, because so much changes between the free plate and the closed box. In theory, they should be relateable, but in practice, two chassis are never identical enough to produce identical results even if you were able to make identical tops. The nature of the linings, their glue area, their stiffness, the side stiffness and mass, the air volume, the back, the bridge, the finish etc.etc. etc. and how they all couple together all interact to alter the performance of the top in a way that is impossible to predict with sufficient accuracy to be useful. Which is pretty much what Alan sees, too:

The problem persists, though, that there is no obvious relationship between the resonant pitches of the top at any stage of construction and the pitches of the resonant modes of the assembled guitar.

...which makes all this tapping and tuning of free plate resonances of doubtful benefit. Indeed, if you watch the video, more emphasis is actually placed on working to dimensions rather than tap pitches.

So what to do?

Alan gets benefit from closing the ring and a half free plate mode. I'm agnostic on this, as the bracing patterns I use mostly seem to produce closed ring and a half modes naturally (which happens much more easily with symmetrical bracing) on the odd occasions I've tested plates just before gluing down. But I also thickness the panels and size the bracing according to the material properties (the elastic moduli and the density) of the wood, which pitches them on the closed box close enough to my preferred mode frequencies that I can trim them to where I want them. Which means you mostly hear the sounds that I designed it to produce. This is a similar sort of technique to that which Martin Schleske uses to make tonal copies of violins using modal analysis, but the violin makers maybe have it easier, as the free plate modes do relate more readily to finished instrument modes, apparently. All the details, analysis, modeling and testing can be found in the usual place.

[Costantino Proietti]

Hello,
for myself first guitar I'm using cedar for the soundboard and madagascar rw for back and sides. The bracing is Torres a bit creative, 7 fans with open harmonic bar. All the struts and fans are cedar except for hole patch that is laminated mahogany. The finished sb weighs 157 grams. Before bracing the tone was D and after braced and scalloped it raised to E. As an apprentice I'm very intrigued by the debates about voicing, but just as insecure where changes on structure could bring. My first thought is that lightening up again the struts, scalloping the harmonic bars , the pitch could move back to an intermediate note. Or could it move to E# o F?

Inevitably to E#.

Sometimes the satire doesn't transmit.

[Alan Carruth]

Trevor Gore wrote:
"The resonant modes are the ones you see with Chladni patterns on a completed guitar, the low order ones being quite efficient radiators quite a way from their centre resonant frequencies. If you want to alter the sound of a guitar, you therefore need to alter the amplitude, bandwidth and centre frequencies of the resonant modes, because very little else produces sound."

Here's where it gets sticky. Certainly most of the power of the guitar, and much of the basic timbre, is produced by the lower order resonant modes that you can see with Chadni patterns on the assembled box. As Trevor says, it's difficult to make two guitars that are 'identical', but with good controls you can match them up well enough so that the lower order modes will be 'the same' within measurement limits. One recent 'matched pair' that I made had all of the lower mode frequencies and shapes matched to within, say, five Hz or less. The impulse spectra were so nearly identical up to 1000 Hz that if they had been pieces of electronic equipment you'd have said they were 'identical'. Yet, in 'blind' listening tests, almost nobody had any problem telling the two apart.

This is one of the paradoxes of the guitar. In many respects the 'quality' of a guitar is determined by how it works in the higher frequency range; above say 1000 Hz. The problem is that the way the guitar works in this range is so complex that the maker has very little direct control over the response in this range except by making changes to the assembled box. Such 'tweaks' are useful for alleviating problems, such as some 'wolf' notes, but otherwise don't have a major effect on overall quality or timbre. And yet, despite inability to directly control 'quality', some makers can more or less consistently build better quality instruments than others. That's the paradox.

One possible solution is that makers have indirect ways of controlling the high frequency behavior. One study suggests that the number and overall 'quality factor' (peak to dip ratio) of high frequency peaks in instrument sound have an effect on perceptions of tone quality. The higher order resonances of the guitar are not effective sound producers, at least at low frequencies and often end up 'stealing' energy. However, in so doing they can alter the peaks and dips of the spectrum. They also change the directional sound output. It has been noted that 'good' guitars tend to 'beam' much of their sound out toward the audience, rather than 'enveloping' the player. This can only be a function of the high frequency modes and their interactionswith each other and the low frequency resonances. It's easy to see a pronounced shift in the direction of sound radiation from the guitar as you go through the pitch of the top 'cross dipole', for example, due to phase shifts between the driving force and the resonances. In the higher range the top becomes something of a 'phased array'. It could also become much more efficient as a radiator near it's 'coincidence frequency', which should be around 3 kHz, just where hearing becomes most sensitive. It doesn't take much power there to produce an audible sound.

I believe that 'tap tones', and 'free' plate tuning by Chladni patterns, which is the 'tech' version of tap tone tuning, works by improving the high frequency response of the plate. From what I've seen the actual frequencies of the 'free' plate modes are much less important than the shapes, such as the 'closed' ring+ mode that Trevor mentions. Tops that have more and better formed high order modes seem to make better guitars.

I am, of course, biased. OTOH, with my damaged hearing I rely on the response of listeners and players more than on my own impressions, and so far the correlation has been pretty good. More direct tests are possible, but difficult. 'Proof' of this conjecture will thus be unlikely in the near future.

[printer2]

The way I see it, taping or Chladni patterns is a method of getting rid of stiffness or weight where we do not want it. Too much stiffness retards the top movement as does too much weight, The idea of removing structure to the point where the tap tone goes away might be a case of reducing the acoustical impedance of the top to where the air load is higher and dampens down the top. When the top is attached to the rest of the box hopefully the top becomes stiffer due to the edges being bound to the sides and with the addition of the bridge.

[Alan Carruth]

Right, 'lumps' of either stiffness or mass show up as distortions in the patterns. 'Well formed' patterns that are active and have narrow bandwidth give a more clear impression of pitch, which is one of the things that 'tap tone' tuners go for. As for tap tones 'going away', I always assumed it was simply that they dropped below the range of hearing. I hadn't thought of the added damping as the plate got more active, but that could well be part of it.

Shaving the braces doesn't usually reduce the mass of the top assembly much. All of the braces together seldom amount to more than about 1/3 of the overall mass of the system; most of the weight is in the top plate itself. Removing material from the upper transverse brace hardly affects the modes of the 'free' plate until you've taken off enough to compromise the structure. In fact, most of the effect comes from removing wood from the braces behind the bridge, which are arguably the least important structurally. On a top that ends up weighing 180 grams or so, of which 40 grams are in the bracing and, say, 9 in the bridge plate, I might remove five grams, more or less, of wood from bracing in the process of tuning.

Working on the bracing of the 'free' plate has a few advanatges. One is that they're easier to get at. My forearms are 'way too big to make brace work on an assembled Classical guitar anything other than torture. Classicals usually don't need much work, but in some cases the removal of a couple of shavings from some braces can make dramatic changes to the modes, which do seem to affect the sound.

Assembled guitars all show pretty much the same series of modes, unless they're really badly made. Fairly large changes in the ones you can see using Chladni patterns don't have a lot of effect on the 'quality' of the tone, unless those changes move a mode onto or off of a played pitch or partial of one. In that case you notice a change in that note, but not much of an overall change in timbre. By the same token, guitars that show different mode shapes or frequencies in the 'free' plates can end up with virtually identical assembled modes, and yet sound much different. It's the old truism that 'free' plate mode frequencies don't correlate strongly with assembled modes. From what I've been able to find out, it seems as though there is a stronger correlation between free plate mode shapes and the tone of the assembled guitar than anything else. I don't worry too much about absolute mode frequencies. Relative pitches of some 'free' top and back plate modes do seem to correlate to a degree with the pitch relationships of the assembled boxes, but it's a complex system, and hard to nail down exactly. Archtops seem much better behaved in that respect, IMO!