Jump to content
Maestronet Forums


  • Posts

  • Joined

  • Last visited

Everything posted by ctanzio

  1. Pure potassium silicate in crystalline form is colorless. But when ground into a powder, it will appear white because the irregular surfaces of the powder grains will scatter light. But this silicate also has a refractive index of 1.5+, about the same as wood, linseed oil and resin based oil varnishes. So when these varnishes are applied to the white powder, it will stop scattering the light and the texture and color of the wood will reappear. If varnishes are penetrating into the wood after you applied this ground, then you most likely used too much oil and/or not enough silicate. You want just enough oil to make a very thick paste so you can evenly spread the thinnest of layers of the silicate onto the wood. Once it dries, it should be clear unless you did not use enough oil to cover all the powder grains. Besides the challenge of getting the right amount of oil to silicate powder, there is also the challenge of getting the grain size fine enough to penetrate the openings in the wood grain. Mineral ground that simply sits on the surface of the wood can have adhesion problems. Additional layers of varnish may fracture or peel off. A downside to this silicate is that it is highly soluble in water. Anything applied to wood that also reacts with water is asking for trouble in humid conditions. But it also supplies another option for applying it. Create a clear, near saturated solution of the silicate in water. Now gently dab the solution onto the wood and let it dry between additional coats. You can finish by gently rubbing the surface to get a uniform white coat. This will guarantee that the silicate will penetrate into the wood grain and thoroughly seal it. Now apply oil varnish as usual to restore the transparency. Be aware that water-based ground applied this way can cause the grain to swell in visually undesirable ways. The challenge with pumice as a mineral ground is that it is a mixture of different colored minerals. It is normally not transparent. So when applied to wood as a fine powder, you are essentially coloring the wood, which means grain detail can be obscured. If you can grind it into a very fine powder, and only apply enough to penetrate into the wood pores without sitting atop the wood itself, you can both seal the wood and perhaps introduce an interesting colored pattern into the wood.
  2. Speculations that taint reputations without some reasonable evidence to back the claims are best avoided or else the Fora itself becomes suspect. That said.... The Dark Reality Behind America’s Greatest Thrift Store Empire | by Alice Minium | Medium Just a starting point for inquiring minds. I would have thought they would simply hang the violin at the local thrift shop. But if one wanted to maximize profits on the sale of free goods, hard to argue with unsupervised auctions over the internet.
  3. The qualitative categories need to have some numerical scale associated with them, like bad=0, good=1, awesome=2, for a computer to generate a scaled output. There is also the question of the inputs. There would have to be a standard way to test each violin that would give you confidence that you are capturing the parameters you think might indicate a violin's performance. Moreover, you don't just play into a microphone and expect the computer to make some sort of sense of it. You would need to give it some way of assessing the sound capture, like breaking it down into a response spectrum and extracting stuff like response peaks, relative strength of the peaks, density of peaks per octave, and maybe a dozen other things I can think of off the top of my head that may or may not be relevant to the sound and playability of a violin. AI that reproduces human adaptability to problems that occur during certain tasks doesn't need a neural net. It is a bit of a different AI problem. Classic examples I worked on years ago from the infancy of AI is teaching a computer how to be a medical expert in blood diseases, or diagnose problems with a fleet of diesel locomotives. The computer programs can accept and organize knowledge from human experts in the form of data points (object, property, value), to assemble and traverse decision trees to guide further testing and assess possible treatments/repairs. They are known as "heuristic" AI algorithms because, in a very real sense, the computers can learn by observing, like humans. Modern versions of these programs now have sophisticated interfaces that let computers gather the data points and do further testing without human intervention, but they still require human experts to train them, or at least spell out clearly defined goals that are quantified in some way. I once had an airplane flight instructor teach one of these heuristic learning programs how to help an inexperienced pilot to land a plane. After a couple of hours of "talking" to the computer using a natural language text interface, he was amazed at just how good the program became at talking someone into landing a plane in real time. We never actually used the program to actually help someone to land a real plane, as it was just a proof-of-concept of the program. The point is one needs clearly defined inputs and goals, all quantifiable in someway, even if it is just probabilities.
  4. AI is a catchall term for any computer algorithm that mimics some aspect of human thinking. It need not be a neural network. The work presented in the Strad article was covered previously on this forum. Here is the thing with neural net algorithms: you have to "teach" the net by changing an input and telling it what it should predict. It then compares what it predicted to what it should have predicted and then adjusts the "weight" it places on its inputs. You have to do this for a lot of different inputs and be able to measure the actual "result" it should predict for each of these cases. Technically, the authors used a neural net to develop a multi-dimensional curve fit of geometry parameters to measured modes. There is nothing really "intelligent" about this approach. It is more a computationally efficient algorithm for handling many inputs, or a small number of inputs with a range of values, to predict a limited number of results. The real question is the following: Is there something one can quantitatively measure that is an indicator of superior violin performance, like tone, bow response and projection. From many threads on this forum, it does not seem like mode predictions will fulfill that requirement.
  5. I agree with your observations about the effect of damping on transient response and decay after lifting the bow. But this doesn't address the idea of low damping giving a louder or more ringing response. Here is something to keep in mind: linear damping is energy loss per cycle of vibration. The vibrating wood is converting a percentage of its energy into heat with each cycle. For a given amount of bow work, a low damped violin will build up sound output to a higher level than a high damped violin. And once the bow is lifted, a low damped violin will vibrate longer and louder than a high damped violin. I cannot comment on your anecdotal story about tapping a Strad. If someone likes the tone of a violin with high damping, they can always bow harder to get a louder sound.
  6. You open the top, tilt back the neck, add a strip of wood, glue the top back, and think a 1 degree change in string angle is more significant than the surgery you just performed on the violin? Can you see why people may be skeptical that you "only changed the string angle"?
  7. How did you change the string angle without changing the bridge height or bridge position? How did you excite the spectrum? Bowing across all the strings using gliss? Tapping the side of the bridge? For a 1 degree change in string angle, you got a rather significant change in dB output. How did you measure that? At a specific frequency from the plot?
  8. This is generally known as the "logarithmic decay" method for determining damping. I created computer programs to process amplitude versus time data from sound recordings and mechanical movement transducers to estimate damping, mostly for the design of structures and safety equipment subject to earthquake loads. Here is the challenge to the use of this type of measurement: real-life vibration waves are complex and one is trying to pick the peak for some number of consecutive cycles. This tends to be easier visually rather than computationally for short time periods, and decay of certain mode frequencies in complex structures can cause "sudden" shifts in the period where the peaks occur. This can mask important damping effects that vary with frequency. Materials that are modelled well by linear damping effects are basically losing a percentage of the vibration energy per vibration cycle. This means that the RATE of energy loss (energy per unit time) is greater for higher frequencies. So general material damping effects work best at suppressing higher frequencies. Damping of lower frequencies usually requires some tuned damping mechanism for each of the prominent lower frequency modes. This is probably why things like the fingerboard, tailpiece, and various wolf suppressors seem to affect specific tonal ranges in the lower frequencies of the violin. It might also suggest why wood with low internal damping seems to give a more brilliant (high frequency) effect, and less of a dramatic effect on the lower frequencies.
  9. In general, string damping increases with a lighter touch. But it is also affected by the angle of the finger to the string. For example, in first position, my first through third fingers contact the string very near the tip and at a sharp angle, so only a small amount of the finger is contacting the string. But if I stretch the third finger for a D# on the A string, the finger is much flatter and more of the finger is contacting the string. There is a distinct drop in loudness unless I compensate with stronger bow action. The length of my pinky is on the small side so that tends to make a flatter contact in first position, and I have the same challenge in maintaining the loudness. I used to tilt the hand a bit at the wrist when playing the 4th finger to get a sharper angle to the string, but I have been trying to keep the wrist quiet. But I am still on the fence on whether or not this is a good practice.
  10. This is more complicated. There is a loudness difference between pressing the string right to the fingerboard with the tip of the finger, and pressing just enough to get a clear note to form. So certainly there is some damping being caused by the finger. However, a stopped E on the A string is also vibrating with less mass than an open E on the E string. So the open E string has a larger potential to exert a higher dynamic force on the bridge than a stopped A. Which effect is the more significant would require some testing. Like clamping the A string to the fingerboard with the fingernail and using the same bow speed an pressure on both the A and the open E.
  11. This is a tough question. It depends on what you want to achieve. If you just want loud and fast bow response, the answer is as small as possible. But the violin might start sounding more like a ringing bell than a violin. Don Noon with his experience with torrified wood might be in a better position to comment on low damping and violin tone.
  12. The simply half power bandwidth formula, r = (f2-f1)/(2fr), is derived by assuming peak power occurs at the resonance frequency. But in reality, it occurs at the UNDAMPED resonance frequency, something that cannot be directly measured from a damped system. For systems that are lightly damped, the damped and undamped resonance frequencies are close enough to use the simple formula when r < 0.05. For larger values of damping, there is an entire field of research dedicated to deriving various formulas to give a better measure of the damping. I believe the formula I quoted was for a mode shape similar to a vibrating cantilever bar. Do a search on the terms "damping ratio", "frequency response function" and "dynamic data" and be prepared for a lot of math and experimental data. There might be a formula that is more relevant to violin modes.
  13. My understanding is the Audacity plot is the dB level of the amplitude of the sound wave. Since power is proportional to the amplitude squared, a -6dB drop in amplitude gives a -3dB drop in power. Let me know if you want me to bore you with the math. It only takes a few steps to derive. If all you want is some relative measure of damping, not a theoretical rigorous measure that one can plug into an equation, the actual dB drop used to get the width of the resonance peak is somewhat arbitrary.
  14. Damping estimates from sound decay measurements pose a challenge to extract useful material information because of the variety of effects that cause the sound to decay: internal wood damping, non-projecting resonators like the tail piece and finger board, and sound radiation into the air. If one could measure the energy loss just from sound radiation, one would probably find geometric conditions, like plate mass and arching, dominate the effect. But it is not realistically possible to separate these different effects from decaying sound recordings because there is no way to physically remove internal damping and non-projecting resonant effects from radiation loss using this method. One thing you can do to estimate internal wood damping, and thus some "quality" measurement of the wood itself, is to do a frequency sweep about a mode resonate frequency, and compute something called the half-power bandwidth ratio. Since the violin body is being driven at a steady state sound output, the radiation effects are canceled, and the damping response becomes that of (mostly) internal wood damping. If you do an Audacity spectral response plot of dB versus frequency, find a resonance peak. Find its dB level. Now subtract 6 from it to get the half power level. Follow the curve on either side of the peak until you reach this half power level and note the frequencies: one higher than the resonate peak and the other lower. Subtract the lower from the higher and divide by the frequency of the resonate peak. This is a guestimate of the internal damping at the resonate frequency. You can repeat this for as many resonate peaks as is possible to get an idea of how material damping varies with frequency. This works for "small" values of damping. A more accurate formula is: damping = {0.5 - [0.25 - 0.0625 x ((f2-f1)/fr)^2) x ((f2+f1)/fr)^2)]^0.5 }^0.5 f1 = lower half power frequency f2 = higher half power frequency fr = resonate frequency The method would also be applicable to a wood billet, but the challenge would be in setting the dimensions to get a range of resonate frequencies in the same range as a violin.
  15. ctanzio


    There are lots of articles accessible over the internet that present a graphic history of the development of the f-hole via pictures of actual violins by famous makers over the centuries. Quite a range of f-hole designs have been successful over the years. But many of them where down-right ugly and ill-proportioned. It does not seem to be lost knowledge, but rather by trial and error makers found that f-holes amplify sound better when slender, and the dimensions place the helmholtz frequency in a range where it actually does something useful. The placement and width of the stem seems to be a compromise between tone and practical matters, like getting a sound post into the violin. Beyond that, Strad and Strad-like f-holes are commonly used because they look so graceful, while also fulfilling their purpose.
  16. ctanzio


    The f-hole dimensions are relative to the volume of air inside the violin and are used to form a tuned Helmholtz Resonator. The tuned frequency is roughly in the middle of the lowest octave playable on the violin because that is the range in which the violin makes sound by a periodic pumping action of the body across the two f holes. For this mode of sound production, the placement of the f-holes is less important than the dimensions. For higher octaves, Helmholtz Resonance and pumping action play an increasingly smaller effect. One can experiment by covering the f-holes with a light material and playing scales. If your f-holes have an overall perimeter length of "standard" f-holes, and the internal volume of your violin is in the range of "standard" 4/4 violins, they should work to produce a loud sound. I am not sure how your conversation with Don went so far off the rails with the talk of pressure differential and sound speed. I believe he was making the point that the f-hole "wings", the square-like tabs that separate the f-hole eyes from the stem, vibrate as one of the natural _STRUCTURAL_ modes of vibration of the violin. NOT as some sort of airplane wing. For some non-trivial range of frequencies, the wings will absorb some of the energy of vibration being inputted by the strings and start to vibrate in a way that produces very little transmitted sound. Thus they may affect the perceived tone of the violin. How much and whether this is good or bad is open to debate, but the effect is not.
  17. ctanzio


    A good starting point for mastering the straight edge and compass to draw violin outlines in various styles: Kevin Kelly Violin Maker (kellyviolins.com) This is not a quick study. But if you take the time to understand and try the method, you can make a drawing to transfer to a wood sheet and cut it into a mold.
  18. For vibrating structures, especially thin plates, there are two major "causes" of dropouts in response curves: 1. Tuned dampener. This is akin to Don Noon's A string afterlength post. The vibrational energy imparted into the violin at the frequency of the afterlength goes into the vibration of the afterlength which generates little radiated sound energy. But this would manifest itself as a distinct ringing under the ear, but not at a distance. I doubt this would be much of an effect that would reveal itself on a spectrum test, but I would have to research it a bit more to speak with confidence. Other parts of the violin that might act as "tuned dampeners" are the neck/scroll, tailpiece and the fingerboard. 2. Normal distribution of natural frequencies. There are a lot of thin plate problems where equations for the mode frequencies can be derived and verified with actual spectrum plot tests. If one plots out the mode frequencies on a log scale, like your spectrum plots, you can see the wide gaps between successive modes, each spaced unevenly from the previous mode, with the frequencies quickly becoming mode densely packed, but with obvious gaps between the denser groupings. In other words, it is mostly an effect of the non-symmetric geometry of the violin. For example, many rectangular plate problems show mode frequencies following this equation: f = B*[ (m/a)^2 + (n/b)^2] where m, n, are independent integers going from 1 to infinity, and a, b are the dimensions of the rectangle, and B is a factor that is related to the stiffness, density, thickness and boundary conditions. If you draw a dot for Log10(f) along a line for a range of values for m and n, you get a distribution of modes that looks a lot like the spacing of modes along a violin spectrum plot.
  19. Your Audacity plot is essentially log(displacement) vs log(frequency). The approximately straight-line roll off at high frequencies is predicted by relatively simple damped harmonic oscillator models, which predict a linear relationship between the logs of these two measures. Moreover, wood is one of those materials that has frequency dependent damping: the higher the frequency, the higher the damping. A closer inspection of response charts shows a roll off that accelerates as the frequency increases, consistent with frequency dependent damping. The highest root frequency played on the violin is around E7. That is ~2600Hz. Any measured response above that is driven by weaker overtones. It should not come as a surprise that the inherent damping properties of wood would dominate at frequencies much higher than this.
  20. Assuming you do all the work yourself... Average quality strings, nut blank, bridge blank, fine tuners, sound post: ~$75 Basic chisel, file, knife, post setter, bridge template, glue pot, glue: ~$150 Crack on the treble side to the fhole should be repaired. Once you add a bridge and sound post and strings, there are significant shearing stresses applied along that line. Most likely it will get worse. Removing the top is the most challenging part of the repair. Gluing the top back on can be tricky and also requires an investment in a set of clamps which is a non-trivial expense. If you think you might enjoy working on your own violins, then this violin might be a good one for practice.
  21. Wood properties like the elastic moduli and yield/fracture points are affected by dynamic loadings, like strain rate. However, these are significant only at high strain rates typically seen during impact and blast events. Except for the occasional showpieces from composers like Wieniawski and Paganini, I doubt a violin will see such high strain rates.
  22. Dammar varnishes are more typically made by dissolving in turpentine or alcohol. Anything that does not dissolve is thrown away. Dammar in turpentine can make a very clear, high quality varnish that dries hard. You might be able to get it to combine with linseed oil by by first dissolving it in turpentine in a ratio of 4 turpentine to 1 dammar by weight. Discard whatever has not dissolved. Then add 1 part linseed oil. Like all anecdotal recipes, your mileage may vary. Also, turpentine is a bit toxic and many find its odor to be obnoxious. You might want to try a resin that is more compatible with linseed oil.
  23. To give an example to reinforce Don's comment about the difference between deflection and stress: Consider a cantilever beam fixed at one end, i.e., a beam sticking horizontally out of a wall, and subject to a point load some distance from the wall. The deflection will be highest at the free end, but the stresses will be highest at the fixed (non-deflected) end. Strain (relative change in deflection per unit length) is strongly related to stress. If everything in some small area is deflecting by mostly the same amount, the area will experience small strain and thus small stress, but can show a dramatic deflection. Deflection diagrams can also misleading unless the boundary constraints are based in some real-life scenario. In the case of a real cantilever beam projecting out of a wall, we know the deflections are relative to an actual, physically fixed end of the beam. So we can intuit something about the deflection and possible strains and stress throughout the beam. But what about a violin sitting on a table top. Is there any point actually physically constrained relative to the deflection of the rest of the violin? Turn it on it side. The deflected shape of the violin does not change. We say the violin is an "unconstrained body" and we must arbitrarily selected a point where all degrees of freedom are specified as zero to stop the numerical solution from blowing up. One would normally illustrate that point on a deflection drawing so one could judge relative deflection among the various parts of the object. As a final point in what has surely turned into an excessively long-winded rant on the science of solid mechanics, notice that the areas of high deflection on the violin diagram are all about the same color. That means the material in those areas have all deflected about the same amount, which means they are most likely under low strain which implies low stress. This is assuming the deflections are taken along the same direction.
  24. I am familiar with Colin's work on mode shapes, but I don't recall any analysis of plate stresses. You would need to provide a reference for me to look at. The term "most of the overtone output" needs further clarification. The shapes of many modes may show significant deflection in the same general area. If this area is relatively thin and far from strong support areas, like the garland, sound post and bass bar, it would not be surprising to see high stresses.
  • Create New...