Official Luthiers Forum!

It's odd: sometimes the software seems to log me off at random, often when I've replied to a message. I did see the pic this time...

That 'smiley' you got is just the lower node line of the 'main top' mode, and every guitar you test, whether good or bad, will do that. Heck, even banjos have a 'ring' mode on the top....

I agree that going from the 'free' modes to the modes of the assembled instrument is not a trivial piece of mathematical work. I was discussing this with a student the other night. He writes software for a firm that does air flow modeling, and he's quite familiar with the state of the art both in fluid and solid models. If you assumed that the properties of the wood were measureable and uniform throughout you could get 'arbitrarily close' to predicting what would happen with software that could run on a good PC, but there would still be some uncertainty. For one thing, the plate boundary conditions are quite difficult to specify, and seem to be something between 'clamped' and 'hinged'. If the wood varies from onepoint to the other, of course, you're in trouble, and it usually does. you could chop the top up into little pieces and test each one, but then you couldn't make a guitar from it. Problems, problems. On the other hand, just because the math is difficult that doesn't mean that the tone of the completed guitar is not in some part determined by the 'free' top conditions.

I have found that if you can control things pretty carefully you can predict what the final mode frequencies will be to a fair accuracy from the 'free' modes. That is, if you start with 'identical' sets of wood, build 'the same' size and shape of guitar, and get the mode frequencies and shapes to match closely, the two instruments will end up with their modes at the same pitches, within the limits of measurement error. This might be trivial, but it's a place to start. I'll note that the assembled pitches don't seem to vary by as much as the 'free' ones.

What's interesting is that even very small differences in the shapes of the 'free' modes, and particularly of the 'ring+' mode on the top, will make an audible difference in the final result, even when the assembled modes look the same and come in at the same pitches. This holds even when the 'free' plate frequencies, masses, and everything else, are held to close tolerances: the shapes seem to matter more. It also seems to hold that two guitars with similar 'free' mode _shapes_ and different 'free' mode frequencies can sound very much alike, provided the relationships between the final assembled mode frequencies are similar.

All of this suggests that the 'free' mode shapes have some meaning, at least in terms of quality control. Again as Bob points out, the free mode shapes are an indicator of the distribution of mass and stiffness in the top, and surely that has to effect the tone?

Hi John,

Good to hear from you. Actually I wasn't referring to the main air mode but to the top and back monopoles. And yes, you are correct in that everything is coupled now and changing one affects the other as well as the main air mode. But to me this is OK because it is closer to reality. I'm more interested in the shape of the modes for the free plate and in the relationship of the frequencies in the finished guitar. So when looking at the frequencies I keep an eye on their relationships and how they change throughout the building process. In Craig's case, he has a finished guitar with the bridge already on, so what he measures now is what it is going to be. Of course, when the bridge isn't attached you have to guesstimate its affect based on experience. Trying to control all of this stuff is not simple and requires trial and error for each model and system you used. But that is the fun of it.

By the way, are you going to Miami or Montreal?

_________________
Randy Muth
RS Muth Guitars Website
RS Muth Guitars Blog
Facebook Fan Page

Looks like we'll all be in Montreal! Al's not on the list for Miami (on their website)

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

See I told you Al could explain it a lot better, but we already knew that.

I have been talking with Al a little bit about some experiment guitars, that I am making. It should be fun as I have a couple of Engineering Students doing an independent (directed) study with me. They are doing lots of wood testing at the moment (both static and dynamic) and will soon play around with free plates modes. Next up will be bracing patterns stuff. At this point it will be fairly loose but if it goes well, I may invite some other students (another semester) to tighten up some parameters so we can look at correlations. We also hope to get some time on a vibration interferometer in April or May.

OK not to Hijack the thread but ...(Randy) yea we are planning to go to Montreal. I exhibited at both Newport 04, and 06 and would return if it was still in RI. Actually I just got an inquiry from someone from those shows in Newport, so even a couple years later ... ya never know. I had thought about not doing any shows this year, but Montreal was although a lot of work, was fun, so we decided to go back. I think a couple of the guitar in my avatar, should be there.

Randy the way I have been testing is very basic . Just holding the guitar by the neck approx 6 inches above my desk in front of the computer speakers. I have the strings dampened with some paper towl then just turn the volume up and scan through the freq range until I can see the most movement from the top. The picture of the top bracing was taken not long before closing the box . The only difference is that the lower legs of the X were taken down to 5/8 from 3/4. All braces are 1/4 inch . The upper transverse braces are 3/8 by 3/4. The top was thinned to .115 no graduating. It is a Jim Olsen brace pattern. My feelings on the top was that it was not very stiff although the bracing stock was. Being a first instrument I just tried to stay on the conservative side. I did note deflection results for both the top and back plates and the bracing stock to give me somewhere to start with # 2. I think the X lap joint might be a little high as I'm getting no sign of saddle rotation even with 13-56 strings. Any way I think it has the potential to sound quite good just need to dial in the top and back a bit better. I really appreciate all the comments and suggestions. Thanks

Craig.

The conclusion I've come to is that if the free soundboard doesn't resonate well..then there is no way it will resonate once you glue it up...since thru gluing it we are basically stiffening it up again. The trick seems to be to make it loose enough to resonate while free as well as once it is glued up....but if you make it too loose you will miss the mark due to a few different reasons.

Cool discussion all around...I think this is where the rubber hits the road with these instruments..!

_________________
http://www.PeakeGuitars.com

And my original post 'disagreeing' with you was because I thought you were promoting the opposite view! Violent agreement indeed!

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Randy I'm really interested in what you said with regards to the monopole. Am I correct in assuming that the monopole is the major contributor to the main top pitch, the cross dipole and long dipole being secondary to this.
I am a bit wary of messing too much with what I have at present for fear of ending up with an expensive pot plant holder or fire wood . At the same time I really would like to warm up the sound as it is very bright at present. Would lowering the height of the X lap joint achieve something similar to scalloping the lower legs as far as dropping the monopole?

Regards

Craig.

Craig,

Yes the monopole is the main resonance frequency of the top where the whole top is pumping in and out. Don't worry about the cross and long dipoles at this point. If you are really interested in learning about all of this, read everything Al Carruth has ever written and then some. A lot of people build wonderful sounding guitars without worrying about any of this stuff. I didn't want to say so, but it does look to me that you were very conservative in bracing the top. This is very natural for your first guitar. If everyone of us could build the perfect guitar their first time, none of us would be in business. This is a long learning process.

You could probably reduce the height at the X and also taper the lower legs, but of course you do so at your own risk. Internet quarterbacking doesn't work. Or you could just say, "Hey I did a pretty darn good job for my first guitar" and be happy with it. It may also warrm up a little over time.

Also, your focus seemed to be on getting the top and back coupled together. You may be aware that there is a whole camp of fine builders who do not do this, but build with stiff backs. Having them couple is not a requirement for building great warm sounding guitars. How the top is built alone will have a far greater affect on the sound.

_________________
Randy Muth
RS Muth Guitars Website
RS Muth Guitars Blog
Facebook Fan Page

Thanks Al, Bob, Alex, Randy ,Parser. The free plate testing is definitly something I will be doing a lot more reading about. I really need some way to work towards consistent repeatable results. Trying to work inside the assembled instrument is not something I want to continue to do if I can at all help it.
Thanks Randy I will proceed with caution

Regards

Craig.

Bob, The main thing that caught my eye in your original post was that it seemed you were suggesting that the dominant modes of an unbound soundboard will be the same modes present in the top once it is glued onto the rest of the box. I would definitely agree that the stiffness data for the unbound top has a lot of bearing on what the top will sound like once assembled..but I think it is important to make the distinction that the modes once assembled will be different from those unbound modes.

I hope this explains my thoughts more clearly..!

Thanks,
Parser

_________________
http://www.PeakeGuitars.com

That single mode (the ring mode) will be there on a glued soundboard. It might not be very efficient, but we wouldn't get much sound out of it if the bridge couldn't move up and down. So that's the sense I'm speaking of modes 'remaining'. Not every soundboard will have a closed ring mode as a free plate, but they all will (or close enough anyhow) when they're on the rims. So the idea is that if it's capable of the diaphraphm-type motion as a free plate, then it'll retain said ability (and efficiently) once bound. The other modes, those dont' really make much intuitive sense to me on a free plate let alone on rims, just that one. So, yeah, I make no claims whatsoever about the rest of 'em (and I can see how thinking that was my intent would jar you a bit, because that would be a silly claim to make!)

This is the one I'm talking about, BTW, if I'm being unclear:
http://www.kettering.edu/~drussell/guit ... /mode3.gif

The rest of 'em...who knows? Last I heard, even He Who Knows had a hard time correlating them to anything hard.

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

That explains it a bit better....I agree that it seems like there could be a similar mode but my gut tells me that the bound plate will exhibit this mode at a higher frequency than the unbound plate. I'll have to keep an eye (and an ear) on this during my next build and see if I can note any similarities. Thanks for your explanation..!

_________________
http://www.PeakeGuitars.com

I have Chladni data on roughly 100 guitars at this point. My data shows the following:

The mode which shows as a ring in the lower bout typically appears at about 250 -260 Hz on my free plates. Once glued to the rims with the back of the guitar on as well, a similarly shaped mode typically appears at about 215-225 Hz.

Mark

You're adding stiffness around the rim, which would move it up...but you're also adding a bunch of mass with the bridge, which moves it down. The back tends to move up, the top moves down. The reasons are my best guesses, but the result (back up top down) is empirically derived.

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Good points! Thanks guys..

My experience is much the same as Marks. Interestingly enough the assembled 'main top' mode usually falls somewhere around 200 Hz on mine, (without the bridge in place) no matter what the pitch of the 'free' plate mode. THe range of variation on the assembled modes is from about 180-230, with the great majority being within 10 Hz of 200. The 'free' plate 'ring+' frequencies range from 207-286. There is no correspondance between 'free' and assembled pitches; some of the highest 'free' pitches have been near the low end of the 'assembled' pitches. 'Closed' free plate modes may change less than 'open' ones, but I'd have to study the data a little more closely to be sure. I have to point out the possibility that the guitars I have data on may all be similar simply because I made them!

Any resonant mode is a 'standing wave' pattern. In a simple system, like a string, a bending wave runs along the string, bouncing off the ends. You get a resonance when the frequency is such that you're giving the thing a kick every time one of those reflected waves comes by, so that it gets reinforced.

On a plate you've got waves running in all directions. Patterns that form 'spokes' on 'free' plates are modes where the edges are fluttering, so to speak: you've got two sets of waves chasing each other around the edge in opposite directions moving faster or slower from one place to another depending on the relatinoship between mass and stiffness. These go away when you glue the thing down becauser the edges can't bend.

'Ring' type patterns are formed by waves bouncing off the edges or off something like the heavy upper cross brace. If the edge is 'free' the reflected wave comes back in the same phase as the incident wave (if I've got it right): that is, if the wave traveling toward the side is 'up', the reflection will also be 'up' in sense. The wave reflecting from a fixed edge, or something heavy and stiff like that brace, will come back in the opposite phase: an 'up' incident wave will give rise to a 'down' reflected wave. The 'ring' patterns that you see are just the places where the incident and reflected waves cancel each other out.

As far as I have been able to tell, you can get a reasonably nice looking ring on any assembled guitar top. It can't be broken up into 'spokes' because the edges are stiff, so the top will sort itself out into a 'ring' mode at some pitch. The tops just seem to work better in some way when the 'free' plate 'ring' type modes are closed. What's hard is to point to excactly how they work 'better'. For one thing, it's hard to accumulate a lot of data when you have to build saleable guitars to get it; that's slow work. I've started some experiments on simpler systems, but even that could use up a lot of time, and one does have to make money once in a while.

Thanks Mark and Al ,This info really helps me a lot . I'm looking forward to putting this into practise with #2.

Just one final question . When testing the free plates what is the best way to support the plates?

Regards

Craig.

Thanks Mark and Al ,This info really helps me a lot . I'm looking forward to putting this into practise with #2.

Just one final question . When testing the free plates what is the best way to support the plates?

Regards

Craig.

Mark Blanchard had an excellent presentation about his extensive knowledge and use of the technique at the Healdsburg guitar show which left me a bit confused but at least aware somewhat. He jots down pattern sketches for every frequency test, using xeroxed diagrams of tops, and files them away. It takes a few years I gather to figure out which shapes are best. It is definitely done before a top is put on (that much I do know). One vague tidbit of information I found interesting to my neophyte mind, was he seemed to imply tops with less "silking" or across the grain strength were more responsive (or maybe better sounding)--which surprised me. But maybe I didn't understand what he said correctly. Someday...

_________________
"Preoccupation with an effect gives it power and enhances the error"
from "Your Owner's Manual" by Burt Hotchkiss.

The standard way to support plates for 'free' plate tuning is to use small blocks of soft foam rubber. I often cut pieces out of 'eggcrate' surface foam, and glue them onto little squares of plywood to keep them upright.

One thing that's important is to move the supports around the where the node lines are. Leaving the pads in an active area will damp the mode some, and often change the pitch and shape. Usually I'll put the pads where I think they ought to go, monkey with the frequency and speaker position until I start to get a mode showing, and then move the pads to the actual lines.

I thought Mark's presentation was pretty clear, but then, I've been at this for a while. It often takes people some time to learn to find modes, and then you have to figure out what they mean, if anything. It's a skill that has to be aquired, like everthing else about guitar making.

I don't think there was anything in my presentation that said that less cross grain silking was desirable. At least I didn't intend to make that statement.

I did say that I feel there is an optimal lengthwise to crosswise stiffness ratio for a given size and shape of guitar. That means that I might consider a top to be too stiff across the grain (relative to its lengthwise stiffness) for a particular guitar.

Mark

Thanks Al

Craig.

I think, if you set up a jig that would clamp one edge of the plate and load the opposite edge, creating a cantilever, if you could then measure the shape of the deformed plate as a function of distance along the cantilever, you could probably back out at least average values for variation in cross-grain stiffness for a given plate.

It would be a good and worthwhile numerical test to examine how a plate responds either to random variations in average properties (statistical uncertainty in mode shape) or to systematic variations such as smooth changes in cross grain stiffness across the plate. Either one sounds like tests your student could do.

_________________
Jim Kirby
kirby@udel.edu

One of the problems with testing cantilevers is that the determination of the properties varies sensitively with the length of the piece in bending. I've been told that the only way to be certain that a piece of wood in not bending in the clamp is to clamp it hard enough to slightly crush the surface. Not much good for a top.

A better way might be to do deflection tests between parallel linear supports. If the supports were 4" apart, and you moved the plate 2" between tests, you'd have overlapping data that might allow you to graph out the variation with some exactness. The crux is whether you could get enough deflection over that small a span to calculate moduli to a usable precision.

We also have to keep in mind that most of the 'shop level' tests we do have an error bar of plus/minus a few percent. Most of us can't measure things like thickness or mass all that precisely, and there are built-in simplifications in the calculations. After a while it just gets eaiser to cut wood.