Official Luthiers Forum!

Size seems to have an important role, as expected... I've tried a few times to bring down the back resonance somewhat closer to the top, being curious to check out the coupling effect, with little luck on my small plantilla. Most of my backs end up around 250 Hz. I could go down to 230 maybe, with scalloping the brace centers in the style of Romanillos, but it is still a long way. At some point I might try to feather the brace ends more, 1/8 instead of 1/4, and thin the rosewood some more. Currently my backs are .08-.1 depending on density....

_________________
Build log

Can someone point me at a source that shows the correlation between free plate tuning and the characteristics of those plates once they have be bound around their perimeter? A PM is fine..

Thanks, Peter Z

Ah ha, that is good to know, Laurent. Thank you!

Those wolf tones might be nice on a weiss.
Just slide your leg to regulate it.
Yes, I am player!

Alan - thank you very much for the information! I have your DVD on plate tuning but have not made the time to watch it yet. I think I need to watch it and then come back and reread this entire thread. It's a lot of information for my inexperienced brain to process.

Peter Z asked:
"Can someone point me at a source that shows the correlation between free plate tuning and the characteristics of those plates once they have be bound around their perimeter?"

I hate to say it, but if you're talking about mode frequencies, there is no simple correlation. It's not that there is NO correlation, it's just that there are so many other variables involved.

I have, for example, made a 'matched pair' of guitars, with the tops cut from the same flitch of redwood, the B&S from the same mahogany, and so on. I matched everything I thought was important as closely as possible, down the the weights and vibration modes of the bridges. When I got the two guitars together they had all of the lower order modes at the same frequencies. Unfortunately, they did not _sound_ the same. Close, but not the same.

It makes some sense to think that the 'ring+' mode on the top in some way 'becomes' the 'ring' or 'main top' mode on the assembled guitar. Of course, when you glue the top to the rim the pitch changes, but we can't predict by how much at this point. It ought to be possible, using some fairly sophisticated software, but even then I have to wonder. I'm pretty sure that the shape of the 'free' plate mode is one of the variables in this, but there are lots of other things that could get into it as well, such as the mass of the rim, the width and flexibility of the liners, and so on.

Sorry I can't be of more help. I wish there were a simple answer, too.

Alan - have you ever done any work with the modes of a braced top once it is glued to the rim (without the back attached) or with a braced top held in some sort of artificial rim that simulates the effect of attaching it to the rim? I think I've seen photos of some kind of artificial rim with a top clamped in it, but can't remember where. In any event, I KNOW I'm not the first to think of the idea.

CharlieT, here's a link to James Watts' site with such a setup. http://www.jameswattsguitars.com/shop%20notes.htm It seems pretty simple to construct.

_________________
Ken Franklin
clumsy yet persistent
https://www.kenfranklinukulele.com

Thank you Ken! That's where I saw it.

The problem is that, from my perspective, a lot of information 'goes away' when you glue the top to the rim. All of the modes that bend around the edges are supressed or changed in shape or shifted 'way up in pitch where you can't see them. That's one reason 'free' plate tuning is so useful: you see stuff that's hard to see otherwise.

You really have to try pretty hard to avoid having the 'usual' series of modes on a assembled guitar. Not only do small changes in the bracing 'wash out', big ones do too! The frequency relatinoships will tend to be different for different sorts of brace patterns, but unless you do something radically different, they won't be totally out of line on any guitar that sounds reasonably good.

Again, I bring up the example of the 'matched pair' I made that sounded noticably different once they were together: all of the assembled modes that I could see were exactly the same in pitch and configuration on both guitars, but one was clearly 'better' than the other. The guitar everybody preffered also had 'better' free plate mode shapes, and that was the only difference I could find between them.

Now, obviously differences in the 'free' modes would not matter if they didn't change something about the way the guitar works when it's together. The problem is in trying to find the relationship.

At the moment, it's all pretty empirical: 'free' plate tuning seems to work, at least for some of us, so we do it. Tim White, formerly editor of the 'Journal of Guitar Acoustics' has said that all systems of guitar acoustics are religions: they work for the folks who believe them, and nobody else can make head or tail of them. At some point we hope to move into the realm of science on this, but that's going to take a while.

Interesting stuff. I have Allan's video and it really only deals with tuning tops, so is not going to answer any questions about backs and how backs couple with the top. Excellent video, and thanks Allan for making it, but don't try and watch it in one session, there is a lot there. I have done a lot of plate tuning in mandolins and have publlished 2 papers about it, but am working on my first guitar, hence the interest. I certainly found Allen's video very useful.

Maybe guitars are more complicated than mandolins (bigger??), but I have made 2 pairs of mandolins with identical free plate tuning frequencies, and they sounded near enough to identical. I played them in a darkened room and had to trurn the light on to work out which one I was playing. The mode shapes do seem to be more consistent in mandolins than guitars, after watching Allen's video, so maybe that explains it. I did just look up the pictures of the modes I took of one pair and the shapes are pretty similar as well as the frequencies.

I have also done correlations of mandolin free plate modes before and after the ribs gave been glued to the top (not published yet). Some of the free plate modes (e.g. the X mode) disappear after the ribs have been glued, or at least I can't measure them. However, there is a strong correlation between the ring mode of the free plate and the ring mode I can measure after the ribs have been glued. There is some scatter, as you expect since different sets of ribs will have differing stiffness, but there is not much doubt that the two ring modes are the same thing.

I have also found a correlation between the ring modes in the top and back and the sound of the completed mandolin. If you match the ring mode of the top and back to within around 5%, then the resulting instrument will sound better than if there is a larger difference. This is published in my secand paper. Both papers are on my web site for those interested.

This is diverting a bit from guitars, but just thought some might be interested.

Yes, very interesting.......thanks for posting that info!

_________________
Formerly known as Adaboy.......

I have not built any mandos (yet), but Darryl's findings with them are very similar to what I've observed on archtop guitars. Flat tops just vary so much....

Yes Allen, I saw your comment on MIMF that if you match the ring mode of the top and back of an archtop then they do tend to sound better. My immediate thought was that an archtop guitar is like a big mandolin, which is not all that surprising.

Peter

Free plate tuning may correlate well with outcomes, but can this be a reason why (that you no longer see the modes when the plate is joined)? You get a bunch of information that appears to become irrelevant once the top is joined to the rims. Kind of like the guy looking for his keys under the streetlamp, no?

_________________
Howard Klepper
http://www.klepperguitars.com

When all else fails, clean the shop.

So, maybe tuning the top and back after attachment to the sides?
I think that's where I'm headed.
Then again,

Peter, you should try to get your hands on David Hurds book "Left Brain Lutherie", he does address this issue and offer some guidance concerning the frequency shift of the back from a free plate to fixed plate.

He post here accasionally, maybe he'll comment himself.

_________________
Jim Watts
http://jameswattsguitars.com

Howard Klepper wrote:
"Free plate tuning may correlate well with outcomes, but can this be a reason why (that you no longer see the modes when the plate is joined)? You get a bunch of information that appears to become irrelevant once the top is joined to the rims. Kind of like the guy looking for his keys under the streetlamp, no?"

Some would say that if the method works, then the information is not 'irrelevant'.

Some years ago Tim White, who edited the 'Journal of Guitar Acoustics' pointed up a paradox: The differences between 'good' and average' guitars seem to be in the high frequency response region, but that's the 'resonance continuum' * where you don't have any direct control over the outcome. So how is it that some folks can consistently make better guitars when they have no direct control over the thing that makes them better?

The free plate modes that 'go away' when you glue the top to the rim are those that have a lot of small vibrating areas. On the completed guitar modes with lots of small antinodal areas are the high frequency modes, up in the resonance continuum. It seems logical to think that by getting those small-area modes to work well on the free plate you're setting up the stiffness and mass distribution that will allow them to work well on the completed guitar.

This could, of course, be a bunch of hooey. It's not too hard to design the experiment to check this out, but it would require some equipment I don't have, and some expertise that would take a while to aquire, most likely. At some point, when I become solvent and notorious (I've given up on rich and famous), maybe I can do it. Meanwhile we can go on faith; 'the substance of things hoped for, and the evidence of things not seen'. As Tim White has said, all systems of guitar acoustics are, in the end, religious.

* As you go up in frequency you tend to get modes forming at more or less regular intervals; say every 50 Hz. In the low octave 50 Hz is the difference between E on the open low string and C, third fret on the A string. Clearly these are far enough apart that they don't couple strongly. However, as you get up around E on the 12th fret of the high E string 50 Hz is just the jump to the next note, more or less: it's only about 6% of the frequency.

Every resonance has a 'bandwidth': a range of frequencies over which it can be driven easily. That 6% frequency difference at E=659Hz is just about the normal bandwidth of a resonant mode of some part of the box or the air in that range. It's likely, therefore, that in that frequency range you're going to have a number of resonances that have overlapping bandwidths: they can each drive the others easily. In this situation it's not possible, even in theory, to say exactly which part of the guitar a resonant peak 'belongs' to. All you can say is that there is motion in this or that part of the guitar, but loading the moving part, for example, may not effect it's frequency very much.

I had this happen to me once on a small guitar I built for my daughter. The 'main top', 'main back', 'main air' and 'neck' modes all came in within a semitone. I kept shaving the back braces trying to get the pitch of that mode to drop, but it wouldn't. It was as if there were an 'exclusion principle' working: whichever part had the resonance closest to a particular frequency gobbled up all of the energy at that pitch, and the back had to make do with the frequency nobody else 'wanted'. Of course, the back was also pushing the other parts into other bins, so to speak. At this pitch, you could not talk about the parts, it was all 'system'.

All you can really say in the resonance continuum is that there are a certain number of resonances per octave, and that they have an average Q value (loss) of thus-and-such. Continuum behavior seems to start to set in at around 600 Hz on most guitars, but, of course, it's hard to be precise. It's certainly in force by 1000 Hz or so.