Jump to content
Maestronet Forums

Something maybe to consider


Johnmasters

Recommended Posts

I am well aware of phase shift in an amplifier. I was more interested in seeing what this fellow is doing by implementing a vactrol into the feedback loop. So far I fail to see how you would utilize this to glean any useful spectra information.

The phase shift angle is a function of frequency, capacitance and resistance which is adjustable. Capacitance and inductance parameters create phase shift due to the reactance of these components, a resistor can only alter the Q of either of these two.

The resistor/capacitor has a time constant. I am not talking about driving a resonance. Maybe Q is part of filter discussion, as in band-pass filters. But this is a constant amplitude filter for phase only. I know about reactance and complex numbers.

I am going to make one and put it between my preamp and amp. Just to listen to solo violin.

Phase shift is meaningless unless referenced to another waveform at a specific point in time. I am talking about the components of a tone. That is the reference.. Especially at the open and close of a note, transients. You could regard the beginning and end of each sawtooth from the bowed string as being the start and end of a cycle. If you build an broadband audio amplifier stage that shifts the phase equally across the spectrum, I realize this, you just reset your clock. The phase shifter depends on frequency of the fourier components of a single tone. you won't hear any difference in the resulting audio unless you compare it to The original un-altered signal; All you have accomplished is delaying the signal. The entire signal shifted uniformly, of course. I am not talking about that. I'm not trying tdiscredit you, (No offence taken) but what do you think you may learn from this?

I will not know a thing until I pass a violin sound through it and put it through my hifi. Perhaps it will imitate a lower quality speaker. Less distinct or fatiguing to listen to. I don't know. It will likely be a small effect. Something "not satisfying."

Link to comment
Share on other sites

  • Replies 58
  • Created
  • Last Reply

Top Posters In This Topic

I Have the Klipsch Forte speakers which have horn tweeter and midrange, 12" cone woofer plus 12" passive radiator for bass. My transient perfects are a three way system I designed using Scan-Speak drivers, 1" dome tweeter, 5" cone midrange, 10" cone woofer. Transient Perfect designs with cone drivers and passive crossovers are a particular problem because the drivers must be offset relative to each other I think I recall the Dahlquist speakers did this. Long time ago. in order to get the sound arriving at one point (10' away in my design) at the same time for all three drivers. That makes cabinet design difficult, particularly if wife approval is required. Ha Ha! Yes I have read about the SAF variable; "spousal approval factor". Note there is sound delay due to the offsets plus due to the roll-offs of the crossover components. So at least here you are only concerned with phase shift at the crossover points. Do you think that within the signal there are frequency-dependent phase shifts in one of the ranges?

The real attraction of horns is that they produce very high spl for the input voltage and dynamic range is incredible. Normal drivers (cones, dome tweeters, etc.) suffer in transients due to their lack of sensitivity; A sudden surge in power heats the voice coil, resistance goes up and the coil heats, increasing the resistance and the thing just does not reproduce the change in spl associated with a transient. This is worst for the tweeter, thus taking the leading edge off of transient signals, so you have a non-linearity to deal with. Horns do transient spl changes spectacularly well. However, the great lengths of horns make it very difficult to use spatial offsets to properly create a transient perfect combined output. Your only reasonable approach is to use a digital crossover and separate amps for each driver, a new level of PITA. Further, tubeophiles object to having opamps between them and the source of output signal. Finally, electrostatic earphones give a reference to HiFi nuts as to what music actually sounds like as only one transducer is needed for the full range of music frequencies. There are no delayed signals to try to properly sum. You pay your money and you take your choice as to your favorite forms of distortion.

This what a friend told me when I asked him about comparing transistor vs tube amplifiers.

Link to comment
Share on other sites

From a scientific / physics point of view, how the heck does one analyse and quantify the chemical change of old wood, that naturally occurs with time ?

This might be of some interest concerning the old sound we always associate woth old instruments.

Chemistry may explain physical changes. Try to measure physical differences and worry about the chemical cause separately. Old sound and old wood may be related only through the use of the common word "old". To me, it is like the intuitive mistake of thinking that a violin "under stress" sounds bad the way people feel when they are "under stress."

Link to comment
Share on other sites

So at least here you are only concerned with phase shift at the crossover points. Do you think that within the signal there are frequency-dependent phase shifts in one of the ranges?

Not just at the crossover points, things have to sum properly over the entire audible frequency range. Not difficult when you are well away from the mid range and effectively only one driver is really being summed. If you mean phase shifts of the signal on the cd, god only knows what went on in the mastering and mixing of the channels to make the source signal for the playback system.

Link to comment
Share on other sites

So at least here you are only concerned with phase shift at the crossover points. Do you think that within the signal there are frequency-dependent phase shifts in one of the ranges?

Not just at the crossover points, things have to sum properly over the entire audible frequency range. Not difficult when you are well away from the mid range and effectively only one driver is really being summed. If you mean phase shifts of the signal on the cd, god only knows what went on in the mastering and mixing of the channels to make the source signal for the playback system.

I know that some purists think that the old platters are "The Vinyl Solution" to sound. Does a digital-analogue converter have properties of phase distortion? I don't know a thing about it. I do not myself detect such small differences. But also, I probably cannot hear much difference in violins after a certain level of quality.

I am not far enough from my Maggies as they are in my shop. I would love to have them in walls (open both sides) on either side of an arch in a wall, upstairs. I DO have a small, 36" pair of old Maggies in a wall upstairs, and they sound very good, with the opposite room adding some breathing room. My shop Maggies are the 6-foot MG 2A. From 1978.

Link to comment
Share on other sites

I know that some purists think that the old platters are "The Vinyl Solution" to sound. Does a digital-analogue converter have properties of phase distortion? I don't know a thing about it. I do not myself detect such small differences. But also, I probably cannot hear much difference in violins after a certain level of quality.

I am not far enough from my Maggies as they are in my shop. I would love to have them in walls (open both sides) on either side of an arch in a wall, upstairs. I DO have a small, 36" pair of old Maggies in a wall upstairs, and they sound very good, with the opposite room adding some breathing room. My shop Maggies are the 6-foot MG 2A. From 1978.

There are d/a converters and then there are d/a converters. Two types are most popular. What I prefer is "zero oversampling" as they seem best with dynamics and an impulse in produces a clean impulse out. On the other hand, there are numerous oversampling schemes which attempt to interpolate between the actual data points on the cd. Some think they give a smoother sound, but I think they introduce some digital hash. This is most readily seen as a pre and post ringing when an impulse function is input. Such ringing may be quite high in frequency (i.e. above 22kHz) and not heard by old guys like me. I think it still causes the edginess that makes the vinyl crowd to dislike cd's. I personally have a large collection of vinyl that I never play, just too much trouble to keep it clean so that you aren't annoyed by clicks and pops. Again, you pays your money..............and if you are addidted to vinyl you get your vinyl and equipment from a pusher, not a dealer :)

Maggies lose a bit of their ambience if you take away their back wall. They are dipoles and depend on that back wall to add the reflections that make the music seem as if it is "live"

Link to comment
Share on other sites

There are d/a converters and then there are d/a converters. Two types are most popular. What I prefer is "zero oversampling" as they seem best with dynamics and an impulse in produces a clean impulse out. On the other hand, there are numerous oversampling schemes which attempt to interpolate between the actual data points on the cd. Some think they give a smoother sound, but I think they introduce some digital hash. This is most readily seen as a pre and post ringing when an impulse function is input. Such ringing may be quite high in frequency (i.e. above 22kHz) and not heard by old guys like me. I think it still causes the edginess that makes the vinyl crowd to dislike cd's. I personally have a large collection of vinyl that I never play, just too much trouble to keep it clean so that you aren't annoyed by clicks and pops. Again, you pays your money..............and if you are addidted to vinyl you get your vinyl and equipment from a pusher, not a dealer :)

Maggies lose a bit of their ambience if you take away their back wall. They are dipoles and depend on that back wall to add the reflections that make the music seem as if it is "live"

I have nice reflection back from the adjacent room through the archway upstairs. It is not so close as 3 feet to a wall, however. Those little maggies sound good. I paid $300 for the pair but had to put in new tweeter windings. Time consuming but not difficult. The factory sells aluminum wire and a very nice emulsion glue that is very easy to use and shrinks up neatly.

I also have a lot of vinyls stored, I would like to record them over to CD... many were bought shortly before I bought a CD and moved my equiptment (and displaced my record player). I like a lot of old recordings, old singers etc.

I can see the hash issue, I don't think I hear it. But over FM radio, I often hear that some recordings are exceptional and some are not. Of course, I assume they are all CD. I don't know if it is the recording engineer or whether they are just different at the digital level.

Can you see any kind of possibility that a violin might have some of the iimpulse ringing distortion? I mean accumulation and superposition over time? Whatever gives presence and clairity to a speaker ought to have possible analogies in a good violin ( ?? ). That is why I mention speakers in a violin forum. It would be nice to find new things the lab folks can measure. Power spectra just CAN'T be all of it, can it ?

Link to comment
Share on other sites

Can you see any kind of possibility that a violin might have some of the iimpulse ringing distortion? I mean accumulation and superposition over time? Whatever gives presence and clairity to a speaker ought to have possible analogies in a good violin ( ?? ). That is why I mention speakers in a violin forum. It would be nice to find new things the lab folks can measure. Power spectra just CAN'T be all of it, can it ?

I can't imagine that, John. The da conversion ringing distortion is non-causal. I can't imagine that in a violin. I can imagine all sorts of wierd energy transfers from mode to mode, but causality just has to be there for a violin. Post impulse ringing I think is part of the non-causal ringing. Sorry, everybody, sometimes you are overcome by the passion of the moment, even obsessively so! :) :) :)

Link to comment
Share on other sites

I can't imagine that, John. The da conversion ringing distortion is non-causal. I can't imagine that in a violin. I can imagine all sorts of wierd energy transfers from mode to mode, but causality just has to be there for a violin. Post impulse ringing I think is part of the non-causal ringing. Sorry, everybody, sometimes you are overcome by the passion of the moment, even obsessively so! :) :) :)

By non-causal, do you mean it is a chaotic event?

Link to comment
Share on other sites

I suspect "phase shift" may mean different things to different people when discussing whole-instrument overtones.

For Violin acoustics, let's state some of the obvious:

(1) Each string's harmonic overtones are a function of its fundamental frequency;

(2) The bowed-string waveform is the realtime "input" to the whole instrument;

(3) The whole-instrument "output" [frequency response] is a function of its "input";

(4) IF there's measured phase shift between string "input" & whole-instrument "output" [frequencies], then there IS an acoustic mismatch.

For instrument design, you could create phase shift(s) anywhere you want.

Jim

There is a phase relation plot to go with the 'impact hammer' sound spectra that show how the modes do relate when it comes to the phase. Eg between the B1- and the B1+ mode there is a cancellation taking place so that there is a deep dip there. In the region between them there is a subtration of the vibrations. Between the A0 and the B1- the phases of the modes are so that the response of the A0 and the B1- add.

This effect will for all practical purposes be the same in any violin. But for the understanding of how the violin modes work, this is important.

In the former referred thread strado mentioned phase relations in mode shapes itself (not his words, but my understanding of it). Symmetric modes, except the first, will radiate less efficiently due to phase cancellations between part of the modes moving outwards while other parts move inwards. Asymmetric modes do also cencel sound, but to a lesser degree. The assymmetry from the bass bar and sound post helps to increase the sound radiation from a violin.

Link to comment
Share on other sites

There is a phase relation plot to go with the 'impact hammer' sound spectra that show how the modes do relate when it comes to the phase. ... The assymmetry from the bass bar and sound post helps to increase the sound radiation from a violin.

Thanks Anders.

I've got a hunch a more straightforward method for sound spectra data analysis is on the horizon - one which relates modes to

String frequencies, harmonics, etc., so phase shifts [from optimum] may be readily understood.

Maybe we'll soon too question why the Cremonese chose as much asymmetry as they did.

Jim

Link to comment
Share on other sites

Perhaps my memory is faulty, but IIRC, "Gibbs Phenomena" is the name given to the wiggles on a Fourier transform that describes the little wiggles at the beginning and end of a square function. Correct me if I'm wrong.

I know what you mean. Isn't that what some call "ringing." An amplifier cannot turn on and off instantaneously. So the ringing is from very high frequencies which are not allowed to add up to make the sharp corners. You said it was not "causal" and that threw me off. I think it is causal and has to do with inadequate put-through of all the frequencies along with their proper phases. I am sure that phase shifting alone would cause a problem with a square wave even if the magnitude of the frequencies made it through.

If I put on my physics hat, it is like the Heisenberg Uncertainty principal... The Schroedinger equation has a wave solution and localizing a particle in real space makes the fourier transform large in the frequency domain. That would be momentum to Erwin.

There is a phase relation plot to go with the 'impact hammer' sound spectra that show how the modes do relate when it comes to the phase. Eg between the B1- and the B1+ mode there is a cancellation taking place so that there is a deep dip there. In the region between them there is a subtration of the vibrations. Between the A0 and the B1- the phases of the modes are so that the response of the A0 and the B1- add.

This effect will for all practical purposes be the same in any violin. But for the understanding of how the violin modes work, this is important.

In the former referred thread strado mentioned phase relations in mode shapes itself (not his words, but my understanding of it). Symmetric modes, except the first, will radiate less efficiently due to phase cancellations between part of the modes moving outwards while other parts move inwards. Asymmetric modes do also cencel sound, but to a lesser degree. The assymmetry from the bass bar and sound post helps to increase the sound radiation from a violin.

I left out spacial radiation in order to think of the intrinsic properties of a violin and its vibration shapes, normal modes and the relations of the phases that drive them. I understand about the radiation wanting assymetry. The idea of an antenna treated with a multipole expansion is the easy way I think of it. I should be careful, it has been 45 years since I studied this !

Dr. Woodhouse sent me a letter just today, and he takes your position, so you are in good stead. But if you say that there is something going on with those modes when tapped with a hammer, then that is closer to what I was thinking. It would always be there no matter what the radiation field was. I had in mind the sawtooth from the bow being like a lot of impacts. (Even though the sawtooth is rounded off.)

Then there is the question of whether the absolute frequencies of the B modes are the important thing, or whether it is their relative spacing in relation to their position on the frequency graph. I suspect the latter.. others take the first view.

Any time you say that something is the same in all violins, I agree. All violins can be identified as violins. Too bad the players don't agree to simply accept their instruments. It is the tiny differences that make for all the discussion.

You're not wrong!

I was not trying to find him right or wrong. Just that perhaps it is not acausal. Doesn't it come straight out of the fourier transform of the square wave and a consideration of how the fourier components are passed imperfectly through the amplifier?

Link to comment
Share on other sites

There is a phase relation plot to go with the 'impact hammer' sound spectra that show how the modes do relate when it comes to the phase. Eg between the B1- and the B1+ mode there is a cancellation taking place so that there is a deep dip there. In the region between them there is a subtration of the vibrations. Between the A0 and the B1- the phases of the modes are so that the response of the A0 and the B1- add.

Anders,

Is this caused by some relation to the A-naught mode? Would you have a single large peak if the violin were struck in a vacuum? I asked before if the two B modes were a single peak with a notch and somebody said they were two separate modes. Now it seems that they are a double mode. Such as in a coupled oscillator. Maybe the distinction is meaningless. But I think I saw this in FEA.

My FEA models show that there are two resonances that look alike. One has the upper bouts bulging in and out, and the other has the bottom bouts doing it. If the violin had the upper and lower parts the same size, maybe one would have a huge single resonance. If one could gradually shrink the upper part, then that resonance would gradually turn into the double peak.

These two modes from FEA were two of the lowest modes, and I think they likely WERE the B modes.

Link to comment
Share on other sites

What you describe doesn't sound like the B1 modes to me. The B1+ and B1- look the same except that the mode shapes are exchanged between the top and back plates, not the upper and lower bouts.

I have seen Bissingers model. It has been some time since I saw my model, so I should redo it. I don't recall how the back interacted with the top. Also, my model has approximations of course.

I recall the Bissinger picture and what you are saying. The accent there is on how the edges of the top and back cooperate to make two modes that interchange the roles of top and back. The picture emphasises the edges. My FEA picture emphasized the centers of the plates. They may be the same thing but look different from different points of view. But I need to go find the models and look at them again.

Link to comment
Share on other sites

Do you have a soundpost in the model, John? There will be a substantial difference betwen the modes with and without the post in. Without the post I think there is no real B1- mode but two modes. If I do not recall incorrectly one symmetric and one assymmetric that with the post combine to the B1-

The B1- will have a free top plate mode like vibration of the back plate, and a mode 2 type vibration in the top. For the B1+ the relation is the opposite if we simplify a little.

Link to comment
Share on other sites

I know what you mean. Isn't that what some call "ringing." An amplifier cannot turn on and off instantaneously. So the ringing is from very high frequencies which are not allowed to add up to make the sharp corners. You said it was not "causal" and that threw me off. I think it is causal and has to do with inadequate put-through of all the frequencies along with their proper phases. I am sure that phase shifting alone would cause a problem with a square wave even if the magnitude of the frequencies made it through.

If I put on my physics hat, it is like the Heisenberg Uncertainty principal... The Schroedinger equation has a wave solution and localizing a particle in real space makes the fourier transform large in the frequency domain. That would be momentum to Erwin.

I was not trying to find him right or wrong. Just that perhaps it is not acausal. Doesn't it come straight out of the fourier transform of the square wave and a consideration of how the fourier components are passed imperfectly through the amplifier?

You're stretching my memory 46 years to my study of the uncertainty principle. Unlike old violin wood, my wooden head is not becoming more elastic. IIRC, the pre-ringing in the wave function was what lead us to question causality, i.e. the wave was indicating some probability that the particle would be somewhere before it could be there. With the DACS, the input of a square pulse into a non-oversampling DAC will output a square pulse starting when it should. With the oversampling DAC, a square pulse input results in a gaussian-like output pulse, but with pre and post pulse ringing in which the small pre-pulse squiggles overlap backwards in time.

Link to comment
Share on other sites

Do you have a soundpost in the model, John? There will be a substantial difference betwen the modes with and without the post in. Without the post I think there is no real B1- mode but two modes. If I do not recall incorrectly one symmetric and one assymmetric that with the post combine to the B1-

The B1- will have a free top plate mode like vibration of the back plate, and a mode 2 type vibration in the top. For the B1+ the relation is the opposite if we simplify a little.

Yes, there is a beam for a post and special nodes was created at the spot behind the bridge to place it. I think I even had strings on it, but it has been a while. I could have overlooks some things.

I have a boatload of old input files, some of which are bad or in mixed units. I need to get them out, sort out the good ones, and discard the bad ones.. Maybe after I get my taxes finished.

You're stretching my memory 46 years to my study of the uncertainty principle. Unlike old violin wood, my wooden head is not becoming more elastic. IIRC, the pre-ringing in the wave function was what lead us to question causality, i.e. the wave was indicating some probability that the particle would be somewhere before it could be there. Except that you would be wrong to expect exactly where it could be... you did not know both the position and momentum exactly at the earlier time... With the DACS, the input of a square pulse into a non-oversampling DAC will output a square pulse starting when it should. With the oversampling DAC, a square pulse input results in a gaussian-like output pulse, but with pre and post pulse ringing in which the small pre-pulse squiggles overlap backwards in time.

I see, a single pulse and not periodic square wave. Yes, that is very strange. But the digitizing must be sampling and something like fourier transform (??) So it is done in real time. I don't know the sampling theorem but I recall trying to study it. It has some analogies of the uncertainy principal. At its core, the tradeoff of bandwidth and precision seems related to the uncertainy principal. There is an equivalent thing with the fourier integrals but does not of course contain the h-bar Plank constant. The uncertainty in question is of the order of 1. But I don't recall either, hope I am not talking through my hat.

At least you cannot sample perfectly finely. You would need an infinite bandwidth of frequencies to do that.

Link to comment
Share on other sites

Except that you would be wrong to expect exactly where it could be... you did not know both the position and momentum Except that you would be wrong to expect exactly where it could be... you did not know both the position and momentum exactly at the earlier time...

That was my point....when classical notions lead to conflicts with what the wave function is telling you, you are forced to give up some of your classical notions, causality being among the first to go.

Link to comment
Share on other sites

Yes, there is a beam for a post and special nodes was created at the spot behind the bridge to place it. I think I even had strings on it, but it has been a while. I could have overlooks some things.

I have a boatload of old input files, some of which are bad or in mixed units. I need to get them out, sort out the good ones, and discard the bad ones.. Maybe after I get my taxes finished.

Ok, I think it is important that the modes the FEA model spits out do look close to what the mesured modes do. The signature modes look pretty similar from violin to violin, so e.g. Bormans animation pictures or Schleskes figures might be a reference. Then the signature mode frequencies could be taken from an instrument with known material parameters, graduations and archings to go as input to the model. Of course the real model shapes need to be close to the FEA model.

I know you are not so interested in this kind of calibration. But this is the direction I would like to take it before I start reading out information on the modes from the FEA model. The calculated modes need to look pretty much right first.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.



×
×
  • Create New...