Another extreme test

[Don Noon]

My workhorse VSO for experiments hasn't seen any action in about 2 years now, but a confluence of things got it out of hibernation for yet another gruesome experiment.  The factors:

1)  Recent discussions about wood aging and degradation  2) I have a MDF back plate created to test my CNC code, and it kinda fits the VSO.  3) For a long time, I have wondered about the influence of the back plate, but never did much testing other than some regraduations, which didn't have a huge effect.

So I have this back plate:  104g, M5=240Hz.  The original back plate, as removed, was 101.5g, M5=352Hz.  The weight is not terribly different, but the stiffness sure is.

Anyone want to hazard a guess about what the difference will be?  Surely the signature modes will be lower, but by how much I won't guess.  My main interest is elsewhere in the tone, perhaps I might be able to see different influences of the top and back.

[Andreas Preuss]

Just for clarification on this experiment:

You are replacing a back plate made of normal wood with a back plate made of MDF, right?

Arching is supposedly approximately the same?

What I have seen on my super light violin project is that changing back plates does not change awfully lot in terms of sound. What is more interesting is how playability and and perception of sound under the ear does change. Softer back plates make the sound more diffuse. At the same time lower resonances get stronger.

All this looks to me that a ‘good sound’ (whatever a maker defines as such) is based on how the top is designed as a membrane. All the rest is a kind of modifying those properties. When making the back a kind of weaker the whole setting of the string angle needs to be readjusted (re calibrated) to make it function to its best again. In that case I would most likely lower the string angle.

i am not too much experienced with reading sound graphs, but there my prediction is that you might end up with a graph of less sharp peaks in the  higher registers. 

In term of wood treatment I am not quite sure how valid this experiment is, because MDF as a material lacks one important factor because stiffness in vertical and horizontal direction is the same. Natural wood is always different in both directions. Wood treatment changes the stiffness in both directions, but never to the degree that they become equal. My experiments with steaming wood (not boiling in water) made maple measurably lighter and tap tones of the uncut planks got a bit higher presumably due to weight loss. On spruce this method didn’t show a significant change.
 

I was always wondering how internal stress in natural wood affects the sound. I am inclined to think that treatments like ponding wood in water relieve stress. There was one scientific paper which compared violins and the violin made with ponded wood showed different all over sound characteristics. (I would need to look it up again, I think it was a paper where Terry Borman participated as a maker) 

Usually my predictions are just the opposite what you come up with. 

[Bill Yacey]

MDF in my experience has pretty poor stiffness, and I imagine it will be very dead acoustically, too.

I look forward to your findings.

[baroquecello]

I love these threads on extreme experiments!

I have my thoughts, but really, I'm only cellist with zilch experience in making, just in playing. PLEASE DON'T TAKE MY COMMENTS SERIOUSLY!

What I'd expect from an MDF back depends on the thickness/graduation. On a thick MDF back, I'd expect the sound to be shrill and piercing, powerful in a really bad way. With a thinner MDF back I'd expect the sound to lose focus and attain a dead sound. On a thick back, I'd expect the bow to feel like ice skating, and on a thin back I'd expect it to lack the possibility to use fast bow speeds. 

For knowing what I'd expect from your particular back, I'd have to have in my hands. I'm assuming one can't regraduate an MDF back, so will you make thicker and thinner versions too?

I could imagine that coating MDF will have a very large effect, expecially if it permeats the material deeply. for instance, I'd expect a Shellac-coated thin MDF back (maybe inside and outside) to work much better than one without.

[Marty Kasprzyk]

It will sound great.  All of the signature modes will be lower in frequency and closer together which will give a rich lower end sound.  The high end will probably be normal.  

The only problems are that it isn't traditional,  looks too plain, it will soak up a lot of varnish and wood dealers will put a price on your head.

[Jim Bress]

Interesting experiment Don! I've been thinking about plate weights and stiffness a lot lately. I my case as it applies to cellos.  Not very fruitful thinking because I've done any testing, nor do I have time to. My guess from all my reading is that response may be better, whereas tone will become less focused. I'm split on whether I think the violin overal will sound darker or brighter. 

I think Andreas point of stiffness being equal for "long grain" and "cross grain" is an important confounding variable.  I'm looking forward to the results.

Cheers,

Jim

[Don Noon]

I like to make some educated guesses about the results before actually finding out what they are, which helps to amplify any errors in thinking and clarify what thoughts need to be changed.

MDF certainly is different.  Density is somewhat higher than most maple, at .75g/cc, and stiffness is isotropic.  Compared to maple, "longitudinal" stiffness is about 25% as much, while "crossgrain" is maybe 30% higher.  Taptones are a mix of stiffnesses and arching, and don't fully capture these differences.  Damping is definitely higher all around in MDF.

The B1- mode appears to consist primarily of back plate longitudinal bending and top plate crossgrain bending, so I would expect this frequency to drop the most.  It was 433 Hz originally, I'll estimate it will go down to 362 Hz.  The B1+ looks like mostly top plate longitudinal bending and back plate crossgrain bending... so it might not change at all from the original 533 Hz.  I say "appears to be" or "looks like" regarding the plate contribution to mode frequencies because that's what is most visible in the mode animations.  However, in-plane stretching can be significant influences, and won't show up in the flapping animations.  That's one thing I hope this test will show.

In the middle frequencies (say 600 - 1400 Hz), the back plate is normally fairly active, so I would expect these resonances to be lower in frequency, and lower in amplitude due to the higher damping.  Above that, the action is mostly in the top, so I would expect the shape to be about the same, but with some loss due to the back damping.  Even though the back might not do much at those frequencies, whatever it does will absorb energy.

BTW, the original back arch was extremely high at 17.8 mm, and the MDF arch is 14.6 mm.  However, I think that the huge difference in material properties will dominate the changes, not the arch height.

Hopefully this thing will be playable, and hopefully the weakness of the MDF won't allow the neck to snap out or droop too much.

[Andreas Preuss]

If I think about it, your experiment seems to aim rather at speed of sound in the material. MDF must have a pretty high damping. Most wood treatments change this property for the better or worse.

 

[Andreas Preuss]

3 minutes ago, Don Noon said:

Hopefully this thing will be playable, and hopefully the weakness of the MDF won't allow the neck to snap out or droop too much.

The neck will drop a lot, because the back plate will bend in the region between upper block and top block. Therefore I'd adjust the fingerboard projection a good 5mm higher than needed for the previewed bridge. If you are nervous about the neck getting loose you can double pin the MDF plate at the top block (or use a metal screw), but I don't see any reason that this can happen unless it is really badly glued.

[Marty Kasprzyk]

3 hours ago, Marty Kasprzyk said:

It will sound great.  All of the signature modes will be lower in frequency and closer together which will give a rich lower end sound.  The high end will probably be normal.  

The only problems are that it isn't traditional,  looks too plain, it will soak up a lot of varnish and wood dealers will put a price on your head.

I forgot to also predict the amplitudes of the signature modes will be higher.

[Don Noon]

Here are the first-day results.  My top-line result:  not as much change as I expected.  Some predictions way off, others close.

I forgot to mention the A0 mode, which dropped from 273 to 264 Hz and gained .6 dB.  This was not a surprise.

B1- frequency dropped from 433 to 408 Hz, about half of what I expected, and amplitued decreased 2 dB.

B1+ frequency surprisingly dropped from 535 to 459 Hz, far more than I expected, and amplitude increased 2.2 dB

The middle frequencies, or "transition hill", became weaker and lower in frequency, about as I expected.

Higher frequencies are a bit different, but generally about the same shape and slightly lower average amplitude, similar to expectations.

Impact spectrum comparison:

I also did a bowed semitone spectrum comparison, with similar results but less resolution.  Here is the bowed semitone scale comparison, string-by-string, with the maple back first and MDF back second:  MDF back comparison.mp3

It appears that in-plane stretching of the back is a more significant factor in the B1+ frequency than I thought.  Perhaps if I had a good Finite Element Model to play with, I could have found that out.  But I don't.

This test also had re-adjusted my ideas about what could be the cause of the "Old Italian Sound"... but not much readjustment.  Just that the back might not be as big of an influence.  I'm still going with the idea of lower damping and loss of some stiffness... mostly crossgrain.

 

[Salve Håkedal]

Thank you!

[Anders Buen]

Maybe similar to what a slab cut back will do?

[Don Noon]

2 hours ago, Anders Buen said:

Maybe similar to what a slab cut back will do?

Depends if you're referring to the MDF back or my comment about the effects of age on a maple back.

MDF would not be similar to slab-cut maple, as the longitudinal stiffness of maple would be the same, but crossgrain stiffness would be slightly lower than quarter-cut.  That is in the opposite direction of MDF, where crossgrain and longitudinal stiffness is the same.

My hypothesis about age, where crossgrain stiffness might be degraded the most, might be be more similar to a slab-cut back... but only in stiffness.  I think the reduced damping of age would be more important, and would not be similar to a modern slab-cut back.  And the back has less effect on tone than a top.

I once made a slab-cut top as a test, and it scored tied for 10th (out of 44) in tone at VMAAI 2015.  It was non-torrefied Sitka on a non-torrefied VSO... maybe if it was torrefied, it would have sounded more like a Strad.  The transition hill was weak and shifted to lower frequencies, but the high end was nothing special.

[Marty Kasprzyk]

5 hours ago, Don Noon said:

Here are the first-day results.  My top-line result:  not as much change as I expected.  Some predictions way off, others close.

I forgot to mention the A0 mode, which dropped from 273 to 264 Hz and gained .6 dB.  This was not a surprise.

B1- frequency dropped from 433 to 408 Hz, about half of what I expected, and amplitued decreased 2 dB.

B1+ frequency surprisingly dropped from 535 to 459 Hz, far more than I expected, and amplitude increased 2.2 dB

The middle frequencies, or "transition hill", became weaker and lower in frequency, about as I expected.

Higher frequencies are a bit different, but generally about the same shape and slightly lower average amplitude, similar to expectations.

Impact spectrum comparison:

I also did a bowed semitone spectrum comparison, with similar results but less resolution.  Here is the bowed semitone scale comparison, string-by-string, with the maple back first and MDF back second:  MDF back comparison.mp3

It appears that in-plane stretching of the back is a more significant factor in the B1+ frequency than I thought.  Perhaps if I had a good Finite Element Model to play with, I could have found that out.  But I don't.

This test also had re-adjusted my ideas about what could be the cause of the "Old Italian Sound"... but not much readjustment.  Just that the back might not be as big of an influence.  I'm still going with the idea of lower damping and loss of some stiffness... mostly crossgrain.

 

How did it sound and play?

I believe a high amount of damping isn't necessarily bad--it should reduce note starting and ending transients which would make fast passages less blurred and it should reduce wolf notes.  

[Don Noon]

11 minutes ago, Marty Kasprzyk said:

How did it sound and play?

I believe a high amount of damping isn't necessarily bad--it should reduce note starting and ending transients which would make fast passages less blurred and it should reduce wolf notes.  

You can hear the scale comparison in the linked MP3 file.  I'd say it's more subdued and mellow, more bottom oriented, and a little less even on the E string.  But basically it's recognizable as the same fiddle.

While high damping might do those things you mention, I think the hidden cost is that the sound out vs. energy in from the player will be reduced, and that's a very bad thing IMO.  Most of the response concepts are based on costant FORCE input, but energy matters a lot to the player.

[Andreas Preuss]

7 hours ago, Don Noon said:

not as much change as I expected.

I told ya. 

[Andreas Preuss]

2 hours ago, Marty Kasprzyk said:

I believe a high amount of damping isn't necessarily bad--it should reduce note starting and ending transients which would make fast passages less blurred and it should reduce wolf notes.  

I think damping is only necessary on the top plate. 
 

For wolf notes I thought it has more to do with the thickness and arching of the top. The lower and the thinner the more likely you will have a wolf, or not?

[Anders Buen]

12 hours ago, Don Noon said:

Depends if you're referring to the MDF back or my comment about the effects of age on a maple back.

MDF would not be similar to slab-cut maple, as the longitudinal stiffness of maple would be the same, but crossgrain stiffness would be slightly lower than quarter-cut.  That is in the opposite direction of MDF, where crossgrain and longitudinal stiffness is the same.

I once made a slab-cut top as a test, and it scored tied for 10th (out of 44) in tone at VMAAI 2015.  It was non-torrefied Sitka on a non-torrefied VSO... maybe if it was torrefied, it would have sounded more like a Strad.  The transition hill was weak and shifted to lower frequencies, but the high end was nothing special.

My experience with slab cut backs is that the B1+ mode tend to lie lower in frequency than it will in normal cut ones. I am sure you have measured your slab cut back insturment and know how that came out.

The crossgrain strength is a bit weaker, and it thus becomes more isotropic than the normal maple plate, I believe. In general maple is more isotropic than spruce to begin with. I think the MDF is quite a bit less stiff in the longitudinal direction. But failry similar for the sideways direction. I may have to look this up, though. And MDF plates can vary too in their properties. With the givenweight it must have been on the thin side of grads, which also is important to bending stiffness.

I do not think old wood loose much strenght unless it is really old and cracked up. The oak in the non alum treated viking ships in our Viking ship museum here in Oslo is believed to have moduluses of about 60% of new wood for the FEA analysis. These ships has been under clay for 800 years. 

Edited May 6, 2021 by Anders Buen
Clarification and spelling corr.

[HoGo]

9 hours ago, Don Noon said:

I think the reduced damping of age would be more important

Is there a study about that? I've assumed the damping of old wood increases as it loses some stifness. But perhaps it's just the idea of old "half rotten" wood in the back of my brain that suggests such thoughts...

[Peter K-G]

13 hours ago, Don Noon said:

B1- frequency dropped from 433 to 408 Hz, about half of what I expected, and amplitued decreased 2 dB.

B1+ frequency surprisingly dropped from 535 to 459 Hz, far more than I expected, and amplitude increased...

 

Not a surprise at all!

As you still do mode experiments, why don't you learn how to set them exactly on the Hz you want and stop predicting?

[sospiri]

On 5/5/2021 at 4:46 AM, Andreas Preuss said:

Just for clarification on this experiment:

You are replacing a back plate made of normal wood with a back plate made of MDF, right?

Arching is supposedly approximately the same?

What I have seen on my super light violin project is that changing back plates does not change awfully lot in terms of sound. What is more interesting is how playability and and perception of sound under the ear does change. Softer back plates make the sound more diffuse. At the same time lower resonances get stronger.

All this looks to me that a ‘good sound’ (whatever a maker defines as such) is based on how the top is designed as a membrane. All the rest is a kind of modifying those properties. When making the back a kind of weaker the whole setting of the string angle needs to be readjusted (re calibrated) to make it function to its best again. In that case I would most likely lower the string angle.

i am not too much experienced with reading sound graphs, but there my prediction is that you might end up with a graph of less sharp peaks in the  higher registers. 

In term of wood treatment I am not quite sure how valid this experiment is, because MDF as a material lacks one important factor because stiffness in vertical and horizontal direction is the same. Natural wood is always different in both directions. Wood treatment changes the stiffness in both directions, but never to the degree that they become equal. My experiments with steaming wood (not boiling in water) made maple measurably lighter and tap tones of the uncut planks got a bit higher presumably due to weight loss. On spruce this method didn’t show a significant change.
 

I was always wondering how internal stress in natural wood affects the sound. I am inclined to think that treatments like ponding wood in water relieve stress. There was one scientific paper which compared violins and the violin made with ponded wood showed different all over sound characteristics. (I would need to look it up again, I think it was a paper where Terry Borman participated as a maker) 

Usually my predictions are just the opposite what you come up with. 

 

5 hours ago, Andreas Preuss said:

I told ya. 

Congratulations Andreas. Excellent.

So the back is made of the prettiest wood we can find?

[Don Noon]

6 hours ago, Anders Buen said:

1) (re: slabcut) The crossgrain strength is a bit weaker, and it thus becomes more isotropic than the normal maple plate, I believe. 

2) I do not think old wood loose much strenght unless it is really old and cracked up. The oak in the non alum treated viking ships in our Viking ship museum here in Oslo is believed to have moduluses of about 60% of new wood for the FEA analysis.

1)  With reduced crossgrain stiffnes and normal longitudinal stiffness, slabcut will be more ANisotropic.  MDF is isotropic, the same stiffness both ways... compared to maple, it's very low longitudinally, but about the same crossgrain.

2)  60% loss of modulus is a huge number, especially if it is longitudinal where the cellulose chains are mostly oriented.  I think aging affects the softer components first, and that would show up in crossgrain stiffness before much happens longitudinally.

5 hours ago, HoGo said:

Is there a study about that? I've assumed the damping of old wood increases as it loses some stifness. But perhaps it's just the idea of old "half rotten" wood in the back of my brain that suggests such thoughts...

No study that I am aware of, just implications.  In hydrothermal processing, it is clear that damping declines, and declines more for stronger processing (implication: this process might be similar to natural aging, but accelerated).  Bruce Tai mentioned hemicellulose loss and decreased EMC with age.  The hemicellulose and water are strong damping components in wood.  Some makers listen for ringing to determine when wood is well aged.  All of this is semi-anecdotal, but it's something.

5 hours ago, Peter K-G said:

As you still do mode experiments, why don't you learn how to set them exactly on the Hz you want and stop predicting?

Because, as I have mentioned many times, exact mode frequencies are not important, so I don't care... except in a theoretical sense in a test, to check my understanding of the physics.

For this particular test, the B1+ went from 535 (normal-ish) to 459 Hz (viola territory and lower by far than any violin I've seen), yet the overall character of the instrument didn't change all that much.  Which reinforces my view of the unimportance of exact mode frequencies, but changes my understanding of the physics.

[Marty Kasprzyk]

3 minutes ago, Don Noon said:

> MDF is isotropic, the same stiffness both ways... compared to maple, it's very low longitudinally, but about the same crossgrain.

.

You could adjust the longitudinal and cross grain stiffnesses to duplicate various woods by cutting grooves in the MDF plate with your cnc machining (subtractive forming) or by gluing on ribs (additive forming) like they do in guitar building.

 

[Don Noon]

1 hour ago, Marty Kasprzyk said:

You could adjust the longitudinal and cross grain stiffnesses to duplicate various woods by cutting grooves in the MDF plate with your cnc machining (subtractive forming) or by gluing on ribs (additive forming) like they do in guitar building.

Yes, I could... but I can't think of any reason WHY I would want to do that.  This test has satisfied my curiosity, and only took me one hour to pull off the back and glue on the new one (but several hours diddling the data and posting on MN).