17 replies to this topic

[Johnmasters]

My name and FEA project was mentioned in posting 370 of the accupuncture thread. Please let me know what I am supposed to be able to do or not do with my FEA models?

Also, I can't see 23 pages of comments on the accupuncture thread. The initial picture is flawed, badly. It shows traveling waves on a closed surface (guitar). Furthermore, a point defect is seen to cast a perfectly sharp shadow.

Problem with the first is that there are reflected waves if the back does not dissipate all of the energy. The second overlooks the situation of diffraction. The understanding and math for these things is nearly as old as Stradivari.

[Wm. Johnston]

It's hard to tell due to the total number of pages in that thread. The problems with the vibrations shown in the video was brought up at least three times in the discussion, and one of those times was a post of yours. Yes, the video's animations are bad.

[James M. Jones]

My name and FEA project was mentioned in posting 370 of the accupuncture thread. Please let me know what I am supposed to be able to do or not do with my FEA models?

Also, I can't see 23 pages of comments on the accupuncture thread. The initial picture is flawed, badly. It shows traveling waves on a closed surface (guitar). Furthermore, a point defect is seen to cast a perfectly sharp shadow.

Problem with the first is that there are reflected waves if the back does not dissipate all of the energy. The second overlooks the situation of diffraction. The understanding and math for these things is nearly as old as Stradivari.

I read it as more of a statement than a question. That being that any attempt to model with computers the acoustic movement of a violin is flawed because wood has way to many variables to corner. A shadow tells me that whoever created the video was not informed at all ...a total fabrication.

[David Burgess]

John, don't worry about it. You're doing fine.

We've been struggling to know how to deal with someone who makes srtong assertions without experience in either violin acoustics, or violin construction and modification.

[Anders Buen]

My name and FEA project was mentioned in posting 370 of the accupuncture thread. Please let me know what I am supposed to be able to do or not do with my FEA models?

Also, I can't see 23 pages of comments on the accupuncture thread. The initial picture is flawed, badly. It shows traveling waves on a closed surface (guitar). Furthermore, a point defect is seen to cast a perfectly sharp shadow.

Problem with the first is that there are reflected waves if the back does not dissipate all of the energy. The second overlooks the situation of diffraction. The understanding and math for these things is nearly as old as Stradivari.

Hi John,
I do not think anybody ask anything from you. I just mentioned your name regarding the FEA models you have sent me. I have been thinking loudly in that thread as part of the ditching of plate tuning techniques that came up there. And I was thinking around some FEA results Colin Gough have done on a violin body similar to yours with a top, back and ribs plus the soundpost varying the rib strength from very weak (low E module, about a tenth of the normal) up to about ten times the normal values and tracking the body modes in small steps.

After a long time off these models I finally dug them up and did a calculation today with a weak rinb, just cutted some zeros off the the E module of the rib material used. I have modified the model a little with a thinner top and model the SP as locking two nodes in the top and back to the same Z movement (or all of x,y and z, I do not remember exactly), to keep the mode number below the limit at 1000.

I can find a B1- mode and some of the expected 'free plate modes'. And I can do that analysis if I find it useful (a similar mode tracking study as Gough have done). I am thinking about it as a sort of support for plate tuning to make sense, that there indeed are some correlation between the free plate properties and the assembled violin modes (An other or alternative way of showing it than my statistical analysis). But we'll see. There are a lot of modes to keep track of. And I have been working on two other projects today. One documenting humidity effects on acoustic measurements on violins (will appear on knutsacoustics.com) and another on bridge/body hills in Hardanger fiddles, which are very clear on instruments with thickish central tops. I have a new light PCB accelerometer now, only 0,3g. I measure point accelerances at the bridge side, and that gave some new intereting insights e.g. regarding the phase of the bridge movement.

I do have an idea on how that Hardanger fiddle bridge moves, but would like to show it in calculations, or possibly in a scaled up physical model using modal analysis. I have a FEA bridge model of a Hardanger fiddle bridge in Abaqus, but made the model and meshed it from within the program and have not been able to get it to run. It has string constraints and loads on it etc, much like what Carl is up to. Your approach programming the input in text probably is much safer John, but it takes some time to learn how to do it. And I am not patient enough to run through the tutorials. But I probably will using some simpler shapes, at some time. Or build the model file like you do in a spreadsheet/text. I could thus mesh the bridge by hand. Possibly the moethod Oliver Rodgers once used for his bridge models.

Edited by Anders Buen, 16 July 2012 - 05:06 PM.

[Johnmasters]

John, don't worry about it. You're doing fine.

We've been struggling to know how to deal with someone who makes srtong assertions without experience in either violin acoustics, or violin construction and modification.

Many thanks for the reinforcement. I am a bit of a paranoid and do not like to think that people are having bad thoughts about me.

Anders, below wants to modify my model to allow for extreamly flexible ribs. That is easy to do in the way he mentions. In order to simplify, I made the ribs of isotropic material, no grain. One can make them as light as air simply by editing the input file I gave him.

Hi John,
I do not think anybody ask anything from you. I just mentioned your name regarding the FEA models you have sent me. I have been thinking loudly in that thread as part of the ditching of plate tuning techniques that came up there.

As I say above, the ribs are of an isotropic material for simplicity. In one model, I had an extra 62x2 nodes to mark out shell elements for linings. If you don't have it, I will try to send it.

My models do strange things in ABAQUS now, something is corrupted, and I get a lot of missing elements.

Using the eigenvalue finder approach will not lead anywhere. The program finds the eigenvectors and normalizes them so that the maximum is 1. Unless you can actually see the ribs move, the motion will have to be assumed to be very small. If you reduce E by a factor of several orders of magnitude, you might start to see things. I am guessing that E would have to be damned small.

I made one model with strings. I could not see the body move even though the strings appeared to move back and forth several feet.

In order to do something useful, I have thought to learn the force-driven version which is called ABAQUS explicit. That looks like work, and I have been too busy to do it.

But....... if one did this, one could see a string driving the box between resonances, and that might be helpful even with the small number of nodes. The top has 300+ shell elements, so that there would still be a lot of modes to mix to give the answer to a driven problem.

[Johnmasters]

John, don't worry about it. You're doing fine.

We've been struggling to know how to deal with someone who makes srtong assertions without experience in either violin acoustics, or violin construction and modification.

Knowing the normal modes is a good thing. But the simple FEA version I used will not tell the Width of the resonance. One has to actually drive the system at a frequency. (which can be, but does not have to be that of a normal mode.)

Seems to me that width of mode resonances is very important for tone, because there are so few down in the bass region, at the start of the response graph. Try making a good speaker using ducts and ports..... it is a nightmare.

Resonances are broadened by damping, as you know. For my personal experiments, I will attempt to add damping to the back by not using a high-efficiency sealer. In other words, keep as much varnish out of the top as possible, with a trick filler/sealer. But don't use this on the back. It would be interesting to see the result, when you consider that many backs were made with unconventional wood with odd grain orientations etc.

[Johnmasters]

I read it as more of a statement than a question. That being that any attempt to model with computers the acoustic movement of a violin is flawed because wood has way to many variables to corner. A shadow tells me that whoever created the video was not informed at all ...a total fabrication.

All that means is that you need a million-dollar FEA package such as Boeing may have. And you would need to do detailed models of wood with its cellstructure. You would need to know, if only to a first approximation, the behavior of the materials. Aluminum is easy, imagine making a 747 out of wood. (Howard Hughes did it !!

Also keep in mind that my FEA experiments only found mode resonances. It did not even find the relative importance of them. To do that, one would need a time-dependent model that was struck with a hammer or such. That is way beyond my skill, but there IS the possibility of reformulating the model in ABAQUS explicit. You can put in a time-dependent force someplace. You should get out the motion of the article. But I have never tried to do it. It looks difficult.

[James M. Jones]

All that means is that you need a million-dollar FEA package such as Boeing may have. And you would need to do detailed models of wood with its cellstructure. You would need to know, if only to a first approximation, the behavior of the materials. Aluminum is easy, imagine making a 747 out of wood. (Howard Hughes did it !!

Also keep in mind that my FEA experiments only found mode resonances. It did not even find the relative importance of them. To do that, one would need a time-dependent model that was struck with a hammer or such. That is way beyond my skill, but there IS the possibility of reformulating the model in ABAQUS explicit. You can put in a time-dependent force someplace. You should get out the motion of the article. But I have never tried to do it. It looks difficult.

Howard Hughes did a lot of things! I won't pretend to understand all the tech modleling or how you create this type of program ...Way over my head.... But I have a feeling that you are headed in the right direction....At least someone is working to shed light... I've learned a lot just watching the simulations.Just the understanding of a 3 dimentional twisting ,bending,vibrating body.as opposed to old model single plane vibration is a real eye opener.

[Johnmasters]

Howard Hughes did a lot of things! I won't pretend to understand all the tech modleling or how you create this type of program ...Way over my head.... But I have a feeling that you are headed in the right direction....At least someone is working to shed light... I've learned a lot just watching the simulations.Just the understanding of a 3 dimentional twisting ,bending,vibrating body.as opposed to old model single plane vibration is a real eye opener.

The FEA is not so much a research tool (as I used it) as a demonstration. A body like a violin can move in all kinds of arbitrary ways.

Amongst all of those motions, a few are all coordinated so that every motion occurs at the same frequency. That means the entire motion is cyclic and coordinated. THESE ARE VERY SPECIAL CASES. (And are called the normal modes)

However, these special cases of motion can be combined in various combinations to produce any of the arbitrary motions. A summation over them can reproduce any arbitrary motion.

That is really all that the special cases are.. and are called eigenfunctions which move in concert with a given frequency. These so-called normal modes "span" the [vector space] of all possible motions.

Knowing these normal mode shapes and frequency are interesting, but it is difficult (impossible for me) to find a connection with the quality of a violin. That is, without correlating data and outcome for a lot of actually made violins. Like most other data, it is somewhat helpful if you know what you want from it. It is not any kind of cure-all.

An analogy: any 3-D vector of velocity for example can be a linear combineation of unit vectors of velocity in the three directions. The set of normal-modes are like unit vectors in a space of dimension whose number is the number of degrees of freedom of the system. Any motion can be seen as a linear combination of the normal mode motions. Sorry to lay this on you, but the idea is essentially the same.. add up components to get the whole.

[MikeC]

What I want to know is if you put two point defects close enough together will the waves become particles?

[Johnmasters]

What I want to know is if you put two point defects close enough together will the waves become particles?

I will assume you are not pulling my leg. The wave-particle idea arrises only in modern physics. Huygen's wavelets and any other wave math apply for all the mechanical waves in a violin.

Any addition of two waves remains a distinct addition. That is called "linear superposition" and it is very important to keep in mind. It is always true for "small" motions.

By the way, the wave-particle duality in QM is open to a lot of interpretations. Nothing is really simple like in the good 'ol days.

[MikeC]

well I was being a little tung in cheek. But It's an interesting question. Does sound have a wave/particle duality. My initial thought is no, because it is just a compression / decompression in a material medium. That's why there is no sound in a vacuum. Matter does of course but does the sound energy traveling through it? But after looking it up on a physics forum... well maybe there is a phonon.
http://www.physicsfo...ad.php?t=413608

Well anyway, I don't think you could recreate the double slit experiment in a spruce violin sound plate... well maybe if you could train Shroedinger's cat to play it. That's what it sounds like when I try to play one. Sounds like a long tailed cat in a room full of rocking chairs.

So in this acupuncture idea how are you supposed to determine on that vast expanse of wood exactly where to place the pinprick. Is there a method or strategy or is it just voodoo. Same with Fry's (I think it was Fry?) poking a coat hanger with a bit of sandpaper on the end through the F hole and scraping a few molecules off the inside 'somewhere' and then the violinist plays it and it sounds so much better and the audience agrees. Well if that isn't delusion and a good example of the placebo effect then what is. And then what about all the dents and dings and varnish wear and scratches from 300 years of use? Those aren't a lot bigger than a pinprick? And they have no effect but the pinprick does?

[Marty Kasprzyk]

well I was being a little tung in cheek. But It's an interesting question. Does sound have a wave/particle duality. My initial thought is no, because it is just a compression / decompression in a material medium. That's why there is no sound in a vacuum. Matter does of course but does the sound energy traveling through it? But after looking it up on a physics forum... well maybe there is a phonon.
http://www.physicsforums.com/showthread.php?t=413608

Well anyway, I don't think you could recreate the double slit experiment in a spruce violin sound plate... well maybe if you could train Shroedinger's cat to play it. That's what it sounds like when I try to play one. Sounds like a long tailed cat in a room full of rocking chairs.

So in this acupuncture idea how are you supposed to determine on that vast expanse of wood exactly where to place the pinprick. Is there a method or strategy or is it just voodoo. Same with Fry's (I think it was Fry?) poking a coat hanger with a bit of sandpaper on the end through the F hole and scraping a few molecules off the inside 'somewhere' and then the violinist plays it and it sounds so much better and the audience agrees. Well if that isn't delusion and a good example of the placebo effect then what is. And then what about all the dents and dings and varnish wear and scratches from 300 years of use? Those aren't a lot bigger than a pinprick? And they have no effect but the pinprick does?

[Marty Kasprzyk]

Maybe dents and dings are actually helpful. I wonder if anybody has ever shot-peened a violin. It helps improve the fatigue strength of metals parts.

[Michael_Molnar]

Maybe dents and dings are actually helpful. I wonder if anybody has ever shot-peened a violin. It helps improve the fatigue strength of metals parts.

Great thought! Get out your pellet guns for target practice. Well, I think that's what hazelfichte (bearclaw) does naturally.

Now, maybe one of our colleagues will start crushing or stamping violin plates. I wonder who would be the first.

Stay Tuned.
Mike

[James M. Jones]

Maybe dents and dings are actually helpful. I wonder if anybody has ever shot-peened a violin. It helps improve the fatigue strength of metals parts.

Sorry, IMO this comparison is a non starter,
Metal has a very different structure from wood.Steel has a dentritic, or tooth like,crystalline structure, while wood has a cellular structure. With steel the crystals can be pushed into new forms without separation to form a stronger material,Shot peening mechanically replaces tension forces in the surface of a workpiece with compressive force, so that surface micro cracks are being pushed together.When steel is formed hot it expands, then on cooling it shrinks,this shrinking is uneven, with the outside cooling much faster than the inside. This shrinking action results in surface tension,peening relieves the tension and replaces it with compression, rather than being pulled apart. Where as with wood the same action will crush or fracture cell walls producing a weaker structure on the surface.Burnishing wood will produce a smoother surface,but only because of the crushing of fibers.

[Marty Kasprzyk]

Sorry, IMO this comparison is a non starter,
Metal has a very different structure from wood.Steel has a dentritic, or tooth like,crystalline structure, while wood has a cellular structure. With steel the crystals can be pushed into new forms without separation to form a stronger material,Shot peening mechanically replaces tension forces in the surface of a workpiece with compressive force, so that surface micro cracks are being pushed together.When steel is formed hot it expands, then on cooling it shrinks,this shrinking is uneven, with the outside cooling much faster than the inside. This shrinking action results in surface tension,peening relieves the tension and replaces it with compression, rather than being pulled apart. Where as with wood the same action will crush or fracture cell walls producing a weaker structure on the surface.Burnishing wood will produce a smoother surface,but only because of the crushing of fibers.

I'm aware of all that. I wondering if the dents and dings on violins affect their sound. Maybe I should slightly restate the question: Has anybody ever dented a viola to see how its sound changes?