The effect of different bridges on tone

Anders Buen · June 3, 2013

I have been puzzled by the sound qualities of different bridges on a violin I recently have glued the top and back plate back on. It is an oldish French, a bit oversized, instrument I have studied in detail many years ago in my master thesis work and I set it up at that time with not much experience (1993). Its bridge from that time was very light 1,37g and its rocking frequency was just below 3kHz. It has sounded good, according to a player I was cooperating with at that time.

The top plate and the back has come loose in some places so I opened it recently and reglued it. At the same time I wanted to try a new bridge for it more in line with later experiences. I wanted to try a heavier bridge and a higher rocking frequency. The strings also needed to come a bit higher.

In my documentation I play scales on the instruments and record the 1/3rd octave band spectra in the same mic position in the same room. This is repeated three times and the avearage spectrum and the standard deviations are plotted for each situation I want to document. The light blue curve is a test from 2008 or 2009 before I reglued the plates. It also had Dominant strings at that time.

Now I have tried three new bridges for it. The first sounded brightest and was made of a low density maple, but the bridge was still too low and I did not reach the high rocking frequency I was aiming for (3,6kHz) so I tried a new one made of a heavier and stiffer material. I reached a higher rocking frequency, just below 3,5kHz but it became on the heavy side, 2,17g. Not by far as "sparking" sound. I tried yet another one, of really stiff wood and reached 3,6kHz and a more moderate weight 2,01g, not far from the average weight of merit bridges from violinbridges.com. now with the right height for the strings, but not the sparking sound I got out of the first new bridge.

I sum up the data on the bridges below. No 1. is the original bridge from 1993, the rest are new now.

No 1: 1,37g 2,97kHz No 2: 1,84g 3,27kHz No 3: 2,17g 3,46kHz No 4: 2,01g 3,64kHz

The graph show bridge 2, with dark blue (the more "sparkling" one), bridge 3, in orange and bridge 4 in red.

I am using a different measurement device for the three new bridges than the old. I should have tested it over with the original bridge, but I am afraid of doing too much harm to the new strings by doing too much changes. The setup for the new bridges are using what I believe is a "Tellefsen set" from wathing the photo of hs del Gesu from the Bergen exhibiton. A mix of brands, Pirastro Olive G 16, Priastro Tonica D, Dominant A and a Westminster 0,26mm steel E.

The differences in the low frequencies might, by part, result from the difference in the strings used or it can also be changes to the body response after the reglueing and resetting the sundpost. However, I am very puzzled by these results. Why do the first new bridge sound so much brighter, even if its rocking frequency is quite moderate?

I wonder if makers fit many bridges on their instruments before they are satisfied with the result?

The pics show the bridges in sequence from above and down (No 1, No 3, No 4) with the bridge on the instrument being the first new one (No 2) in the table.

Experiencces with Hardanger fiddle bridges also give the impression that the qualities of the bridges are very important. The body modes are important too, but does not influence the entire higher frequency spectrum of a violin like a bridge does.

Christopher Jacoby · June 3, 2013

I usually use a test bridge for each violin, one I keep around for each model (I only build on two or three moulds), and get a feel for the spirit of the sound. The stiffness of the bridge seems to make more difference than the carving, as long as one stays within reasonable thicknesses between the kidneys, ankle thickness, etc.

After I've played it with the test bridge, I know how much wood I want to try and leave above the heart, and how stiff a blank to choose. I know a few makers that seal the sides of softer bridges with CA glue...

Anders Buen · June 3, 2013

The solution with a reference bridge seems like a good idea. I have tried the "CA glue trick" and use it on Hardanger fiddle bridges too if the wood or bridge needs some stiffening.

I wonder if it is possible to salvage that brighter bridge by raising the G string only by a small pyramid or go to the more difficult step of rising the entire bridge with something under the feet. The latter method might change the sound more than the first method. - Does not look too good, but if the sound is better, why not?

I am testing the bridges rocking frequency by fixing the feet in a vice and record the sound after flicking the bridge sideways with the fingernail.

All the new bridges have a wide waist to keep the rocking frequency high. But maybe the combination of the mass and rocking frequency has to match something in the violin to work well?

Berl Mendenhall · June 3, 2013

Anders, it seems that sometimes changing bridges can change the sound quite a bit and sometimes very little. I try to cut all my feet and ankles the same or as close as I can get. I don't usually mess with narrow or wide waists. I do change the thickness of the top edge where the strings are. Do you think the difference in sound is mainly caused by different density in the maple. Just some thoughts.

GlennYorkPA · June 3, 2013

I have been puzzled by the sound qualities of different bridges on a violin I recently have glued the top and back plate back on. It is an oldish French, a bit oversized, instrument I have studied in detail many years ago in my master thesis work and I set it up at that time with not much experience (1993). Its bridge from that time was very light 1,37g and its rocking frequency was just below 3kHz. It has sounded good, according to a player I was cooperating with at that time.

The top plate and the back has come loose in some places so I opened it recently and reglued it. At the same time I wanted to try a new bridge for it more in line with later experiences. I wanted to try a heavier bridge and a higher rocking frequency. The strings also needed to come a bit higher.

In my documentation I play scales on the instruments and record the 1/3rd octave band spectra in the same mic position in the same room. This is repeated three times and the avearage spectrum and the standard deviations are plotted for each situation I want to document. The light blue curve is a test from 2008 or 2009 before I reglued the plates. It also had Dominant strings at that time.

Now I have tried three new bridges for it. The first sounded brightest and was made of a low density maple, but the bridge was still too low and I did not reach the high rocking frequency I was aiming for (3,6kHz) so I tried a new one made of a heavier and stiffer material. I reached a higher rocking frequency, just below 3,5kHz but it became on the heavy side, 2,17g. Not by far as "sparking" sound. I tried yet another one, of really stiff wood and reached 3,6kHz and a more moderate weight 2,01g, not far from the average weight of merit bridges from violinbridges.com. now with the right height for the strings, but not the sparking sound I got out of the first new bridge.

I sum up the data on the bridges below. No 1. is the original bridge from 1993, the rest are new now.

No 1: 1,37g 2,97kHz No 2: 1,84g 3,27kHz No 3: 2,17g 3,46kHz No 4: 2,01g 3,64kHz

The graph show bridge 2, with dark blue (the more "sparkling" one), bridge 3, in orange and bridge 4 in red.

I am using a different measurement device for the three new bridges than the old. I should have tested it over with the original bridge, but I am afraid of doing too much harm to the new strings by doing too much changes. The setup for the new bridges are using what I believe is a "Tellefsen set" from wathing the photo of hs del Gesu from the Bergen exhibiton. A mix of brands, Pirastro Olive G 16, Priastro Tonica D, Dominant A and a Westminster 0,26mm steel E.

The differences in the low frequencies might, by part, result from the difference in the strings used or it can also be changes to the body response after the reglueing and resetting the sundpost. However, I am very puzzled by these results. Why do the first new bridge sound so much brighter, even if its rocking frequency is quite moderate?

I wonder if makers fit many bridges on their instruments before they are satisfied with the result?

The pics show the bridges in sequence from above and down (No 1, No 3, No 4) with the bridge on the instrument being the first new one (No 2) in the table.

Experiencces with Hardanger fiddle bridges also give the impression that the qualities of the bridges are very important. The body modes are important too, but does not influence the entire higher frequency spectrum of a violin like a bridge does.

Anders,

Have you never been equally puzzled by effect of different bows on sound quality?

I was expecting you to jump in on the thread about bows but got a resounding silence from the harmonic analyzers and the bow makers.

Glenn

Oldtimerr · June 4, 2013

Anders, Sorry I cannot answer your question from the standpoint of a violinmaker but as a player for over 60 years I can state that both the bridge and the bow can make a world of a difference, even up to a positive difference of 100%. OT

Peter K-G · June 4, 2013

My experience is that the bridge can't change the basic timbre of the violin. But it is like tuning a F1 car. (I prefer Lotus) A good violin can be even better.

I have also tested different bridges and adjusting on the same violin. It seems tuning bridges takes a lifetime to learn too.

Anyhow I think it goes hand in hand with the body modes. If the violin is balanced you can install a high performance bridge. Otherwise you have to start adjusting the bridge to filter out unwanted frequencies.

One of the most unpleasant thing is when the bridge hill extends past 4 kHz (B1+ > 560 Hz). This can be adjusted to some extent by tuning the bridge, but you allways end up with filtering out something else too.

I have found that the thickness of the upper edge is the most important for the "sparkling" (focus) of the sound. You can notice the difference by testing same types of bridges with edge thickness from 2 to 1 mm.

Marty Kasprzyk · June 4, 2013

The soultion with a reference bridge seems like a good idea. I have tried the "CA glue trick" and use it on Hardanger fiddle bridges too if the wood or bridge needs some stiffening.

I wonder if it is possible to salvage that brighter bridge by raising the G string only by a small pyramid or go to the more difficult step of rising the entire bridge with something under the feet. The latter method might change the sound more than the first method. - Does not look too good, but if the sound is better, why not?

I am testing the bridges rocking frequency by fixing the feet in a vice and record the sound after flicking the bridge sideways with the fingernail.

All the new bridges have a wide waist to keep the rocking frequency high. But maybe the combination of the mass and rocking frequency has to match something in the violin to work well?

Hi Anders,

Have you tried the quick experiment of putting tiny wood wedges in the bridge slots to prevent all of the bridge rocking?

If nothing happens I don't understand why. If the violin response does change I don't know why either.

Don Noon · June 4, 2013

Have you tried the quick experiment of putting tiny wood wedges in the bridge slots to prevent all of the bridge rocking?

If nothing happens I don't understand why. If the violin response does change I don't know why either.

I had been thinking of re-cutting the bridge on my #12, to lower and lighten it for my own personal playing, and this thread got me to do it. And while I was there, I tried wedging the bridge slots.

The lightening consisted of lowering the strings ~.5mm for fiddle playing, and I also thinned out the ankles from 5.4mm to 4.0mm. To blend this smoothly with the lower edge of the bridge, the legs had to be thinned slightly. All these mods brought the bridge from 2.14g to 1.94g.

The wedges weighed only ~0.02g, almost nothing, so any tone effect would be from stiffness changes only.

The lightening appears to have had the most strengthening effect in the 2-5 kHz range, with a fairly steep cutoff above that. According to theory, one would expect all higher frequencies to be strengthened with a lighter bridge, so perhaps the thinning of the ankles and legs added flexibility and lowered the cutoff frequency. Maybe not... it's just a hypothesis that might work.

Wedging the gaps in the bridge would in theory get rid of almost all bridge flex, and therefore reverse the isolation effect. That appears to be what happened, as the frequencies above 5 kHz see the most increase.

I have looked in the past at tonal effects due to bridge flexibility, and it seemed to take a huge, radical change in stiffness before anything obvious showed up in the response spectrum. So I haven't bothered to try measuring bridge rocking frequency. I just try to take out as much wood as possible while keeping sufficient strength. I suppose if I wanted to go for a muted, mellower sound, I might leave more wood at the top and take it out of the waist and legs, but I don't recall ever wanting to go that way.

Marty Kasprzyk · June 4, 2013

I had been thinking of re-cutting the bridge on my #12, to lower and lighten it for my own personal playing, and this thread got me to do it. And while I was there, I tried wedging the bridge slots.

The lightening consisted of lowering the strings ~.5mm for fiddle playing, and I also thinned out the ankles from 5.4mm to 4.0mm. To blend this smoothly with the lower edge of the bridge, the legs had to be thinned slightly. All these mods brought the bridge from 2.14g to 1.94g.

The wedges weighed only ~0.02g, almost nothing, so any tone effect would be from stiffness changes only.

12 bridge mods.jpg

The lightening appears to have had the most strengthening effect in the 2-5 kHz range, with a fairly steep cutoff above that. According to theory, one would expect all higher frequencies to be strengthened with a lighter bridge, so perhaps the thinning of the ankles and legs added flexibility and lowered the cutoff frequency. Maybe not... it's just a hypothesis that might work.

Wedging the gaps in the bridge would in theory get rid of almost all bridge flex, and therefore reverse the isolation effect. That appears to be what happened, as the frequencies above 5 kHz see the most increase.

I have looked in the past at tonal effects due to bridge flexibility, and it seemed to take a huge, radical change in stiffness before anything obvious showed up in the response spectrum. So I haven't bothered to try measuring bridge rocking frequency. I just try to take out as much wood as possible while keeping sufficient strength. I suppose if I wanted to go for a muted, mellower sound, I might leave more wood at the top and take it out of the waist and legs, but I don't recall ever wanting to go that way.

Hi Don,

I had always pretty much accepted Dunnwald's idea that the real high frequency range created "harshness" and that this was bad and that the amplitudes there should be minimized.

However maybe a little of the high frequency stuff (above 5000Hz in Ander's above graph) gives the "sparkle" that Anders likes. Perhaps there is an optimum amount that differs for various individuals' tastes.

I can't remember: Did you ever try your heat treatment methods on bridges?

Anders Buen · June 4, 2013

One of the puzzling things was that the impact hammer spectra did not show so clear differences while the differences showed up clearer from in the played spectra..

The light blue curve is recorded using an omni mic, the rest is a Zoom H4n wav recorder directed 45degrees upwards and away from the player. The spectra are averages of the two mic signals. I do not know if slight details in the mic dorection may play a role or not. Should do a run with an omni mic and one of our companys Norsonic N 140 units and see how it compares to the wav recorder spectra. The Zoom have cardiodids angled in a 90 degrees cross.

I think the bridge mass play a role, but i have no idea why the first new bridge give much more in the "bridge hill" region. But need to check up the measurement technique first. Should do a run with the original bridge too. If the better high frequency response persists, then I wonder if there are something matching between the bridge and the ireland area of the fiddle.

Bridges are, no doubt, very important. And I think they need to have their feet fitted very precisely. Adding masses to their top could be a test too.

Don Noon · June 5, 2013

Hi Don,

I had always pretty much accepted Dunnwald's idea that the real high frequency range created "harshness" and that this was bad and that the amplitudes there should be minimized.

However maybe a little of the high frequency stuff (above 5000Hz in Ander's above graph) gives the "sparkle" that Anders likes. Perhaps there is an optimum amount that differs for various individuals' tastes.

Here is a clip of Ehnes on the Lord Wilton... first as recorded, then with a 12dB dropoff starting at 5kHz and up, then a 12dB amplification of that range. It is interesting to note that the spectrum of the full recording, unequalized, shows a relatively smooth attenuation of ~18dB/octave above 4kHz or so, and these equalizations put a kink in that smooth dropoff. To me, the attenuated version sounds pleasant but dull, and the amplified one sounds way too edgy.

Track 30 equalized.mp3

So this tells me that balance is the important thing... having just enough of the high frequencies sounds best.

I can't remember: Did you ever try your heat treatment methods on bridges?

Yep. However, it is primarily just to get a nice color. There should be some sonic difference, at least due to the slightly reduced density. I haven't tried to measure any acoustic differences, though.

Peter K-G · June 5, 2013

Here is a clip of Ehnes on the Lord Wilton... first as recorded, then with a 12dB dropoff starting at 5kHz and up, then a 12dB amplification of that range. It is interesting to note that the spectrum of the full recording, unequalized, shows a relatively smooth attenuation of ~18dB/octave above 4kHz or so, and these equalizations put a kink in that smooth dropoff. To me, the attenuated version sounds pleasant but dull, and the amplified one sounds way too edgy.Track 30 equalized.mp3So this tells me that balance is the important thing... having just enough of the high frequencies sounds best. Yep. However, it is primarily just to get a nice color. There should be some sonic difference, at least due to the slightly reduced density. I haven't tried to measure any acoustic differences, though.

This is the reasons why I also keep going on with my research! It sounds delicate. Even if the amplified piece sounds a little edgy it is not harsh. Here lies "The secret" because the cremonse violins sounds all very different but they sparkle.

Violadamore · June 5, 2013

Referencing a recent discussion on the bass thread, does anyone have any numbers for what effects ammonia (& etc.) treatment has on bridges? I carve some of my own from local woods, and am rather curious about this.

Michael_Molnar · June 5, 2013

I guess what Don says is that there is an optimum response for a bridge. Just increasing the high frequencies does not optimize the sound.

Marty Kasprzyk · June 5, 2013

Here is a clip of Ehnes on the Lord Wilton... first as recorded, then with a 12dB dropoff starting at 5kHz and up, then a 12dB amplification of that range. It is interesting to note that the spectrum of the full recording, unequalized, shows a relatively smooth attenuation of ~18dB/octave above 4kHz or so, and these equalizations put a kink in that smooth dropoff. To me, the attenuated version sounds pleasant but dull, and the amplified one sounds way too edgy.

Track 30 equalized.mp3

So this tells me that balance is the important thing... having just enough of the high frequencies sounds best.

Yep. However, it is primarily just to get a nice color. There should be some sonic difference, at least due to the slightly reduced density. I haven't tried to measure any acoustic differences, though.

Hi Don,

Thanks for the recordings.

The sound does seem to go from "dark" to "bright" as the centroid of the spectrum moves to higher frequencies.

But Anders used the term "sparkle" and I might be making a mistake in assuming this is the same as "bright". Sparkle could imply a rapidly changing color and/or intensity whereas bright describes a steady-state condition.

The opposite of "sparkle" might be "dull".

If rapid change is important then maybe the damping of the system plays a role which made me think your high Q treatment might be beneficial for bridges. This probably wouldn't be noticed in a FFT tap test spectrum.

Does anybody drop bridges on a hard surface and measure their ring time?

Don Noon · June 5, 2013

One of the puzzling things was that the impact hammer spectra did not show so clear differences while the differences showed up clearer from in the played spectra..

One thing that struck me in the response plots was the ~4dB difference in the "sparkling" bridge over a very wide range of frequencies. That is a huge difference in amplitude and power, and I can not presently think of a physical mechanism of the bridge that would account for it.

One other thing... the "rocking frequency" does not show any obvious corresponding feature in the response curve. This is something I also observed in my testing, trying to verify Woodhouse's predictions of envelope curves of bridge mass and stiffness. I think it is somewhat similar to tap tones, in that the edge restraints are varying. The island and body of the violin presents a far different interface than a heavy vise, and the interaction between them is not simple.

I guess what Don says is that there is an optimum response for a bridge. Just increasing the high frequencies does not optimize the sound.

As usual, I think it's much more complicated.

"High frequencies" are not all created equal; 2-4kHz will have a vastly different tonal influence than 5-10kHz. Try Audacity equalization on a solo violin and hear it for yourself.

I'm not even sure what "response for a bridge" means yet. I does seem clear to me that reducing the mass at the strings (reduced height, thinner, lower density) will in general increase the power at higher frequencies, and reduced stiffness (waist, legs, ankles, softer wood) might lower the frequency at which the response tends to drop off.

And "optimum" would depend on the character of the instrument you're starting with, and the preferences of the player.

Anders Buen · June 5, 2013

I have done a new test with the assumed to be "brighter bridge" (No 2) and the original bridge (No 1) using the Zoom recorder in the same position. The brown curve is a new test of the same bridge as the dark blue run, and the violet curve is the original bridge using the same test equipment and strings as the other tests.

I am disappointed to see that the brown and dark blue curve are not on top of each other. But there are something parallell in the curves. And the original bridge seem to give more output in the 2,5kHz range than bridge no 2, the assumed "brighter bridge".

Clearly I do not seem to be able to drive the instrument with the same force between the takes, as can be seen in the lower frequencies. It can also be the fact that I had to turn down the amplification on the recorder in one of the tests, maybe the orange run, due to overload at one of the notes. So there might have been some clipping in the dark blue test.

Testing the bridges in short sequence with the recorder in the same pos might be a possible variation reducing method. Right now the melacholy will dominate for a while..

I think the relatie wide "bridge body hill" is something Woodhouse does predicts. The higher the losses at the feet, the wider the hill. It is way narrower with the bridge feet clamped in the vice.

Peter K-G · June 6, 2013

I know you don't like my "nonsense" but...

Anders, something happened after correction/reglueing that has nothing to do with the bridge. I have experienced this too

Sure different bridges has effect on the higher frequencies. The "spakling" Cremonese sound however happens, even if there is a lot of dBs after 4 kHz. The key to the Cremonese sound is the curve shape after 4 kHz!

This is one of my many discoveries over the last year in my research.

You can also see this from Don's recording and equalizing (128 FFT size). They all sparkle! (original is best)

Don Noon · June 6, 2013

The key to the Cremonese sound is the curve shape after 4 kHz!

I don't think so. According to my research, it is:

-very strong, full response in the range of 1.5 - 4.5 kHz

-not very strong in the range of .8 - 1.3 kHz

Above 4.5 kHz I would characterize more as hiss and sizzle, rather than the more musical "sparkle". There may be some more of that in the old fiddles, but if so, I'd consider that more of an artifact rather than a key.

The lower frequency range (around 1 kHz) can add power, but at the expense of a more muddy or crude sound.

Michael_Molnar · June 6, 2013

Don,

I am a visual person. Can you show this in a graph with examples?

Mike

Peter K-G · June 6, 2013

I don't think so. According to my research, it is:

-very strong, full response in the range of 1.5 - 4.5 kHz

-not very strong in the range of .8 - 1.3 kHz

Above 4.5 kHz I would characterize more as hiss and sizzle, rather than the more musical "sparkle". There may be some more of that in the old fiddles, but if so, I'd consider that more of an artifact rather than a key.

The lower frequency range (around 1 kHz) can add power, but at the expense of a more muddy or crude sound.

I Agree with what you say, but there is an indirect connection from the body modes to the curve after 4-4,5 kHz. I have been searching for patterns between the body modes deltas and dB:s and beginning to see how it works. My focus is on the body modes and to figure out more in details how they build up the Cremonese sound that "sparkle". And as you say above certain Hz it's more sizzle, which is also a characteristic of the Cremonese. The "sparkle" comes from the lower Hz combinations but it's in harmony with the bridge hill region.

Plowden Meditation:

Titian Meditation:

Willemotte Meditation (Not so much curved and you can here it also, least Cremonese sound to my ears):

Il Cannone Paganini Concerto 1 Solo:

Joe Swenson · June 9, 2013

From vocal recordings, the "sparkle" your ears detect results from boosting the frequency content at around 8 kHz. Might be interesting to see how the spectra compare in this region.

After playing my first violin in the white and then varnishing, that range seemed to be where the varnish had the most damping effect. The white violin was very crisp sounding and the varnish really mellowed it out. So it easy to understand how the Cremonese varnish could be the secret to that brilliance.

Sign In

The effect of different bridges on tone

Recommended Posts

Anders Buen

Christopher Jacoby

Anders Buen

Berl Mendenhall

GlennYorkPA

Oldtimerr

Peter K-G

Marty Kasprzyk

Don Noon

Marty Kasprzyk

Anders Buen

Don Noon

Peter K-G

Violadamore

Michael_Molnar

Marty Kasprzyk

Don Noon

Anders Buen

Peter K-G

Don Noon

Michael_Molnar

Peter K-G

Joe Swenson

Create an account or sign in to comment

Create an account

Sign in

Recently Browsing 0 members

Browse

Activity

Store

Support