Official Luthiers Forum!

I have only completed four guitars to date (just finishing #5 & #6) and with each I have just reduced the thickness of the top to a point where it "felt" right when bending in my hands. No real science, only a good guess based off what Serge DeJonge taught me (he didn't believe in tap tones or measuring deflection. He just gave us a good idea of what the finished thickness range should be for each particular wood).

I am a scientist in my paid job (still working on making building guitars a paying proposition) and being such I am now looking for a way to quantify what I've been "feeling". I've read some threads concerning Chladni patterns, tap tones and deflection testing but it all still seems a little like Voodoo to me. I just ordered John Mayes DVD on voicing tops and hope this will help (Thanks Hesh for the tip you gave in the "Do you measure while tap tuning?" thread).

Just wondering if any of you guys could enlighten me as to just how you go about testing deflections. Do you test the raw tops halves prior to joining at a predetermined thickness? After joining? Test each brace? The top after bracing? I'm a little overwhelmed , can I get some input from those who already gather this kind of data.

Thanks... Greg

[quote="GregE"] Do you test the raw tops halves prior to joining at a predetermined thickness? After joining? Test each brace? The top after bracing?

Greg,

I do all of these. I pre level the spruce unjoined or profiled and test it and then at each stage in between gluing it on the guitar. I test each brace also for stiffness and weight. When I get a large batch ready I test them all and weed out about half usually and then retest them before using them. It is most likely a little overkill and may be more an exercise in paying close attention than actual useable data. I try to make sure that I make any notes on the actual tactile feel of the wood at each stage also. In terms of weights and thicknesses, I think I first test rough tops at .14 and backs at .12 with a constant weight ( a little clock!) across 18" of span. I do the braces all at 5/16 x 5/8 across 18". The way I see it, especially at first, is that there is no "too much" information to have. Whether I ever reference it later or not, at least I could if I wanted to. As others have said too, noting the feel of the wood and very deliberately paying attention to how it feels in your hands and how it looks, smells, works and cuts/bends is equally as important as the hard data you can write down.

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

I am really amazed comparing my latest 2 guitars. One is a light and thin spruce, with light cypress B&S, the other one is a thicker and denser spruce, so it is heavier than the other twofold - AND the back is heavy rosewood. They are braced with the same wood, same way, and share a lot in common, but the light guitar sounds a little more open with more crystalline trebles while the heavier guitar is somewhat muted, even if it shows a lot of potential. But the old one has already half year of daily play in.
My point is that even if the stiffness is quite different - is sufficient to press around the top, the lighter guitar is quite flexy while the new one is tough as nails - there is a WIDE range between them, I could stick 2 other guitars in between and still feel the flex differences, the sound difference is small, you would need to pay some attention to tell them apart.
So what really matters is the weight to stiffness ratio not stiffness in itself. You might get a better ratio leaving a stiff dense top thicker than average, even if the general advice is to make it thinner !!

_________________
Build log

I am interested in this topic as well. Can someone give a short tutorial of HOW you actually do deflection testing?

thanks
Richard

Sure. I only measure the long grain stiffness. Some measure both, but this is my approach.

Build a jig like this. In mine the rails are 18" apart on center. I got a Starrett dial indicator on ebay for $15. The weight is a standard 5lb dumbell. It was a little heavy so I sanded it until I got it to weigh exactly 5 lbs. The weight isn't important. You can choose any weight you like.






When you use the jig, make sure your soundboards are glued up square like this one (different shapes will alter your results, so for consistancy sake try to keep them like this).



I also take density measurements and record this data as well. I am seeing an overall trend that the higher the density, the greater the stiffness. There are exceptions to this of course, but it is a strong enough correlation that I have come to expect it.



Put the soundboard in the rig like this. I put a mark around the pin so I can get back to the same place over and over. I usually try to mark it over the centerline.



Zero the guage (remember to bonk the sound board a few times and reset it until it stays pretty stable)



Apply the weight and notice the deflection. I try to place the weight as close as I can to the pin and on subsequent measurments I try to get it as close as I can back in the original location.





Then sand the soundboard a little thinner and remeasure. Remember the cube rule applies, so a small amount of material removed will make a big difference. Especially as you get close to your target. Do this until you hit your target deflection. Depending on your build style and the weight you use to deflect with will determine how much is "enough" and how much is "too much".

_________________
Brock Poling
Columbus, Ohio
http://www.polingguitars.com

that is exactly the same density I got from Shane's tops.

_________________
Build log

Yeah, I find most of the Lutz comes in around 7 g/in^3

_________________
Brock Poling
Columbus, Ohio
http://www.polingguitars.com

I only have a small scale and weight the offcuts, some came at 410 (about 6.8) others at 420 Kg/m3 which is about 7 in/g indeed.
But someone else who bought a batch of Lutz from Mario measured them at 450 Kg/m3 and that is Sitka like.

I have an Austrian Alpine top at about 410 too, and a larger numbers of top samples plus brace wood billets from Rivolta (italian spruce) which tend to be about 370 - 380 Kg/m3.

_________________
Build log

I do similar to Brock except I have the gage under the wood and my weight is round and it finds its own center.

_________________
"An adventure is only an inconvenience rightly considered. An inconvenience is an adventure wrongly considered." G. K. Chesterton.

I do it the same way as Brock. 18 inch spread and a dial indicator. I think I use the same weight. Some would say it would be better with a bar weight for even distribution. I am sure they are right but my system works for me.
My dial indicator is digital which is easier for my simple brain...
I also use an arm that is adjustable but can be locked in position. It was $15 at a local hardware store

_________________
Andy Z.
http://www.lazydogguitars.com

Brock:

Thanks bud, that's a great tutorial on deflection testing.

Now, would you be will to give up the secret data that you collect? How much deflection do you have as a target under these conditions? I would imagine that it varies quite a bit with respect to the shape and size of the guitar being built.

Again, great shots and explanation, Thanks

Greg

It isn't a secret as much as it is a hesitation to put that number out there without explaining the whole thought process I go through for voicing. It is all interconnected.

Understand the point of deflection testing is to try and remove some of the variability in the stiffness of each top. There is no magic to it. It is just another tool in our bag of tricks.

But, to answer your question. I think if you go between .200" and .250" you will be in a good place to start experimenting. I am constantly tweaking things and trying thinner tops, thicker tops, lighter braces, taller braces, different brace configurations etc. I don't think there is one magic way to get a good sounding guitar.

I will tell you that I have pushed the limits pretty hard and have used tops in the .300 - .350" range and they work fine. It is a completely different sound from a heavier top. Now I am experimenting with slightly thicker tops and lighter bracing.

Other things I am tweaking on are bridge plates and backs.

My goal in these experiments is not so much to build a "perfect" guitar, as much as it is to continually gain a better understanding of how to dial in (to the extent that it can be controlled) the sound I want on each build.

I think in the end... the "trick" is to build a LOT of guitars and to carefully pay attention to each one.

_________________
Brock Poling
Columbus, Ohio
http://www.polingguitars.com

Greg
It really isn't a secret. Each set up is a bit different. Each guitar shape requires a different thinking
I use the drum head principle so I want a greater deflection for a smaller body and vice versa.

here is some of MY data in MY setup
Groden SJ 6 string Lutz AAA 44 0.215
Groden SJ 6 string Lutz AAA 46 0.177
Groden SJ 6 string Lutz AAA 80 0.143
Groden SJ 6 string Lutz AAA 125 0.118
Groden SJ 6 string Lutz AAA 178 0.104
Groden SJ 6 string Lutz AAA 211 0.096
Groden SJ 6 string Lutz AAA 242 0.093

Here is on from an L-OO I went for a greater deflection. Smaller body

Lutz L-OO Lutz/AAA 64 0.153
Lutz L-OO Lutz/AAA 75 0.143
Lutz L-OO Lutz/AAA 120 0.127
Lutz L-OO Lutz/AAA 169 0.11
Lutz L-OO Lutz/AAA 241 0.096
Lutz L-OO Lutz/AAA 275 0.091
Lutz L-OO Lutz/AAA 290 0.087
Lutz L-OO Lutz/AAA 330 0.083

Sometimes you will find that the top deflects differently on the other side. Also the width of the plate DOES make a difference. So you need to keep track of that data as well. You need to be consistant.
Remember my data reflects my building and my desired tone characteristics. How you brace the guitar will obviously have huge effect.

It is like what Brock said. There are so many variables so there is no magic number. Our numbers are reasonable starting points but don't forget it is just one factor amongst many!!!! Bracing, sides, backs etc etc

Hope this helps

_________________
Andy Z.
http://www.lazydogguitars.com

Can anyone share density numbers? It would be interesting too, at least for me.

_________________
Build log

Spruce comes in between 6 - 8 g/cm^3

I routinely see red spruce above these numbers, but 6 - 8 covers the lion's share of the density numbers.

_________________
Brock Poling
Columbus, Ohio
http://www.polingguitars.com

5 lbs? 10 lbs? You guys are such woosies.
http://mysite.verizon.net/nostberg/chapter2/step7.htm

_________________
Here is what a Parlor Guitar is for!
http://www.youtube.com/watch?v=oEa8PkjO6_I

Well, that is maple ...

Ok, off to ruin some more wood....

_________________
Here is what a Parlor Guitar is for!
http://www.youtube.com/watch?v=oEa8PkjO6_I

One way to look at all of this is that we're trying to make the top as light as we can, but stiff enough to resist the bridge torque over the long term. Obviously, the bracing goes a long way toward taking the torque load, but my feeling is that I like to see a certain 'balance' between the stiffness of the bracing and the top; each should be taking it's part of the load. That means you want a certain deflection under a given load, and the top thickness is chosen to give you that deflection.

From beam theory we find that, for a given length and width of top the stiffness will vary as the Young's modulus (E) and the cube of the thickness. Double the E value and the stiffness is doubled, double the thickness and the stiffness is _eight_ times as high, all else equal. Of course, you can do the math the other way: start with a top of a certain size, supported to give a certain free span, load it with a known weight, and the deflection will give you a reading of the E value for bending/stretching/compression in that direction.

I use another approach, in part because I have the apparatus, and in part because it gives me some other information. If you know the dimensions of the piece, and the mass, the frequencies of the fundamental bending modes in the lengthwise and crosswise directions can be used to obtain the E values. You can also measure the half power band width and get the Q value; a measure of 'loss' in the wood. I've been measuring all of my top wood for a few years now, and am starting to find some interesting stuff.

One of the _most_ interesting things is that there is a linear realtionship between the density and the Young's modulus along the grain for _all_ of the soft woods I've tested. Soft wood with a density of 300 kg/m^3 will have a lengthwise E value of about 6000 megaPascals, and wood with a denstiy of 500 kg.m^3 will come in at about 17,000 mPa, +/- 10%. This includes samples of Western Red Cedar, Engelmann spruce, European spruce, Red spruce, Sitka spruce, and Western Hemlock. I have checked with other folks who have done similar measurements, and they affirm the relationship.

Cross grain stiffness, when graphed similarly, forms an almost perfect scatter. Ah well; you can't have everything.

If you make the simplified assumption that it is the lengthwise stiffness that resists the bridge torque, than you can use this data to arrive at a working thickness for your tops so long as you have a known data point to start from. If you made a guitar with known wood, and it worked well, you can use that to scale the thicknesses of all other tops for the same model, based on the density or the E value data from testing. I'm only just starting to get this sort of data, so I can't say as much as I'd like right now. At any rate, I suspect that what works for my desighns might not be true for yours, simply because of differences in the way we handle bracing and so on.

You will note, of course, that since E values vary as the density, and stiffness goes as the cube of thickness, a low density top will tend to end up lighter than a high density one for a given stiffness, assuming it falls more or less 'on the line'. This is not to say that you want to send back all of the dense wood: sometimes a bit of weight in a top seems to be useful. I'll note that Red spruce (AKA 'Adirondack') is usually on the dense end of the chart, and it's possible that the extra mass, and higher impedance, help to keep it from 'breaking up' under hard strumming.

Needless to say, there is a lot to do in following up this appraoch. Give us a hand!

Fascinating as always Al.

This gives me an idea. For those who are familiar with some of the distributed computer processing projects such as Folding@Home (unlocking the gene code) and SETI@home (search for extratesticals...) where home computer users donate time with their computers to crunch numbers why not create a distributed project to support your research? Instead of computer time you could provide us with the methodology that you want us to use and we could gather data and pass it on to you.

This might be the first time that we as a forum have had the opportunity to make a contribution to understanding the science of how a guitar works.

What do you think?

I'm quite possibly off-base here, but I'd guess that the majority of the work is in the physical collecting of the data rather than just sheer number crunching, but...

If we could have an agreed to regimen for collecting an agreed to set of data, it seems to me that the OLF could be extremely beneficial by providing a central repository for said data. It would be a relatively straightforward job to put a MySQL open source database together to be hosted on the OLF server and also an import page that would allow importing of either spreadsheets or text files. Data could then be exported through a set of canned SQL stored procedures and formatted as text or .csv files that could be directly read by spreadsheets for those that want to further manipulate the centralized data. This all might merit a separate thread, but I can see some real possibilities in combining efforts on this.

_________________
Jim Howell
Charlotte, NC

Jim buddy that is what I meant - for us to collect data for Al and others to use using a prescribed method. No computer time/use would be required for the collectors. SETI and Folding are just examples of distributed efforts.

I like the central suppository idea.

I deflection test my top plate before installing the rosette/bracing etc. I could give you that information, but I think it would be pretty useless to anyone because the top plate is just a small part. The bracing, bridge plate, bridge, sides, neck block and back all work together. To give an example, I started with a deflection of .250", but I have changed my bracing and now my deflection goes as high as .500", with my target generally between .400" and 450". These tops unbraced are very floppy and give an ugly sounding thud of a tap tone. Once braced, voiced, and put on stiff sides and a responsive back, it is completely different.
I don't see how a data base of deflection/stiffness values would be very helpful when it is such a small part of the overall picture.

_________________
"An adventure is only an inconvenience rightly considered. An inconvenience is an adventure wrongly considered." G. K. Chesterton.

Steve--

My thoughts are that a database of values needs to include more data than just deflection data. A list might include density, standard deflection test, thickness, Q value, Young's Modulus, etc. We could also include a photo with a scale layed on the top to display the general grain pattern. Those that understand more fully please chime in!

In my mind, a standard testing apparatus needs to be defined in a way that is both useful and relative easy to build. This may include multiple test rigs and all people that wish to contribute would not have to do the "full Monty". People could contribute what they feel comfortable in testing. The testing procedures could be standardized and then put into a downloadable .pdf format for general perusal.

_________________
Jim Howell
Charlotte, NC

Actually I had something else in mind.

Since Al can only test so many tops I thought that if he prescribed a methodology that we all could accomplish that does not require NASA equipment to do we could contribute data to Al's research. I never intended the database to mean a thing to any of us.

In return we help Al figure out more about how a guitar works. And you know what that means since Al is our go-to-guy for explaining complex concepts. Al would in turn, when he learns more, let us know what is on his mind. Perhaps a book might result. In any event we would all be that much more ahead.

I'm game.

Steve buddy when you said that your tops prior to bracing don't ring well would you liken this to a drum head prior to being tensioned?

WOW Guys!

I didn't expect this kind of response to this thread. The idea of us all collecting data with a standard method of operation is a great idea! I've been in science for years and it always seems like you can never gather too much data, you never know what kind of discover might be gleaned from the smallest observation.

On my building course with Serge DeJonge, he had a set guideline for finished top thicknesses and for bracing to accompany it. In general spruce was 2.2 - 2.4 mm (0.087" - 0.094") and cedar was 2.8 - 3.0 mm (0.110" - 0.118"). I don't have the bracing measurements in front of me right now. For him it didn't seem to matter what shape/size of guitar, the tops fell in this range for finished thickness and the bracing was altered to compensate for the particular guitar shape/size. Again, since I am such a rookie, I'm sure this is a simplified view of what actually went on in his head. He had an uncanny ability to just bend a piece of wood and tell you if it felt right.

I REALLY appreciate all the feedback on this topic. We should work on agreeing on a set method of collecting this data, I'm sure the whole OLF community could benefit from such an approach.

If any other experience builder would like to chime in on this topic, that would be great.

Thanks again... Greg