Technical analysis

The monitoring system is called "TV-holography" and uses a powerful green Argon LASER and a holography optical measuring system containing a video camera for monitoring of the resonances in the wooden body of the instruments. Each resonance was exitated by a sinousodial sound signal (a note with no overtones in it) from a loadspeaker or a piezoelectric crystal waxed to the violin.

Resonances

Typical forms of vibration is typically related to each resonance. So different instruments have more or less the same forms of vibration, though not with the same activity. Thus they can be named as seen in the figure to the right. The resonances also are different in how long they are in activity after exitation is shut off. If you hit a bell it sings for quite a long time, wood sounds somewhat more "dead". Different wood sounds also a bit different depending on the internal damping in it. If there is a crack in the belly it will sound different with a shorter decay time.

I use my ears and six adjustable tuning forks while working tonally with an instrument and the plates. That is a fast and reliable method, but it may be difficult to distinguish between tones of higher frequencies when the instrument is glued togehter. The reason for that is because the air resonance dominate the sound and mask other resonances while tapping and listening. The decay time becomes lower and and frequencies tend to lie higher up, both making identification more difficul. Then analyzing the impulses using a computer with a sound card is useful, like WinMLS, or a program I have made in MATLAB. Then the resonances of the instrument appear as hills and walleys as seen in the figure unederneath.

The figure show a resonace curve from a violin my grandfather Anders Buen sr. has "avstemt". The first four resonances are A0 (the air tone), and the first important structural resonances T1, C3 and C4. You may see how they appear in the figures to the right. (Those more familiar with looking at such curves may see that the high frequency response seem very low. This is because the excitation is by snapping with the finger nail at the bridge side, a signal with much energy in the lower frequencies, and much less at higher frequencies. The input force is thus not flat, or accomplished for, as it would be when using a controlled input signal like an impact hammer. The method does however give important information about frequencies and levels of the resonances below, say 1kHz).

The figure to the left show a comparison of two instruments grandfather A. Buen sr. has "avstemt", a special tuning process for the plates of the instrument. The sound that is used for these curves you can listen to here: Avstemt french violin (green curve). And the other: Avstemt violin from 1965 (red curve) The placement of resonances are quite similar. The red curve should be shifted up, as that instrument is slightly more resonant than the green one. Tthe finger snap was slightly harder at the green.

Scientists knows how the sound spectrum should look like from a good violin but not so much about how we can achieve this spectrum. They should have even and high resonances for the lower tones under ca 1100Hz, vallies around 500Hz and 1500Hz and a broader top around 2300-3000Hz. This last top makes the instrument easier to be heard over other instruments or background noise.

The frequency responses you see here is from two world class instrumnts made by Joseph Guarnerius delGesu in 1738 and 1742. The blue curve is an instrument after the famous norwegian violinist, Ole Bull, and is presently played by the concertmaster in Oslo Philharmonic Orchestra. The other is owned by a company in London. They are very similar in frequency response, but the London Guarneri seem to have a bit more response in the high frequency region around 3-4kHz. The measurements are done with WinMLS and was done by Anders Buen in 2003. (The excitation was by snapping at the bridge side (somewhat gentle!). This "snap signal" contain more bass than high frequency sound, and thus the measured violin response is not balanced as it would using an impact hammer. With a measured force spectrum one may compansate for the uneven force spectrum afterwards. The price difference between the "force transducers": 1. a finger snap and 2. a snap by an impulse hammer. is some USD4000. The time needed for the instrument setup is also about a factor 1000 in favour of the finger, 3 seconds as opposed to half an hour)

On the hardangerfiddle there is done little on the acoustics and quality. Recently I have compared frequency responses of some 34 Guarnerius and Stradivaris with about 30 famous hardingfiddles.

On wood and compensation for differences

Generally a light weight and resonant wood is assumed to be good. But if the body is too sensitive to the string signal, a natural sound mightl not be produced. The violin may also collapse or deform over time when there is not enough strength in the plates. Thin plates make the instrument resonant, but less strong against string pressure. Balancing these requirements well is a "key performance index" for a luthier.

The violin maker usually determine the relationship between mass and stiffness in the plates by bending them in different directions ore by tapping and listening. Physically both gives measurable quantities. We all do measurements with our senses in our everyday life, so there is nothing special about this..

The materials differ widely in their ability to produce sound, so the luthier should be able to pick out the best material to start with and compensate for the differences from the norm in some way. A method of doing that is by adjusting the frequencies of the plates of the instrument during the construction process.

"Vibration modes in violins"

The measurement system

Resonances

On wood and compensation for differences

Further reading