Official Luthiers Forum!

Jerry,
Good to see I’m not the only tonofobe bothered by this.
“Chasing” is mostly what I have done also.
As far as I can tell, hiding the dead note is all we can do. But I think it is related to the response of the instrument as a whole so you can’t manipulate the wolf out of existence without some other consequence or compromise in tone or volume or structure.
I think some of the variables you can do are changing the mass of your bridge, that changes the wolf in relation to your top, and, the size of your sound hole, that changes the main air response which affects the wolf or dead notes.
I’m thinking if the dominant tap tone of the top is different from the back and both are different from the main air tone, your dead notes should be less dead. I may be barking up the wrong tree.
This is a work in progress for me and I appreciate your input and experimentation.


Wade

_________________
Wade
Nashua, NH
http://www.wadefx.com

boatpaddle wrote:
"It's a balancing act where one thing always effects another. "

Bingo.

The thing about tuning the top and back to work together is that it can spread the response out. Done properly, it's about as close to a free lunch as you'll get.

When you talk about a resonance, there are two things that are important: the pitch and the 'Q value'. The Q, or 'quality factor', refers to the width of the peak of the resonance, and is a measure of the losses involved. Something like cardboard, that has high losses, has resonances that are spread out over broad bands, which is why the tap tone doesn't have a very well defined pitch. With low losses all of the energy of the resonance is confined to a small range of frequencies, and you hear a nice pitched tone.

Q can be defined by the 'half power band width'. If you look at the peak of a resonance it should be something like a Gaussian curve. The frequency at the peak is what we call the 'resonant pitch'. If you go to 3 dB down, or 70.7% of the amplitude on either side of that peak, you will define a frequency band in which the power is at least 1/2 the maximum. The peak frequency over the band width is the Q value. High losses tend to drive down the height of the peak more than they do the activity off peak, so the whole curve is flatter. The band width is greater in this case, and the Q value lower, so Q is a measure of losses.

One of the problems with the guitar is that we just don't have the power to work with that the violin or piano do, and we can't afford to waste any. Thus, we want to keep the losses low. But low losses imply that the resonances will be sharply defined, and this means that some notes will be much stronger, and have less sustain, than others.

One way around that is to tune different resonances so that they are just within each other's half band width. If you're talking about the top and back, for example, this will make it easy for the top to drive the back. As the top approaches it's peak the back starts to 'steal' a lot of energy from it, so the top amplitude does not raise as fast as it would have. As the top starts to drop from it's peak, the back reaches it's own, but since the top is not vibrating as much, the back doesn't get to as high an amplitude as it otherwise would have, either. The result is that you get an output spectrum that has a broad double peak, with a dip in between. The maximum height is less, but the area under the curve, which is the 'total available horsepower' is greater overall. You have three good strong notes, rather than one real honker. And, best of all, you have done this without actually adding to the losses in the system.

The trick is that you have to do it just right. If the two resonances are too close together the back may be able to steal energy faster than the top can provide it, in a sense. The sound ends up coming and going at whatever the difference frequency is. This can be a nasty wolf, although it's easy to fix once you've found it. The semitone seperation seems to be close enough to work well, and far enough apart to be on the safe side. Again, a matter of balance.    

Alan,
i follow you up to the last paragraph there. Your words, by the way, are
always top notch and my attention always peak when I see you have given
a hearty response.

So "The semitone separation seems to be close enough to work well." For
some reason, right now, I want to believe you mean an interval of a
flatted fifth. Is that right? It would make sense with what you were
saying about the range where the top starts to lose its power then the
back beginning to gain its power. I can see the logic in that.

But i also seem to recall that folks tune the top and back anywhere from
1/2 to a whole step apart. That is much tighter than the flatted fifth
interval, but I couldn't help to think that sometimes that might mean
semi-tone as well? I'm confused.

Is there a source of methodically procured data that might show these
relationships?--in graphical or other form? I'd love to spend a year doing
some hard core research on top/back coupling, but just don't have the
time right now.

Thanks you guys! good questions and responses!

Here's a great site about guitar acoustics with lots of data: http://www.phys.unsw.edu.au/music/guitar/

Alan, I wonder if you've seen this study and am interested in what you think about it...

There are a number of different 'tuning' schemes for relating the back and top main tap tone pitches, and they all work well for somebody. Each one has a different chracteristic timbre, and you simply pick what you like and go with it. I've never tried a flatted fifth interval, but it could work.

I first learned about the semitone seperation from Fred Dickens, who worked at Bell Labs and made really nice classicals on the side. He used some circuit modeling software at work to figure out the relationships that would give the kind of spectra that he was seeing in guitars that he liked. It turned out that having the 'main back' tap tone a semitone higher than the 'main top' gave the results he thought were best.

If you look at Fletcher and Rossing's 'Physics of Musical Instruments', fig. 9.9, you'll see that the D-28 they tested had the main top and back modes much less than a semitone apart, and a D-35 more like three semitones. A Kohno 30 classical had them about two semitones apart, and a Conrad came in at more than six semitones difference, with what I'd call a very loose top. It's iompossible to say whehter any of these makers was trying for any particular relationship, of course, but I doubt it in the cases of the production boxes.

I had a classical guitar in for repair once with a whole thicket of buzzes. After we got the fret buzz, the nut buzz, and the saddle buzz, there was still one we couldn't find, on G# of the low E string. This was about the pitch of the 'main air' resonance, and the 'main top' resonance was just an octave higher, with the 'main back' tap tone seven Hz above that. What we were hearing was the difference frequency between the main top and main back resonances on the octave harmonic of the played note, which were strongly driven by the large changes in air pressure. A small amount of added weight at the bridge location dropped the 'main top' mode to 11 Hz below the 'main back', and got rid of the buzz.

I once had a fellow bring in two steel string guitars that he had, one with great bass and 'flat' trebles, and the other with nice clean 'fat' trebles but a 'thin' bass tone. He wanted to figure out how to get one or the other of them to work all across the range. Since it seems to me that good trebles are more a matter of the way the top works, while bass comes from the rest of the box, I thought it might work to re-tune the back on the one with the thin bass. Sure enough, the one that had the good bass tone had the semitone seperation, while the other back was much stiffer. Shaving the back braces down got the tap tones in line, and amounted to a bass tone graft without harming the clartiy of the trebles. This was right after I'd spoken with Fred about all of this, and it was nice to get such a clear demonstraton.

I look in on the New south Wales website every once in a while; they are doing some good work there.

gburghardt, I'm not sure if you're clear on the terminology or not but a semitone is a half step, such as the interval from "c" to "d flat", a flat fifth such as "c" to "g flat" would be 6 semitones.

_________________
David White, Toronto

"All my favourite singers can't sing."

[QUOTE=drwhite]... a flat fifth such as "c" to "g flat" would be 6 semitones.[/QUOTE]
It's often called a "tri-tone"--two minor thirds stacked together (C to E flat/E flat to G flat).

Some poeple like to tune the back a fourth or fifth higher than the top, and Geza might be thinking of a semitone from that.