Official Luthiers Forum!

Now that I have 15 or so builds behind me, I'm trying to better control my results. With the exception of a way over braced parlor, I've been pretty successful with sound quality. But there have definitely been some standouts and some that worked better for one style of playing over another. I'm generally building medium to smaller bodied (OM, OO and parlor). Interestingly, my first acoustic is still the best one out there and consistently gets peoples' attention.

I don't build enough instruments to get that intuitive sense or feel of what is going to sound good that comes from years and dozens/hundreds of instruments. So I'm working on getting more scientific with the process to try and dial in both my intuition and sound. My notes are more detailed now and I'm taking a lot more calibrated measurements of stiffness, thickness, etc so I can look back and see what worked.

Part of this is trying out spectrum analysis and storing the results. I'm working on a couple of tops right now and finally got the Luthier Spectrum Analyzer software downloaded and running. Now I need to understand what I'm seeing. And how those numbers compare to what others are targeting. Is my Fundamental too high? Are the other peaks making sense? How does my decay compare?

I realize there is a incredible book out there that is on my wish list, but I'm going to have to wait for some spare cash before I can jump into that! So I'm reaching out for input.

Here's where I stand right now on an OM top, Western Red Cedar. I should say that I've been following Robbie Obrien's excellent voicing lessons to get to this point. The top has definitely opened up and gotten more musical as I've moved along, with some nice initial attack and some nice complexity coming in after it. But is that showing up in the spectrum analysis? I don't want to push it too far. I probably do have some room to take more off some braces if your input pushes me in that direction.

Thoughts and discussion would be welcome. And the Luthier Spectrum Analyzer is a nice little system once you get a feel for how to run it.

_________________
Tony Thatcher
Bozeman, Montana

You need to specify some more stuff. Is that the 'free' top, not on the rim, or is it mounted? If mounted, is the box open in back or closed? If 'free' how is it supported, and where was it tapped? Mic position?

Hi Alan. Thanks for watching over me as I wade into the deep end of this!

Free top. Hooked over my thumb through the sound hole and held about 5 inches in front of a cheap mic. I can't seem to get the software to "hear" the input from my decent mic through my Zoom audio interface. There's probably some ambient noise in there as a result.

I'm open to any and all input on how to do this consistently.

_________________
Tony Thatcher
Bozeman, Montana

Depending on the exact taping and holding points, and the location of the mic, the big peaks you are seeing up to 250 Hz or so are the pitches of some of the main low-order resonances. Without really exact specs about the measurement conditions it's hard to say which is which with any confidence. It's likely that the two lowest peaks, below 125 Hz, are the lowest torsion mode, and the fundamental lengthwise bending mode. The big peak at about 225 is most likely to the 'ring-and-a-half' or 'ring+' mode, assuming you have not taken the braces down much from 'rough'.

In many respects what you're doing here is old fashioned 'tap tone' tuning, using a mic and spectrum analyzer instead of your ear and brain. The measurements, particularly of pitch, are more 'accurate' than most of us get by ear, but because you're relying on measurements of single taps it's clumsier, in the sense that you'd need to record and analyze a bunch of taps to determine which areas of the top are active at what pitches. If you hold and tap it's easy to move both the tapping and holding points and figure that out fairly quickly, but, again, most of us are not as good at determining an exact pitch.

What I do is a further 'tech' elaboration of this, using 'Chladni patterns' to actually look at the shapes of the resonances. This, it seems to me, is where the real information is. The exact pitches of the modes don't tell you as much as you'd expect, but the shapes and relationships between modes can be useful information. This is particularly true in establishing the 'balance' between the brace stiffness and the top that seems to contribute to things like 'clarity' in the sound.

This sort of testing used to be pretty difficult to get set up for when I got started with it in the '80s; we had to build our own hardware signal generators. Now it's a piece of cake. Thanks to the advances in electronics and software you can get a decent signal generator for free (if you have the tablet or cell 'phone), and all you need beyond that is a 12W-15W amplifier, some foam pads, and glitter or sawdust.

A friend of mine, and his wife, both retired software developers, wrote an Android app he calls 'Luthier Lab'. Mostly it was for his own use, but I think it was partly a favor to me. They determined that it had no real commercial potential, so they're just giving it away. It has a spectrum analysis tool, and also a signal generator. Along with that you get drawing tools to lay out shapes and brace patterns, fret calculators, and a setup to archive each build through pictures that you can label, and text notes. He keeps adding little goodies to it.

Some useful math:

The Gore/Gilet books (your obvious reference) cost about $350. But if you buy a StewMax membership for about $40, you get to buy the books at 50%, or around $175. And you get free shipping on everything you buy, not just the books. The benefits of a StewMax membership might be a closer call for someone not buying the Gore/Gilet books, but for some planning on buying the books, it is an obvious way to save some money.

Yes the incredible book out there is going to answer most of your questions.

However before finding the bucks for that...it is I think more useful to look at the resonant frequencies of the completed instruments rather than the free plates.

The first three resonances: the main air resonance, normally around 100 hz, the main top (monopole) at around 200 hz and the back sort of around 200 and something.

First you don’t want any of these on a scale note. Secondly if you have a good sounding guitar where are those frequencies sitting? Perhaps you can aim to get them on the next build.

The great value of the G an G book is that they show how you can change those main frequencies but there is info on that freely available elsewhere on the net.

There are many useful discussions on this stuff on the ANZL forum (not wishing to divert people from the OLF!)

Good luck, this is not easy stuff to get your head round. Dave

As Dave m2 says, it's the pitches of the assembled modes that matter the most. The 'free' plate pitches are not good predictors of the assembled modes, simply because there are too many variables involved. All the parts of the assembled box react on each other and the exact mass and stiffness of the parts changes the result.

If you make two guitars that are 'identical'; same size and shape, sister cuts of wood, careful control over the construction, and matched free plate mode frequencies and shapes, the low-order assembled modes will be the same within very close limits. Relax any of those conditions and things change. And even with 'matched' pairs they won't sound exactly the same. My latest such pair had virtually identical spectra below 1000 Hz, but in blind listening tests everybody could tell them apart. Small changes in the density and stiffness of the top wood in local areas altered the output in the 2-4 kHz range, were hearing acuity is high, and people picked up on that.

From what I can tell the 'free' top mode shapes are related to those high frequency modes in the assembled box. Guitar tops that have lots of well formed 'free' plate modes tend to work better in the high frequency range, which tends to correlate well with high marks for 'quality' of sound; 'clarity', 'color' and so on. The low order modes that Dave mentioned seem to predict the 'character' of the sound to some extent: the difference between a Dread and a Parlor, for example, or even a Dread and a Super Jumbo.

Dave wrote:
"Good luck, this is not easy stuff to get your head round. "
Amen! I've heard it said that the only physics that is more complex then acoustics is quantum mechanics. In many respects the hallmark of a good guitar is it's complexity; they're deliberately complicated, and that's what gives you the range of tone color and dynamics. Many of the traditional features, such as the shape, and the location of the sound hole, seem to have been settled on because they add complexity to the sound. The old boys didn't know that in the way we can, but when they hit something that worked they were smart enough to stick with it.

Thanks all! I've got some studying to do! Any links to online material would be appreciated. I've been having trouble separating the trees from the forest.



Sent from my SM-G973U using Tapatalk

_________________
Tony Thatcher
Bozeman, Montana

There's a thread on the Acoustic Guitar forum, the 'Custom Shop' page, about a build by Mark Blanchard, that gets into plate tuning some in more recent posts. There are good pictures showing some of the modes of the top in question, along with shots of the bracing and measurements of the top thickness graduations and brace profiles. You could do worse than to emulate his work.

I wrote a three-part series of articles for American Lutherie magazine that came out in '91 and '92(!). We've learned some things since then, but it's still good basic information. Those should be available as reprints in the 'Big Red Book' series.

I was also recruited as the 'talent' for a plate tuning video some time back. It's been a while since I heard anything from the producer, but it may still be available. I tuned the tops of several steel string guitars, some easy, and some not so much, just to show the variety of things you might run into. It's four hours of me saying:"Well, that didn't work!", but I'm assured by some of the folks who've gotten it that they learned something anyway.

I don't know of much in the way of on-line tutorials, though.

Tony, I recently built a guitar and used Alain's software to help me think about what I was doing. Short story, this was a classical built completely (top, back and sides) out of Douglas fir. The Spectrum Analyzer gave me the confidence that my wood choices were totally bogus and that I could see it becoming more "musical" as I closed the box and worked on the bracing. The guitar came out quite good - I posted some pictures and an audio clip a while back.

I won't say the Analyzer helped me do any specific operation, it just told me where the air and some of the body resonances were and that made me feel good about what I was doing.

ps - I have also been using the software to compare several of my finished guitars and I run a before and after spectrum when I do any significant change to a guitar.

Tony, your wave graph (top one) does not look typical. Possibly caused by your mic or some windows setting.
Look in windows audio settings and make sure there are no effects, gain, etc. added. Or try a different distance for the mic.
For a braced top the wave chart should be more "structured". Like the attached.

This does not mean the spectrum is inaccurate. Your main top is around 225 like Al mentionned. Adding the bridge should increase it a bit. Depending where you want your top to be, you can carve the braces a little more or not.
If you do so then activate the "KEEP" mode. Reload your saved graph or record a new one before carving. Do some adjustment and record a new spectrum on the same graph with the KEEP mode ON, you can then see what direction it is going, what peaks are going up or down or shifting frequency, etc. You can label the curves ( by clicking on the curve color in the panel at the right) to keep track of what is what.
The decay is a measure of how long the top resonnate from the tap to almost no sound. There is no magic number, but again you can measure if what you do improove or not the resonnance.
Good luck with your build.

Alain:
Are you sure those first two recordings aren't clipping? You'd expect a tap to start loud and fall off exponentially.

The third one, in blue, is showing some 'beating' between resonances that are close in pitch.

Yes they are clipping, this is why they are labelled "too close".
I am sorry may be I should have explained better.
These are all from the same instrument. A ukulele IIRC. I recorded that some months ago already, just to demonstrate the influence of the distance on the recording quality.
A gain adjustment would probably also produce similar results.

EDIT: After having a second look at Tony's graph, I realised that his view is zoomed! So possibly this is why it look strange to me. Once unzoomed it is probably closer to the typical shape. Nevertheless I still think that holding the mic a bit closer or applying some gain would be beneficial.
Note however that all the curves on my graph had the same fundamental at 161/160.5 Hz. So the distance is not really critical for the spectrum analysis. You just get a nicer wave chart!

Now there's an interesting idea for analysis software... have it detect the tap sound and then print the spectrum on screen immediately, so you can move and tap repeatedly and look at each one in rapid succession without having to stop and fiddle around on the computer. Maybe add some forward and back buttons so you can page back and forth between the most recent taps. And for building a reference library, have a picture of a guitar top where you can drag and drop icons for the hold point and tap point, and save that along with the audio/spectrum data.

The "Autorun" on LSA will do just that for you.

I'm also early on in the journey, but by pulling hard on the G & G books as well as Somogyi's books (and a little from Companio) I would have to say that my experience also suggests that the real answers are found in the fully assembled and strung samples.

That said, I do tap and record the lowest frequency peak on my tops and backs to make sure they are hitting the same pitch as my target instrument. That usually gets me very close once assembled. I've found that the bracing shape and thickness work determine the color of the tone where the target frequencies get the "personality" of the instrument right.

FWIW, I recently finished my first Parlor after building several Jumbos and tried to apply a similar approach. I hold the top at the edge where the transverse brace and x brace meet and tap where the bridge will be. The parlor's three lowest peaks on a braced top, ready to glue on the rim were ~77, 142, 229.

Ah; that's what the 'too close' and 'too far' referred to! It makes sense now. I may be slow, but I'm not too quick....

I use a dB meter as my mic, and can change the gain settings easily. I like to record the sound with the mic in the same position relative to the guitar every time, so I use different gain settings and tap strengths to get it right.

I've always felt that more information is better, so long as you don't fall into the trap of spending all of your time getting it. The tap spectrum on a finished guitar is an excellent 'snapshot': lots of information that you can get quickly, even if it's not everything you'd like to know. Once I've gotten that I look at the Chladni patterns of the top and back, and find the lowest 'neck' resonance and the 'main air' resonance. That all gets included on a chart that goes into the folder for that guitar. It makes it fairly easy when you want to see what you did in the past, either to copy it or avoid it.

How do you measure neck resonance?

Lots of information to digest. For starters I need to dial in my mic setup and figure out why the mic running through my audio interface to the computer is not being "heard" but LSA. The computer recognizes it, along with my DAW. And LSA shows it on the list of available mics, but it is not getting any input.

On the topic of free plate vs closed box. I understand that tap tuning with a spectrum analyzer is a fancy way of using your ear. But a pro who can tap tune by ear must be hearing something that they like and understand will result in a decent guitar when the box is closed and all the work is done. So it would be interesting to get a better understanding of what a decent top should sound like in a visual manner, along with hearing it (I've been recording each step in the process of voicing this top).

I've watched a number of videos of pros voicing their tops and each one seems to go about it slightly differently. But likely each are listening for similar elements depending on what their goal is for a guitar. I've seen some of the methods that some of you are using to "close" the box without actually closing it by clamping the top to the back/sides. Interesting, but I'm not sure I want to go down that path right now. I've also come to the conclusion that - for now - I am not really interested in building active backs and trying to couple the top and back. Too many variables for me to try and keep track of at this point in time.

I'll see if I can get some more consistent taps going and post them. I also have a second top that I'm just about to brace, so that will provide another opportunity to explore this.

Cheers!

_________________
Tony Thatcher
Bozeman, Montana

meddlingfool asked:
"How do you measure neck resonance?"

It's actually the lowest whole-body 'bar' resonance; the 'first corpus' (C1) mode, but since the neck bends a lot it's often referred to as the 'neck' resonance. The guitar is vibrating like a xylophone bar, and there are two stationary node lines for this mode: one somewhere around the nut or first fret, and the other across the wide part of the lower bout. I usually support the guitar on foam blocks at the wide part of the bout and with a bock under the neck near the head junction, and put strips of foam under the strings to damp them. I get it high enough off the bench to slip a speaker under it, and position that under the neck block. I use the signal generator to drive it, and simply touch the top of the headstock lightly to feel the vibration. If you want to get really precise you could stick on a small piece of steel as an armature, and use an electric guitar pickup (coil and magnet) to read the motion of the top of the head.

The less 'tech' version is, to hold the guitar up by pinching the neck at the upper end between your thumb and finger. Use your finger to damp the strings, and allow
the guitar to hang as freely as you can. Tap on the back of the headstock with the ball of your finger (a 'soft' tap, as opposed the 'hard' one from your nail), and listen with your ear next to the head (don't get a string end in the ear!). You should hear a fairly clear low-pitched tone.

Usually this mode is too low in pitch to really affect the sound of the guitar. It's often down around C below the low E pitch, but sometimes it's higher. This is more common on classical guitars with wide 12-fret necks, and particularly if the neck is tapered so that it's deeper at the body end.

If the 'neck' resonance is high enough in pitch it can couple with the main air resonance. If you think of the guitar supported on the bench, as the head and tail block move 'up' the neck block is moving 'down'. The top of the box is being compressed along it's length. Since tops are normally at least a bit arched, if only due to string pull, compressing the top lengthwise causes it to puff up a bit, and suck a little air in at the sound hole. Half a cycle later the air gets pushed out. Since this is what's happening with the Helmholtz-type 'main air' resonance the two can couple; the neck pushes are, and the pressure changes in the box help push the neck. When this happens the body/neck vibration can act as a 'flywheel', storing some energy and feeding it back into the air. As I say,I've only seen this once or twice on a steel string, but it's not uncommon on classicals, particularly the better ones.

This shows up in the tap spectrum as a double peak around the 'air' frequency. The peak height is lower than it would have been without the couple, but the response is spread out over a wider band of frequencies. Instead of one strong note at the 'air' pitch you get two or three that are not quite as strong. This is, in itself, an advantage, since that really strong note is the 'guitar wolf' that sucks all the energy out of the string and converts it to sound. s it's twice as powerful for half as long. You don't notice the extra power, but you do notice the lack of sustain.

There is another, more subtle, advantage. The area under the coupled curve is greater than it would be without the coupling: the 'gain bandwidth product' is larger. I other words, there's more available horsepower in the lowest range. This often seems to show up as a particularly 'rich' or 'dark' timbre in the bass, and seems in some ways to 'put a floor' under the sound.

Note that the pitch match has to be really exact for this to work. Since the head moves a lot in this mode the weight of the head and the machines can be important if the neck stiffness is such that the match is 'close'. Adding or subtracting an ounce or so can really make a difference. I think this may be why some people really hear a difference when they swap out machines and others don't. There can be enough difference in the density and stiffness of neck wood to matter even when the dimensions are held to close tolerance.

There are, of course, higher order 'neck' modes, but they're even more variable in pitch and harder to control. Also, it's not clear how they would affect the tone unless one of them happened to have a strong antinode (active area) just where you were fretting that note, and even then.... Considering how much more flexible the top is, I'm not sure it would be productive to put much effort into those higher order 'neck' modes.