William Fry Internal Scraping Method

[Stephen Perry]

10 hours ago, Oded Kishony said:

Scraping or removing wood from the surface of the instrument in some random way, even if it is an 'educated' guess, and hoping for a positive outcome is like putting your elbows on a piano keyboard and expecting to hear a Bach fugue-not likely.

That is precisely right.  The things I do in setup and in subsequent tweaking are anything but random.  I always do things in the same order, and look for specific spots that respond in specific ways.  These are generally present, and usually in the same general area.  Fry's work seemed to look at picking an area based on some principle, rather than observation of what that particular instrument is doing.  I'm always trying for some specific goal I can hear, even if minor.  It's more getting things out of the way of the instrument doing what it wants than imposing some model on it.  

Won't substitute for arching, graduation, and proper setup, certainly.  I doubt any magic scraping or treatment of any kind would!

[Steve Rodriguez]

Sorry for the late input into this topic.  This is my first post to actually give input. 

I am a complete amateur having only built one violin recently, and am usually only on the receiving end for help from everybody.  However, maybe this can be of some help. 

Before I started building my violin, I experimented re-graduating two thick plated old German ebay fiddles trying out a few methods; one incorporating all the things I could described in that Fry book.  I removed the bass bar and along with reducing the thickness of the plates, I painstakingly mapped out and incorporated everything I could learn in that book and was real anxious to see what would happen after I put the violin back together again. 

What I found was that it sounded better, but after I completed regraduating and playing the second violin not using the Fry method, I began to think that the first violin's improvement in sound seemed like it was due more to the overall reduction of thickness of the plates alone.  My first violin I made from scratch sounds better to me than those two regraduated violins, but anyway this is just my opinion. 

Maybe I did something wrong in implementing those features, but that is how it turned out for me.  

I did get some good plate graduation experience from working on those two ebay fiddles though.

Regards,

Steve

 

[Steve Rodriguez]

I should clarify that I’m talking more about using his graduation map than scraping a finished violin.  Sorry about going down that bunny trail.

Steve

[Lars Silen]

I have been using internal sanding for a number of years now, essentially the same method as Fry's but using much better tools and probably better understanding. I think people that say that a small amount of sanding/scraping don't have any audible effect simply don't know what they are talking about.

Random sanding/scraping/re-graduation won't do anything positive to the instrument but could easily kill the sound.

A small example of internal sanding.:

I returned to adjusting a violin built a few years ago. Doing some test playing I noticed that the instrument was fairly even ... all strings had roughly the same loudness but the tone was generally dull.

Doing some tests I guessed that the f-holes wings were not working properly. It is easy to check this by knocking from a wing to the corresponding maximum vibration area. If the pitch of the knock tone differs then, to me, it means that we don't have a proper acoustic/mechanical match between the areas and this means that the wing doesn't get vibration energy and thus doesn't work properly.

When doing comparisons I usually play a half tone scale from the low G up to B on the E-string but with a short pause between each string. This allows me to look at spectra for the strings separately or to do a spectrum for all strings combined.

In the present case knocking from the wing towards the maximum vibration area (I call it the D-area) in SW (neck is north) it turned out that the bass side f-hole's knock tone was clearly higher than the corresponding main area in SW. This means that the inner side of the f-hole and the wing are slightly too thin ... which isn't a good starting point because it is difficult to add wood to the f-hole/wing. The solution is to increase the pitch of the  D-area and leave the f-hole and wing alone.

The question is then: Is the result a homeopathic kind of adjustment or is it clearly audible? I measured a half tone scale before the adjustment and after ... and the main frequency range that was changed was from 2500 - 6000 Hz. The increase in volume in this range was close to 10 dB in some areas which means that the change is very clearly audible. In a sense the wing works as a tweeter.

The first picture shows the situation after the adjustment. This is part a half tone scale played in the first position. The program used is Audacity.

The second picture is a similar half tone scale played in the same room with the same set up before acoustical adjustment.

The third picture is the difference between the two spectra. Notice the strong increase in power for 2500-3500 Hz and a similar boost around 6000 Hz. The sound of the working D-wing is easy to hear when compared to the original sound.

Technical data about the recording equipment:

Sound card: Zoom R8 @44.1 kHz

Microphone:  Rode NT1 large membrane mic.

The mic was positioned roughly perpendicularly to the violin in the direction out from the e-string side f-hole.

/Lars Silen   (physicist like Fry but I think Fry didn't know the whole instrument well enough)

[edi malinaric]

55 minutes ago, Lars Silen said:

I have been using internal sanding for a number of years now, essentially the same method as Fry's but using much better tools and probably better understanding. I think people that say that a small amount of sanding/scraping don't have any audible effect simply don't know what they are talking about.

Random sanding/scraping/re-graduation won't do anything positive to the instrument but could easily kill the sound.

- snip -

In the present case knocking from the wing towards the maximum vibration area (I call it the D-area) in SW (neck is north) it turned out that the bass side f-hole's knock tone was clearly higher than the corresponding main area in SW. This means that the inner side of the f-hole and the wing are slightly too thin ... which isn't a good starting point because it is difficult to add wood to the f-hole/wing. The solution is to increase the pitch of the  D-area and leave the f-hole and wing alone.

- snip -

 

 

Hi Lars - thank you for your post.

Two points...

- for clarity, would it be possible to plot the "difference" graph with the negative values pointing downwards?

- if one added a couple of coats of varnish over the "too thin" area how would it affect the frequency?

(when painting plywood sailplanes there is a definite increase in tap frequency after respraying)

cheers edi

[Don Noon]

Lars,

I have some serious difficulty with what you presented.

1- Bowed scales, in my experience, vary a lot just from the inability to bow exactly the same every time.  If you can show that the spectrum difference is very low from one take to the next, with hours or days between recordings (however long the modification takes), then there would be more credibility.

2 - Overall levels are vastly different between before and after, implying the bowing was quite different.

3 - Just looking at the level at 600 Hz, I the two graphs differ by nearly 10 dB.  However, that does not show on the "difference" graph.  Why?  Are the before/after graphs reversed from your description (first=modified, second=unmodified), or the "difference" plot being before minus after?

[Lars Silen]

2 hours ago, Don Noon said:

Lars,

I have some serious difficulty with what you presented.

1- Bowed scales, in my experience, vary a lot just from the inability to bow exactly the same every time.  If you can show that the spectrum difference is very low from one take to the next, with hours or days between recordings (however long the modification takes), then there would be more credibility.

2 - Overall levels are vastly different between before and after, implying the bowing was quite different.

3 - Just looking at the level at 600 Hz, I the two graphs differ by nearly 10 dB.  However, that does not show on the "difference" graph.  Why?  Are the before/after graphs reversed from your description (first=modified, second=unmodified), or the "difference" plot being before minus after?

Don,

You are absolutely correct (1) in that it is difficult to bow exactly in the same way. The difference plot has a offset estimated from the big major peak at 300 Hz this is simply to make the plot easier to read. Another complication and difference between before/after is also how well the pitches match between the two scales. Playing a half tone scale in tune is not easy .

(2) Recordings were done on different days in a "standard" setup. Even fairly small changes in distance between player and microphone may influence the recording level. Small changes in the angle between the major axis of the violin in relation to the microphone may of course also influence especially the high frequencies because because high frequencies are more directed. On the other hand the ear is fairly well fixed in relation to the violin when playing and the ear hears the same difference as the mic.

(3) I realized that the Maestronet system showed the pictures in another order than loaded. I corrected the text very fast but you clearly have been reading the old version. You have good eyes and understand what you see  !

(For other readers)
Absolute level differences between two otherwise  identical spectra are of course seen as an offset between two curves  that have the same  general form.
Obviously modifications may both make some resonances stronger and suppress other. The answer to Edi is shown below:

I realize that I read the difference incorrectly. There is actually a fairly strong boost also in the area around 600 Hz. What I see as the "zero level" is the area at ca. 300 Hz at roughly -10 dB. The offset used is incorrect but as such not interesting.

How much material was removed from the D-area to cause the changes?

I estimate that I sanded an area 50x50mm roughly 300 turns back/forth to match the pitch of the knock tone to the pitch of the f-hole side and the wing. A long time ago I measured how fast sanding removes wood. The result then was ca. 0.1 um/one movement back-forth using a set of magnets that is different from what I use today but probably fairly close. The inner working magnet uses 80-grit sand. From this an estimate could be that roughly 3/100 of an millimeter was removed. The area sanded is typically thin so the thickness changed by perhaps 1.5%. The modification is too small to be measured by ordinary thickness measurement instruments but my view is that even a small change like this is absolutely significant when we are talking about mechanical resonances.

A warning!
Changes are slow when sanding the G, D, A, and E-areas. Without any risk it is possible to sand those areas say 30 - 50 strokes between test playing. No big risk for fast major damage. When adjusting the sides if the f-holes never sand more than ten strokes between test playing.  When sanding the wing, for example the inner wing on the bass side (bridge to wing tip), one has to be very careful. Sand a maximum of  five strokes and then re-test. It is very easy to sand this area too much. Compare this to the tuning of an accordion reed. Tuning is done by very lightly carefully scraping the reed, sanding or using even a very small grinder is far to rough. The violin f-hole wings are extremely small parts where the resonance changes very much even for extremely small changes.

Notice!

My experience is that the effect of sanding initially is very strong but it will partly disappear within minutes/hours. This means that a modification in many cases has to be repeated a number of rounds until the modification is fixed. A positive aspect of this is that sanding "too much" isn't a catastrophe because if the instrument is left alone for a few days it may be just right due to the backing of towards the old state. I don't know the exact reason for this effect but I guess that sanding one of the wood surfaces will lower the hardness of the sanded surface. Fairly fast the sanded surface will oxidase and harden which is heard as a partial undoing of the sanding.

The reader probably notices that the method is to adjust a tuned and fully playable instrument. Only very small changes are made between test playing. Because the process is fast (sand/play/sand 30 - 60 seconds between steps) the brain still remembers the timbre before a modification which means that it is possible to learn by ear where to fix specific problems.

 

[Danube Fiddler]

On 8/13/2018 at 4:45 PM, Don Noon said:

As a scientist, I found gaping holes in Fry's logic big enough to drive a truck through.

As a maker, I found that even fairly severe regraduations only had limited tonal effects, so I didn't think that little piddly scraping as described by Fry was even worth following up... in addition to the problem of it being logically  faulty.

I focus on the wood and the arching mostly, graduations less so.  Then after it's put together: setup.  If there is excessive stiffness to the instrument, then I'll open it up for messing with thicknesses... but that been necessary on my last several instruments, and certainly not something I'd attempt thru a soundhole.

Very interesting discussion !

At first my personal experience : scraping ( changing graduations)  really matters, while I at the same time agree with your observation, that some typical character of a violin can/will remain the same within the altered sound. Perhaps I could compare with the fact, that your wifes voice will sound different in a phone call, but you will recognize her voice - something remains the same. May be the violin remains the same while any kind of graduation - tuning, but its "clothes" are changing, which can make a considerable difference.

I don´t know the Fry method, I rather like outside- tuning or re-opening the instrument, this also is, because I don´t like sandpaper on violin surfaces very much.

Generally the effects of random - tuning or an experienced/ learned tuning could be extremely different. 

As you told, your reason for any kind of grad-intervention would be a quite wrong stiffness. Since there are so fine methods like this of Dr. Harris to control single - plate-stiffness - did these methods fail in predicting the final total stiffness of the assembled instrument ?

 

[Lars Silen]

Builders interested in trying internal re-graduation can try it using the tool below. This is a magnetic sanding tool where one magnet 10 mm diameter and 2 - 2.5 mm thickness is glued to the plastic support using super glue. Let the glue dry/harden for a few hours before use. The working internal magnet is an identical magnet with abrasive sand glued to the side attracted by the outside working magnet. I use 80 grit sand that (found on ebay using the search "sand blasting 80 grit"). Apply super glue on the magnet face and pour a small amount of sand over it and let it dry thoroughly. Suitable magnets can be bought on ebay. Search for "super magnet". For the tool base given you want magnets with 10 mm diameter and thickness 2-3 mm. A thicker magnet gives a stronger pressure and faster sanding.

How is the tool used:

Use a clean paper on top of the varnish, under the outer working tool,  to protect the outer surface and drop the working magnet into the instrument. By placing the outside magnet close to the insert point you will catch the internal magnet immediately. Move the outside magnet to the area to be sanded, it will drag the working magnet to the sanding location. You can feel the pressure on the sanding magnet by trying to lift the outer magnet.

To avoid scratching the f-hole when removing the inner working instrument turn the outer magnet on its edge which will also turn the inner magnet n its non abrasive edge for safe removal.

The base used for the outer part of the sanding tool is shown in the picture. Unzip the zip file "fiol_slipverktyg.zip" and 3d print the resulting stl file.

 

fiol_slipverktyg.zip

[Danube Fiddler]

2 hours ago, Lars Silen said:

Builders interested in trying internal re-graduation can try it using the tool below. This is a magnetic sanding tool where one magnet 10 mm diameter and 2 - 2.5 mm thickness is glued to the plastic support using super glue. Let the glue dry/harden for a few hours before use. The working internal magnet is an identical magnet with abrasive sand glued to the side attracted by the outside working magnet. I use 80 grit sand that (found on ebay using the search "sand blasting 80 grit"). Apply super glue on the magnet face and pour a small amount of sand over it and let it dry thoroughly. Suitable magnets can be bought on ebay. Search for "super magnet". For the tool base given you want magnets with 10 mm diameter and thickness 2-3 mm. A thicker magnet gives a stronger pressure and faster sanding.

How is the tool used:

Use a clean paper on top of the varnish, under the outer working tool,  to protect the outer surface and drop the working magnet into the instrument. By placing the outside magnet close to the insert point you will catch the internal magnet immediately. Move the outside magnet to the area to be sanded, it will drag the working magnet to the sanding location. You can feel the pressure on the sanding magnet by trying to lift the outer magnet.

To avoid scratching the f-hole when removing the inner working instrument turn the outer magnet on its edge which will also turn the inner magnet n its non abrasive edge for safe removal.

The base used for the outer part of the sanding tool is shown in the picture. Unzip the zip file "fiol_slipverktyg.zip" and 3d print the resulting stl file.

 

fiol_slipverktyg.zip

Very elegant !

[Lars Silen]

22 hours ago, edi malinaric said:

Hi Lars - thank you for your post.

Two points...

- for clarity, would it be possible to plot the "difference" graph with the negative values pointing downwards?

- if one added a couple of coats of varnish over the "too thin" area how would it affect the frequency?

(when painting plywood sailplanes there is a definite increase in tap frequency after respraying)

cheers edi

I have thought about doing internal varnishing bet I have never really found the need for it. Probably a layer of varnish on either side would change the sound in some way ... ow much? I have no idea. I have also thought about using my own special paint made from tungsten oxide (extremely heavy) to influence the vibrations of too thin areas. My understanding though is that the stiffness of the wood is much more important than the mass of the vibrating are.

So ... worth trying but I wouldn't have too high hopes .

[edi malinaric]

42 minutes ago, Lars Silen said:

I have thought about doing internal varnishing bet I have never really found the need for it. Probably a layer of varnish on either side would change the sound in some way ... ow much? I have no idea. I have also thought about using my own special paint made from tungsten oxide (extremely heavy) to influence the vibrations of too thin areas. My understanding though is that the stiffness of the wood is much more important than the mass of the vibrating are.

So ... worth trying but I wouldn't have too high hopes .

Hi Lars - my thoughts about the use of varnish in fine-tuning were kick-started when someone wrote that some varnish "wear patterns"in old Italian violins just didn't make sense. They postulated that maybe the final tuning efforts involved modifying the outside varnish by adding and removing varnish in selected areas to modify the acoustic response.

In the example I mentioned - of aircraft dope altering the pitch of wooden sailplanes. It was the D-box section of the wing - typically formed from 1.5mm plywood. As in instruments, one attempts to keep the coat as thin as possible - so mass change would be negligible. However the dope does shrink as it dries - introducing a tensile stress on the outer surface of the ply - which is the probable cause of the change in pitch.

Some years back I purchased a violin "in the white" from a pawn shop - mainly because of the case. The poor violin had been painted over with white PVA paint. I think that I'm going to have some fun recording tap tones on the front and back and them see how they change as I scrape away the paint.

cheers edi

[Don Noon]

5 minutes ago, edi malinaric said:

As in instruments, one attempts to keep the coat as thin as possible - so mass change would be negligible. However the dope does shrink as it dries - introducing a tensile stress on the outer surface of the ply - which is the probable cause of the change in pitch.

It doesn't work that way.

[David Burgess]

12 minutes ago, edi malinaric said:

 

In the example I mentioned - of aircraft dope altering the pitch of wooden sailplanes. It was the D-box section of the wing - typically formed from 1.5mm plywood. As in instruments, one attempts to keep the coat as thin as possible - so mass change would be negligible. However the dope does shrink as it dries - introducing a tensile stress on the outer surface of the ply - which is the probable cause of the change in pitch.

What happens after it's been humidity cycled a few times? Does the tension remain?

[edi malinaric]

21 hours ago, Don Noon said:

It doesn't work that way.

Hi Don - I'm always happy to learn - what way does it work?

 

21 hours ago, David Burgess said:

What happens after it's been humidity cycled a few times? Does the tension remain?

Hi David - unfortunately I don't know. It was just a curiosity that I noted at the time.

After recovering the wings with fabric and returning the glider into service one loses touch with it and returns to enjoying flying your own machine.

In carrying out AIrworthiness Inspections on wooden gliders one spent a lot of time tapping glued joints to locate glue failures. On a pair of ASK7 wings I spent one whole evening tapping every joint in every built-up rib (think model aeroplanes) - and marking every "hollow" sounding joint with a piece of 20mm x 20mm red insulation tape. When the Committee, who'd being hassling me to "just get the job done", saw the number of  suspect joints, they backed off and allowed me to complete the repairs in peace. Last I heard it is still flying - 30 years later.

cheers edi

[Don Noon]

29 minutes ago, edi malinaric said:

Hi Don - I'm always happy to learn - what way does it work?

Generally, static stresses don't change mode frequencies of a structure.  The exception most brought up is strings... where stress DOES change the frequency, but that's an extreme case where the string has essentially zero structural stiffness, is perfectly  straight, and the stress is applied by an external framework.  

[David Burgess]

I can see how the application of a resin might turn wood into sort of a "stressed skin" material, raising the frequency, at least temporarily.

[Evan Smith]

1 hour ago, Don Noon said:

Generally, static stresses don't change mode frequencies of a structure.  The exception most brought up is strings... where stress DOES change the frequency, but that's an extreme case where the string has essentially zero structural stiffness, is perfectly  straight, and the stress is applied by an external framework.  

How would you explain steel drums?

The stress is within the frame itself.

How easy is it to confuse stress and tension?

Are they one and the same?

Musical saws?

[Don Noon]

23 minutes ago, David Burgess said:

I can see how the application of a resin might turn wood into sort of a "stressed skin" material, raising the frequency, at least temporarily.

I can't.

6 minutes ago, Evan Smith said:

How would you explain steel drums?

The stress is within the frame itself.

How easy is it to confuse stress and tension?

Are they one and the same?

Musical saws?

Steel drums and musical saws are all about the curvature, not stress.  Tension is one specific form of stress.  Compression and shear are other common forms.  Being investigated by the FBI is another form, but less common.

[Evan Smith]

6 minutes ago, Don Noon said:

I can't.

Steel drums and musical saws are all about the curvature, not stress.  Tension is one specific form of stress.  Compression and shear are other common forms.  Being investigated by the FBI is another form, but less common.

Hey You!

Don't get smart with me.

[Don Noon]

2 hours ago, Evan Smith said:

Hey You!

Don't get smart with me.

Can't help it... I was born this way.

[David Burgess]

5 hours ago, David Burgess said:

I can see how the application of a resin might turn wood into sort of a "stressed skin" material, raising the frequency, at least temporarily.

 

4 hours ago, Don Noon said:

I can't.

Then how would you explain these increases in frequency, from applying surface coatings?

http://jpschmidtviolins.com/MVA_filler.pdf

 

[Don Noon]

4 hours ago, David Burgess said:

Then how would you explain these increases in frequency, from applying surface coatings?

http://jpschmidtviolins.com/MVA_filler.pdf

These are all CROSSGRAIN samples, which aren't very stiff, and therefore coatings can improve stiffness (if longitudinal grain samples were used, the results would not look very good).

Coated samples sanded down to original uncoated weight (approx) also showed frequency gain... which I would attribute to either the coating impregnating the surface a bit, or making measurements after the thickness sander heated up the samples and drove off the moisture.  I don't see anything about careful humidity control and stabilization of these samples.

It has nothing to do with static stresses in the coatings.  It's all about stiffness, weight, and thickness.

[Emilg]

3 hours ago, Don Noon said:

These are all CROSSGRAIN samples, which aren't very stiff, and therefore coatings can improve stiffness (if longitudinal grain samples were used, the results would not look very good).

Coated samples sanded down to original uncoated weight (approx) also showed frequency gain... which I would attribute to either the coating impregnating the surface a bit, or making measurements after the thickness sander heated up the samples and drove off the moisture.  I don't see anything about careful humidity control and stabilization of these samples.

It has nothing to do with static stresses in the coatings.  It's all about stiffness, weight, and thickness.

Don, as an experiment i re-topped a violin which had a quite weak sound (#3). I used a piece of spruce i had from my guitar building days for bracing. It was quite wide grained but had a highish RR of 15.1.

The free plate M5 however turned out disappointingly low: 280 Hz w/o bb and 327 w. bb. at 68 grams.

I coated the inside with a few layers of eggwhite, maybe less than 1g. M5 was now 340 Hz with bb, an increase of 13 Hz!

So what you say makes sense: the crossgrain stiffness was low because of the wide graines and the eggwhite sort of fixed that. The longitudinal stiffness was probably uneffected by the coating. I guess together they support M5.

The violins sound was totally transformed. It is now the loudest violin i have, with a strong E which i hope will mellow down a bit. G and A are strong too, as is the D, which on my other violins is sometimes weaker in the lower positions. G, D and A also sound good, but A from 5th pos and up and E need to develop (sound stiff/harsh).

To sum it up: i'd say the top makes up 90% of the sound and power. I can't say what the eggwhite did soundwise, but it helped stiffen the top (M5) quite a bit.

[David Burgess]

4 hours ago, Don Noon said:

1. These are all CROSSGRAIN samples, which aren't very stiff, and therefore coatings can improve stiffness (if longitudinal grain samples were used, the results would not look very good).

2.Coated samples sanded down to original uncoated weight (approx) also showed frequency gain... which I would attribute to either the coating impregnating the surface a bit, ...

3. ...or making measurements after the thickness sander heated up the samples and drove off the moisture.  I don't see anything about careful humidity control and stabilization of these samples.

4. It has nothing to do with static stresses in the coatings.  It's all about stiffness, weight, and thickness.

1. Don't some plate modes involve crossgrain stiffness?

2. I agree that there would have been some impregnation of the surface, and I made a mistake using the term "surface coatings", when the article refers to them as "fillers".

3. Samples which hadn't been thickness sanded after application of the filler showed a frequency increase too.

4. Perhaps not. Two of the fillers which contract the most (in my experience), egg white and casein, were not the top performers at raising frequency. I'd tend to attribute it more to the stiffness of the filler material itself, or to the filler creating better adhesion between the wood fibers than there was in the untreated wood.