Measure the frequency response of an acoustic guitar body?

[thefsb]

I have two Yamaha guitars, an FG830 budget-priced dreadnought and an LJ56 top of the line jumbo. I am interested to measure the practically relevant differences in the frequency response of the two. I have a UMIK-1 and some experience with REW. I also have the DAW Reaper which can record and analyze signals.

Characterizing each guitar in absolute terms, as Amir does for example, seems too ambitious. But if I use the same uncalibrated DIY measurement technique on both guitars I may still be able to see interesting differences. Tone generation and impulse excitation of the sound-box both seem promising and I have tried both.

For the impulse response I damped the strings and recorded at the sound-hole while tapping the body top behind the bridge. I wasn't able to figure out how to average the FFTs of the series of taps in Reaper. Read below about what I did so far.

For tone excitation I used REW and a pair of headphones attached to the guitar as shown. Crude but it was all I could think of with what I have available.

And the results with a setup similar to this using eight sweeps per guitar.

This is interesting enough to pursue. Need to be careful drawing conclusions from this given the wonky test rig but it's still interesting to compare peaks and valleys.

I have several questions for the measurement nerds of ASR.

Got any ideas for how better to excite the guitar for a sweep in REW? A shaker transducer that attaches to one of the bridge holes perhaps. Any idea how to make something like that for modest $$?

How can I obtain frequency response curves from a series of impulse responses? Reaper provides a dynamic spectrum display. A snapshot in time of it looks like this but I want a way to combine the responses from 20 or 30 taps and produce one static curve.

I'm positioning the mic close to the sound hole. That's convenient for SNR but might not be very relevant to what the guitar sounds like. Maybe I should mic it more like one would in a studio. Or with the mic positioned relative to the instrument where the players ears would be.

Any other ideas?

 

[Trif]

For a studio approach, try placing the mic about a foot and a half (45 cm) from the lower bout and aim it at the soundhole. The soundhole is basically a poorly tuned bass port*. Remember, it's a mic. You can listen to what it's picking up. Less room sound will give you clearer graphs.

Would you consider installing a Barcus-Berry?

As far as excitation goes, I think using headphones is brilliant. Compared to other possibilities, they are ruler flat, and padded! I would not clamp them around the guitar though, as that will affect the resonance. The ideal place to inject a signal, of course, would be the bridge. But as the focal point of vibration, anything attached there will alter the top's resonance, so I'd suggest the nut.


* "poorly tuned" from a loudspeaker's point of view, that is. For making music, it's just what it should be. (The number of violinists who think the sound comes out of the f-hole is truly discouraging. No, dear, the sound comes from the top plate, which surrounds the hole.)

 

[bluefuzz]

Any other ideas?

My other hobby is building acoustic guitars so I have some experience of this … ;-)

You don’t want to put the mic into the soundhole. Better to position it about 50 cm away from the top pointing at the bridge. In fact you don’t really need a fancy measurement mic at all to do impulse response measurements of guitars (although it won’t hurt). I just use the internal mic of my iPhone or iPad. I have found the iPad/iOS app ‘FFT’ from Studio Six Digital better for this kind of thing than REW.

If you are on Windows the free Visual Analyser is also excellent for doing this kind of analysis.

You don’t need to do frequency sweeps either - just whack your guitar (gently!) on the bridge and record the result. I use a wine cork impaled on the end of a chopstick as my ‘whacker’. Very high tech!

If you want chapter and verse on doing this kind of testing take a look at Australian luthier Trevor Gore’s book Contemporary Acoustic Guitar Design and Build. It’s pricy but is undoubtedly the bible of scientific guitar building and testing. Trevor usually hangs out on the ANZLF forum and often answers questions. There are many searchable threads about impulse testing guitars and how to go about it on the ANZLF forum.

Otherwise, you’re welcome to ask me if you need further details.

 

[thefsb]

For a studio approach, try placing the mic about a foot and a half (45 cm) from the lower bout and aim it at the soundhole. The soundhole is basically a poorly tuned bass port*. Remember, it's a mic. You can listen to what it's picking up. Less room sound will give you clearer graphs.

Roger that.

Would you consider installing a Barcus-Berry?

I would prefer not to. I dislike the sound of acoustic guitars with pickups attached. I removed the Baggs IBeam that was in the LJ56. Yuck! I'm really an electric player who recently got into playing acoustic again and I like the sound of the instrument in my ears as I play. I just want to understand it a bit better.

As far as excitation goes, I think using headphones is brilliant. Compared to other possibilities, they are ruler flat, and padded! I would not clamp them around the guitar though, as that will affect the resonance. The ideal place to inject a signal, of course, would be the bridge. But as the focal point of vibration, anything attached there will alter the top's resonance, so I'd suggest the nut.

Yeah, that's a good point. I'm curious about the sound of both guitars (especially the LJ56) around the 190 Hz because the fundamental of the F# and G around that frequency decays almost instantly leaving a tinny, flaccid tone. Phones as I have them attached may well change all that.

* "poorly tuned" from a loudspeaker's point of view, that is. For making music, it's just what it should be. (The number of violinists who think the sound comes out of the f-hole is truly discouraging. No, dear, the sound comes from the top plate, which surrounds the hole.)

I love being around musicians.

 

[thefsb]

My other hobby is building acoustic guitars so I have some experience of this … ;-)

You don’t want to put the mic into the soundhole. Better to position it about 50 cm away from the top pointing at the bridge. In fact you don’t really need a fancy measurement mic at all to do impulse response measurements of guitars (although it won’t hurt). I just use the internal mic of my iPhone or iPad. I have found the iPad/iOS app ‘FFT’ from Studio Six Digital better for this kind of thing than REW.

If you are on Windows the free Visual Analyser is also excellent for doing this kind of analysis.

You don’t need to do frequency sweeps either - just whack your guitar (gently!) on the bridge and record the result. I use a wine cork impaled on the end of a chopstick as my ‘whacker’. Very high tech!

If you want chapter and verse on doing this kind of testing take a look at Australian luthier Trevor Gore’s book Contemporary Acoustic Guitar Design and Build. It’s pricy but is undoubtedly the bible of scientific guitar building and testing. Trevor usually hangs out on the ANZLF forum and often answers questions. There are many searchable threads about impulse testing guitars and how to go about it on the ANZLF forum.

Otherwise, you’re welcome to ask me if you need further details.

Great info! Thanks.

VA and ANZLF forum provide plenty of material to keep me going.

I chose a fondue fork to skewer the cork, which I think is from a bottle of bourbon rather than wine.

 

[thefsb]

I just remembered that I made a video of the F# suck

 

[bluefuzz]

around the 190 Hz because the fundamental of the F# and G around that frequency decays almost instantly leaving a tinny, flaccid tone.

This happens when you have several resonances too close to each other which can often happen on responsive guitars. Often known as wolf notes on violin family instruments. If it’s at ~190 Hz it probably means the main top monopole is too close to twice the frequency of the main air resonance which is typically around 90-100 Hz. It’s better if the main top is around 170-180 Hz and the back around 4 semitones higher.

 

[thefsb]

This happens when you have several resonances too close to each other which can often happen on responsive guitars. Often known as wolf notes on violin family instruments. If it’s at ~190 Hz it probably means the main top monopole is too close to twice the frequency of the main air resonance which is typically around 90-100 Hz. It’s better if the main top is around 170-180 Hz and the back around 4 semitones higher.

That is nearly beyond my comprehension at it stands. I may be able to figure it out.

But, in the case of my LJ56 that I demonstrate in the video, is this a feature of the design of the instrument? or a feature of this particular example?

 

[bluefuzz]

That is nearly beyond my comprehension at it stands. I may be able to figure it out.

The third image you posted at the top shows it very well actually. That’s the sort of ‘mountain range' you want to see. The first peak is the main air, the second ‘twin’ peak is probably the top monopole coupling with the back. Such a double peak indicates two resonances being repelled from each other because they are too close. Not necessarily bad, but something you ususally want to avoid.

is this a feature of the design of the instrument? or a feature of this particular example?

Most guitars of ‘normal’ size, i.e. O, OO, OOO, have similar resonances. A drednaught obviously a bit lower and a parlour guitar a bit higher, but actually not by much. For the main air resonance we are talking between F and A most commonly. Factory built guitars are typically built to spec, i.e. the top is 2.8 mm thick, X-braces 10 mm high and 6 mm wide etc. Due to wood, even of the same species, being of quite variable physical properties (modulus of stiffness, damping, mass etc.) the resulting factory guitar will have a random range of resonances. A stiff top coupled with stiff braces and a stiff back will result in a very ‘tight’ instrument that won’t do much. Sometimes a factory guitar built to spec can be great if the wood used happens to fall into the ‘right’ relative properties. This is where a luthier-built instrument has the potential to be ‘better’ than a factory guitar because the luthier (if she knows what she’s doing) can place the resonances in the right place to achieve a resposive instrument.

 

[thefsb]

Most guitars of ‘normal’ size...

Very interesting. My take aways are ...

1. The answer to my question is: it could be either. A bit disappointing since the benefits of a Lx56 over the models priced one rung below it (Lx36) is supposed to be that it is made entirely by one master builder.
2. Understanding and being able to recognize what I want from a particular instrument might take fair bit of learning. There's a clear analogy to the audiophile market. A lot of the high-end products are sold on pride of ownership, optics, and bafflegab. Technical understanding can help one cope with the bafflegab.
3. Choosing a luthier to build a custom instrument might be tricky because one would want to figure out if a luthier knows how to adjust each instrument's response but even if she does, she may talk about the adjustments and their effects in a different language.

 

[thefsb]

The number of violinists who think the sound comes out of the f-hole is truly discouraging. No, dear, the sound comes from the top plate, which surrounds the hole.

Just a second. What's going on with the sound port that's become quite popular on nylon guitars and is starting to appear on steel-string acoustics?

 

[jae]

 

[bluefuzz]

Just a second. What's going on with the sound port that's become quite popular on nylon guitars and is starting to appear on steel-string acoustics?

Not a lot IMO. Some guitarists regard the port as a kind of extra ‘monitor’ - and yes some sound does come out the soundhole and side-port if there is one (just like some mid and high frequency tones can leak out the port on a loudspeaker). But the port will raise the pitch of the main body resonance so you want to make the main soundhole smaller to maintain that pitch. I think sideports are a bit of fashionable placebo for the most part but they can be useful as another exit for retrieving that missing plectrum you dropped in the soundhole … ;-)

 

[Trif]

Just a second. What's going on with the sound port that's become quite popular on nylon guitars and is starting to appear on steel-string acoustics?

Good question.

A quick oogle of the 'net will reveal pretty much what bluefuzz said. A few are trying to increase the bass response (one, uh, "luthier" even added a tube to the port) while most are presenting it as a "monitor" to help you hear while you play in noisy surroundings.

No one mentioned the issue of acoustic feedback on stage, so apparently this is a fad among parlour players.

I have a personal preference in these matters. I started selling Martin guitars in 1975. That leads to a (biased) opinion, "If these turkeys had decent rosewood, they wouldn't need to put a hole in the side to increase the highs."

Basically, it is a tertiary sound source. The top, the soundhole, and then the port (to actually address your question). The back is likely louder, but it's facing the wrong way. The top remains the primary transducer that allows the strings to vibrate the air.

Not sure why, but at this point I feel it appropriate to point out that while everyone focuses on the bridge as a source of vibration, the nut actually moves further and it has a two foot baseball bat for leverage on the guitar body (again, there's a curve at the base of the neck as it widens to meet the guitar body. look at a trumpet's bell. same curve. same purpose. transformers are everywhere! it's all impedance matching).

 

[thefsb]

transformers are everywhere

Except for MC cart SUTs, which have gone out of fashion. But there appears to still be a market for Lundahl and Jensen in microphones.

 

[thefsb]

I tap tested according to @bluefuzz description using a phone with its buit-in mic.

The LJ56 with the missing fundamental between F#3 and G3, i.e. ~190 Hz, as demonstrated in the video above.

The FG830 with a less severe case of a similar pathology.

That was interesting. I tested the LJ56 second and it was striking how much louder it was in the tap test.

Assuming the trough I highlighted with the yellow arrow corresponds with the experience playing those F# and G notes, it seems this test might help point to what's going on but is no substitute for testing by playing since the notch doesn't look severe given the overall contour of the terrain it appears in and because there's nothing suggesting a problem in the FG830 curve.

However, I was impressed how little the red max sample-hold curve changed after the first tap on subsequent taps.

 

[thefsb]

Not sure why, but at this point I feel it appropriate to point out that while everyone focuses on the bridge as a source of vibration, the nut actually moves further and it has a two foot baseball bat for leverage on the guitar body

I'm sure there are those who can answer this very well but based on my experiment I'd guess that luthiers concentrate on exciting the bridge for a couple of reasons. 1) I don't see much energy below 150 Hz from the body from when tapping the nut, and 2) Maybe luthiers don't want the modes of the neck complicating the picture.

Comparing these to the above pictures, which would you guess is which?

 

[thefsb]

View attachment 146653

A wolfnote eliminator on a cello?

Curiosity aroused.

 

[jae]

A wolfnote eliminator on a cello?

Curiosity aroused.

Does playing at different positions from the bridge change it much? What about when using more left hand pressure or two fingers? Different amplitudes?

An 'eliminator' of this type will not work well with strings under higher tension, such as those of a violin, on top of the fact there is not too much distance/vibration between the bridge and cordier anyway (which seems like an issue with most guitars...). With bass guitars I have seen people add mass or different placements of clamps (including guitar "sustain clamps") at positions on the headstock which would serve to change its mass/impedance and increase the sustain that way, so that could be one experiment. Perhaps you could also experiment with some blu tack on the body, which is a common thing to do with violins.

 

[mixsit]

Acoustic -and electric player here, read through the thread a couple of times but still not clear what it is you're trying to get from this?
Thanks