Official Luthiers Forum!

I got a chance to use my new Maddell tone generator (thanks Mike) on a top I am working on tonight and I have some questions for those of you who are familiar with this process. I swept from about 50-400 and found 4 frequencies that were very distinct. They are shown below. The top row was the first time through and then when I got to 240 I used the sticky tack to guide my brace shaving and successfully closed the circle. Then I went back through the frequencies I had mapped earlier to see how they changed. The dots are where the foam pads were.



I seem to remember that in Mark Blanchard's presentation, he was mapping more than 4 frequencies on each top. I was surprised that that was all I found. I also remember hearing something about trying to find a frequency where the node went up and around the soundhole like a horseshoe. I thought I would see something like that but I can't see which of these could do that. Also, when I tried to test the back the vibrations made all of the poppy seeds dance right off of the back. Is there a trick when using a substantial dome to the back to get them to stay on? Is a different material more apt to remain on the plate but still be able to move to the nodes? And do most of you test the back as well as the top? Finally, is there a correlation between how close the closed circle is to the edge of the plate and how the guitar sounds? I realize that each person has their own data, but is something that you follow? Thanks very much.

_________________
Burton
http://www.legeytinstruments.com
Brookline, MA.

The most important one is there, anyhow, in the ring mode. A lot of the modes only appear at very tight, specific frequencies and you can blow right by them unless you're going up very slowly in frequency. Glitter doesn't dance off the back as much, probably because it's flat in cross-section rather than round like the poppy seeds.

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Those modes look pretty good to me. The frequencies are lower than mine usually are, but then I build a pretty stiff top.

The "horseshoe" mode should be in between the one at 151 and the one at 237. Hold the speaker either right at the tail block position or right below the soundhole.

I would like to see some of the higher modes. The cross dipole is an important one to me. Based on your other modes it should be in the upper 300s or low 400s.

Sometimes if your speaker happens to be positioned right over a node line when you sweep through a mode's frequency, the mode may not become very active. Try moving the speaker around a lot s you sweep slowly through the frequencies.

Using a material that isn't round will make dealing with curved plates easier. If you think backs are hard, try it on an archtop plate !! I use table saw dust. It doesn't roll around and I have a seemingly endless supply of it.

It is not really necessary to show the foam block position. It is a good idea to position them on a node line, but having done that, the position will not have any significant effect on the mode shape or the frequency and so it isn't important data.

Your mode shapes are very symmetrical. Are you using a symmetrical bracing pattern ??

Mark

Burton (or Mark), could you explain?

_________________
Dave
Milton, ON

Uhhhh... ya..... I was wondering about that too.

?????

Mark

Thanks for your replies! I will go back and sweep slower through the frequencies, I am very interested in seeing what else I can find. After you said it Mark I remembered that you had advocated table saw dust, I will try that too. I am using symmetrical bracing. It is a little unorthodox, but not too much and has served me well. I'll post a photo when I get home this evening. The sticky tack thing is something I found in an old MIMF thread. Someone said that adhering a little of that blue tack stuff to the brace would act as a mass increase which would mimic a stiffness decrease fairly well. I.e. if the sticky tack changes the nodes in a productive way, remove brace material there. I didn't have to do any radical brace shaving but it worked pretty great for me. Removing a little from where I put the tack did change the node line almost exactly the same way. I was surprised when I read it that people don't advocate it, it made me think it had been proven problematic or something, it was a thread from 6 or 7 years ago I think.

_________________
Burton
http://www.legeytinstruments.com
Brookline, MA.

Alan Carruth has mentioned the use of Blu-Tac or some other type of non-staining poster adhesive. Appears that it mimics shaving a brace to decrease stiffness by altering the stiffness/mass ratio of that area. Seems counterintuitive, but it does appear to work.

_________________
now known around here as Pat Foster
_________________
http://www.patfosterguitars.com

Isn't that interesting?! ...Learning all the time!
Thanks.

_________________
Dave
Milton, ON

What Mark said.

237 for the 'ring+' is not too low: I've had it below 230 a few times on steel strings with no problem.

I use aluminum glitter, which is what I learned from Carleen Hutchins. DON'T use the cheap plastic kind: talk about static cling! Augh!

As Mark said, the dipole mode would be a good one to look for. There are a couple of others you might try to find, if only for the practice. Each mode tells you something about the mass and stiffness distribution of the top, and as we go on we should learn more about what the 'other' modes mean.

That makes sense...the natural frequency (f) is always proportional to (k/m)^.5 where k is stiffness and m is mass. Adding mass and decreasing stiffness both result in a lower note and vice-versa.

_________________
http://www.PeakeGuitars.com

Where does one get this aluminum glitter

I am really ignoratant about this chladni thing. What exactly are you trying to accomplish by changing the patterns? Are you aiming for increased volume of the guitar, improved balance, improved "tone", or some combination. How do these patterns relate to the goal?

It occurred to me that the node shapes might not relate well to volume since it seem to me that the patterns indicate a relative stiffness that you are looking at. Any light you guys can shed to bring me up to speed would be appreciated.

John

John, I dont know much abut this either , but what I think they are trying to do is have complete patterns, that is nodal lines with no "broken" or incomplete lines , which , in theory shows the plate to be vibrating at its potential .... Jody

I got a quick moment last night to record some more frequencies and found 2 above 300. They were as unsymmetrical as the first ones were symmetrical! I didn't have time to find the lower Horseshoe one but I will look tonight hopefully. I may post them because I am not sure which is the cross dipole.

John, I am a beginner into this stuff but I can give you a beginner answer. From what I understand, the real reward of the testing is apparent through time. After many guitars of the same size being tested you start to recognize similarities in the patterns of the ones that sounded better and search out wood and similar bracing nuances that those guitars shared. There are some very general ideas that seem to be universal for most (but not all?) guitars. Finding the mode that shows a circle shape in the lower bout is one of these and that is what I was attempting to do, to change the bracing so that that circle appeared closed. I did find that the more closed it got, the better the tap tone was. I think of it as a way to trust your ears with your eyes in that sense. For someone who has lots of experience using them I think it is safe to say you can see quite a bit about the distribution of stiffness on the plate (even more on an unbraced plate) without doing deflection testing. Probably doing both though and keeping records is the fastest way to begin to see how they match up. As a supplement to tap tuning and deflection testing(whether with your hands or instruments), it seems to be an excellent tool to maintain the high quality of the best instruments you can make. I hope that helps and hopefully others will fill in where this description is lacking.

Al, I will get some of that glitter, thanks.

_________________
Burton
http://www.legeytinstruments.com
Brookline, MA.

Good craft stores will have aluminum glitter. The stuff from Wal-Mart or wherever is often plastic. I get it in three colors and mix them, so it always shows up no matter what color wood I'm testing. One of these days I'm going to use a different color glitter for each mode, and super glue it on. It might make the final French polish a bit tricky, but it sure would look cute!

Most people who use Chladni testing are after 'better' tone, whatever that is. More importantly, though, we're after more consistent tone. If you get the same mode shapes on two similiar guitars (and the shapes seem to count for more than the frequencies, within reason), they will sound more alike than if the mode shapes are different. That's where the record keeping comes in: you record what mode shapes and frequencies you got, and how you (or your customers) liked the guitars, and after a while you figure out what works and what doesn't. Since you probably hear better than I do, and almost certainly have somewhat different tastes, you might find that what you like to see in the modes is not the same as what I do.

As Mark pointed out in his excellent talk last summer at H'burg, the modes are just 'visible tap tones': like written language as opposed to spoken. That's why they're so helpful. I can't remember the tap tones of the guitars I made five years ago, but I don't have to: I've got all of that stuff written out in a file folder. All I have to do is pull the file and I'm good to go on a copy, if that's what I want. Even better, I can pull all the folders on all of my best guitars and see what they all had in common. I haven't done the amount of 'data mining' that Mark has, but I've learned a lot over the years.

All of this begs the question of why 'closed' ring-type modes seem to make 'better' guitars. Nobody knows for sure! We have a couple of theories or models that seem to make some sense; the most commonly cited one is that, since the modes will have to be closed on the completed guitar top, getting them to close on the free top just gets the top to do what it has to because it 'wants' to. There are fancier ways of putting it, but that's the essence. It's hard to develop data that woud 'prove' that one way or the other, so at the moment we're sort of stuck with that. It's probably best just to think of this as a quality control technique for now, and not get too het up over the theoretical underpinnings.

This is beginning to make some sense. I appreciate everyone's patience with a neophyte.

I noticed that the patterns were only closed at 240 and 237. At least that seems true according to my understanding of what constitutes a closed pattern. If this is true, then are you looking for closure at any old frequency, or do you want it to be at or near 240?

The closed ring mode generally only happens at one frequency on a free plate.

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Bob:
I think he meant '240 before making some adjustments' and '237 after'.

Closed is closed, and I take what I can get...

There are actually a couple of different pieces of information here: the mode _shapes_ tell you something about the distribution of mass and stiffness within the plate, in a relative way. That is, if the bracing is too light or too heavy for the top, or the brace angles are 'wrong', the modes won't close. The _frequencies_ tell you something about the ratio of mass and stiffness. If the plate is light or stiff it will tend to have higher mode frequencies. I'd worry a bit about a light weight top that had low mode frequencies. That, in fact, was the problem with my first 'sandwich' top, and, sure enough, it didn't hold up too well. I should have known, but, as Feyman said, the easiest person for you to fool is yourself. Next time (I keep telling myslf) I'll trust the numbers more.

Of course, I'm going to cop out on specifying what might constitute 'light' or 'low frequency'. I tend to be pretty conservative as a builder: so long as I can get the tone I want, I'd sooner over build than under build. Thus I haven't had too many that self-destructed, so I can't really say what the limits are. Maybe, as time increases my bolditude I'll learn.

It also seems to be true that there are a few different modes that can be coaxed to form more or less closed ring+ shapes, or other kinds of 'closed' modes. When I made my first 'semi-matched' Small Jumbos, with 'normal' asymmetric and 'double-X' bracing, they both had modes at about 260 and 285 Hz. On the 'normal' braced top the ring+ was the mode at 260, and the other was a bit of a mess, while on the 'double-X' top it was the other way around, with the 285 mode being the ring+. Neither guitar had an actual closed ring: I could not get the 'normal' bracing to close, and was committed to profiling the 'double-X' the same way, as far as possible, to get a good comparison. Both were very close to closed, and both sounded good, if slightly different. In another instance, I built a classical that had a 'ring' mode (the one below the 'ring+' in pitch) that formed a sort of open ring+: so there were two ring+ modes, sort of. I'd never seen that before, and I put it together that way to find out what happened. It was the only guitar I've ever had rejected on the basis of sound: it was uneven. I was able to fix it, and the new owner is very pleased, but it was _not_ a lesson I'll forget in a hurry.

Let me reiterate: 'closed is closed'. I've seen quite a variety of 'ring' shapes, running from 'squares' and 'diamonds' all the way through to lovely round rings, and they've all worked pretty well so long as they were closed. I've been trying to run an experiment with a somewhat simpler system than a guitar (making guitars takes so long!) and the preliminary data from that suggests that there may not be much benefit once you get past 'closed', but there seems to be a pretty good difference between 'a little open' and 'just closed'. It's entirely likely that there is a tonal difference between a round ring and a diamond, but it might take a while to figure out what that is, and why.

Here's a sample set of fairly typical looking modes for one of my braced tops. The bracing is symmetrical and similar to the Larrivee pattern. Sometimes it is easier to find the modes if you have an idea of what shapes you are looking for.

It's a OO size spruce top, 14 1/4" lower bout.

Thanks for that input guys.

Mark, if you do not get that set of nodes when you first begin your test, I assume you begin to shave braces to try to achieve one of those configurations. Do you pick one particular frequency when you are shaving and testing to zero in on a particular chladni shape, and if so which one. Do you add the sticky tack material to the braces to help you spot which ones need shaving?

TIA for any help you can give.

John

First of all a correction.... the mode at 81 Hz should be 181 Hz.

Those shapes are what the modes looked like when I finished carving the braces. During the carving process I routinely check them to see how things are shaping up. I don't look at all of them every time but I do look at all of them now and then through the carving process just to make sure none of them are getting out of line.

Mostly I monitor the fifth one (239), the sixth one (257) and the tenth one (458). 6 and 10 are the only ones that really have the potential to form "closed" shapes. The others are more of a concern to me in terms of their frequencies. The frequencies can help me get a handle on how stiff the plate is in the areas that are the most active in each of the modes. There is also a directional aspect to each mode. The first mode (91) mostly indicates stiffness in the same direction as the grain. The fifth one (239) mostly tells me about the stiffness across the grain in the lower bout, especially the area below the soundhole.

The trick, of course is to know what distribution of mode shapes and frequencies will indicate a mass and stiffness distribution that will in turn produce a guitar that sounds the way you want it to. There is no universally "correct" mode set. But there can be a set that tends to produce guitars that you will like.

What I try to do is correlate what I see in the modes with what I hear in my guitars and then use that information to help keep me on track to building in the good stuff and avoiding the bad stuff. The only way I know of to figure out which is which is to build a bunch of guitars, keep records and look at he data to see what the great guitars had in common and what the not so great ones had in common.

Mark

Mark, this is very interesting.
Can we relate some of these frequencies to the different modes of the guitar. Like cross dipole, monopole, etc.?

Sure. Generally speaking a free plate that shows cross grain modes that have frequencies that are higher than usual relative to the long grain modes, will tend to produce a guitar that has a cross dipole that is higher than usual compared to the long dipole. The opposite is true as well. If the cross grain mode frequencies are low relative to the others on the free plate , then the cross dipole on the finished guitar will tend to be lower relative to the other modes.

I have found that lengthwise to crosswise stiffness ratio that is intrinsic to the top wood itself will tend to telegraph right on through to the modes of the finished guitar. It is possible to shift the ratio somewhat with the bracing, but only so much. The properties on the wood itself tend to dominate. You have to do fairly drastic things with the bracing to overcome the inherent properties of the top plate itself. At least in terms of what shows up in the modes.

Mark

I'll note that there is no _simple_ and _direct_ correspondance between the 'free' plate mode pitches and the pitches of the modes of the assembled instrument. The relationships Mark talks about are true in general, but not of a nature that would allow you to predict from the free mode frequencies exactly what the assembled guitar will do within a few Hz.

If you make two instruments that are well matched as to wood, design, weights, bace patterns, and so on, and the free plates have similar mode shapes and pitches, the modes of the assembled guitars will be very similar, or even 'the same', as far as you can find them using Chladni patterns. The more things you change, the less likely it is that the final results will be similar. Using a different back and side wood, for example, could alter the assembled top mode pitches, even if the tops were as alike as you could get them. You have to remember that the modes of the assembled guitar are modes of the _whole_ instrument; what shows up on the top is effected by what the back does. Even different binding or liners could have an effect.

That's why all of this plate tuning stuff is on somewhat shaky theoretical ground. There is no question but that the mass and stiffness distribution that sets up the free plate modes has an effect on the final tone, but getting from point A to point B is mathematically and theoretically a nightmare. There's too much going on, and even a very complete computer model usually needs a bit of of 'tweaking' to get close in the low range. Once you get to about 600-800 Hz it's a real mess.

On the bright side, the _exact_ pitches of the assembled modes don't seem to matter all that much, within reason. I find that when the 'main top' mode goes above about A#=233 Hz, the sound tends to be too 'forward' for me, with too much 'seperation'. OTOH, one I made like that for a good jazz player was just what he wanted, although he said that it looked at him funny if he played a plain G chord. Iirc, the 'ring+' on that one was pretty high, above 285 Hz. I don't like to go much below about 210 Hz for the 'ring+' mode, although the sound penalties are not as pronounced in that direction.

I looked up the records on a dozen 12-fret 000s I've made. The free plate ring+ modes ranged from 207-265 Hz, and the assembled 'main top' modes were from 173-232. The one with the highest assembled mode had a free top pitch of 256 Hz, another, with the free top pitched at 264 ended up with the 'main top' mode assembled at 201. It's just not all that simple. I wish it were.

Mark,

Thanks for posting those modes, It is always nice to have a baseline to start from and have an idea where to look. As I looked further on this top the modes became less symmetrical and a little jumbled. I also found with those I could change the frequency by up to 6 or 7 Hz by changing where the blocks were without radically changing the shapes. There were some in between modes that I could find definite activity in but not get defined enough shapes to emerge also. I think I am going to add some classical style braces in the lower bout running from the tail block to the sides and see what they change. I can always chisel them off. Anyway, here is what I have got. I think the one at 336 is the cross dipole but my angled lower transverse brace is defining the angle? I am not sure about this.

_________________
Burton
http://www.legeytinstruments.com
Brookline, MA.