sam9 said:
I don't know if the answer exists. If you think about it, music instruments are nothing but harmonics generator. A piano sounds different from a drum because they generate different harmonics / harmonic composition.
so I would be surprised if all the instruments in the world suddenly genrate mostly 2nd or even harmonics.
Has anyone done spectrum analysis on musical instruments? I know that the new york times did quite extensive analysis on pianos and they might have employed some scientific methods rather than myth in their analysis.
But it would be interesting to see how a piano's FFT compares with that of a violin, for example, and how much harmonics are there.
If an instrument does generate a lot of odd harmonics, what does that say about amp distortion?
cunningham said:
I would be surprised if the component costs, incrmental or absolute, associated with a high-end amp costs more than 5% of its total costs.
high-end amps are like perfume, or environmental mediation: the overwhelming majority of your purchase prices goes into non-sound enhancing venues,
"Profile A would be the least objectionable. Not because of the absolute figures of distortion but because of the relative relationship between even and odd order harmonic distortion. "
Or would it be B, because .oo1% is not perceptable? I don't know what the currently established limit of perceptability is, as every decsade or so a zero gets added in front. Still there must be one, it is unreasonable to assume there is none.
So construct a Case C such that each figure has one more leading zero. Now which is least objectionable?
Construct case D with one more leading zero -- and so on.
Ultimately one of the two propositions must be true
1 There exists some level of distortion regardless of the harmonics below which it is pointless to go because it is not perceptable. Thus no further sonic improvement is possible with regard to distortion. (This is limited to operation below clipping or otherwise within the inherent limits of the amp.)
or
2 There is some distribution of harmonics that sounds better than zero distortion, thus lowering distortion is pointless because the lower you go the worse it sounds (Assumes, the distribution of harmonics is constant.)
The point is: is zero distortion possible and if so is it desireable assuming you don't screw up something else.
Here is a "softer" question. Select a distibution of harmonic artifacts that sounds "good". Now magically reduce each artifact proportionately. The same profile exists only lower. Does this sound better or worse?
Or would it be B, because .oo1% is not perceptable? I don't know what the currently established limit of perceptability is, as every decsade or so a zero gets added in front. Still there must be one, it is unreasonable to assume there is none.
So construct a Case C such that each figure has one more leading zero. Now which is least objectionable?
Construct case D with one more leading zero -- and so on.
Ultimately one of the two propositions must be true
1 There exists some level of distortion regardless of the harmonics below which it is pointless to go because it is not perceptable. Thus no further sonic improvement is possible with regard to distortion. (This is limited to operation below clipping or otherwise within the inherent limits of the amp.)
or
2 There is some distribution of harmonics that sounds better than zero distortion, thus lowering distortion is pointless because the lower you go the worse it sounds (Assumes, the distribution of harmonics is constant.)
The point is: is zero distortion possible and if so is it desireable assuming you don't screw up something else.
Here is a "softer" question. Select a distibution of harmonic artifacts that sounds "good". Now magically reduce each artifact proportionately. The same profile exists only lower. Does this sound better or worse?
Hi,
In a perfect and purely theoretical world it probably would but...
Since no active device is linear enough to achieve these kinds of figures all by itself, we'd need to apply NFB and other trickery in all kinds of ways to get there.
"There" being the measured figure here.
The problem with NFB is that it can give rise to other artefacts that, while they won't readily show up in usual measurement procedures, they still alter the way we perceive the result in a not so subtle way.
My main point being is that the numbers alone, as we currently obtain them, don't paint the whole picture.
IOW, we aren't measuring what we perceive no matter how powerful that same set of measuring tools has become.
See above.
I think such amplifiers exist for a number of years already yet they're still not convincing soundwise.
Why?
Is it because we're not measuring all the different kinds of distortion the human ear is capable of picking up?
Why is it we can immediately tell the difference between a live piano in say, a room and the reproduction of a recorded one in that same room, for instance?
Are our auditory senses really limited to the 20Hz-20KHz band?
What makes our brain distinguish between the real and the recorded sound?
Distortion, surely. Only far more complex than we're currently willing to accept IMHO.
Many people seem to prefer a certain pattern of distortion over no distortion (measured as we currently do) at all.
Again, why?
Is this quest for infinitesimally low measured distortion figures rubbing the music of that quality allowing us to fool ourselves into thinking this is indeed closer to the real thing?
If so, then what is it we're killing while going for this low distortion target? What is killed in the process and how can we avoid it?
Possible, probably not in the near future...
Desirable provided nothing else is messed up in the process, yes absolutely.
I can only assume that's what we all want:
a reproduction of an event that took place in another place and time in such a way that we're led to believe we're reliving the event that happened elsewhere as if we'd be present.
Cheers,
In a perfect and purely theoretical world it probably would but...
Since no active device is linear enough to achieve these kinds of figures all by itself, we'd need to apply NFB and other trickery in all kinds of ways to get there.
"There" being the measured figure here.
The problem with NFB is that it can give rise to other artefacts that, while they won't readily show up in usual measurement procedures, they still alter the way we perceive the result in a not so subtle way.
My main point being is that the numbers alone, as we currently obtain them, don't paint the whole picture.
IOW, we aren't measuring what we perceive no matter how powerful that same set of measuring tools has become.
See above.
I think such amplifiers exist for a number of years already yet they're still not convincing soundwise.
Why?
Is it because we're not measuring all the different kinds of distortion the human ear is capable of picking up?
Why is it we can immediately tell the difference between a live piano in say, a room and the reproduction of a recorded one in that same room, for instance?
Are our auditory senses really limited to the 20Hz-20KHz band?
What makes our brain distinguish between the real and the recorded sound?
Distortion, surely. Only far more complex than we're currently willing to accept IMHO.
Many people seem to prefer a certain pattern of distortion over no distortion (measured as we currently do) at all.
Again, why?
Is this quest for infinitesimally low measured distortion figures rubbing the music of that quality allowing us to fool ourselves into thinking this is indeed closer to the real thing?
If so, then what is it we're killing while going for this low distortion target? What is killed in the process and how can we avoid it?
Possible, probably not in the near future...
Desirable provided nothing else is messed up in the process, yes absolutely.
I can only assume that's what we all want:
a reproduction of an event that took place in another place and time in such a way that we're led to believe we're reliving the event that happened elsewhere as if we'd be present.
Cheers,
How to tailor harmonic magnitude spread? Like we wanted 3rd is 10% of 2nd harmonic? Or they are not "Tailorable" at all? Just have to follow the nature of tubes, transistors, etc?
A differential pair can cancel 2nd harmonic. But the residue of it, fedback to the differential raises higher order harmonics.
Simple question. How to reduce 3rd harmonic? What cct do that?
A differential pair can cancel 2nd harmonic. But the residue of it, fedback to the differential raises higher order harmonics.
Simple question. How to reduce 3rd harmonic? What cct do that?
http://www.gedlee.com/distortion_perception.htm
In particular check out the second power point presentation. It has most of the details.
Apparently it's all about masking:
1) "Distortion by-products that are created upward in frequency are likely to be less perceptible (masked to a greater extent) than those that fall lower in frequency."
2) "Distortion by-products that lie closer to the excitation are less likely to be perceived than those that lie farther away (masking is a localized effect ¡¦it mostly occurs in the vicinity of the masker)."
3) "Distortion by-products of any kind are likely to be more perceptible at lower signal levels than at higher signal levels. (Less masking occurs at lower signal levels)"
Leading to this hypothesis:
1) "The masking effect of the human ear will tend to make higher order nonlinearities more audible than lower order ones."
2) "Nonlinear by-products that increase with level can be completely masked if the order of the nonlinearity is low."
3) "Nonlinearities that occur at low signal levels will be more audible than those that occur at higher signal levels."
Then there's a presentation of a new metric with a small test indicating that this metric correlates with perception much better than THD or IMD.
In particular check out the second power point presentation. It has most of the details.
Apparently it's all about masking:
1) "Distortion by-products that are created upward in frequency are likely to be less perceptible (masked to a greater extent) than those that fall lower in frequency."
2) "Distortion by-products that lie closer to the excitation are less likely to be perceived than those that lie farther away (masking is a localized effect ¡¦it mostly occurs in the vicinity of the masker)."
3) "Distortion by-products of any kind are likely to be more perceptible at lower signal levels than at higher signal levels. (Less masking occurs at lower signal levels)"
Leading to this hypothesis:
1) "The masking effect of the human ear will tend to make higher order nonlinearities more audible than lower order ones."
2) "Nonlinear by-products that increase with level can be completely masked if the order of the nonlinearity is low."
3) "Nonlinearities that occur at low signal levels will be more audible than those that occur at higher signal levels."
Then there's a presentation of a new metric with a small test indicating that this metric correlates with perception much better than THD or IMD.
Hi guys. First of all, I in no way should be giving any answers regarding amplifier design. So I won't. However the original question did ask (basically) why people like tubes over the numerically superior SS amplifiers of today. Which some consider to be more accurate and less colored.
In short...I don't consider them superior because they may have better specifications. In fact, I'm not likely to ever consider SS superior in sound reproduction over tubes at all. But, the same can be said about tubes as well...because (as some others have said) there's some well designed good sounding tube amps...and some well designed good sounding SS amps. Any given SS amplifier just isn't an indicator of what all other SS amplifiers will sound like. So how can it be useful saying that SS amplifiers are better than tubes, or that tubes are better than SS. One piece doesn't describe the rest...and who can say which (individual) piece has reached the pinnacle of the potential of SS or tube might have. (even tho tube amps do seem to be a mature design..there have still been improvements such as better designed OPT's which led to not only better measurements and coverage of the extremes, but better sound as well.)
The above seems as silly (to me) as implying that SS isn't colored, while tubes are. Tube amps can be extremely colored...and various tubes (or circuits) can lessen the degree of coloration. But coloration does exist within tubes amps. This is not unlike SS amps tho...SS amps are colored otherwise they would sound alike. Many do not. Coloration variation of intensity or quality of tone. Even within higher priced SS amps they can sound very different from one another. Naim, Linn, Levinson, , NAD, Krell, Plinius, Coda, McCormack, Conrad Johnson on and on...they really sound very different from one another. I'm not talking about instances where they may be pushed beyond their design capabilities. I'm talking about voicing. They're just colored differently from one another.
This is probably the main focus of most of the tube Vs SS arguments I've read. Clipping and distortions characteristics of the amps when pushed near, to or passed levels in which they were designed or intended for. Near as I can see, is that this argument has no basis in many of the discussion comparing sound of an amplifier (SS or tube) when used appropriately. The above (looks to be) contrasting what happens when a tube amp is placed in an ill matched system Vs what happens to a SS amplifier which is pushed beyond what it's able. If this is the case, for either the tube or SS amp, then you really need to look at getting a new (higher powered tube or SS amp) or get more efficient speakers so your amps aren't pushed into area where they weren't designed to operate.
Some of you guys (the real designers out there) would know this better than I...but what do the distortion numbers look like on an average (decently designed) 2A3, 300B or 845 tube amp look like when they're using only 1/8, 1/4 and 1/2 of they're available power. Conversely what do they look like in the same for SS. Do you (the real designers out there) feel that this one area is where a significant difference can be found between tubes and SS?
If there are differences in sound characteristics when tube amplifiers aren't clipping...then why do so many discussions seem to look at this point as one of the primary topics of discussion. The dark tube amplifier sound could be looked at in the same way. This isn't indicative of how a tube amp sounds...so why imply that it is...and why use "this" (distorted when pushed too hard and cut off frequency extremes) sound as a comparison. Don't know?? perhaps some of you do
thanks,
dan
In short...I don't consider them superior because they may have better specifications. In fact, I'm not likely to ever consider SS superior in sound reproduction over tubes at all. But, the same can be said about tubes as well...because (as some others have said) there's some well designed good sounding tube amps...and some well designed good sounding SS amps. Any given SS amplifier just isn't an indicator of what all other SS amplifiers will sound like. So how can it be useful saying that SS amplifiers are better than tubes, or that tubes are better than SS. One piece doesn't describe the rest...and who can say which (individual) piece has reached the pinnacle of the potential of SS or tube might have. (even tho tube amps do seem to be a mature design..there have still been improvements such as better designed OPT's which led to not only better measurements and coverage of the extremes, but better sound as well.)
The above seems as silly (to me) as implying that SS isn't colored, while tubes are. Tube amps can be extremely colored...and various tubes (or circuits) can lessen the degree of coloration. But coloration does exist within tubes amps. This is not unlike SS amps tho...SS amps are colored otherwise they would sound alike. Many do not. Coloration variation of intensity or quality of tone. Even within higher priced SS amps they can sound very different from one another. Naim, Linn, Levinson, , NAD, Krell, Plinius, Coda, McCormack, Conrad Johnson on and on...they really sound very different from one another. I'm not talking about instances where they may be pushed beyond their design capabilities. I'm talking about voicing. They're just colored differently from one another.
Bas Horneman said:
This is probably the main focus of most of the tube Vs SS arguments I've read. Clipping and distortions characteristics of the amps when pushed near, to or passed levels in which they were designed or intended for. Near as I can see, is that this argument has no basis in many of the discussion comparing sound of an amplifier (SS or tube) when used appropriately. The above (looks to be) contrasting what happens when a tube amp is placed in an ill matched system Vs what happens to a SS amplifier which is pushed beyond what it's able. If this is the case, for either the tube or SS amp, then you really need to look at getting a new (higher powered tube or SS amp) or get more efficient speakers so your amps aren't pushed into area where they weren't designed to operate.
Some of you guys (the real designers out there) would know this better than I...but what do the distortion numbers look like on an average (decently designed) 2A3, 300B or 845 tube amp look like when they're using only 1/8, 1/4 and 1/2 of they're available power. Conversely what do they look like in the same for SS. Do you (the real designers out there) feel that this one area is where a significant difference can be found between tubes and SS?
If there are differences in sound characteristics when tube amplifiers aren't clipping...then why do so many discussions seem to look at this point as one of the primary topics of discussion. The dark tube amplifier sound could be looked at in the same way. This isn't indicative of how a tube amp sounds...so why imply that it is...and why use "this" (distorted when pushed too hard and cut off frequency extremes) sound as a comparison. Don't know?? perhaps some of you do
thanks,
dan
Try this
Hi Darkfenriz,
For tube sound try this scheme:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=20686
Don
Hi Darkfenriz,
For tube sound try this scheme:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=20686
Don
Hi Darkfenriz,
You may search for Erik Pritchard patents,
Alternative approach is here
http://www.diyaudio.com/forums/showthread.php?postid=407896#post407896
You may search for Erik Pritchard patents,
Alternative approach is here
http://www.diyaudio.com/forums/showthread.php?postid=407896#post407896
Since I feel I've met some scepticism and misunderstaning when I raised
the topic of distortion, I suggest a little DSP
simulation(look into atachment). If you are a happy owner or are
able to reach MATLAB workspace please make the simulation
(you can do it as an m-file) as follows
(it works on the 'c:/1.wav' file and plays
the original and then the enhanced wave):
%gentle distortion
w=wavread('c:/1.wav',2097152);%or any wave you want
a=[-1:.0001:1];
n=0.8*a-0.28*a.^2+0.16*a.^4-0.04*a.^8;
plot(a,n);
n=0.8*w-0.28*w.^2+0.16*w.^4-0.04*w.^8;
e=n;
e(1,1)=0;
e(1,2)=0;
d=n;
d(1,1)=0;
d(1,2)=0;
%first order Butterworth filter
%lowpass Fp=800Hz
for a=2:2097152
e(a,1)=0.053972856762248569*n(a,1)+-0.89205428647550289*e(a-1,1);
e(a,2)=0.053972856762248569*n(a,2)+-0.89205428647550289*e(a-1,2);
end
q=3*e+1.5*n;
%lowpass again- cuts slightly the over 12000Hz freqs
for a=2:2097152
d(a,1)=0.57081533689866215*q(a,1)+0.14163067379732439*d(a-1,1);
d(a,2)=0.57081533689866215*q(a,2)+0.14163067379732439*d(a-1,2);
end
wavplay(w,44100);%original wave
wavplay(d,44100);%distorted+filered
% I like the gentle old-time crunch indeed
the topic of distortion, I suggest a little DSP
simulation(look into atachment). If you are a happy owner or are
able to reach MATLAB workspace please make the simulation
(you can do it as an m-file) as follows
(it works on the 'c:/1.wav' file and plays
the original and then the enhanced wave):
%gentle distortion
w=wavread('c:/1.wav',2097152);%or any wave you want
a=[-1:.0001:1];
n=0.8*a-0.28*a.^2+0.16*a.^4-0.04*a.^8;
plot(a,n);
n=0.8*w-0.28*w.^2+0.16*w.^4-0.04*w.^8;
e=n;
e(1,1)=0;
e(1,2)=0;
d=n;
d(1,1)=0;
d(1,2)=0;
%first order Butterworth filter
%lowpass Fp=800Hz
for a=2:2097152
e(a,1)=0.053972856762248569*n(a,1)+-0.89205428647550289*e(a-1,1);
e(a,2)=0.053972856762248569*n(a,2)+-0.89205428647550289*e(a-1,2);
end
q=3*e+1.5*n;
%lowpass again- cuts slightly the over 12000Hz freqs
for a=2:2097152
d(a,1)=0.57081533689866215*q(a,1)+0.14163067379732439*d(a-1,1);
d(a,2)=0.57081533689866215*q(a,2)+0.14163067379732439*d(a-1,2);
end
wavplay(w,44100);%original wave
wavplay(d,44100);%distorted+filered
% I like the gentle old-time crunch indeed
Attachments
ATTENTIONthanks for the hint!
I simulated your circuit in a circuit simulating software and the distortions look fine I mean there's mainly even order harmonics, mostly 2nd.
Would be nice to simplify the circuit. And it would be great in my opinion if one could build an easy circuit where you could regulate easily the character of the distortion, to vary between even and odd order, and high and low order harmonics...
Cheers!
Dominique
I simulated your circuit in a circuit simulating software and the distortions look fine
I mean there's mainly even order harmonics, mostly 2nd.Would be nice to simplify the circuit. And it would be great in my opinion if one could build an easy circuit where you could regulate easily the character of the distortion, to vary between even and odd order, and high and low order harmonics...
Cheers!
Dominique
hi
here's the heavier version of the matlab dsp distortion:
%%%%%%%%%%%%%%%%%%%%%%%%%%
%heavy distortion
w=wavread('c:/1.wav',2097152);%or any wave you want
a=[-1:.0001:1];
n=0.6*a+0.4*(-1.4*a.^2+0.8*a.^4-0.2*a.^8);
plot(a,n);
n=0.6*w+0.4*(-1.4*w.^2+0.8*w.^4-0.2*w.^8);
e=n;
e(1,1)=0;
e(1,2)=0;
d=n;
d(1,1)=0;
d(1,2)=0;
for a=2:2097152
e(a,1)=0.053972856762248569*n(a,1)+-0.89205428647550289*e(a-1,1);
e(a,2)=0.053972856762248569*n(a,2)+-0.89205428647550289*e(a-1,2);
end
q=17*e+0.8*n+0.8*w;
for a=2:2097152
d(a,1)=0.42736706578659844*q(a,1)-0.14526586842680311*d(a-1,1);
d(a,2)=0.42736706578659844*q(a,2)-0.14526586842680311*d(a-1,2);
end
wavplay(w,44100);%original wave
wavplay(d,44100);%distorted+filered
% I like the gentle old-time crunch indeed
here's the heavier version of the matlab dsp distortion:
%%%%%%%%%%%%%%%%%%%%%%%%%%
%heavy distortion
w=wavread('c:/1.wav',2097152);%or any wave you want
a=[-1:.0001:1];
n=0.6*a+0.4*(-1.4*a.^2+0.8*a.^4-0.2*a.^8);
plot(a,n);
n=0.6*w+0.4*(-1.4*w.^2+0.8*w.^4-0.2*w.^8);
e=n;
e(1,1)=0;
e(1,2)=0;
d=n;
d(1,1)=0;
d(1,2)=0;
for a=2:2097152
e(a,1)=0.053972856762248569*n(a,1)+-0.89205428647550289*e(a-1,1);
e(a,2)=0.053972856762248569*n(a,2)+-0.89205428647550289*e(a-1,2);
end
q=17*e+0.8*n+0.8*w;
for a=2:2097152
d(a,1)=0.42736706578659844*q(a,1)-0.14526586842680311*d(a-1,1);
d(a,2)=0.42736706578659844*q(a,2)-0.14526586842680311*d(a-1,2);
end
wavplay(w,44100);%original wave
wavplay(d,44100);%distorted+filered
% I like the gentle old-time crunch indeed
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.