Tap tones

[FiddleMkr]

I am at the “graduating plates” phase now. My process has been to get the plates close to the specified thicknesses and weights, then put their tap tones at a reasonable distance apart. 
       Because I have heard an uproar of objection concerning tap tones, and before I get too far, I want to ask. What should I pay the most attention to? And at this point is there anything more important than the tap tones? (Since the wood quality and archings are done) 

[H.R.Fisher]

  If your using the chladni method i have found that numbers are not as important as making sure that the tea leaves or whatever granules you use are very active/excited                especially mode 5 and mode 2. Other than that  I don't concern myself about too much about numbers.  This has been my experience thus far, but i'm still learning.

 

[Don Noon]

10 hours ago, FiddleMkr said:

What should I pay the most attention to?

Everything else.  Wood, arching, grads, weights.  

10 hours ago, FiddleMkr said:

And at this point is there anything more important than the tap tones? (Since the wood quality and archings are done) 

Everything else, as above.  You can measure and record taptones for educational purposes, but if you try to "tune" to some particular spec, you will have to diddle with grads, weights, or some other parameter that is likely more important, and possibly screw things up.

My opinion, obviously... but based on my experience, testing, and data.

[Andreas Preuss]

16 hours ago, FiddleMkr said:

And at this point is there anything more important than the tap tones? (Since the wood quality and archings are done) 

What is the weight of your top plate?

If it is too heavy this would be a major concern and eventually a good reason to make a new top.

Otherwise you can ‘scan through’ the top plate in search for too much material. I use sometimes simply a light source to see the denser areas. I don’t care if further scraping creates slightly irregular thicknesses. Half a gram weight reduction in 10 spots adds up to 5 gram. This can be around 6-7% of the total weight and should not alter the tap tones significantly. However I prefer to do this after f-hole and bass bar, aiming at a weight of 60-65g.

 

Edit: concerning tap tones we have the tendency to look alone on the frequency. I think it is by far more instructive to listen to how a tapped plate reacts. 

Edited November 6, 2023 by Andreas Preuss
Addition

[FiddleMkr]

8 hours ago, Don Noon said:

Everything else.  Wood, arching, grads, weights.  

Everything else, as above.  You can measure and record taptones for educational purposes, but if you try to "tune" to some particular spec, you will have to diddle with grads, weights, or some other parameter that is likely more important, and possibly screw things up.

My opinion, obviously... but based on my experience, testing, and data.

Then how do you know when to stop carving? Do you just ignore the tap tones and stop when the graduations and weights are (reasonably) correct? 

[Don Noon]

2 hours ago, Andreas Preuss said:

I don’t care if further scraping creates slightly irregular thicknesses. Half a gram weight reduction in 10 spots adds up to 5 gram. This can be around 6-7% of the total weight and should not alter the tap tones significantly. 

I have yet to see any correlation between plate taptones and quality of the completed instrument (excluding way-the-hell-out-there bricks or overly thin plates).  You can find a vague correlation with signature modes, but try showing that signature modes matter to quality of the completed instrument (excluding pathologically extreme examples).

25 minutes ago, FiddleMkr said:

Then how do you know when to stop carving? Do you just ignore the tap tones and stop when the graduations and weights are (reasonably) correct? 

That would be my recommendation.  I keep taptone data just in case there is something that matters.  So far, the only thing I can get out of the data is noise, except for the few times I followed the taptones into the weeds and things got worse.  Most of the time, it's meaningless.

[Michael Richwine]

2 hours ago, FiddleMkr said:

Then how do you know when to stop carving? Do you just ignore the tap tones and stop when the graduations and weights are (reasonably) correct? 

The best maker I know, with instruments in orchestras across the world, goes largely by the feel or flex of the plates in his hands. He is very finicky about the arching and choice of wood, and uses a consistent graduation pattern, but generally stops carving when the flex feels right.

[FiddleMkr]

1 hour ago, Don Noon said:

I have yet to see any correlation between plate taptones and quality of the completed instrument (excluding way-the-hell-out-there bricks or overly thin plates).  You can find a vague correlation with signature modes, but try showing that signature modes matter to quality of the completed instrument (excluding pathologically extreme examples).

That would be my recommendation.  I keep taptone data just in case there is something that matters.  So far, the only thing I can get out of the data is noise, except for the few times I followed the taptones into the weeds and things got worse.  Most of the time, it's meaningless.

I don’t doubt your assertion that things got worse after following tap tones into the weeds, but how did you know that things got worse? Did you test completed instruments that had tuned plates verses ones without? Or did you take pairs of plates (tuned and not) and put them on the same rib assembly, and test them? (And by “test” I mean judging by ear if the instrument sounded good or not.) 

     The reason I ask is because I only make one or two instruments per year, and I don’t have enough years left to figure this out myself. Again, I’m not doubting your assertion. I just want to learn how to make a good fiddle. 

[FiddleMkr]

19 minutes ago, Michael Richwine said:

The best maker I know, with instruments in orchestras across the world, goes largely by the feel or flex of the plates in his hands. He is very finicky about the arching and choice of wood, and uses a consistent graduation pattern, but generally stops carving when the flex feels right.

I remember you saying this before, and I started trying to get a feel for this with my last (couple of) pairs of plates. It can be a little scary, particularly to twist a two piece plate. (It’s a good way to find out how good of a job you did gluing it together.)   
      It would be interesting to know what the correlation is between the resistance to bending and a good sounding instrument. And if this could be measured.

Edited November 6, 2023 by FiddleMkr
Added last sentence

[Marty Kasprzyk]

12 hours ago, H.R.Fisher said:

  If your using the chladni method i have found that numbers are not as important as making sure that the tea leaves or whatever granules you use are very active/excited                especially mode 5 and mode 2. Other than that  I don't concern myself about too much about numbers.  This has been my experience thus far, but i'm still learning.

 

For many years Robert Spear used the old fashioned glitter tests with a loudspeaker blasting sound to the bottom side of a horizontal plate in a "cup up" position with the glitter on the plate's inside surface.  A frequency generater was used to vary the sound frequency until strong vibration patterns emerged.

He found the mode 2 and mode M5  frequencies but he believed that the height of the glitter bounce was very important and he was looking for sharp narrow node lines for the M2 and M5 modes which was an indication of good glitter bouncing.

The glitter bounce height is obviously a measure of how much the plate is vibratiing and therefore how much sound could be produced.  Spear wanted to ensure the instrument was adequately loud and he wasn't very concerned with hitting target M2 and M5 target frequenices like Carleen Hunchins had advocated.

What Spear was doing was a subjective measure of the plate's "mobility" or "admittance" which is the ability of the plate to move.  The reciprocal of admittance is impedance Z which is how difficult the plate is to move.  A stiff S and heavy m plate has a high impedance from   Z= (Sm)^.5   

The historic way of bending and twisting the plates gave a "feel" for the plate's stiffness and it can be argued that this works very well with a lot of experience and no equipment is needed but for record keeping it is helpful to have a actual numeric measurement.  The plate mode's stiffness S can be calculated from the mode frequency f and plate mass:

S=mf^2 

Spear was trying to make sure his plates weren't very stiff or heavy so they could vibrate easily.  

Newer ways of measuring plate admittance have been developed what use laser vibrometers or accelerometers, impact hamers and computer software which generate numerical measurements rather than the subjective "glitter bounce" height that Spear used.  Modern modal analysis is used to show the M2 and M5 mode shapes that Spear's glitter had showed.

These new techniques were used in the recently completed  Bilbao Project in Spain's Bele violin making school https://www.bele.es/en/bilbao-project-introduction/  where top and back plates were made with various levels of impedances https://www.bele.es/en/making-tops-backs/ using the basic impedance equation Z= fm where f is the mode frequency and m is the plate's mass.

So the Bilbao Project had mode impedance targets rather than just mode frequency targets that Hutchins had promoted.  The plate mass is also important.

Combination of these different plates were assembled into five different violins which were carefully studied in blind listener tests.  All this stuff is great work but the equipment to do this is too expensive for most violin makers.

I think similar results could be obtained by using the simple and inexpensive glitter tests to get the M2 an M5 frequencies and by also using their plate weights to do the impedance calculations.

 

 

 

[FiddleMkr]

2 hours ago, Marty Kasprzyk said:

For many years Robert Spear used the old fashioned glitter tests with a loudspeaker blasting sound to the bottom side of a horizontal plate in a "cup up" position with the glitter on the plate's inside surface.  A frequency generater was used to vary the sound frequency until strong vibration patterns emerged.

He found the mode 2 and mode M5  frequencies but he believed that the height of the glitter bounce was very important and he was looking for sharp narrow node lines for the M2 and M5 modes which was an indication of good glitter bouncing.

The glitter bounce height is obviously a measure of how much the plate is vibratiing and therefore how much sound could be produced.  Spear wanted to ensure the instrument was adequately loud and he wasn't very concerned with hitting target M2 and M5 target frequenices like Carleen Hunchins had advocated.

What Spear was doing was a subjective measure of the plate's "mobility" or "admittance" which is the ability of the plate to move.  The reciprocal of admittance is impedance Z which is how difficult the plate is to move.  A stiff S and heavy m plate has a high impedance from   Z= (Sm)^.5   

The historic way of bending and twisting the plates gave a "feel" for the plate's stiffness and it can be argued that this works very well with a lot of experience and no equipment is needed but for record keeping it is helpful to have a actual numeric measurement.  The plate mode's stiffness S can be calculated from the mode frequency f and plate mass:

S=mf^2 

Spear was trying to make sure his plates weren't very stiff or heavy so they could vibrate easily.  

Newer ways of measuring plate admittance have been developed what use laser vibrometers or accelerometers, impact hamers and computer software which generate numerical measurements rather than the subjective "glitter bounce" height that Spear used.  Modern modal analysis is used to show the M2 and M5 mode shapes that Spear's glitter had showed.

These new techniques were used in the recently completed  Bilbao Project in Spain's Bele violin making school https://www.bele.es/en/bilbao-project-introduction/  where top and back plates were made with various levels of impedances https://www.bele.es/en/making-tops-backs/ using the basic impedance equation Z= fm where f is the mode frequency and m is the plate's mass.

So the Bilbao Project had mode impedance targets rather than just mode frequency targets that Hutchins had promoted.  The plate mass is also important.

Combination of these different plates were assembled into five different violins which were carefully studied in blind listener tests.  All this stuff is great work but the equipment to do this is too expensive for most violin makers.

I think similar results could be obtained by using the simple and inexpensive glitter tests to get the M2 an M5 frequencies and by also using their plate weights to do the impedance calculations.

 

 

 

I couldn’t find the results. Perhaps there was testing to determine how good or bad the instruments sounded, with plates tuned this way or that, and I missed it. Or was the objective just to make a loud instrument? Or a plate with maximum amplitude at M5 and 2?

 

[Andreas Preuss]

5 hours ago, Don Noon said:

I have yet to see any correlation between plate taptones and quality of the completed instrument (excluding way-the-hell-out-there bricks or overly thin plates).  You can find a vague correlation with signature modes, but try showing that signature modes matter to quality of the completed instrument (excluding pathologically extreme examples).

In my oversimplified logic more flexible plates give the player more possibilities to take control of the sound in dynamics but also for different bowing techniques. So it is not directly correlated to things you see in a sound graph. 

[Torbjörn Zethelius]

7 hours ago, FiddleMkr said:

I’m not doubting your assertion. I just want to learn how to make a good fiddle. 

Pray.

[LCF]

6 hours ago, Marty Kasprzyk said:

For many years Robert Spear used the old fashioned glitter tests with a loudspeaker blasting sound to the bottom side of a horizontal plate in a "cup up" position with the glitter on the plate's inside surface.  A frequency generater was used to vary the sound frequency until strong vibration patterns emerged.

.......

Many ( too many) years ago I had a job  maintaining stringed instruments for an Education department music initiative across many schools. Mostly these were the other kind of 'usual', cheap Chinese fiddle outfits. Uggggghhh. I also repaired and rebuilt a lot of the usual types of usuals which were donated, found in cleanouts, inherited, bought from goodwill shops ... This was at a time when the ideas of Carleen H were hot off the presses and my time wasn't worth much so I had a more or less free hand to test those ideas and to learn without doing significant damage. It definitely was possible to increase responsiveness by eg regraduating soundboards which were originally very thick and roughly made also the backs were often thick planks and reworking those alone made a huge difference. Specific combinations of mode frequencies, back vs top 'by the book' did not  lead to any outstanding results. My impression is that smoothness of graduation, the weight and the overall pitch and responsiveness are the important parameters. Of course I couldn't do much about the arching.  For the same target pitch of eg mode 5 some plates sent the glitter bouncing off into orbit and on others it barely moved at all. These latter usually ended up not so good. At a very non-specific level that is in  agreement with this plate impedance concept but it was far from being a well controlled experiment. 

 

 Nowadays I'd rather ask the question what makes some violins sound so awful. Apart from those nearly all the rest will work and sound OK for someone. There must be something(s) which make a difference or else all fiddle makers should go out business.

 

 

6 hours ago, Marty Kasprzyk said:

I think similar results could be obtained by using the simple and inexpensive glitter tests to get the M2 an M5 frequencies and by also using their plate weights to do the impedance calculations.

I read that glitter has been outlawed in the EU as being a microplastic pollutant. Good riddance! Tea leaves are much nicer, or poppyseed. 

[Marty Kasprzyk]

4 hours ago, FiddleMkr said:

I couldn’t find the results. Perhaps there was testing to determine how good or bad the instruments sounded, with plates tuned this way or that, and I missed it. Or was the objective just to make a loud instrument? Or a plate with maximum amplitude at M5 and 2?

 

I don't know if a final report has been written yet.  The attached video is a presentation by Claudia Fritz which describes the listener and player tests.  Some conclusions are given at about the 1:14 time.  https://www.youtube.com/watch?v=jnq1JdnV8Dg

 

My impression is that there seems to be a big scatter in preferences between listeners.  There also is a big scatter between players what players like. And there is a big scatter between players and listeners, and scatter between the rooms being used for all the testing.  All of this is interesting to me because I'm scatter brained.

 

[LCF]

Plate tuning:

 

Knock knock!

Who's there?

(No answer.)

[Marty Kasprzyk]

9 hours ago, Marty Kasprzyk said:

For many years Robert Spear used the old fashioned glitter tests with a loudspeaker blasting sound to the bottom side of a horizontal plate in a "cup up" position with the glitter on the plate's inside surface.  A frequency generater was used to vary the sound frequency until strong vibration patterns emerged.

He found the mode 2 and mode M5  frequencies but he believed that the height of the glitter bounce was very important and he was looking for sharp narrow node lines for the M2 and M5 modes which was an indication of good glitter bouncing.

The glitter bounce height is obviously a measure of how much the plate is vibratiing and therefore how much sound could be produced.  Spear wanted to ensure the instrument was adequately loud and he wasn't very concerned with hitting target M2 and M5 target frequenices like Carleen Hunchins had advocated.

What Spear was doing was a subjective measure of the plate's "mobility" or "admittance" which is the ability of the plate to move.  The reciprocal of admittance is impedance Z which is how difficult the plate is to move.  A stiff S and heavy m plate has a high impedance from   Z= (Sm)^.5   

The historic way of bending and twisting the plates gave a "feel" for the plate's stiffness and it can be argued that this works very well with a lot of experience and no equipment is needed but for record keeping it is helpful to have a actual numeric measurement.  The plate mode's stiffness S can be calculated from the mode frequency f and plate mass:

S=mf^2 

Spear was trying to make sure his plates weren't very stiff or heavy so they could vibrate easily.  

Newer ways of measuring plate admittance have been developed what use laser vibrometers or accelerometers, impact hamers and computer software which generate numerical measurements rather than the subjective "glitter bounce" height that Spear used.  Modern modal analysis is used to show the M2 and M5 mode shapes that Spear's glitter had showed.

These new techniques were used in the recently completed  Bilbao Project in Spain's Bele violin making school https://www.bele.es/en/bilbao-project-introduction/  where top and back plates were made with various levels of impedances https://www.bele.es/en/making-tops-backs/ using the basic impedance equation Z= fm where f is the mode frequency and m is the plate's mass.

So the Bilbao Project had mode impedance targets rather than just mode frequency targets that Hutchins had promoted.  The plate mass is also important.

Combination of these different plates were assembled into five different violins which were carefully studied in blind listener tests.  All this stuff is great work but the equipment to do this is too expensive for most violin makers.

I think similar results could be obtained by using the simple and inexpensive glitter tests to get the M2 an M5 frequencies and by also using their plate weights to do the impedance calculations.

 

 

 

I should have emphasized that one reason why top and back plates are arched and used in the "cup up" direction in the glitter tests is because the glitter doesn't fall off of the plate's edges when it bounces all over.  

The glitter tests don't work very well for assembled violins because the glitter just rolls off of the plates before it has a chance to form any node lines.  The modern modal analyis methods using impact hammers and accelerometers works much better.

 

 

[LCF]

2 hours ago, Marty Kasprzyk said:

I should have emphasized that one reason why top and back plates are arched and used in the "cup up" direction in the glitter tests is because the glitter doesn't fall off of the plate's edges when it bounces all over.  

The glitter tests don't work very well for assembled violins because the glitter just rolls off of the plates before it has a chance to form any node lines.  The modern modal analyis methods using impact hammers and accelerometers works much better.

 

 

In effect you can only watch the glitter dance if the instrument is not assembled. This highlights the problem of attempting to work free plates in such a way that a specific outcome is accomplished for the assembled instrument. There was an interesting photo via of your links of an edge holding device for free plates. Maybe that would help? We used to do something like this for guitar family instruments. There was a cut-out in a sheet of mdf slightly smaller than the soundboard , or back, which was then clamped in place with approx 8mm soft foam gasket between the board and the mdf. That is useful because guitar ribs are not compliant and tend to not take place in sound production in fact it's better if they don't so the top vibration is similar in the jig to the finished instrument. Guitars are like banjos with two heads. What a terrible thought.  For violins this is not the case. 

As far as exploring the vibrations of whole assembled violins there are many discussions on MN to follow up. It doesn't have to be sophisticated. A combination of tapping and holding in various places together with the use of a cheap microphone and a computer running free software is good enough. 

Then you can contemplate whether the whole body modes give you much  useful information. 

[David DEROY]

8 hours ago, Marty Kasprzyk said:

I should have emphasized that one reason why top and back plates are arched and used in the "cup up" direction in the glitter tests is because the glitter doesn't fall off of the plate's edges when it bounces all over.  

The glitter tests don't work very well for assembled violins because the glitter just rolls off of the plates before it has a chance to form any node lines.  The modern modal analyis methods using impact hammers and accelerometers works much better.

 

 

it seems to work fine also on the outside , photos taken on  Instagram 

[Don Noon]

7 hours ago, LCF said:

Then you can contemplate whether the whole body modes give you much  useful information. 

The Strad3D project has been available for 17 years now (!), and has more mode information than you can ever get in your shop.  I can't tell you how many hours I have spent studying the modes and contemplating them, but it's a LOT.

My result:    Interesting (for a thechnonerd like myself), informative, but not really useful in any practical way for making a good violin, and I am not aware of anyone who has been able to do so.  The plate modes, being only distantly related to a few of the lower body modes, are even less useful for making a good violin*.

*If you have no idea what the density and stiffness of your wood is, then taptones and weight can be of some use in deciding when to stop thinning a plate.  I have used that for regrads of thick student violins.  But for tonewood from reliable sources, I think using weight and thickness are all you need... or calibrated hands for the flex test.

[Davide Sora]

I don't feel like saying that tap tones are complete garbage, they just don't have the importance and reliability that people would like them to have. The problem is that they start to make sense when considered within a personal database relating to a certain model, and after having accumulated a certain number of plates with data on material properties, plate properties, and arching shapes, and knowing their effect on the sound of the finished violins. At this point, the tap tones are data as valid as any other and, therefore useful, although like all other numerical data, not essential for a truly expert luthier. As many people say in these cases, Stradivari did not need any numbers to make his violins, but not all of us have the centuries-old experience to which he had access, so we must try to make do with the few years of our personal experience. So anything that can help is welcome, as long as you are able to give it the value it deserves, without expecting miracles from every single piece of data you can get.

[Don Noon]

1 hour ago, Davide Sora said:

The problem is that they start to make sense when considered within a personal database relating to a certain model, and after having accumulated a certain number of plates with data on material properties, plate properties, and arching shapes, and knowing their effect on the sound of the finished violins. 

That's a big problem... especially if you don't make a lot, try out different archings, grads, weights, use varying wood, and don't have a good way to evaluate the sound of the finished instrument.

Folks who have made a "certain number" of instruments don't seem to ask, "What should I use for a taptone target?"  It's just the beginners.  And for them, I would suggest paying attention to everything else first (wood, arching, grads, weight) and only using taptones to start their database to see if they can make any sense of it later.  I'm jujst reporting that I can't see much useful in my database of taptones for ~30 instruments, but I experiment with all the other variables.

[Davide Sora]

32 minutes ago, Don Noon said:

That's a big problem... especially if you don't make a lot, try out different archings, grads, weights, use varying wood, and don't have a good way to evaluate the sound of the finished instrument.

Folks who have made a "certain number" of instruments don't seem to ask, "What should I use for a taptone target?"  It's just the beginners.  And for them, I would suggest paying attention to everything else first (wood, arching, grads, weight) and only using taptones to start their database to see if they can make any sense of it later.  I'm jujst reporting that I can't see much useful in my database of taptones for ~30 instruments, but I experiment with all the other variables.

Yes, many other data are more important than tap tones, which are ultimately just a consequence. But sometimes they can give guidance on how much is expected from a raw piece of wood, and how much of its measured properties remain in the processed plates. Not to mention the possible errors in measuring properties. That's certainly not the case for you, but I've seen people blindly rely on numbers that seemed suspiciously optimistic to me. Tap tones would reveal these kinds of errors to some extent

[David Alex Tricot]

On 11/6/2023 at 4:47 PM, Michael Richwine said:

The best maker I know, with instruments in orchestras across the world, goes largely by the feel or flex of the plates in his hands. He is very finicky about the arching and choice of wood, and uses a consistent graduation pattern, but generally stops carving when the flex feels right.

I would be interested in learning how to get good flexibility. Any idea how and where to learn that? What is good flexibility, what is not good...

[Marty Kasprzyk]

4 hours ago, Davide Sora said:

>

Stradivari did not need any numbers to make his violins, but not all of us have the centuries-old experience to which he had access, so we must try to make do with the few years of our personal experience

>

What percentage of Strad's violins were really good?   If you made enough violins some of them might be good. Even a dumb squirel can sometimes find a nut.