Top and back tap frequencies on my assembled F5 build

[HogTime]

Note: I know many think tap tuning is a bunch of hooey. I don't know enough to cast a vote. However, I'm posting this for those that might be interested.

I tried Siminoff's tap tuning method on the top when I was doing my build. Had some good preliminary results when shaping the tone bars. Then things became inconsistent and I stopped.

Decided today to measure the frequencies of the top and back since it's now assembled in-the-white (but no strings). I held it by the neck and tapped, roughly in the center, 3 times with Siminoff's recommended felt covered wooden mallet.

Attachments are an MP3 sound file and two PDFs with the frequency spectrum of the loudest tap. I used the PC program Sound Forge 9 to do the analysis. At the bottom of the page you see that the peak frequency of all three taps was the same.

Front: 308 Hz
Back: 306 Hz

Siminoff's book says the goal is a D#4 (311.13), so I'm pretty close. FYI, I don't intend to try any tweaking.

[Chris Baird]

When the instrument is glued up in the white and you bonk it you get the fundamental air resonance coupled with the lowest body modes. There is some change that can occur with the flexibility of the wood, however, most of this tap is governed by internal air volume and sound hole size and doesn't have much to do with how you carved your top and back.

[Tavy]

I'm with Chris here -the big broad peak you're seeing is almost certainly the Helmholz - also I tend to record 20-30 taps in rapid succession and then select and analyse the whole lot together - it gives the FFT algorithm more to chew on and produces a cleaner spectrum that's easier to understand (IMO), picked that up from a violin building site, not sure how Roger does it...?

If you also have spectra of the dis-assembled top, top on ribs etc, then you may be able to pick out how those resonances have moved (broadly speaking) when the box was closed out. But it ain't easy!

[fscotte]

I think you have to find a way to isolate one plate and the internal air resonance when trying to check the tap tone of assembled plates.

[HogTime]

Thanks for the comments.

I did this on a whim, rather than really trying to analyze anything. It is what it is and that's the way it will remain.

I did note that Roger's Tap Tuning book called what I measured the "air chamber tuning".

[Chris Baird]

When you glue one part to another, the modal lines go throughout the whole thing. So when you glue your plates to your rims you have a whole new global set of modes that vibrate through all the parts. There is no longer a "back mode" or a "top mode" as the plates had when they were separate. Some mode dominate certain parts, but, they are all global when the parts are glued together. When the body is glued up you have "body modes" and no longer have "plate modes".

[Dave Cohen]

308 Hz and 306 Hz are indistinguishable as far as mandolins are concerned, which is as it should be. When you excite the normal modes of an object, it doesn't matter where on the object you excite them; they should be the same regardless. The only exceptions to that are if you happen to bonk right at a nodal line of one of the modes, in which case that particular mode will be weak or absent, but all of the other modes will be excited as usual. So,.. if in doubt, bonk the instrument in a few different locations on the top plate and acquire spectra for each location.

When you bonk your assembled mandolin and listen to it, you will hear predominantly the Helmholtz or first air resonance. But if you bonk the instrument with your computer mic a meter or so away, collect the time domain data, and ft to get an audio spectrum, you should see peaks for all of the modes. If you don't have much resolution, you will likely see only one peak for the (main mode + Helmholtz mode) interaction. With a bit more resolution, however, you will see that the large peak is split into three (very occasionally four) peaks. The first main (aka (0,0) or "trampoline") mode should peak anywhere from about 250 Hz to nearly 300 Hz or even above that. The Helmholtz air mode will usually peak at around 285-300 Hz, and the upper main mode will peak anywhere between 350 Hz to over 400 Hz. There may be some confusion with that one, as the sideways rocking (aka (1,0)) mode may start to appear as low as 400-450 Hz. The lower main mode is the plates moving antiphase to each other, and that is the one largely responsible for pumping air in and out of the soundholes. The upper main mode is the plates moving in phase (accelerometers on the plates would be antiphase), and at least in theory shouldn't pump much, if any, air.

There are always pitfalls and tricks. If your instrument is strung up, the strings will steal energy from the body modes at their open string frequencies, and you may be able to observe that. To avoid that, just stuff a piece of soft foam between the strings and the fretboard to completely damp the strings. That, or just remove them altogether, as they don't alter the body or air modes much if they are not in motion, or better yet, not even there. For the main modes, the top plate and back plate motions are pretty well coordinated. When you get to the higher modes, the top and back plates may start to undergo different motions from each other. I and many others have seen that with guitars as well.

http://www.Cohenmando.com

[fscotte]

I'd like to revisit this thread and add my two cents.

I have done some bonk testing on my Gibson A5 without strings. I lay the mandolin top down onto the bed and tap the center of the backboard several times and record the taps with a small mic just a few inches away. I use Audactiy FFT to analyze the taps. I do the reverse and lay the mandolin on its back and tap the top plate.

There are dominant frequencies for each plate. The top taps a strong 536 Hz and the back taps a strong 565 Hz.

Interestingly, I did this with a recent "almost completed mandolin" in the white. The top I had tapped at 552 Hz, and the back was at 618 Hz. However, the back was still in need of sanding. As I sanded the back, it went down to to 562 hz where I stopped as it was very pleasant sounding at that point.

So what "mode" is being dominant here? Is this possibly the "ring mode" or something else? My "ring mode" for the top plate assembled in the rim without the back on was originally 545Hz to 558 Hz, depending upon humidity. So it's interesting that this same frequency range showed up on the top plate in my above experiment.

[Dave Cohen]

The "ring mode" as Alan Carruth, Peter Coombe, and others refer to it, is a free plate mode. You won't see a "ring mode" at that frequency in an assembled mandolin. From the frequnecies you have given, you are likely seeing peaks for the sideways rocking or (1,0) mode, depending on what type of mandolin you have made.

In all of the interferometry I have done on assembled mandolins, I have only seen a mode with a circular node once. That was in a back plate, and was up around 1.3 kHz.

http://www.Cohenmando.com

[fscotte]

I wonder if the 1,0 mode is something that can be tracked consistently in subsequent mandolins. I've only been using this mode to finalize the back plate for two mandolins now and am hoping it remains a viable mode to follow in future mandolins. I really don't know.

[Dave Cohen]

The (1,0) or sideways rocking mode is a strong mode, though not a strong radiating mode. Any sound radiation from that mode would have to be dipole radiation, except there is no sideways dipole air mode near that frequency. The strong monopole modes are the two (0,0) modes, at around 250-300 Hz and 350-420 Hz or so. The Helmholtz air mode, typically around 280 Hz, interacts strongly with those two. The two (0,0) modes and the Helmholtz give a big peak. If you have better than the usual 44 kHz resoution in your sound card, you may be able to pick out individual peaks. Although, if you scrape or sand and bring the (1,0) mode down in frequency, you will be bringing the (0,0) modes down as well. You just won't know how much.

http://www.Cohenmando.com

[Pete Jenner]

This thread made me want to do some bonking so I did.
I muted my strings and bonked my mandolin. First the top then the back then the two combined.

Can anyone tell me why there is a peak at 90Hz? Is it an air resonance? Dave?

[calden]

Pete: Me, too!

I was inspired to bonk away - on a couple of mandos I am building. This is the graph from a bonk I did on a spruce top that is 80% carved - the middle is about 8-9 mm and the edges are down to about 4 mm, leaving room for a little more sanding. Ignore the "peak" value - that is just where I left the cursor when I did the screen shot. The real peak value is at about 325 hz., so I figure I have some material to remove yet.

Am I on the right track with using this technology? I had no idea Audacity could analyze the frequency like this.

[Tavy]

This thread made me want to do some bonking so I did.
I muted my strings and bonked my mandolin. First the top then the back then the two combined.

Can anyone tell me why there is a peak at 90Hz? Is it an air resonance? Dave?

I would suspect noise: either from the PC, or the mic ?

Calden: up the value of the sample size to the max, if it's too small everything is so blurred you can't tell much of anything unfortunately

[Pete Jenner]

I would suspect noise: either from the PC, or the mic ?

Ah I suspected it could be the laptop mic. Should be easily testable.

[Dave Cohen]

Pete, you have already gotten good suggestions to look for artifacts. I've looked down in that frequency range for beam-like bending motions in mandolins, found none.

http://www.Cohenmando.com

[calden]

Calden: up the value of the sample size to the max, if it's too small everything is so blurred you can't tell much of anything unfortunately

Tavy:

Okay, I redid this. In the prior plot I posted I had set the Axis to Log instead of linear, which is why it was blurred rather than spiky.

Am I doing it right? I thunked the top with my knuckle, several times, at about where the bridge will be, and recorded it. Then I selected the strongest waveform without apparent clipping and plotted that.

Here is the new plot, set to a linear axis, but this was since I had glued the top to the sides, so the new frequency spike, it seems, is now up around 700 hz.

This is the correct way to use the frequency measuring function of Audacity to measure the tap tone, right? If not please correct me.

Thanks,
Carlos

[Pete Jenner]

Carlos, you should have it set to logarithmic. Increase the size to the maximum.

[Pete Jenner]

Pete, you have already gotten good suggestions to look for artifacts. I've looked down in that frequency range for beam-like bending motions in mandolins, found none.

http://www.Cohenmando.com

Thanks Dave.