Lets build something that looks like an L5 archtop

[Freeman Keller]

What most people do is to hold the top at what they think is a neutral spot (a "node") and tap near where the bridge will be. They listen to the note that comes off the top. Siminoff uses a microphone and runs it thru a strobe tuner to try to see what that note is. Bourgeois wants the note to be "musical". Others want a strong fundimental or maybe lots of partials. It actually pretty hard to tap something today and try to remember what it sounded like yesterday or last week. Is the note going up or down, is it more musical or less.

Enter technology. I happen to have a beta copy of some software that will capture the note and let us look at it and analyze it. It will present the note in the time domain (attack, sustain, decay) over a period of time that its ringing, and it will take a snap shot of the note and present its frequency content. This is called a Fourier Transformation and is a basic part of wave theory - that every continuous wave can be represented by the sum of a bunch of sine waves. I talked about it in the Classical thread in posts 91 an 92 and other, I'm not going to repeat it here

https://www.tdpri.com/threads/my-classical-challenge.1036434/page-5

Here is how it works and what it looks like. This is the spectrum of the top at the very beginning of the project. It has been cut to rough shape, joined and the outside is roughly carved. I'm holding it near the waist and tapping at the bridge area with my carefully calibrated knuckle (I would use my finger tips but I keep my nails kind of long and don't want to damage the wood). It makes a nice "bonk" sound

OK, here is what you are seeing. The top graph is the time domain view of the bonk - the computer is sitting there listening, at about 1200 msec it hears something above its threshold and starts recording. You can watcht the note decay over the next couple hundred msec.

The lower graph is the frequency spectrum of that note at its maximum value. The peaks show both the frequency and the relative amount - its got lots at 239 and 275 hz, as well as many others.

Remember something interesting. The strings of a guitar range from E2 at 82 hz to E4 at 329 hz, E5 at the 12th fret of the 1st string is 659 hz. So the significant thing right now is that all of these peaks are well within what we expect to hear from a guitar.

Bonk

 

[Freeman Keller]

During the time I've been working on this I've been bonking away on the top. Lots of bonks all stored away on my little lap top. Some trends appear as I go back over them, I won't bore you with a lot of graphs but here are a couple that are representative

In this one the top has been carved inside and out to a uniform 1/4 inch thickness

The time domain response is longer - it seems to have more "sustain". I now have a big peak at 433 hz - thats pretty close to A4 on the high E string. Some other peaks seem better defined.

Here it is after carving the top closer to Benedetto's specs, adding the f-holes and the braces before they were carved at all

The significant things are that the 430 hz peak is still there but look at the ones at 570 and 607 - those tell me the top is really really stiff (again). And of course it is, I just glued big fat braces on it.

Next I've bound the f-holes, thinned the top a little more (which makes it less stiff) and shaved the braces to about 1/2 inch high. Remember that making a brace less tall reduced its stiffness by the cube of the height, if I make it half as tall it becomes 1/8 as stiff.

Again, the significant thing is that the big peak has moved down in frequency as it should and I have a lot of nice complexity in the 600 to 800 hz range. At this point I've decided that I'm done doing what is called "free plate tuning" and its time to glue the top to the rim, then add the back.

Here is the closed box

The time domain graph is stronger and sustains longer - it looks like I'm getting some interaction between the top and the back. But WHOA, what is going on at 117 and 228 and there abouts? I had almost none of that before, now those are the dominate notes. They are actually two different A sharps (or Bb's) at 116 and 223 hz - 1st fret of the 5th string and 3rd fret of the 3rd string. These are big time guitar notes!

The answer to where did they come from is really simple - those are the so called "main air mode" and its first harmonic - this is the amount of air in the box and what its doing when I bonk the top. This is the 16 inch lower bout and 3 inch depth. The main air mode of a dreadnaught is pretty well known at G2, this is a few semi tones higher.

The nice thing is, this is totally what was expected.

So what does this all mean? Darned if I know but I'll take some guesses. The main air mode is the single most important mode for an acoustic guitar. A# seems pretty OK, a bigger box would be lower but I think its going to have a decent midrange. Its not going to be a bass cannon, I can accept that. That 433 peak is still there but much lower. I was guessing that was the main top mode but that may also be the two peaks at 575/600. What the others mean I don't know.

So that is my little bonking story. As the wise old man said, "if you can't dazzle them with brilliance baffle them with BS".

 

[crazydave911]

The only thing I looked for was tone and did it sustain. The only archtop I ever built was sycamore back and sides and a redwood top, and yes I used tone bars. My saddest regret is selling it, it was the loudest, with tone guitar I ever made. And now, even with electrics I tap for tone and sustain. I've tried tuning a few times for flats or sharps but usually forget. But if can't FEEL it when I play it unplugged, I know it isn't worth a damn though I seldom have trouble giving the dead beasts a home. But if the pickups are ALL I hear, I don't want it

 

[Moodivarius]

Very interesting read Freeman.


I did the tone tap when I built my violin.


Taken from my website on making the violin:

When getting very close to the proper thicknesses, I would hold the the back by the upper left corner between my thumb and fore finger, lightly tapping the middle of the plate with my knuckle. Listening to the tone of the plate, I would take more wood out to raise the tap tone to the desirable frequency of the back plate. The desired frequency should be 1 1/2 semitone higher than the belly. E for the belly and F# for the back.

http://www.moodivarius.com/carving_back.htm


It was an interesting process.


Scott

 

[Telekarster]

Fantastic Freeman!!!

 

[blackbelt308]

Very cool experiment, Freeman! If at each step you "bonk" the top multiple times, is the decay and spectrum reasonably consistent?

Kept thinking: "there once was a note, pure and easy"!

 

[Boreas]

Great work Freeman!

You likely don't have the equipment to do this, but it would be interesting to temporarily attach a pure tone generator/transducer that would vibrate at controllable frequencies like is often used in amp repairs. The generator produces a (variable) tone to the transducer which in turn starts the top vibrating. With an infinite range of tones generated, one would likely encounter an infinite range of responses from the top. If you think about it, your calibrated bonk would likely be different than mine. But if you subject the top to the typical dominant vibration frequencies produced by each string, you would expect to see differing responses with each tone. And then, should we analyze the top's response to chords? I suspect Martin and other big manufacturers have done this type of testing, but as you say, what does it actually tell us?

What is my point? Dunno. We agree that tweaking a top can indeed alter the final tone, but is the "calibrated bonk' method the ideal way to test? Will my knuckle produce different dominant frequencies from the top? So many variables come into play in creating acoustic instruments - probably some we don't often consider - temp/humidity/age of wood, etc.. What does a finish do to the frequency response amplified by the top? Will these bonking responses remain consistent over the life of the instrument, or do they start and continue to change with age? So many questions...

 

[Freeman Keller]

Very cool experiment, Freeman! If at each step you "bonk" the top multiple times, is the decay and spectrum reasonably consistent?

Kept thinking: "there once was a note, pure and easy"!

The patterns at each step were pretty consistent, but I would do a dozen or so bonks and save several that were pretty much the same. When I embedded them in the above discussion I tried to select ones that were representative.

Great work Freeman!

You likely don't have the equipment to do this, but it would be interesting to temporarily attach a pure tone generator/transducer that would vibrate at controllable frequencies like is often used in amp repairs. The generator produces a (variable) tone to the transducer which in turn starts the top vibrating. With an infinite range of tones generated, one would likely encounter an infinite range of responses from the top. If you think about it, your calibrated bonk would likely be different than mine. But if you subject the top to the typical dominant vibration frequencies produced by each string, you would expect to see differing responses with each tone. And then, should we analyze the top's response to chords? I suspect Martin and other big manufacturers have done this type of testing, but as you say, what does it actually tell us?

What is my point? Dunno. We agree that tweaking a top can indeed alter the final tone, but is the "calibrated bonk' method the ideal way to test? Will my knuckle produce different dominant frequencies from the top? So many variables come into play in creating acoustic instruments - probably some we don't often consider - temp/humidity/age of wood, etc.. What does a finish do to the frequency response amplified by the top? Will these bonking responses remain consistent over the life of the instrument, or do they start and continue to change with age? So many questions...

You are aware of the process of making "Chladni patterns" which is a time honored method of studying classical instruments like violins and classical guitars. A loud speaker connected to a frequency generator is suspended above the instrument, the frequency is slowly swept across the range of interest. Some sort of small powder, glitter or sawdust or whatever, is sprinkled on the top. At certain frequencies the glitter will move away from some parts of the top and settle at others - where it settles there is no motion (nodes) and the clear areas are the anti nodes where there is maximum motion.

Chladni patters were used a lot in the study of violins and I have watched Alan Carruth do it with guitars. It does tell you two things - the frequencies where parts of the guitar want to vibrate (resonate frequencies which are my peaks) and the areas of the top that are moving or not (which my method doesn't show).

The problem with Chladni patterns is that I can't see how you use them to achieve a "better" sounding guitar. With tap tuning at least you can tell if resonate frequencies are changing as you work with the top and braces.

I saw a lecture by a classical guitar builder where he attached a little transducer to the bridge of a guitar and drove it with various frequencies. He had a calibrated microphone suspended above the guitar to record the response

He also has some accelerometers stuck to the top and if you notice, the guitar is suspended so the back isn't dampened.

 

[Boreas]

The patterns at each step were pretty consistent, but I would do a dozen or so bonks and save several that were pretty much the same. When I embedded them in the above discussion I tried to select ones that were representative.



You are aware of the process of making "Chladni patterns" which is a time honored method of studying classical instruments like violins and classical guitars. A loud speaker connected to a frequency generator is suspended above the instrument, the frequency is slowly swept across the range of interest. Some sort of small powder, glitter or sawdust or whatever, is sprinkled on the top. At certain frequencies the glitter will move away from some parts of the top and settle at others - where it settles there is no motion (nodes) and the clear areas are the anti nodes where there is maximum motion.

Chladni patters were used a lot in the study of violins and I have watched Alan Carruth do it with guitars. It does tell you two things - the frequencies where parts of the guitar want to vibrate (resonate frequencies which are my peaks) and the areas of the top that are moving or not (which my method doesn't show).

The problem with Chladni patterns is that I can't see how you use them to achieve a "better" sounding guitar. With tap tuning at least you can tell if resonate frequencies are changing as you work with the top and braces.

I saw a lecture by a classical guitar builder where he attached a little transducer to the bridge of a guitar and drove it with various frequencies. He had a calibrated microphone suspended above the guitar to record the response

View attachment 928836

He also has some accelerometers stuck to the top and if you notice, the guitar is suspended so the back isn't dampened.

No, I wasn't aware of that process. I am but a simple, armchair enthusiast... But thanks to you, I can research it a bit more!

 

[2groggy]

Can we make any association between free-plate-tuning to glued-to-the-box tuning?

 

[Freeman Keller]

Can we make any association between free-plate-tuning to glued-to-the-box tuning?

Yes and no. As I see it there are three stages of tuning the top:

Free plate, without and with braces. The top is not attached to the sides, you hold it up, tap it and shave a brace.

Top glued to rim, back off. That constrains it somewhat, but essentially its still "free". You can still get at the braces to shave them but it s more difficult. Obviously if you glue the back on first as Benedetto suggests you can't do this.

Box closed, that introduces the "main air mode", the top is pumping air in and out of the sound hole(s), the back and sides are coupled to the top. Its pretty hard to do any brace shaving at this point, however people do reach into the sound hole on flat tops and work on braces. Some people will thin there perimeter at this point, Benedetto talks about working on the recurve. If I do any more it will be pretty minor.

Do I see an association between the free plate and closed box - absolutely. That big peak at 575 or 580 is there in both situations and I believe that is the so called main top mode - the entire top is moving as one unit. If Boreas was looking at a Chladni pattern at that frequency he would see no glitter anywhere in the center of the top, it would all have moved to close to the rim - the entire top is moving.

I'll add that Chladni patters can be done with the top glued to the rim or the box close. You need to be able to support the top solidly so you can't just hold it in front of the loud speaker. With the back off you can still shave braces.

 

[Freeman Keller]

OK, the bonking has been fun, lets get back to building guitars. In fact, lets build a neck

With the exception of a few Fender style necks, everything I build is more Gibson or Martin style. Slanted headstock, big tall heel, double acting truss rod with the adjuster either in the head or the heel depending on the style of guitar. Almost all of my necks are mahogany, it is such a nice stable easy wood to work with, and it will match the back and sides. Years ago I built a few one piece necks (meaning sawn out of one piece of wood with a separate fretboard) but now days I assemble them out of several pieces cut from a single board. The built up necks are much less wasteful of wood and only a tiny bit more work.

Benedetto shows building a three piece laminated neck, that is fairly common on archtops - the laminations can show fine worksmanship and are very strong and stable. I start with a piece of one by three mahogany from any one of several luthier supply places. Long ago I adopted 16 degrees as my head stock angle - it really doesn't matter but if you standardize then you can make some jigs and use them over and over.

There are two ways of doing a scarf joined head stock and I've done both. One puts the head piece on the bottom of the neck stick, one puts it on the end of the stick. There are slight advantages of each, I've come to prefer putting it on the end. Making the cut is kind of funky because one piece will be flipped over and I can never remember exactly how long to make it. I find it helpful to make a little sketch, note that the 8-1/2 inch measurement that will become the top of the head is in the middle of the piece you cut off.

I've got it laid out on the mahogany board. Ideally I would set my miter fence for my band saw at 16 degrees and make the cut but my saw is too small for that angle. Instead I just clamp the board to a square block and do the cut by eye. This is one of the reasons I checked my saw for square back before I started

The head piece is thinner than the neck stick, this is a good time to do it

Clamp the two pieces together and clean up the cut with a little plane. At this point everything needs to be perfectly square

Its also very important that the glue face is perfectly flat. That seam will show and you want it to look nice

Over the years I have tried several schemes for gluing and clamping the head. The problem is that when you apply glue the joint is very slippery and if you remember your high school physics, a wedge is one of the very powerful simple machines. Short story is that you can't just clamp them.

A forumite gave a neat trick that I've used on the last several necks - pin the head piece to the neck stick with a couple of tooth picks. You could use brads or nails or even screws but I don't like any metal where I'll be cutting and routing. I try to put them in a place that will be routed away

I actually ended up with four tooth picks but didn't take a picture. Glued the head to the neck and now it doesn't want to slide around. You can see two of the tooth picks, I added that other two before I got the clamps fully tightened

And here it is the next day

 

[Jim_in_PA]

From bonking to getting neck-ed...what a forum!

Seriously, I'm really enjoying this project.

 

[Bob J]

Yeah, I’m sure we’re violating some forum guidelines here!

 

[Freeman Keller]

That last picture in post #192 shows a potential problem with gluing the head to the end of the neck stick. You want the plane of the top of the neck to be perfectly flat as it transitions from the neck stick itself to the head. The wedging action wants to force them apart which means the head will slide down the ramp - that is what the tooth picks are there for. I try to drill the holes so the head piece is standing a bit tall so if it does move I can still plane it flat. In this case it came out almost perfect. I hope that makes sense for any of you thinking about making a scarf joined neck

The next part is easy - establish the nut location and measure back to the heel. Determine how much wood will be sticking inside the guitar (depends on what kind of joint you will be using) and wack the excess neck off. Cut it up into little blocks and stack the heel

 

[mistermikev]

During the time I've been working on this I've been bonking away on the top. Lots of bonks all stored away on my little lap top. Some trends appear as I go back over them, I won't bore you with a lot of graphs but here are a couple that are representative

In this one the top has been carved inside and out to a uniform 1/4 inch thickness

View attachment 928581

The time domain response is longer - it seems to have more "sustain". I now have a big peak at 433 hz - thats pretty close to A4 on the high E string. Some other peaks seem better defined.

Here it is after carving the top closer to Benedetto's specs, adding the f-holes and the braces before they were carved at all

View attachment 928583
The significant things are that the 430 hz peak is still there but look at the ones at 570 and 607 - those tell me the top is really really stiff (again). And of course it is, I just glued big fat braces on it.

Next I've bound the f-holes, thinned the top a little more (which makes it less stiff) and shaved the braces to about 1/2 inch high. Remember that making a brace less tall reduced its stiffness by the cube of the height, if I make it half as tall it becomes 1/8 as stiff.

View attachment 928585
Again, the significant thing is that the big peak has moved down in frequency as it should and I have a lot of nice complexity in the 600 to 800 hz range. At this point I've decided that I'm done doing what is called "free plate tuning" and its time to glue the top to the rim, then add the back.

Here is the closed box

View attachment 928586

The time domain graph is stronger and sustains longer - it looks like I'm getting some interaction between the top and the back. But WHOA, what is going on at 117 and 228 and there abouts? I had almost none of that before, now those are the dominate notes. They are actually two different A sharps (or Bb's) at 116 and 223 hz - 1st fret of the 5th string and 3rd fret of the 3rd string. These are big time guitar notes!

The answer to where did they come from is really simple - those are the so called "main air mode" and its first harmonic - this is the amount of air in the box and what its doing when I bonk the top. This is the 16 inch lower bout and 3 inch depth. The main air mode of a dreadnaught is pretty well known at G2, this is a few semi tones higher.

The nice thing is, this is totally what was expected.

So what does this all mean? Darned if I know but I'll take some guesses. The main air mode is the single most important mode for an acoustic guitar. A# seems pretty OK, a bigger box would be lower but I think its going to have a decent midrange. Its not going to be a bass cannon, I can accept that. That 433 peak is still there but much lower. I was guessing that was the main top mode but that may also be the two peaks at 575/600. What the others mean I don't know.

So that is my little bonking story. As the wise old man said, "if you can't dazzle them with brilliance baffle them with BS".

just wanted to say I thoroughly enjoyed this post. Great post.

 

[Freeman Keller]

just wanted to say I thoroughly enjoyed this post. Great post.

I'm glad you enjoyed it.

 

[netgear69]

Great stuff interesting to see how an archtop is made

 

[Freeman Keller]

When you stop to think about a guitar neck to body joint its a pretty amazing structure. There are at least five axis of rotation and angle controlled by the joint, on a typical acoustic guitar with 165 pounds of string tension it might have close to 80 inch pounds of moment and over 800 psi of compression stress. And all it has to do is hold the neck in perfect alignment so that each of 22 frets will be just the right distance from 6 little strings. Show the neck joint a little respect.

Just as there are a couple of kinds of neck joints for electric guitars there are at least three for acoustic instruments. Nylon strung guitars with 80 or so pounds of tension usually have a "Spanish heel" joint, the sides fit into slots in the neck which is also the head block. Spanish heels are pretty permanent but rarely need to be adjusted - if you want to see one find my Fir Classical thread.

Some modern guitars, including Taylor and Art and Lutherie, have their necks bolted on - either butted against the end of the guitar or mortised into them. The design is simple, fool proof, easy to work on and if you aren't happy with the geometry take it apart and fix it. If you want to see a bolted M&T neck joint find my Mahogany 00 thread.

And then there is the dovetail. The dovetail is an elegant wood working joint that tightens itself by wedging action, there is no hardware in the joint. In fact it really doesn't need glue, but a small amount is added to help. However dovetails are figity to get the geometry right and can be a pain to make them tight. William Cumpiano called them "the joint from hell" and for many first time builders they are.

I used dovetails for my first two guitar builds (they were Martin clones) and managed to get them together and aligned the way I wanted them, but most of the guitars that I built after that have been bolted M&T. I reset guitars with dovetails and used them on a couple of other scratch builds, and by now I'm starting to feel like I understand them.

There are two good reasons to use a dovetail on this guitar. First, its what would be expected on an acoustic archtop. Yes a few people (Ken Parker again) do other joints but most archtops will have a dovetail. Second reason is pretty compelling - its about the ONLY joint that will work. If you don't have a sound hole so you can get your hand inside the box its pretty hard to do a bolted joint. So, dovetail it will be.

We have already made the female part of the joint (oh oh, someone's going to start making jokes again), let me say that differently, we have already made the mortise. I used a head block that had the pocket already cut and simply opened up the sides and top

To make the tenon I first measured the angle of the top to the front of the guitar

Its a hair more than 4 degrees, call it 4-1/2. Transfer that to the neck heel

There are two lines (kind of hard to see), one is the end of the tenon and one is the cheeks of the heel itself. Cut the neck off at the end of the tenon at 4-1/2 degrees.

A long time ago I made a jig for routing dovetails. The neck goes into the jig and is clamped at 4-1/2 degrees (this shim is actually the piece I cut off the end of the neck)

Clamp this critter to the work bench so I can get at it

And route the tenon with a dovetail router bit

Take the neck out of the jig and relieve the area next to the tenon, remember that the front of the guitar is slightly curved

Here is the rough fit of the tenon into the mortise

The bearing surfaces are the heel cheeks and the angled part of the dovetail. I don't want the end of the tenon hitting the body - if I ever have to take the neck off thats where the steam will be injected.

I feel pretty good at this point - the joint still needs a bunch of work but its not too bad.

 

[pixeljammer]

This is really interesting. Thanks for posting it all.