Hi,
I will have to come back one day.
If I get the wolf far enough from the door I intend to do a full "Silk Road" Tour from Xi'an (Chang An) to 'stambul (Constantinople) by sidecar motorcycle. It would take me through Tashkent as well...
This is unsurprising, though sad. Much of the advanced capabilities the Sovied Union possessed where lost when it broke up...
Though back in former east germany we used to joke that russian microelectronic circuits where especially service friendly, as you could get inside them to fix them.. ;-)
I agree on using very fast parts and that we need to watch interstage non-linear loading (capacitive and trans-resistive) in solid state equipment, it is one of the big differences when comparing Tube equipment, feedback being employed or not.
Ciao T
I will have to come back one day.
If I get the wolf far enough from the door I intend to do a full "Silk Road" Tour from Xi'an (Chang An) to 'stambul (Constantinople) by sidecar motorcycle. It would take me through Tashkent as well...
This is unsurprising, though sad. Much of the advanced capabilities the Sovied Union possessed where lost when it broke up...
Though back in former east germany we used to joke that russian microelectronic circuits where especially service friendly, as you could get inside them to fix them.. ;-)
I agree on using very fast parts and that we need to watch interstage non-linear loading (capacitive and trans-resistive) in solid state equipment, it is one of the big differences when comparing Tube equipment, feedback being employed or not.
Ciao T
Last edited:
You claim , it is unaudible, but substraction test show (e.g for tube amplifier with THD or IMD e.g 0,5%) it clear audible. The same empirical test show no audible artefacts (e.g. for SS amp with THD and IMD about 0,003%) But somebody will claim it is audible. Without no proof, only "I can hear it",masked in music, but unmasked the same signal is for the same subject unadible. You believe, it is possible?
It is impossible (show me one proved , measured examle of such amp.) to built and amplifier, that measures e.g THD at 20kHz 0,01% and CCIF 19+20kHz (at the same level) e.g.0,5%. In spectra 2nd HD for sole 20kHz will be the same in proportion as for 20kHZ spectral line from CCIF measurement.
jan, I concur that you are describing the difference labeled here
http://waltjung.org/PDFs/SID_TIM_1.pdf
http://waltjung.org/PDFs/SID_TIM_2.pdf
http://waltjung.org/PDFs/SID_TIM_3.pdf
as “soft TIM” (rise in distortion with the onset of slewing) and “hard TIM” (distortion during actual slew limiting of the amplifying device)
Regards
George
It seems that russians are especially non-talented in micro-circuits (similarly like in playing soccer), while RF transistors were not bad, and those manufactured 30-40 years ago were quite close to some modern RF Mitsubisi and NXP products.
When at the end of 1970-th a soviet fighter plane was landed in Japan due to pilot's disloyalty, japanees military engineers were terribly lothing over control system of the plane, based on tubes.
At the end of soviet epoch I personally "tried to help to russians with microcircuits" (joke), by developing computer soft for simulation of the processes of ion implantation doping, of the mechanisms of point defects production and non-equilibrium annealing. But all these tools finally have gone to Germany, Forschungs Zentrum Rossendorf near Dresden.
About definition of "transient distortions", here is my absolutely non-scientific understanding. When an amp receives very "sharp" stimulus, a front of the square wave for instance, it needs some time period for completing some transient behaviour, and while it is "occupied" by fighting with the sharp stimulus, all other signal detailes, falling in to this short period of amp's insanity, are strongly damaged (distorted).
Last edited:
Hi,
So is an 0.5dB difference in level. However, neither I nor anyone else listens to their music after passing it through a subtractive test. I am not as such interested in the result of just listening to the residual.
Moreover, in most cases we have a definition of distortion at rated power. With tube Amplifiers that have a simple monotonic distortion the distortion at normal listening levels will be considerably lower than the levels at maximum.
I doubt anyone would claim that 0.003% HD/IMD as such is audible. Yes are we only talking about HD/IMD? No.
I am unsure what you are attempting to say.
My point is simple.
1) Have the original undistorted music signal.
2) Have the same signal exposed to a distortion producing mechanism that produces 3% 2nd HD and 0.3% 3rd HD (and no higher order ones) at "full scale".
3) Have the same signal exposed to a distortion producing mechanism that produces 001% THD resulting only from crossover distortion (e.g. un-biased Class B Output Stage) at "full scale".
4) Have a large number of people compare the two resultant signals in a blind test, asking them to rate the three resulting signals according to how distorted they perceive each track, compared to the original unmodified track.
If 1 & 2 are rated as essentially identical with the control and 3 is rated highly distorted (which is how things will play out and have played out in a number of tests I am familiar with) it is safe to conclude that the distortion present in 2 is inaudible, while that present in 3 is audible (not that this breaks new ground).
This is not the same as saying that isolated from the original and scaled correctly in loudness the pure distortion residual will be inaudible. Nor is there is any point to establishing this, as now the human ear is not exposed to the original signal, which would produce the very distortion inside the ear that would have masked the distortion, nor is the actual object of the exercise (namely the music) still present.
So, your contention that the isolated distortion, without accompanying signal will be audible is correct, it is however equally meaningless, especially if we understand how the human hearing works... It is an answer looking for a relevant question, without finding one.
Again, you completely miss my point. Could it be the language barrier?
Let me be clear.
If we measure THD at a given frequency for a simple (non-looped feedback system or one with the correct application of NFB) we can derive (as in calculating the 2nd, 3rd etc. order sum and difference products) from the HD Spectrum a very close approximation of what we will measure both in terms of total IMD and individual products for a signal with the same peak value as the single tone HD measurements and composed, though we must of course account for distortion of distortion.
What I did not suggest is that the measured values will be identical for THD and CCIF IMD measurements. Though with the tube and hybrid gear I design I find that SMPTE IMD and THD closely correlate, even at fairly low levels of (T)HD.
Also, if I measured 0.01% THD at 20KHz and then 0.5% CCIF IMD while using a signal with the same peak value than for the HD measurements I would be getting quite worried...
Ciao T
So is an 0.5dB difference in level. However, neither I nor anyone else listens to their music after passing it through a subtractive test. I am not as such interested in the result of just listening to the residual.
Moreover, in most cases we have a definition of distortion at rated power. With tube Amplifiers that have a simple monotonic distortion the distortion at normal listening levels will be considerably lower than the levels at maximum.
I doubt anyone would claim that 0.003% HD/IMD as such is audible. Yes are we only talking about HD/IMD? No.
I am unsure what you are attempting to say.
My point is simple.
1) Have the original undistorted music signal.
2) Have the same signal exposed to a distortion producing mechanism that produces 3% 2nd HD and 0.3% 3rd HD (and no higher order ones) at "full scale".
3) Have the same signal exposed to a distortion producing mechanism that produces 001% THD resulting only from crossover distortion (e.g. un-biased Class B Output Stage) at "full scale".
4) Have a large number of people compare the two resultant signals in a blind test, asking them to rate the three resulting signals according to how distorted they perceive each track, compared to the original unmodified track.
If 1 & 2 are rated as essentially identical with the control and 3 is rated highly distorted (which is how things will play out and have played out in a number of tests I am familiar with) it is safe to conclude that the distortion present in 2 is inaudible, while that present in 3 is audible (not that this breaks new ground).
This is not the same as saying that isolated from the original and scaled correctly in loudness the pure distortion residual will be inaudible. Nor is there is any point to establishing this, as now the human ear is not exposed to the original signal, which would produce the very distortion inside the ear that would have masked the distortion, nor is the actual object of the exercise (namely the music) still present.
So, your contention that the isolated distortion, without accompanying signal will be audible is correct, it is however equally meaningless, especially if we understand how the human hearing works... It is an answer looking for a relevant question, without finding one.
Again, you completely miss my point. Could it be the language barrier?
Let me be clear.
If we measure THD at a given frequency for a simple (non-looped feedback system or one with the correct application of NFB) we can derive (as in calculating the 2nd, 3rd etc. order sum and difference products) from the HD Spectrum a very close approximation of what we will measure both in terms of total IMD and individual products for a signal with the same peak value as the single tone HD measurements and composed, though we must of course account for distortion of distortion.
What I did not suggest is that the measured values will be identical for THD and CCIF IMD measurements. Though with the tube and hybrid gear I design I find that SMPTE IMD and THD closely correlate, even at fairly low levels of (T)HD.
Also, if I measured 0.01% THD at 20KHz and then 0.5% CCIF IMD while using a signal with the same peak value than for the HD measurements I would be getting quite worried...
Ciao T
Hi,
The Tubes in the (I believe it was a Mig-23) had some advantages at the time, when the planes where expected to be around nuclear strikes, which tend to fry solid state electronics.
I used to have a russian Raduga Colour TV (fitted with a Pal Decoder so I could watch west german TV). It used tubes in the various output stages driving the picture tube.
As a result, when you had a flashover, the TV just kept going. If the same thing happened to an all solid state east german colour TV that used the same picture tube the results where spectacular.
I was able to run the picture tube in my Raduga with a lot higher Anode voltage as a result, giving a much brighter and sharper picture...
Ah, yes, to "Silizium Tal" (Silicon Valley in German). One of the few post reunification industry successes...
Yes, it essentially about what happens when the amplifier "looses control".
Of course there are those who insist that this should never be allowed to happen.
Ciao T
The Tubes in the (I believe it was a Mig-23) had some advantages at the time, when the planes where expected to be around nuclear strikes, which tend to fry solid state electronics.
I used to have a russian Raduga Colour TV (fitted with a Pal Decoder so I could watch west german TV). It used tubes in the various output stages driving the picture tube.
As a result, when you had a flashover, the TV just kept going. If the same thing happened to an all solid state east german colour TV that used the same picture tube the results where spectacular.
I was able to run the picture tube in my Raduga with a lot higher Anode voltage as a result, giving a much brighter and sharper picture...
Ah, yes, to "Silizium Tal" (Silicon Valley in German). One of the few post reunification industry successes...
Yes, it essentially about what happens when the amplifier "looses control".
Of course there are those who insist that this should never be allowed to happen.
Ciao T
What always wondered me is the lack of tests that take care of the essence of the music: attack, decay, sustain and release are features of every single musical note. According to Encyclopaedia Britaanica: "Because many musical instrument families have similar spectra, there must be other factors that affect their tone quality and by which their tones can be distinguished."
What are the test that measure the distortion of these important time parameters?
In addition, others musical features are "Vibrato (a periodic slow change in pitch) and tremolo (a periodic slow change in amplitude)": what are the tests that check the fidelity of the audio device in reproducing such other musical characteristics? These are "transient" features, according to my thinking.
I don't know of any specific test on this subjects, but my knowledge is quite limited. Someboby here, maybe, could help to make light on these factors.
For "distortion" I mean, for example: difference in the attack and decay time parameters between input and output "transient" signal. I'm not speaking about rise time or slew rate here. But, this "distortion" can not be easily controlled, since the transient response is made by the sum of every single components of the audio chain: you could have a "transient perfect" amp, but when connected to a preamp the sum of the transient response of the two devices will be no longer be "perfect".
Last edited:
I believe it is important to think about what these music attributes mean in terms of signals to the amp. What is tremolo else than a variation in level? What is an attack else than a suddenly starting waveform? And 'suddenly' in music terms - assuming for instance that the music signal is limited to max, say 30kHz. If you would make an amp that has a full power BW of 30kHz without slew rate limiting, I can't see any mechanism that would 'distort' the attack.
I know that we don't listen to slew rate; we listen to music. But an amp 'has no idea' what music is. The amp amplifies electrical signals. We, as designers, need to specify and verify the electrical signal responses that are necessary and sufficient for an amp to faithfully reproduce music, within the audibility thresholds of the ear.
I am convinced that we can do that and that the best designers do it fully well. Some, like Bob Cordell, even build purpose-designed test equipment to test esoteric distortion mechanisms like Phase Intermodulation, to make sure that they are far below what can be reasonably be assumed to be audible.
Does that mean all amps sound the same? Unfortunately, no, they don't. But the best ones out there have no audible difference between their input and output signals.
Does every listener prefer that amp? Probably not; there's too much difference in preferences.
jan didden
Jan,
If this is the case, then all the best ones should sound all also identical to each other, right?
And, do they?
Ciao T
If this is the case, then all the best ones should sound all also identical to each other, right?
And, do they?
Ciao T
If this is literally the case then the amp is not an amp at all, its a piece of wire with no (or imperceptible) gain. So given this apparent incongruity, how did you determine that this was so - experimental details please?
You need a very fast signal for this to happen- faster than you are likely to see in a normal bandwidth limited audio signal. The cure is a well degenerated LTP It's quite easy to simulate this, or measure it directly with a wide bandwidth dual trace scope. A well designed amp will recover in us. You get more ringing and crap on your speaker cable than that.
Last edited:
You do perfectly know what I meant, and so does anyone else here who has an IQ larger than their shoe size. So why this statement?
jan didden
While I can't really see the 30Khz BW on most musical attacks, it seems like a safe number, by a wide margin.
But can a given amp really do that on complex loads across its bandwidth and at various power levels? For me, that's the crux of the problem. It's harder than it seems. And I don't see a lot of amps measured like that - good ones or bad ones. If we did, we might have a better basis to judge them on specs alone.
But it's not the case. However, many amps do sound more alike than different. Even an amplifier fetishist like me will admit that.
By "audible", I take it that loudspeakers have entered the equation. Which ones? Maggies? Puppies? Sound Labs? Headphones?
There's a thought, turn phase shift and slew audible, instead of the wave.
(think me just failed the mondopoint scale test)
Hi,
I agree. Incidentally also on the "more alike but different" bit.
Yet we have a dichotomy.
Jan, whom I do respect asserts that the best amplifiers show no difference between input and output (within reasonable bounds I would add), I think using what we call "null test", right Jan.
Yet you, I (and I suspect Jan as well) find that these amplifiers that produce no appreciable difference between input and output do appear to produce different sounds in our systems.
This is then either illusion, experimental error or magic, or we have something happening that is as of yet unexplained but real.
Ciao T
I agree. Incidentally also on the "more alike but different" bit.
Yet we have a dichotomy.
Jan, whom I do respect asserts that the best amplifiers show no difference between input and output (within reasonable bounds I would add), I think using what we call "null test", right Jan.
Yet you, I (and I suspect Jan as well) find that these amplifiers that produce no appreciable difference between input and output do appear to produce different sounds in our systems.
This is then either illusion, experimental error or magic, or we have something happening that is as of yet unexplained but real.
Ciao T
I can't speak for Jan, But I would say "No". That's too broad a brush.
I would say that many amps that measure the same THD and FR response (within reason) do not sound the same. But that's only because those two measurements are not the whole picture. That's old news.
The term of bandwidth limited function, does not imply that this function is slew-rate limited.
Ageev's theorem, proven in 1957, states:
arbitrary changing bandwidth limited function, has time periods of a limited duration, at which the function wil exceed any high slew-rate limit.
I already mentioned this theorem at various forums and threads. First reaction always was that it is impossible. But later on, many people accepted, that the slew-rate is simply a matter of probability, infinite number of harmonics, being sumed, can eventually lead to very high slew rates. I could imagine this, that there is crest-factor for amplitudes, and there is similar factor for slew-rates.
The square-wave test signal, has a high, but quite a limited slew rate, and I am aware of not so many amps, that could nicely do 200 kHz square wave. What is important, the amp should make minimum influence on the low-level signals, passing through the system at the moment of sharp transients.
Last edited:
Hi,
Sorry to be persistent, but either it is:
"All amplifiers that show no appreciable difference between input and output signals also sound the same"
OR
"Despite not showing appreciable difference between input and output signals, such Amplifiers CAN show different sound quality subjectively"
It is one or the other. No broad brush, a simple boolean.
Two statements, where, if one is true the other must be false.
Ciao T
Sorry to be persistent, but either it is:
"All amplifiers that show no appreciable difference between input and output signals also sound the same"
OR
"Despite not showing appreciable difference between input and output signals, such Amplifiers CAN show different sound quality subjectively"
It is one or the other. No broad brush, a simple boolean.
Two statements, where, if one is true the other must be false.
Ciao T
This is indeed important. Music may be very complex, but it's just a single value level at any one moment. The complexity comes from the rapid changes of that momentary signal level. If we say, that a signal is bandlimited to 30kHz, there are NO components in that signal that are 'faster' than 30kHz (lets assume for the sake of this discussion a 30kHz brick-wall low pass).
So, when my amp can reproduce anything up to 30kHz, full power, with high linearity, what else is there to worry about?
Now the interesting thing becomes: how can we verify that the amp does reproduce anything up to 30kHz full power with high linearity. But that's a manageable engineering problem. That's what engineers are good at: you give them a target, they get there.
jan didden
- Status
- Not open for further replies.