I recently finished building my first EL34 single ende tube amp. I must say it sounds rather good. I dare to say it sounds really good. I have had many amps and speakers and also worked in a hifi store for several years. Last weekend I went to a friend of my friend's father who used to be a professional audio technician for several decades and has worked with several great artist along the way. So if one person knows sound, it must be him. He also builds spuikers for a living.
When I took the amp to his place, he was very impressed with the sound. Ofcourse the low end lacks in comparison with a powerfull solid state, but he very much liked the openess, transparency and detail my amp produced. He also said how easy it is to listen to. Like no listening fatigue at all. Later in that day a friend of him als visited (also audiophile) and confirmed what he heard. He was also very impressed with my amp. My friend (also a musician) was there also and was very impressed by the sound and noted how clean and free of distortion it sounded. He said the music has a lot of body to it and the instrument seperation is stellar. He did not want to stop to listen to it. What had to be a 1 hour demonstration turned to a 3 hour listening session.
The square waves look absolutely awfull on the scope. They are something like the picture below. Note that this is not the actual screenshot of my scope. I have no picture of my own scope and am writing this post from outside of home. But the square wave looks like this across almost all of the frequency range. It is absolutely symmetrical. There is less ringing though in my case.
What the... &*^!!8768?!?!!?
Like how is this even possible? I would expect a ton of ringing and simple bad sound being produced by my amp. It should induce a ton of fatigue and distortion to the sound based on the measurements on my scope, yet it yields very good results while actully listening to it.
The wrong way to measure
Measuring the voltage swings on the speaker output posts either with an artifical load (resistor) or a real speaker load. We would only measure the voltage swing, but not the actual pressurewave we as humans perceive.
The right way to measure
Place a microphone in the room where your listening position is. Ideally the microphone would have the same characterisics as the human ear. A "perfect" micrphone would still yield unusable results as it would posess super human capabilitys and thus yieliding inacurate results as to simulate what our ears are hearing. We will capture the pressurewave and the actual way our ears perceive it and thus yielding an accurate representation of the amps and speaker's performance.
Why does the apparent overshoot sound good in real life?
My theory is that it has to do with the way we are converting energy. All energy in the universe transmutes. In our case we are transmuting electrical energy into kinetical energy by changing the magnetic flux in the speaker's magnet. Because the movement of a speaker includes a time to speed up and time to slow down, this already causes distortion in the sound. An ideal speaker would move in a instant. Ideally in planck time. But this is not possible in practice.
But what I think happens is that the oveshoot, which is a higher voltage, makes the speaker move faster on the initial movement therefore causing it to move more "accurate" thought the amp itself does not produce a correct signal.
I like to see it as follows: Whe we want to accelerate a car from 0 - 60 Mph we would need some engine power. Imagine my car is able to produce 100 Hp. If my goals were to be to reach the 60Mph as accurately as possible, in ideal circumstances I would only need to appy lets say 20% throttle to achieve it in 3 seconds. But because of air resistance and other factors I need to apply 100% of throttle to get there in 3 seconds and that go back to 20% to maintain. Though I have a "overshoot" of throttle, my real life performance would be close to a square wave than what my throttle input actually is.
Should we than throw the old way of measuring out?
No. Absolutely not as it will still provide valuable insight in the technical limitations of an amplifier. For example when it starts to really clip. And it will not make you deaf when you use an artificial load (resistor)
So what do you think? Is thare any sense to my theory? Am I missing something? Or could I be onto something?
Let me know!
P.S. Any square wave above 10Khz ~ish is a sine anyway due to the fact use a DAC at 44Khz sampling rate..
When I took the amp to his place, he was very impressed with the sound. Ofcourse the low end lacks in comparison with a powerfull solid state, but he very much liked the openess, transparency and detail my amp produced. He also said how easy it is to listen to. Like no listening fatigue at all. Later in that day a friend of him als visited (also audiophile) and confirmed what he heard. He was also very impressed with my amp. My friend (also a musician) was there also and was very impressed by the sound and noted how clean and free of distortion it sounded. He said the music has a lot of body to it and the instrument seperation is stellar. He did not want to stop to listen to it. What had to be a 1 hour demonstration turned to a 3 hour listening session.
NOW HERE COMES THE CATH:
Yesterday I wanted to measure my amps distortion and also measure some other things. I was baffled by the results. I attached my prongs to the speaker output posts with a real speaker load.The square waves look absolutely awfull on the scope. They are something like the picture below. Note that this is not the actual screenshot of my scope. I have no picture of my own scope and am writing this post from outside of home. But the square wave looks like this across almost all of the frequency range. It is absolutely symmetrical. There is less ringing though in my case.
What the... &*^!!8768?!?!!?
Like how is this even possible? I would expect a ton of ringing and simple bad sound being produced by my amp. It should induce a ton of fatigue and distortion to the sound based on the measurements on my scope, yet it yields very good results while actully listening to it.
My theory:
My theory is that we are measuring the amplifiers performance all wrong this whole time. Why? Because we are not measuring what our ears hear! Simple as that. We have to think of another way to properly measure what we as humans are actually hearing, and that is the pressurewave produced by the speaker, and not the voltage swings that are going into the speaker.The wrong way to measure
Measuring the voltage swings on the speaker output posts either with an artifical load (resistor) or a real speaker load. We would only measure the voltage swing, but not the actual pressurewave we as humans perceive.
The right way to measure
Place a microphone in the room where your listening position is. Ideally the microphone would have the same characterisics as the human ear. A "perfect" micrphone would still yield unusable results as it would posess super human capabilitys and thus yieliding inacurate results as to simulate what our ears are hearing. We will capture the pressurewave and the actual way our ears perceive it and thus yielding an accurate representation of the amps and speaker's performance.
Why does the apparent overshoot sound good in real life?
My theory is that it has to do with the way we are converting energy. All energy in the universe transmutes. In our case we are transmuting electrical energy into kinetical energy by changing the magnetic flux in the speaker's magnet. Because the movement of a speaker includes a time to speed up and time to slow down, this already causes distortion in the sound. An ideal speaker would move in a instant. Ideally in planck time. But this is not possible in practice.
But what I think happens is that the oveshoot, which is a higher voltage, makes the speaker move faster on the initial movement therefore causing it to move more "accurate" thought the amp itself does not produce a correct signal.
I like to see it as follows: Whe we want to accelerate a car from 0 - 60 Mph we would need some engine power. Imagine my car is able to produce 100 Hp. If my goals were to be to reach the 60Mph as accurately as possible, in ideal circumstances I would only need to appy lets say 20% throttle to achieve it in 3 seconds. But because of air resistance and other factors I need to apply 100% of throttle to get there in 3 seconds and that go back to 20% to maintain. Though I have a "overshoot" of throttle, my real life performance would be close to a square wave than what my throttle input actually is.
Should we than throw the old way of measuring out?
No. Absolutely not as it will still provide valuable insight in the technical limitations of an amplifier. For example when it starts to really clip. And it will not make you deaf when you use an artificial load (resistor)
So what do you think? Is thare any sense to my theory? Am I missing something? Or could I be onto something?
Let me know!
P.S. Any square wave above 10Khz ~ish is a sine anyway due to the fact use a DAC at 44Khz sampling rate..
Last edited:
I'm not technically versed enough to actually address your theories, but based on my own experience I do have a few thoughts:
1) That square wave isn't actually that bad. I've seen much worse. ;-)
2) A little overshoot on the leading edge is fine, in fact preferable to my ears.
3) Ringing is a different matter and speaks to stability. If your little amp uses feedback you might want to tame the ringing with some tweaks to the compensation network. If your amp sounds good now, it will sound even better if it's stable, especially in the super-sonic frequencies.
4) Distortion is overrated. ;-) My homebrew 20wpc Williamson monoblocks have .3% THD at 1 watt and 10% THD at 20 watts. In other words, they are exactly in line with the way an original Williamson was expected to perform. They are endlessly listenable, never ever fatiguing. Perhaps a more telling measurement is IMD, which has more effect, I think, on listenability, in terms of clarity and accuracy.
In terms of how we measure amps, honestly, this is an old argument that goes back to John Atkinson's measurements of amps like the Cary 300B and Wavelength, wherein he scratches his head and says that despite the measurements, the amp sounds great. Actually, it goes back further than that. A lot of the "golden age" magazine articles from the 1950's discuss the disparity between measurements and listenability, and those guys were hardcore EEs. Generally speaking, if an amp gets the midrange right, a lot of other sins can be overlooked.
1) That square wave isn't actually that bad. I've seen much worse. ;-)
2) A little overshoot on the leading edge is fine, in fact preferable to my ears.
3) Ringing is a different matter and speaks to stability. If your little amp uses feedback you might want to tame the ringing with some tweaks to the compensation network. If your amp sounds good now, it will sound even better if it's stable, especially in the super-sonic frequencies.
4) Distortion is overrated. ;-) My homebrew 20wpc Williamson monoblocks have .3% THD at 1 watt and 10% THD at 20 watts. In other words, they are exactly in line with the way an original Williamson was expected to perform. They are endlessly listenable, never ever fatiguing. Perhaps a more telling measurement is IMD, which has more effect, I think, on listenability, in terms of clarity and accuracy.
In terms of how we measure amps, honestly, this is an old argument that goes back to John Atkinson's measurements of amps like the Cary 300B and Wavelength, wherein he scratches his head and says that despite the measurements, the amp sounds great. Actually, it goes back further than that. A lot of the "golden age" magazine articles from the 1950's discuss the disparity between measurements and listenability, and those guys were hardcore EEs. Generally speaking, if an amp gets the midrange right, a lot of other sins can be overlooked.
What is the rise-time of the square wave, how many times does the output ring, and what is the frequency of the ringing?
It matters because real music is not necessarily fast rise-time square waves. Thus real music may not excite any ringing at all. What you may find is that if you use swept sine wave as a test signal then you will probably find that at some high-ish frequency the amplifier gain starts to increase (need to watch the input signal amplitude too in case your signal generator output isn't constant with frequency). As you continue to raise the sine wave frequency at some point a peak output amplitude will be observed. Any further increase in frequency beyond that point should start decreasing in amplitude. (maybe safer to do this a low power)
So, it depends on where the frequency response starts increasing. Is it in the audio band or well above the audio band? Also, how much is the gain increase at the peak?
EDIT: One reason an amplifier might not ring with a resistive load, but show ringing with a speaker load can be because of speaker cable capacitance. Also, since its a tube amp it could be the output transformer that is ringing.
Last edited:
I'm not sure I can answer the question. ;-) I assume it's a question of slightly extending the HF response, transient response or just subjective taste. I take Dave Gillespie's work over at Audiokarma as a guideline and he generally leaves a little overshoot at the leading edge when correcting amplifier issues.
I do have a schematic if you want, but I am not looking for fixing it. I am trying to understand as to why it is the overshoot that people are actually looking for. I did some further research and it seems to have to to with harmonics which would make sense to me.
It looks about the same in a resistive load.
I see multiple comment on the fact that apparently it is this very overshoot that has to do with “pleasant” sound.
I have made a screenshot from my digital scope. It shows no ringing at all, only the overshoot.
This is a screenshot of a 400Hz square wave signal into my speakers measured at the speaker taps. My analog scope shows the exact same waveform.
So what does this overshoot show. Amount of harmonics?
I have made a screenshot from my digital scope. It shows no ringing at all, only the overshoot.
This is a screenshot of a 400Hz square wave signal into my speakers measured at the speaker taps. My analog scope shows the exact same waveform.
So what does this overshoot show. Amount of harmonics?
Attachments
Too much treble and or not enough bass. This assumes you are using 10x probes and they have been calibrated with the scope's calibration square wave.
I found some resistors for a resistive load. Did some testing and the results are below. This is a 1000Hz square wave. It is quite different to the wave generated when coupling speakers.
Switching between negative feedback or not makes absolutely no difference.
I decided to check the DAC with headphone and line output lying around and decided to use them also. And guess what. The ringing patterns are identical. Only difference being the HF being louder in my amp. So that might be something I might try to fix to get a better balanced tone. Please check my screenshots.
8590 is taken from the amp. 8591 is taken directly from the dac.
Attachments
Last edited:
Interesting- i figured it was from the speakers. Poorly damped voice coil resonance? Are these some open baffle high-efficiency types?
Personally, in my experience significant overshoot sounds strident or fatiguing, just a small bit or slightly rounded leading edges is ideal. I don't think I've ever heard of it referred to as a source of pleasant sound.
Personally, in my experience significant overshoot sounds strident or fatiguing, just a small bit or slightly rounded leading edges is ideal. I don't think I've ever heard of it referred to as a source of pleasant sound.
A complete and accurate schematic is worth 1,000 words, and at least 1 hour of analysis of the possible effect(s).
Please post a schematic.
The loudspeaker you used is not listed, but I bet that it is like most loudspeakers:
Various impedances, perhaps from 6 Ohms to 25 Ohms, depending on the frequency.
And, its impedance is Resistive, Capacitive, Inductive, LCR, LR, and CR, depending on the frequencies.
The load line is straight, circular, and elliptical, at the different frequencies (All of those).
Resonance is present, and will make most amplifiers appear to "ring" at those frequencies.
Expect that when the amplifier drives a loudspeaker.
Go back and send the square wave to the amplifier, but load the amplifier with an 8 Ohm Non-Inductive Load resistor. Then report what the square wave shape looks like.
Test at 100Hz (will have a fast rise, and then have a curved downward slope Test at 1kHz, and test at 10kHz.
Using a fast rise 10kHz square wave generator, then . . .
If you see some moderate Ring" that is 20us from ring top to ring top, then . . .
Congratulations! You have 'ringing' at 50kHz, from the excess frequency spectra of the fast rise square wave generator that includes more than 50kHz.
A Redbook CD player output signal should not have more than 22.05kHz; there is no CD 50kHz signal to "activate" the 50kHz "ringing", so no 50kHz "ring", not on the 8 Ohm load, and most likely no 50kHz on the loudspeaker either.
After you post a schematic, we can look to see if there is something obviously wrong with the amplifier (I suspect there is nothing wrong).
Then, you will be able to bring your amplifier to a local audio club for a competitive amplifier listening shootout.
Sleep well at night, the arts and sciences of signal sources, amplifiers, and loudspeakers are generally well known.
"Enjoy the Music!" (with thanks and respect to the magazine reviewer who first wrote those wise words).
Please post a schematic.
The loudspeaker you used is not listed, but I bet that it is like most loudspeakers:
Various impedances, perhaps from 6 Ohms to 25 Ohms, depending on the frequency.
And, its impedance is Resistive, Capacitive, Inductive, LCR, LR, and CR, depending on the frequencies.
The load line is straight, circular, and elliptical, at the different frequencies (All of those).
Resonance is present, and will make most amplifiers appear to "ring" at those frequencies.
Expect that when the amplifier drives a loudspeaker.
Go back and send the square wave to the amplifier, but load the amplifier with an 8 Ohm Non-Inductive Load resistor. Then report what the square wave shape looks like.
Test at 100Hz (will have a fast rise, and then have a curved downward slope
Test at 1kHz, and test at 10kHz.Using a fast rise 10kHz square wave generator, then . . .
If you see some moderate Ring" that is 20us from ring top to ring top, then . . .
Congratulations! You have 'ringing' at 50kHz, from the excess frequency spectra of the fast rise square wave generator that includes more than 50kHz.
A Redbook CD player output signal should not have more than 22.05kHz; there is no CD 50kHz signal to "activate" the 50kHz "ringing", so no 50kHz "ring", not on the 8 Ohm load, and most likely no 50kHz on the loudspeaker either.
After you post a schematic, we can look to see if there is something obviously wrong with the amplifier (I suspect there is nothing wrong).
Then, you will be able to bring your amplifier to a local audio club for a competitive amplifier listening shootout.
Sleep well at night, the arts and sciences of signal sources, amplifiers, and loudspeakers are generally well known.
"Enjoy the Music!" (with thanks and respect to the magazine reviewer who first wrote those wise words).
The speaker I have are of the Piega P4L type. It is a two-way speaker with a ribbon tweeter. The efficiency is say 89db, so nothing “high efficiency” when compared to some Klipsch ones. The ribbon tweeter might be what causes the overshoot during the real load maybe as it happens in the higher frequency domain where the tweeters are active.
Some further questions:
1. Can we state that my amp produces no significant overshoot into the resistive load therefore being a “good” amp. It is only when we connect to the speakers that the overshoot happens, but there is no fatigue when listening therefore being okay in real use environment.
2. How do I increase the bass to be level with the higher frequencies haha.
Some further questions:
1. Can we state that my amp produces no significant overshoot into the resistive load therefore being a “good” amp. It is only when we connect to the speakers that the overshoot happens, but there is no fatigue when listening therefore being okay in real use environment.
2. How do I increase the bass to be level with the higher frequencies haha.
Roger Wilco! The ringing might be something I cannot get out of the equation as of now since it also shows when coupling to my DAC. I have no dedicated sine wave generator (yet).A complete and accurate schematic is worth 1,000 words, and at least 1 hour of analysis of the possible effect(s).
Please post a schematic.
The loudspeaker you used is not listed, but I bet that it is like most loudspeakers:
Various impedances, perhaps from 6 Ohms to 25 Ohms, depending on the frequency.
And, its impedance is Resistive, Capacitive, Inductive, LCR, LR, and CR, depending on the frequencies.
The load line is straight, circular, and elliptical, at the different frequencies (All of those).
Resonance is present, and will make most amplifiers appear to "ring" at those frequencies.
Expect that when the amplifier drives a loudspeaker.
Go back and send the square wave to the amplifier, but load the amplifier with an 8 Ohm Non-Inductive Load resistor. Then report what the square wave shape looks like.
Test at 100Hz (will have a fast rise, and then have a curved downward slope
Using a fast rise 10kHz square wave generator, then . . .
If you see some moderate Ring" that is 20us from ring top to ring top, then . . .
Congratulations! You have 'ringing' at 50kHz, from the excess frequency spectra of the fast rise square wave generator that includes more than 50kHz.
A Redbook CD player output signal should not have more than 22.05kHz; there is no CD 50kHz signal to "activate" the 50kHz "ringing", so no 50kHz "ring", not on the 8 Ohm load, and most likely no 50kHz on the loudspeaker either.
After you post a schematic, we can look to see if there is something obviously wrong with the amplifier (I suspect there is nothing wrong).
Then, you will be able to bring your amplifier to a local audio club for a competitive amplifier listening shootout.
Sleep well at night, the arts and sciences of signal sources, amplifiers, and loudspeakers are generally well known.
"Enjoy the Music!" (with thanks and respect to the magazine reviewer who first wrote those wise words).
Lets consider the wave-shape of the 400Hz square wave of Post # 10.
The spectrum of a 400Hz square wave, the amplitudes of the fundamental and all the odd harmonics: is 1 @ 400Hz, 1/3 @ 3x 400Hz = 1200Hz **, 1/5 @ 5x 400Hz = 2000Hz, etc.
That "ring" looks like a half cycle of 1200Hz **. No surprise to me.
A typical 2 way loudspeaker may have a slightly broad high impedance from about 1kHz to 2.5kHz, that is resonant in that frequency range.
Post a schematic, please.
The schematic allows us to analyze the circuit, estimate the damping factor, look at the different negative feedback path(s), which tube will clip first (like driver or output tube), etc.
The spectrum of a 400Hz square wave, the amplitudes of the fundamental and all the odd harmonics: is 1 @ 400Hz, 1/3 @ 3x 400Hz = 1200Hz **, 1/5 @ 5x 400Hz = 2000Hz, etc.
That "ring" looks like a half cycle of 1200Hz **. No surprise to me.
A typical 2 way loudspeaker may have a slightly broad high impedance from about 1kHz to 2.5kHz, that is resonant in that frequency range.
Post a schematic, please.
The schematic allows us to analyze the circuit, estimate the damping factor, look at the different negative feedback path(s), which tube will clip first (like driver or output tube), etc.
wg_ski,
It is not amplifier overshoot, it is not amplifier instability.
I have measured many simple tube amplifiers that look like post #10.
They look essentially perfect on a load resistor, but just like that on a typical 2-way loudspeaker.
Especially amplifiers that do not have global negative feedback, and that have a low to moderate damping factor.
The intentional filtering is in the 2-way loudspeaker might be a simple 6dB/octave crossover at about 1200Hz, as a real good example.
Sometimes, it is a resonance in the bass/mid driver, even if there is no bass/mid crossover.
Your Milage May have Varied.
Just my opinions and experience.
It is not amplifier overshoot, it is not amplifier instability.
I have measured many simple tube amplifiers that look like post #10.
They look essentially perfect on a load resistor, but just like that on a typical 2-way loudspeaker.
Especially amplifiers that do not have global negative feedback, and that have a low to moderate damping factor.
The intentional filtering is in the 2-way loudspeaker might be a simple 6dB/octave crossover at about 1200Hz, as a real good example.
Sometimes, it is a resonance in the bass/mid driver, even if there is no bass/mid crossover.
Your Milage May have Varied.
Just my opinions and experience.
Last edited: