Play-in Testing

[Don Noon]

Rather than plug up the "what's on your bench" thread, I think there's enough material for a separate one.

 

Here's a summary:

Start with a high-powered compression horn driver, machine a cone adapter to get from the voice coil to the "stinger"

Attach with high-temperature silicone

Here it is with a toothpick "stinger", glued to a setscrew and threaded into the cone, for replaceability reasons

And the setup, with the stinger glued to a tiny aluminum piece which is held to the bridge by string tension

And the 40-pound box to contain the deafening noise

 

This is all driven by a mini MP3 player and a power amplifier, so I can put together any sort of soundfile to pump through the violin.

 

I selected my Violin#5 for testing, as it has had about 2 years to settle in and dry out, but has seen very little playing in the last year or so, probably no more than an hour, and almost zero in the last 6 months.  It also has unprocessed wood, which may or may not matter, and has a relatively weak high end that I want to try to bring up.

 

For baseline data, I took an impact spectrum (of course), but also have a set of driver files that includes white noise, sine sweep, sawtooth sweep, sawtooth semitone scales, and sine pulses.  I may not need all that stuff, but it's easy data to take... just push the button.

 

Surprisingly, I got some interesting data before I even started the play-in phase of the test.  Yesterday (day -1) I naturally played around with the system, seeing what it could do, and then took the baseline dataset.

Today (day 0) I took the dataset again, and then took another dataset to check for repeatability.  Everything was exactly the same for all 3 datasets, nothing touched or moved.

 

Here is the comparison of the white noise response.  Yesterday's data is measurably different than today's, and today's 2 tests are nearly identical

 

This is the difference between them... yesterday vs today, and  today vs. 1 minute later:

 

Perhaps there is some difference in the driver's ferrofluid coolant viscosity between the tests on different days, but I'd expect the difference to be more uniform.  The variable nature of the difference I interpret as something going on in the instrument structure, mostly.  I'm very happy to see that the spectral difference is extremely close on repeated tests, especially considering how variable the response is... every little peak and dip stays right where it is, to an amazing degree.  So I am quite confident I'll be able to measure anything that changes.

[Anders Buen]

Did you monitor the humidity or moisture content of that fiddle the different days?

[Don Noon]

I didn't record the RH yesterday, but I'm doing that from now on.  I doubt it was much different then... some rain yesterday, and today.

 

A few more details on the present test.  This is the last impact spectrum before getting started, showing the strong low end and very strong transition hill (~950 - 1300 Hz for this fiddle). 

The higher frequencies are a bit choppy, and overall relatively weak.  That's the way it sounds in playing, too... big bottom end, midrange power, but a bit uneven and muddy.

 

So, making the assumption that played frequencies will get stronger over time, I tailored a noise spectrum to exercise the high end, and also the gap between the B1+ resonance and the transition hill.

 

So, that's the energy I'm putting into the fiddle at the moment, at deafening levels (but only a few watts RMS, I think).  We'll see if there are any changes that correspond to the input spectrum.   There are other ideas I could try out, but this was my best first guess.

[~ Ben Conover]

The cure to all things electronic :

[Don Noon]

The cure to all things electronic :

 

Sure, a cure if you have an electronic illness. 

 

However, it doesn't do anything to answer the questions.

[Don Noon]

After a few hours in the shop with this thing, even with the box quieting it down, I decided I couldn't stand much more of the noise.  So I moved it to a spare bedroom.  I took a calibration reading before and after the move, and there are some changes.  Almost certainly the different room acoustics made a difference in the readings after the move.  The reading before the move showed increased strength in the 1300-1600 Hz range, but random changes elsewhere.  Hmmm.  I also noticed that the driver had warmed up to 87F from the ambient 66F.  That would affect coil resistance and ferrofluid viscosity, for more confusion factors.

 

The one thing I know is that the readings are exceptionally repeatable if everything is kept exactly the same, untouched, without much time elapsed.  I'll have to see how sensitive the setup is to little things, like the slight jiggling of puting the heavy box on and off, and repositioning the microphone.  I do have to move the microphone to put the box on, but I'm careful to put it back within a  mm or two of where it was.  There's also temperature and humidity to look into.

[jezzupe]

Sure, a cure if you have an electronic illness. 

 

However, it doesn't do anything to answer the questions.

Don on top of everything else we need you to prove for us, we now need you to raise Heifetz from the dead, piece of cake

[~ Ben Conover]

Before the electronic music came from America, we had peole like Delia Derbyshire doing some very nice stuff, such as Dr Who.
Not to mention the wax recordings by Heifetz.

[~ Ben Conover]

This link explains :
http://www.youtube.com/watch?v=XAMa9_B5MEY

[David Burgess]

Don, what about using an impact hammer to test for changes, to take possible variations in the driver out of the picture?

 

I use a simple gravity powered one which is quite repeatable. Set it a certain distance from a marked spot on the bridge when it's hanging free, pull it back agaist a stop, and let 'er rip. Need to catch it after the first bounce so it only strikes once, but that's not hard.

[saintjohnbarleycorn]

interesting stuff don, thanks for taking the time to post your info.

[Berl Mendenhall]

Great work Don, very interesting.  Thanks for sharing.  

[Don Noon]

Don, what about using an impact hammer to test for changes, to take possible variations in the driver out of the picture?

 

I really would prefer the in-process tests, to look for changes without disturbing the setup.  At the moment, I think the driver variability may not be an issue, and a few fairly simple tests should confirm that (today's agenda).  I do have my "before" impact test, so I can double-check the in-process changes later, when the test is over.

[Violadamore]

Thanks, Don, extremely interesting

 

I'm wondering if work in this area of "electronic drivers" might lead to a standard test protocol to compare violins?  Talk about an idea to start a riot with...........  

[Don Noon]

Thanks, Don, extremely interesting

 

I'm wondering if work in this area of "electronic drivers" might lead to a standard test protocol to compare violins?  Talk about an idea to start a riot with...........  

Nope.  The calibrated impact hammer is much better than this at characterizing instruments, and even that isn't going to replace good players.

 

However, for the nitpicky, there may be some use for electronic drivers to look into the finer points of transient behavior.  Maybe.

[jezzupe]

My concern is the heating up of the coil while inside the box, it seems that when all contained in the box that the coil could act like a heater, and after time could change the interior MC and or give variable readings based on being tested "cold" and then ":hot"?

[jezzupe]

By the way I predict that you will find noticable difference, and furthermore that these differences, if checked, are temporary and or if a "played in fiddle" is allowed to sit for a long period of time it will return to a "un warmed up state" and that yes it could be "re warmed"

[David Beard]

Nope.  The calibrated impact hammer is much better than this at characterizing instruments, and even that isn't going to replace good players.

 

However, for the nitpicky, there may be some use for electronic drivers to look into the finer points of transient behavior.  Maybe.

 

 

Your dirver arrangement might allow exploration of some of the less static characteristics of violin tone.  

 

I'm guessing that by transients you mostly meaning the beginning of a single note.  However, there are also interesting phenomona in the change from one note to another. 

 

Better violins often seem to give an impression of 'hanging on' to the prior note, or at least making this affect available to the player.  It feels like a fatness of sound that carries from note to note.  It feels like one can develop a resonance in the first note which carries on into the new note.

 

Another playing phenoma which seems to distinguish better instruments is a sense that the instrument favors harmonicity, that the instrument almost pulls toward harmonous playing in the way resonances develop in the tone and preferentially support some combinations over others.

 

These are two characteristics which IMO help distinguish better insturments and set them appart from lessor creations.   I didn't describe these very well, but hopefully others will recognise the phenomena.   

 

Your driver arrangement makes it possile to exploring these and other time related aspects of violin behavior.  Other than col legno and pizzacato, it isn't normal for the violin to be driven from a strike of energy.   Violins are normally sounded by a sustained and pitched driving vibration. While such sustained and driven stimulation is likely too complex to be the favorite choice in some experiments, it's also much closer to normal playing.  

[Don Noon]

Better violins often seem to give an impression of 'hanging on' to the prior note, or at least making this affect available to the player.   

Sounds to me like it could be an effect of lower damping.  However, I've also heard it said that good instruments stop sounding immediately when you stop bowing... the opposite effect.

 

 

Now for some more (yawn) test data.

I had established that repeatability was exceptionally good for tests done consecutively, without disturbing the setup in any way.  I wanted to know if the microphone could be repositioned without causing different readings, and if I could put on and take off the heavy sound insulating box accurately.  The results:

Again, the untouched tests were extremely repeatable.

 

Microphone placement doesn't cause any problem (there are marks on the baseplate where the mic goes).

 

However, the setup is very sensitive to physical assault, as seen by the difference when I accidentally bumped it with the box.  The disturbance to the measurements appeared to be mostly temporary, but I'll have to be very careful if I want decent readings.

 

If I'm very careful not to bump anything, the disturbance isn't too bad (last test).  I think that with practice, I can handle the box carefully enough.  Until my back gives out.

[David Beard]

Sounds to me like it could be an effect of lower damping.  However, I've also heard it said that good instruments stop sounding immediately when you stop bowing... the opposite effect.

 

Yes.  A good fiddle doesn't railroad the player into one kind of voice or color.  Instead, it gives the sense that many colors and effects are readily accessible.    

 

Even though your driver opens up many possibilities, it still might not be sufficient to examine some effects that are readily available in live playing.  

 

There isn't a simple input output relation between what the player feeds in and the results the instrument produces.  Rather, the player is listening to achieve a particuar efffect.  Many physical input factors can be dynamically adjusted to open up and sustain a desired effect.  In a good player's hands, a bowed string doesn't just allow dynamic adjustment of pitch and volume content of the input, but also adjustment of things like impedance and dampening (manipulating the bowhand weight and contact point i.e). 

 

Lab replication some of the effects important to a player's experience of instruments might require some sort of computerized system to listen to the sound output and dynmically adjust the driver to achieve and sustain targeted effects.  

[David Burgess]

Deleted, probably irrelevant.

[Don Noon]

Datapoints at 2 hours and 23 hours look to me like something could be changing, but not well correlated to the energy spectrum being put in.  It's only around 1 or 2 dB... not game-changing, but should be audible.

 

I'll gather a few more days of data to get a better picture, and to make sure it isn't related to slight weather changes or some other accidental effect.

[Anders Buen]

You could use a control instrument from the same environment. Would it be possible to compare the temperatures on both? The control could go in the same case, but only as a passive element.

Sound proofing of a box like that takes covering the inside with a porous absorber of sufficient thickness. 40-50mm mineral wool or rubber foam with open cells would do the job. Cover all surfaces.

[Wm. Johnston]

It would be interesting to know how much power is being dissipated by the violin, since energy loss to radiation is pretty low in violins shouldn't the main energy loss be through heating the wood? It's probably a pretty small temperature change for a violin in normal playing though. Just thinking and wondering out loud, I'll go away now.

[David Burgess]

It would be interesting to know how much power is being dissipated by the violin, since energy loss to radiation is pretty low in violins shouldn't the main energy loss be through heating the wood? It's probably a pretty small temperature change for a violin in normal playing though. Just thinking and wondering out loud, I'll go away now.

Probalby need to weigh the violin before and after to make sure that the moisture content remained the same. Keeping the humidity the same in the environment might not cut it if the wood's heating a bit. I'd expect that the air inside the box is warmer (therefor dryer) than ambient too