The caracteristic sound of an amplifier (or CD player for that matter), is based on the harmonic distortion. The statement that there is a transistor with "good sound" is both true and false, actually 99% false.
The human ear percepts third harmonc distortion as a "foul" sound, the second harmonic however makes the amplifier sound "warm". Tube amplifiers have very large second harmonic distortion (often1-10%), thats`why they have a "warm and soft" sound.
You could go the effort and tracking down transistors that are expensive, and are used in Krell, Electrocomaniet and so on, but the results will be minute.
There are many more and better ways to make your amp sound
the way you like, just remember that third (odd harmonics generally) has an ugly and harsh sound.
You could with a little effort design an amplifier in the 0.001% THD region with general purpose transistors (BC546, BD139, TIP35), and make it sound like a million!!
The key is using good constant current source and an active mirror load at the differential input (if you want a warmer sounding amplifier, just replace the current mirror with resistors, this will increase the second harmonics).
OK, a lot of talk here...the bottom line is, its the circuit topology that desides the THD and S/N. I a wilson current source would make a differental input pair consisting of BC546/BC556 transistors sound far better than replasinbg the wilson with a Transistor-diode current source and using 2SC2240 transistors (however I would use the 2SC2240/2SA970 anyway because they have higher Vce tolerances, and they have higher beta (a wilson curren source with that also uses these, will have an output impedance in the 10-50Mohm region).
The human ear percepts third harmonc distortion as a "foul" sound, the second harmonic however makes the amplifier sound "warm". Tube amplifiers have very large second harmonic distortion (often1-10%), thats`why they have a "warm and soft" sound.
You could go the effort and tracking down transistors that are expensive, and are used in Krell, Electrocomaniet and so on, but the results will be minute.
There are many more and better ways to make your amp sound
the way you like, just remember that third (odd harmonics generally) has an ugly and harsh sound.
You could with a little effort design an amplifier in the 0.001% THD region with general purpose transistors (BC546, BD139, TIP35), and make it sound like a million!!
The key is using good constant current source and an active mirror load at the differential input (if you want a warmer sounding amplifier, just replace the current mirror with resistors, this will increase the second harmonics).
OK, a lot of talk here...the bottom line is, its the circuit topology that desides the THD and S/N. I a wilson current source would make a differental input pair consisting of BC546/BC556 transistors sound far better than replasinbg the wilson with a Transistor-diode current source and using 2SC2240 transistors (however I would use the 2SC2240/2SA970 anyway because they have higher Vce tolerances, and they have higher beta (a wilson curren source with that also uses these, will have an output impedance in the 10-50Mohm region).
Yep that sounds right. Even so, you shouldn't be able to tell the difference between a good tube and good SS amplifier if the distortion is low enough. It's also true that most of the sound comes from the circuit around the transistors. I use TIP35/TIP36 in almost all of my amps. They'll sound as good as I make them sound, and recently I've been able to make them sound very very good. Also because they're included in the feedback loop, I can pull a TIP35, replace it with a TIP41, TIP31 or 2N3055 and there's no difference in sound (power handling excluded). It's mostly in the circuit, garbage in means garbage out. Transistors that work in the 50Mhz range aren't going to give you any advantage if your circuit sounds like crap at 1Khz.
By the way, what exactly is a current/active mirror?
By the way, what exactly is a current/active mirror?
The sound often called as a "transistor sound" often results (according to my experience) from intermodulations of audio signal with RF interference or D/A residuals at CD output, even if they are very low in magnitude. This occurs also in case you build an amp with THD as low as 0.0005% with only 2nd harmonic line in the spectrum. I have made a lot of experiments with this phenomenon, amps with absolutely same THD and IMD specs (hardly measurable) did sound quite different according to HF residuals filtration. Also - the signal transfer on higher current level (50 Ohm load) brings great improvement to sound quality (grain disappears, bass is tough, midrange and treble clear and localization improved). Bruce Candy must have found the same, as Halcro dm58/68 use also 50 Ohm inputs and current inputs in addition to standard 50k inputs.
You folks are thinking about the problem, but you are not thinking it through. First of all, 3'rd harmonic is actually almost as acceptable as 2'nd harmonic. How do we know? Because 3'rd harmonic is the only distortion normally measurable on analog magnetic tape, and its value is usually between .1% and 10% depending on output level. How is it that we can, or could listen to analog tapes without crying out in pain?
It is the HIGHER ORDER ODD HARMONICS that are the big problem. For example, 7th and 9th harmonic. 8th harmonic should be OK, within reason.
Transistors generate much more higher order distortion than do tubes. This is because of the curvature of their transfer function, due to very nonlinear Gm. Tubes also have some nonlinearity in gain, but they work on a different principle and change less over current, and usually have less distortion, and it will be of lower order in general.
It is the HIGHER ORDER ODD HARMONICS that are the big problem. For example, 7th and 9th harmonic. 8th harmonic should be OK, within reason.
Transistors generate much more higher order distortion than do tubes. This is because of the curvature of their transfer function, due to very nonlinear Gm. Tubes also have some nonlinearity in gain, but they work on a different principle and change less over current, and usually have less distortion, and it will be of lower order in general.
Hi,
If you'd replace the words "third harmonic" distortion with odd order harmonic distortion I'd agree with the above.
Tube amplifiers do not have large amounts of "second", let's call that even order, harmonic distortion per se. Those that do present such a harmonic distortion will tend to sound unrealistically warm and rich and will be called by a lot of people of having a "tube" sound.
While I do appreciate what tubes do right, I do refuse to accept that there is a "tube" sound as such.
Naturally tubes and semiconductors distort in a different way when overloaded, sent into clipping and alot of this characteristic behaviour has led to the believe that a tube amp seems much more powerful than its' equivalent semi-conductor amp.
In reality it's just the clipping behaviour of the tube amp that when sent into clipping produces more even order harmonic distortion which the human ear perceives as more related to the signal than distortion containing predominantly odd harmonics.
When clipping occurs with semi-conductor amplifiers it occurs much more abruptly and subjectively seems as if the amp has ran out of steam and distorts in a horrible manner. (Odd order harmonic distortion)
Which ever one prefers, it's still distortion nonetheless.
Cheers,
If you'd replace the words "third harmonic" distortion with odd order harmonic distortion I'd agree with the above.
Tube amplifiers do not have large amounts of "second", let's call that even order, harmonic distortion per se. Those that do present such a harmonic distortion will tend to sound unrealistically warm and rich and will be called by a lot of people of having a "tube" sound.
While I do appreciate what tubes do right, I do refuse to accept that there is a "tube" sound as such.
Naturally tubes and semiconductors distort in a different way when overloaded, sent into clipping and alot of this characteristic behaviour has led to the believe that a tube amp seems much more powerful than its' equivalent semi-conductor amp.
In reality it's just the clipping behaviour of the tube amp that when sent into clipping produces more even order harmonic distortion which the human ear perceives as more related to the signal than distortion containing predominantly odd harmonics.
When clipping occurs with semi-conductor amplifiers it occurs much more abruptly and subjectively seems as if the amp has ran out of steam and distorts in a horrible manner. (Odd order harmonic distortion)
Which ever one prefers, it's still distortion nonetheless.
Cheers,
Surely everything we perceive as 'sound' is distorted. If musical instruments didn't add harmonic distortion they would all sound the same. Air has a non-linear transfer characteristic, our ears aren't linear and neither is the way our brains process 'sound'.
Suddenly amplifier distortion doesn't seem so bad.
Suddenly amplifier distortion doesn't seem so bad.
I both agree and disagree on that one, but I still insist on the odd (and especially the third) harmonics being much more unpleasant than the 2nd. I`ve read a couple of textbooks that also agrees on my opinion... But could of course be wrong, since I haven`t really got great experience on this, nevertheless, I know how to decrease harmonic distortion in amplifiers, and I`m pretty good at it. And that`s a start!
Tube generally have much more harmonic distortion than semi`s, because they have an much more unlinear transfer curve. I`m pretty sure everyone agrees on that....
Again, the distortion in transistorts depends much much more on the topology, and how the input and output impedances "harmonize" with eachother (for instance, in the output transistors, it`s very important to have the input capacitance in mind, which also determines the input impedance for various frequencies). I don`t know if that came out the way I meant it, it sounded very good in norwegian...
Regardless, everyone must agree on, that distortion should be as low as possible around 3kHz, it`s around this area that the human ear is most sensitive. The key is to design the amplifier to have a P1 point above above 1kHz, it`s impossoble to have a harmonic distortion carachteristic that is a straight line through the audio spectrum. No matter what we do, the distortion will rise with higher frequenzy above the P-point.
Tube generally have much more harmonic distortion than semi`s, because they have an much more unlinear transfer curve. I`m pretty sure everyone agrees on that....
Again, the distortion in transistorts depends much much more on the topology, and how the input and output impedances "harmonize" with eachother (for instance, in the output transistors, it`s very important to have the input capacitance in mind, which also determines the input impedance for various frequencies). I don`t know if that came out the way I meant it, it sounded very good in norwegian...
Regardless, everyone must agree on, that distortion should be as low as possible around 3kHz, it`s around this area that the human ear is most sensitive. The key is to design the amplifier to have a P1 point above above 1kHz, it`s impossoble to have a harmonic distortion carachteristic that is a straight line through the audio spectrum. No matter what we do, the distortion will rise with higher frequenzy above the P-point.
What`s the source of the higher THD tube amplifiers then?
I really want to know, since the sources of my knowledge around tubes generally are schoolbooks and amplifier construction manuals...
I`ll quote G. Randy Slone:
"There are a variety of reasons that many audiophiles prefer vacuum tube audio power amplifiers to solid state designs. The most probable reason is that tube amplifier naturally produce (again, see transfer curve) a relatively high percentage of second-order harmonic distortion. In musical terms, a secon order harmonic is always one octave above the fundamental frequency, so the tones can never become dissonant with the program material. The sonic effect of adding second order distortion is similar to hearing two musical tones instead of one, simulating a perception of body or character to the original program material."
I have over 9 yars of experience in electronics, and I can`t honestly remember once, anyone or any textbook claiming otherwise....
My goal here isn`t to provoke you tupe-lovers, I think this subject is extremely interesting, and I really enjoy discussing these things. It makes me seeing things a little bit in perspective..
I really want to know, since the sources of my knowledge around tubes generally are schoolbooks and amplifier construction manuals...
I`ll quote G. Randy Slone:
"There are a variety of reasons that many audiophiles prefer vacuum tube audio power amplifiers to solid state designs. The most probable reason is that tube amplifier naturally produce (again, see transfer curve) a relatively high percentage of second-order harmonic distortion. In musical terms, a secon order harmonic is always one octave above the fundamental frequency, so the tones can never become dissonant with the program material. The sonic effect of adding second order distortion is similar to hearing two musical tones instead of one, simulating a perception of body or character to the original program material."
I have over 9 yars of experience in electronics, and I can`t honestly remember once, anyone or any textbook claiming otherwise....
My goal here isn`t to provoke you tupe-lovers, I think this subject is extremely interesting, and I really enjoy discussing these things
. It makes me seeing things a little bit in perspective..Richard C said:
Hm, I am not so sure I would say that instruments cause
distorsion. The harmonic spectrum of them is inherent to them
and is desirable, since we seem to enjoy intrumets sounding
different. As for the distorsion in air and the ear,
this is always present, whether we listen to real life
sound, speach, music, cars etc. or audio equipment, so I would
guess our brain has
learnt to compensate rather well for these forms of distorsion.
Distorision in audio equipment, on the other hand is a very
new phenomenon, and we have had only about a century
to try adopting to that (evolution is slow in humans). Besides, since the various equipment sound different, we don't get a fair chance to adopt to it.
For instance, im a disturbed by the sound of cars, I have a big
parking lot outside my kitchen window, but at least they sound
as I am used to, not distorted by audio equipment. Two floors
above me there is a man playing the grand piano, but apart
from him being a rather good musician and play music I can
stand or even like, I clearly hear, despite two concrete floors
in between, that it is the sound of a real instrument, sounding
if not quite like in a concert hall, so at least normal. On the other
hand, the neighbours one floor above me obviously has a
very crappy audio equipment, probably a cheap home theater
setup, and it is very disturning since it sounds so very wrong
and confuses my brain.
PanzerLord said:
I used to believe so too, thinking that the tube hype is only
about people wanting a particular manufactired sound. After
readin a post by Torsten/Kuei at Audioasylum I am not so
sure anymore. I know next to nothing about tubes. However,
as I recall, his claim was that he only approved of certain
types of tubes and only when used in certain ways, and in
these cases they had an almost perfect linear transfer
characteristic. I don't know if it is true, but it made me rethink
that maybe there is something to this tube stuff after all.
Yes, I have heard two very fine amplifiers from the same stable, one tube and the other transistor.fdegrove said:
At levels less than overload, both amplifiers were remarkably similar, and on the short listening session the tube amp came out as being slightly blacker between notes and not much else difference - perhaps slightly 'sweeter' overall.
IOW both amplifiers were musically very good indeed, and just about any listener would be perfectly happy with either one.
Yes, overload behaviour is mission critical and tube amps do not by definition always sound less bad than overloaded transistor amplifiers.
The above mentioned transistor amplifier was designed to incorporate benign overload behaviour and it shows - there are no transient instabilities or rail sticktion (latchup) nasties and the overload sonics of this transistor amplifier are indeed nicely benign and inoffensive - it just quietly alerts the listener to turn down a bit in a similar manner to an overloaded tube amp.
Eric.
Harmonics
I think it's interesting that so much technical thought goes into reproducing music and how to describe how "good" the reproduction is from the electronics point of view without music being given much of a chance to defend itself.
Here's my two cents then:
Music is played from scales. But those scales were generated FROM the harmonics of a fundamental note (440 Hz - A). Now music theorists consider "good" sounds and consonant and "bad" sounds as dissonant (which if used right can be good...) The most consonant sounds are the octaves and then the harmonics starting from lowest order first that aren't octaves of previously encountered harmonics.
Anyway to put it simply, the 2nd and 4th harmonics are octaves of the fundamental, according to music they should actually reinforce the root note being played (which is a possible reason why the even order clipping distortion of tubes sounds "warm")
The 6th harmonic (still even order,adding to that warm sound) is called a 'fifth'.
What's key is that the 6th harmonic is actually just an octave of the 3rd harmonic. If the 6th adds to the warm sound, the 3rd HAS to as well.
The point is, since harmonic distortion energy tends to be concentrated in the lower harmonics first even order is less likely to contan energy in harmonics that start to be dissonant. But the even vs odd stereotyping really is over generalized.
Why is odd worse than even in the end? Because odd order harmonics will reach dissonant tones faster, but that doesn't make all odd order harmonics "bad"
--
Danny
I think it's interesting that so much technical thought goes into reproducing music and how to describe how "good" the reproduction is from the electronics point of view without music being given much of a chance to defend itself.
Here's my two cents then:
Music is played from scales. But those scales were generated FROM the harmonics of a fundamental note (440 Hz - A). Now music theorists consider "good" sounds and consonant and "bad" sounds as dissonant (which if used right can be good...) The most consonant sounds are the octaves and then the harmonics starting from lowest order first that aren't octaves of previously encountered harmonics.
Anyway to put it simply, the 2nd and 4th harmonics are octaves of the fundamental, according to music they should actually reinforce the root note being played (which is a possible reason why the even order clipping distortion of tubes sounds "warm")
The 6th harmonic (still even order,adding to that warm sound) is called a 'fifth'.
What's key is that the 6th harmonic is actually just an octave of the 3rd harmonic. If the 6th adds to the warm sound, the 3rd HAS to as well.
The point is, since harmonic distortion energy tends to be concentrated in the lower harmonics first even order is less likely to contan energy in harmonics that start to be dissonant. But the even vs odd stereotyping really is over generalized.
Why is odd worse than even in the end? Because odd order harmonics will reach dissonant tones faster, but that doesn't make all odd order harmonics "bad"
--
Danny
john curl said:
But tubes do the same. Have a look at tube amps reviewed by Stereophile and you will see that they also do have 8th, 10th harmonics at the spectrum, their spectral response is like ridges and valleys. But what they actually do - they are quite insensitive to HF intermodulations and they mask. Masking is the big trouble. Most of producers do not try to make low distortion components, but they tune the audio chains to mask the problems, not to solve them. Transitor and OpAmp gears are more sensitive to the signal quality and cable phenomena - that's all. Once you solve it you will get to higher league compared to "pleasant sounding" amps. Direct comparison shows that "pleasant sounding" amps bring their own coloration, hide the details, unfocus the soundstage and bring something like "tail" behind the sound, also unable to reproduce "silence" under (between) tones.
Azira,
very intersting analysis. I have never thought of that before,
and cannot remeber anybody else making this comparison. It
is probably not easy to draw any straightforwards conclusions
from it, but at least it might indicate the usual "odd is good even
is bad" distinction is way too simplistic, which I suppose most
agree on anyway, but not for the same reasons.
Hm, I just recall PRR recently wrote that musical instruments
tend to have harmonics that are not perfect multiples of the
fundamental but slightly off. If that is true, it complicates the
issue further.
very intersting analysis. I have never thought of that before,
and cannot remeber anybody else making this comparison. It
is probably not easy to draw any straightforwards conclusions
from it, but at least it might indicate the usual "odd is good even
is bad" distinction is way too simplistic, which I suppose most
agree on anyway, but not for the same reasons.
Hm, I just recall PRR recently wrote that musical instruments
tend to have harmonics that are not perfect multiples of the
fundamental but slightly off. If that is true, it complicates the
issue further.
In the immortal words of Mark Twain, who remarked that he'd never met a man he didn't like, I'm tempted to say that I've not read a single post in this thread I didn't agree with.
In point form, here's my take:
1. The musical scale favors chord structures with minimal discord, and this means H2 and H3 and maybe H4 are just fine. But beyond that, we rapidly enter into discordant sounds, precisely because music is played in complex chord structures and harmonics beyond H3 quickly degenerate into discord.
2. Now, music is also about mood - and discord is a part of this manipulation. Rikard Strauss, Rachmaninov, many of the tone poem composers, even Sibelius and the later Beethoven String Quartets, ALL exhibit strong discord from time to time. This music is not everyone's cup of tea, but it is deeply evocative to me, and clearly the discordant balance is very carefully structured.
3. Since music is based around 'pleasing' chord structures, and the ear is so extraordinarily sensitive to so-called 'discord', it follows that intermodulation 'sum and difference' products, which have no musical relationships at all to any tonic scale I know of, will be pivotal in corrupting recorded renditions of performances.
4. Thus, on this basis are identified two separate issues; one harmonic overtones, produced by the non-linear amplification process; and sum/difference intermodulations, which also result of non-linearities in the process, and seem particularly troublesome with time-smear distortions, such as those generated in slow global negative feedback loops. As an aside, some of the amplitude modulated jitter problems in the digital domain create intermodulation products, which are very nasty.
So, our amplification process should keep all harmonic generation low order, with particular emphasis on avoiding H5 and higher, and it should be clean, and as linear as possible, to avoid the blight of 'sum/difference' intermodulations, which typically give a muddy, vague sound. I guess this comes back to the tired old advice we've all heard a million times, 'As linear as possible, and no global feedback'. Of course, this is as practical as riding a beaver across the Sahara, but never mind, we hear it constantly anyway.....
That said, when one listens to a good quality system from afar, and then a genuine musical instrument, like Christer's Grand Piano in the apartment above, the skilled listener can clearly pick up the difference. Why is this, and once we identify it, could this be the key to producing truly realistic sounds?
Of course, to open another can of words, the dreadful truth is that the recording process is probably as guilty of corrupting the music as the playback - and arguably more so.
Cheers,
Hugh
In point form, here's my take:
1. The musical scale favors chord structures with minimal discord, and this means H2 and H3 and maybe H4 are just fine. But beyond that, we rapidly enter into discordant sounds, precisely because music is played in complex chord structures and harmonics beyond H3 quickly degenerate into discord.
2. Now, music is also about mood - and discord is a part of this manipulation. Rikard Strauss, Rachmaninov, many of the tone poem composers, even Sibelius and the later Beethoven String Quartets, ALL exhibit strong discord from time to time. This music is not everyone's cup of tea, but it is deeply evocative to me, and clearly the discordant balance is very carefully structured.
3. Since music is based around 'pleasing' chord structures, and the ear is so extraordinarily sensitive to so-called 'discord', it follows that intermodulation 'sum and difference' products, which have no musical relationships at all to any tonic scale I know of, will be pivotal in corrupting recorded renditions of performances.
4. Thus, on this basis are identified two separate issues; one harmonic overtones, produced by the non-linear amplification process; and sum/difference intermodulations, which also result of non-linearities in the process, and seem particularly troublesome with time-smear distortions, such as those generated in slow global negative feedback loops. As an aside, some of the amplitude modulated jitter problems in the digital domain create intermodulation products, which are very nasty.
So, our amplification process should keep all harmonic generation low order, with particular emphasis on avoiding H5 and higher, and it should be clean, and as linear as possible, to avoid the blight of 'sum/difference' intermodulations, which typically give a muddy, vague sound. I guess this comes back to the tired old advice we've all heard a million times, 'As linear as possible, and no global feedback'. Of course, this is as practical as riding a beaver across the Sahara, but never mind, we hear it constantly anyway.....
That said, when one listens to a good quality system from afar, and then a genuine musical instrument, like Christer's Grand Piano in the apartment above, the skilled listener can clearly pick up the difference. Why is this, and once we identify it, could this be the key to producing truly realistic sounds?
Of course, to open another can of words, the dreadful truth is that the recording process is probably as guilty of corrupting the music as the playback - and arguably more so.
Cheers,
Hugh
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.