Modifying a closed box (classical)

Edzard · Post by **Edzard** » Wed Sep 23, 2015 7:45 pm

Hello All,

I build a few years ago (BG..

) a classical guitar with a (rather) stiff Lutz top, maple back/sides and Romanillos bracing pattern.

The bridge was made from African Blackwood. I made it a bit smaller (16 x 25 cm) to keep the weight on the "low" side (19 gr).

The top is around the bridge area around 2.4 mm, descending to 2.1/2.0 mm around the edges.

The guitar sounds good, but lacks a bit of power.

Because the bridge split a little around the bass side I thought it was a good moment to "fine tune" the top.

The following VA-picture (AG...haha) is taking with the bridge off.

98 Hz and 238 Hz (370 Hz?).

What's your take on this guitar? A lot of "sanding" to get the peaks down?

Best regards,

Edzard

Trevor Gore · Post by **Trevor Gore** » Thu Sep 24, 2015 6:22 pm

A few questions first:

What's the back made of, how thick is it and what frequency does it give when you tap the back?

Do you have a spectrogram of the guitar with its original bridge on?

kiwigeo · Post by **kiwigeo** » Thu Sep 24, 2015 11:16 pm

The 238Hz peak looks like it might be a back peak....if it's the main T(1,1)2 top peak then it seems very high. Like Trevor suggests...tap the back and see what peak(s) you get.

Edzard · Post by **Edzard** » Fri Sep 25, 2015 2:24 am

Thanks Guys,

Trevor, the back is from Maple. It's around 2.5/2.4 mm. I totally forgot to make spectrogram with the bridge on.

I can't upload a picture (Could not upload attachment to ./files/775_cb2370524add7a529aaee94c9b1be39f.) probably some system configure error,

But the peaks are at 280 Hz, 400 Hz and 650 Hz.

Thanks again,

Best regards,

Edzard

Trevor Gore · Post by **Trevor Gore** » Fri Sep 25, 2015 8:54 am

It would seem that you have the top and back panels rather thick and stiff, which will also make them heavy and so of low monopole mobility.

A main top resonance T(1,1)2 of 238 Hz is rather high. Typical is more like 190Hz when the bridge is on. I was concerned that the peak at around 225Hz might be the T(1,1)3, which would have meant that the back is tuned to a lower pitch than the top, which is never a good idea. However, with the back frequency at 280Hz you seem to have a non-live back to your guitar. I would have expected a higher T(1,1)1. Has the guitar got a smaller than normal sound hole, or a fretboard overhang?

When I have used Lutz in classical guitars, the tops finish up in the range 2.0 to 2.1mm in the centre and thinner around the edges. So that might give you some ideas of what to do next!

If you can, it would be good to measure the monopole mobility so you can compare it to the values in the book and also measure your progress. Of course, you will be doing this without the bridge on, but it should give you an impression of how things are going.

kiwigeo · Post by **kiwigeo** » Fri Sep 25, 2015 6:08 pm

Edzard wrote:
I can't upload a picture (Could not upload attachment to ./files/775_cb2370524add7a529aaee94c9b1be39f.) probably some system configure error,

Edzard,

PM me if youre having problems. You can email me the image and I can upload it for you.

Martin (Mod)

kiwigeo · Post by **kiwigeo** » Fri Sep 25, 2015 6:10 pm

Trevor Gore wrote:
When I have used Lutz in classical guitars, the tops finish up in the range 2.0 to 2.1mm in the centre and thinner around the edges. So that might give you some ideas of what to do next!

I've made a couple of fan braced classicals with Shane Neiffer's Lutz and the tops ended up with similar thicknesses. 2.4mm is definitely on the thick side.

kiwigeo · Post by **kiwigeo** » Sat Sep 26, 2015 8:54 am

VA plot for back tap testing on Edzards Classical:

Edzard · Post by **Edzard** » Mon Sep 28, 2015 7:00 pm

Hello Guys,

Thanks again Martin for uploading the missing picture.

I've started sanding this weekend. Bridge area is now around 2.1/2.2 mm descending to 2.0/1.9 around the edges.

I took an other spectrum picture but found no drastic changes. Maybe a few Hz lower then before.

The top is now (ofcourse) more flexible then before. I can move it (pumping) with my thumb in the bridge area without massive force.

Is there some (linear) relation between sanding down the top and lowering the frequencies. Or does it "pop" down when you hit the "sweet spot" with regarding the desired/optimal thickness?

I would expected a much lower frequencie (or more change) with this sanding down.

Best regards,

Edzard

Trevor Gore · Post by **Trevor Gore** » Tue Sep 29, 2015 7:52 pm

That's a rather strange looking VA plot compared to the first, but for the moment I'll assume the peaks are in the right places.

I've never had to do a major re-thickessing of a top, but have shaved braces a few times. In those situations, it takes a long time for much to happen, then it starts to happen very quickly. What you'll find is that the top is now more sensitive to added mass. So whereas x amount of mass might have dropped the T(1,1)2 by say 10 Hz, it may now drop it 20 Hz. That's one of the reasons why it's good to collect a lot of data (tap response, bridge rotation, monopole mobility etc.) before starting work then a bit of arithmetic will give you a good handle on a course of action and how far to go

If you have the top down at ~2.1mm, I'd be inclined to stop there. What are the physical dimensions of the braces?

Edzard · Post by **Edzard** » Sat Oct 31, 2015 1:41 am

Hello All,

So I thinned the top down to 2.0 mm around the bridge area and 1.9/1.8 mm around the side.

When I record the frequency there are no changes. Strange? At least to me it would make sense that you saw some (major) drops.

And one from the back. Back is now around 2.2 mm.

It's a 7 fan bracing scheme.
3 braces in the middle are about 6 mm high and 2.8/3 mm wide.
Then one on each side of 4.5 mm high 2.8/3 wide.
The outer one is 4 mm high about 2.8 wide.

The width of the lower bout is 360 mm (36 centimeter). Height is about 87 mm. So a smaller box overall.

Rosette hole is 89 mm.

Problem is that I can't reach with my hand in the box to take down some bracings. Simply to big/fat.

Does this (maybe) mean that you can't change a guitar that much when it's "character" is defined when choosing the top/bracing/closing the box is done?

Best regards,

Edzard

kiwigeo · Post by **kiwigeo** » Sat Oct 31, 2015 12:21 pm

Edzard,

How much effect on peak frequency you see by thinning around the perimeter will depend alot on how close to the edge of the top your bracing is. If the bracing is close to the perimeter of the top as I suspect is the case with your guitar then you'll see little change after thinning. On my falcate braced classicals there is at least 30-35mm between the end of the primary and secondary braces and the edge pf the soundboard.

Trevor Gore · Post by **Trevor Gore** » Sat Oct 31, 2015 1:17 pm

Edzard, can you try putting ~20grm of Blutack (or similar poster putty) in the bridge position (assuming the bridge is still off) and post the VA plot?

Edzard · Post by **Edzard** » Mon Nov 02, 2015 9:32 pm

Thanks for your time and trouble Trevor and Martin. It's appreciated!

I have some kneaded eraser. Weight of one package is 21.7 grams.

OK. Now a picture without everything. Just for the record.

When I put the kneaded eraser on the top like a small bal on the bridge area I get the following frequency response (168 Hz):

When I place the eraser like a bridge form (see above picture) I get the following frequency response (193 Hz):

What happend? I'm baffled...

And a picture so you can see the outer bracing. The space between the lower bout and the first bracing is about 38/40 mm.

Thanks again guys!

Best regards,

Edzard

Trevor Gore · Post by **Trevor Gore** » Mon Nov 02, 2015 11:50 pm

Edzard, one of the things that's puzzling me is that every time you post a VA plot everything above ~200Hz seems to change. However, you left your "hammer" in one picture and that has given me an idea. I think you're only exciting low frequencies due to the nature of your hammer. The theory goes something like this: you can only get frequencies excited by a hammer strike that have a period longer than the contact duration of the hammer strike. So, for example, if the hammer is in contact for 1 millisecond, you will excite frequencies up to 1kHz. If the hammer is in contact for say, 5 milliseconds, you will only see frequencies up to 200 Hz with any degree of accuracy.

I use a piece of low loss "rubber" cut from a 45mm diameter "super ball" (one of those kid's balls that are really bouncy). It's mounted on the end of a bamboo satay skewer, so that hammer has a very low resonant frequency (miles away from what we're interested in measuring), but a very short contact time because it's light and bounces quickly. Might be worth trying a different hammer.

Do you have Chladni equipment?

Edzard · Post by **Edzard** » Tue Nov 03, 2015 8:25 am

Hello Trevor,

I made a hammer with an eraser. Here's my setup.

: DSC00061.jpg (148.75 KiB) Viewed 23901 times

: DSC00062.jpg (75.54 KiB) Viewed 23901 times

: DSC00064.jpg (61.79 KiB) Viewed 23901 times

I made 4 guitars with different tops (2 lutz and 2 euro spruce). They sound all very good (if I can say so).
I took a finished one and made some measurements too.

First the one without bridge and the new hammer.

Then with the "bridge" attached.

Then one from the working one guitars.

I had only time to make one Chladni pattern.

: DSC00067.jpg (53.39 KiB) Viewed 23901 times

Thanks again.

Best regards,

Edzard

Trevor Gore · Post by **Trevor Gore** » Wed Nov 04, 2015 12:48 am

There's something about those plots that doesn't look right, which is why I'm holding off on giving you a point of view. I had a mess around with my system this evening to see if I could reproduce some of the lack of repeatable features that you seem to have in your plots. I turned up the gain on the mic and whilst I could still see the T(1,1)1 and T(1,1)2 the plots looked "not right" in a similar sort of way. So one last check and if that checks out OK we'll work on the plots you have. Can you check the gain on your mic? One easy way is to record some music using the settings you have currently and listen for distortion on the playback. Whilst you're at it, it's probably worth checking for noise filters and other extraneous processing that may be happening that may be messing with things.

dshaker · Post by **dshaker** » Wed Nov 04, 2015 8:01 am

I'm suspicious of peaks at exactly 100 hz and 200 hz. I am wondering if there is noise pollution at those peaks. It might be interesting to see a spectrum of the background noise, with no thumping of geetars.

Edzard · Post by **Edzard** » Wed Nov 04, 2015 8:05 am

I played a bit tonight with the configuration of my system but it seems I'm drifting further away from getting good predictable/reliable results.

I turned of the internal mic (although the samsung was the input device) and turned the volume back to 30%.

Maybe it's a problem with the system/software. I have a surface pro3, Windows 10, VA 2009 and a Samsung c01u USB mic.

Problem is that I have only windows 10 laptops and they start blowing the fan at one point when working. Only the Surface works some hours without blowing.

I will do some more testing tomorrow Right now it's getting late over here and I'm missing something obvious. You are correct about getting this right, otherwise the whole exercise/working with your system/book is pointless. Very frustrating but true!

Thanks again.

Regards Edzard

dshaker · Post by **dshaker** » Wed Nov 04, 2015 10:40 am

When I was having noise pollution problems, I
- used a microphone with a long cable
- turned off the lights (mostly fluorescents)
- piled pillows on top of my laptop to kill the disk noise
- put the mic in a nest of pillows with the only open space pointing at my soundboard

That helped.

Trevor Gore · Post by **Trevor Gore** » Wed Nov 04, 2015 1:29 pm

Might be worth trying one of the later versions of VA. Just moving VA 9 from XP to Win7 threw up some compatibility problems for me.

Edzard · Post by **Edzard** » Wed Nov 04, 2015 8:19 pm

I have the VA program configured conform your paper "Technical note on collecting spectrograhic data".

FFT size is 16384 and Frequency sampling is 11025 (Bandwidth is 0.512 Hz / Spectral line resolution is 0.67 Hz) and buffer set to 10.

What I noticed on my Surface (I7 Intel processor) and other newish laptops is that the sample rate/time to record the hit is very slow.
If you hit to fast (say a hit per 1 seconds) you see a distortion on the live screen. Like the program doesn't know what to do if it reads 2 hits after each other.

So last night, in my pyjamas (what a hobby!), I changed the Frequency sampling to 22050 (double) and got the following results. With the eraser bridge attached.