Different amps have different frequency response. It may measure flat using an 8 ohm resistor, but with a speaker load it is no longer flat. Higher impedance at certain frequencies make some amps output higher voltage (louder). Therefore, different amps sound different. Frequency response is king.
It has nothing to do with harmonic distortion, which humans cannot really hear according to a lot of research done on the topic. So this topic is really not so mysterious. The fact that it there is so much debate over it when it is easily explained never made much sense to me.
It has nothing to do with harmonic distortion, which humans cannot really hear according to a lot of research done on the topic. So this topic is really not so mysterious. The fact that it there is so much debate over it when it is easily explained never made much sense to me.
Yes. Valve amps (and simple class-A solid state amps), tend to have higher output impedance. The way they handle the speaker load is unique. This delivers unique, typical sound signature. When this is "right", many people will like/prefer this bass signature. I did in the past, now I don't.
I think there are still many that have not been explained.
One feature of valve amps (and simple class-A amps) is the dominant H2 distortion. I like this in the past. Because of the "tube/sweet" signature and the ability of the H2 to mask other problems in the sound. But now, once the other problems are solved, the H2 becomes liability.
One example, where it is not yet explained imho, is related to the HF garbage beyond 20kHz (also related to very high order distortion). Theoretically this is inaudible. But in my experience, there is effect, directly or not...
And BTW, you know my hypothesis... And my opinion regarding distortion thresholds for human mind/brain.
Last edited:
When we talk about valve amp difference with solid state:
(1) Remove H2 differences.
(2) Remove Zout differences.
(3) Remove high order distortions, by comparing class-A to class-A (so no crossover distortion as well). And as you noted from past research, only a few people can differentiate low distortions in an ABX, but my hypothesis says that it may affect people unconsciously.
(4) Soft clipping?? I have no experience. I don't believe this will have big effect. But I don't know why people have too much questions but have too little energy to find the answers (by doing experiments).
(5) Remove instability due to wider bandwidth?
I think I need this explained: What is the effect of circuit simplicity to sound?
As we know, valve amps are almost always simple circuits. There is always that unique "transparency" in a simple circuit. But where does this "transparency" come from? What variables are affected by simple circuit? Phase?
The test: simple circuit versus complex circuit (multiple stages) with equal distortion and noise. Let's check the measured differences, and check with ABX if there is audible differences (not using general ears but one with good ears).
Last edited:
True. Some don't aim for hi-fi.cotdt said:
True. Some don't aim for hi-fi.
True, partly. It is unsurprising that amps which don't aim for high quality sound reproduction will sound different from each other. One of the signs of high quality in reproduction is that such amps will all sound very similar to each other.
It is true that frequency response and harmonic distortion are separate issues. We can hear harmonic distortion, but we are less sensitive to it than to minor frequency response problems. The big snag is that you can't normally have some harmonic distortion without also having some intermodulation distortion - and that we can hear more clearly than harmonics. We often measure harmonics because they are easy to measure, but we understand that in doing so we are also estimating IMD levels.
There are lots of things in audio which are not mysterious, yet people either cannot or will not believe them. Examples are: all competent hi-fi amps sound very similar, all competent cables sound the same (i.e. no sound at all!), bits are bits.
There is no simple relationship between apparent circuit simplicity and perception of sound. Some people fondly imagine that a 'simple' circuit (i.e. one they consider simple - which probably means few components) must soud better; others believe that a 'complex' circuit (i.e. one they consider complex - probably lots of components) must be needed to achieve low distortion (or whatever). I believe that they are both wrong. What is needed is a simple relationship between input and output, and this can be achieved in lots of different ways. We want the voltage relationship to be linear, and the frequency response to be flat; these are simple requirements.Jay said:
Yes, but what happen when we can hear down to microscopic level? Very similar is not the same with same or equal. In this case, I don't need to believe anything...
Big question for you: What do you think of CFP output stage? Do you think it is possible to design a competent hi-fi amp with that topology?
Yes, that sounds very logical. But don't you think that everything is simple when we look at it from macroscopic view? Let's take a look at line level circuits. FR is not an issue. Linearity? It is imo not that simple. If only anyone is interested to conduct a test... (My hypothesis is that you cannot enforce perfect linearity when you use a lot of stages).
That is why I said "similar". Obviously, the closer the amps get to perfection the more they will sound alike.Jay said:
The CFP output stage can be quite good, provided it can be made stable. Using the right quiescent current can be important - most people do not realise how low this should be. The crossover region can be a bit bumpy, but in my view a well-designed CFP running at the right quiescent current can be used for hi-fi. I'm not sure why you think this is a "big question".
I don't understand your comment.
Perfect linearity is impossible, so in that sense you are right. Sufficiently good linearity can be obtained by using a sufficient number of stages, probably combined with some sufficiently well-designed negative feedback.
Can you give an example of a competent hi-fi amplifier with CFP output?
Because I can hear better than most people. CFP amps sound different than most hi-fi amps, so according to your definition, CFP amps are not hi-fi...
But I have never been disappointed with CFP amps. It is "different" but likeable. Things that make you think that, once you can see the big picture, and realize that hi-fi is "so far in the sky", audio is just about preference...
No. Neither can I give an example of a competent hi-fi amplifier with Darlington output or any other solid-state output topology. The reason is quite simple: for some years I have only been interested in DIY amplifiers, and only valves.Jay said:
Not necessarily. Two issues: was the CFP quiescent current set correctly; what if the CFP ones are hi-fi and all the rest not? The correct quiescent current for a CFP output typically lies in the 5-20mA range, which is far lower than many people realise. Douglas Self found this and wrote about it in Wireless World and his books; I subsequently explained it in an article for WW - my simple formula accurately predicted his numerical results. My guess is that many CFP amps are set with far too high a current; perhaps partly due to 'first watt' ideas, and parly due to ignorance of what is required.
When you say that CFP amps sound different, I assume you mean that they sound different to you. To what extent do they sound different before you realise that you are hearing a CFP amp?
Audio is certainly about preference - that is, using my definition of 'audio' which is when people want a certain sound rather than reproduction. Genuine hi-fi is not up in the sky; it is being routinely delivered. We know this because almost nobody can distinguish between different hi-fi systems without peeking. Note that hi-fi does not mean that nobody can tell any difference, because there will always be a few people with exceptional hearing - although not the thousands who fondly imagine that they have exceptional hearing, but inconveniently lose this skill when properly tested.
Aint all valve amps have high THD??? If THD is your measure for hi-fi, then your amp is not hi-fi then?
That's why I asked for an example. That will quantify qualitative statements.
You believe that all hifi amps should sound similar. Yes I agree, that is the case. But no CFP amps I have built or heard fall into the "similar" group.
Most of my listening experiences are not controlled nor blind. But I'm sure that my ears do not trick me because I'm also experienced with blind listening. I didn't say that I know the characteristics of CFP amps sound. I said that CFP amps sound different with other hifi amps, in blind listening.
Do CFP amps have higher output impedance? I don't know about this, but I think that from the sound I perceive, it may have high Zout. And then may be slightly more linear? But how can it be if the quiscent current is so low??? Then the last one, may be slightly more second order distortion???
But you are aware that people cannot distinguish between THD 1% and THD 0.001% without peeking. So by that fact, all of them should be equally hi-fi.
But I'm curious about your valve amps. How genuine they are in sound reproduction?
Hmm, maybe. For decades, for example, the scientifically ideal headphone response was dead flat until we eventually learned otherwise. Until that revelation anyone vigorously expressing a preference for headphones with massive upper midrange emphasis they would have been ridiculed as unscientific. In hindsight it's an incredibly obvious oversight.
IMO still much to be learned.
actually, I was thinking of the 2004-2009 "reboot*" - in my opinion James Edward Olmos' singular best in a great body of work, and James Callis gave far more nuance to a previously one dimensional caricature than would comfortably fit into our preconception of "the bad guy" .
*writers did a masterful job of creating several additional layers to the meaning of that word.
This was definitely appointment TV for a while, - too bad the Caprica prequel "failed to launch" as well - I guess they just got too creative for their own good?
There's something almost eerie about vacation walkabouts on the Vancouver False Creek and Coal Harbour waterfront, or the amazing Sun Yat-Sen memorial garden and thinking - isn't that were ...... happened?
As to the original thread topic - I've built / owned / heard enough thermionic valve audio devices in the last 20 years to realize that they can sound very good, or not - but a well designed SS / Chip / class D amp can certainly cost a lot less both to own and to operate. All the 5 or so tube amps in my current possession are "resting" in storage, or in state of incomplete "renovation"
Last edited:
That depends on what you mean by "high". Valve amps don't usually have zeros after the decimal point in their THD figure, but few people believe that zeros are necessary.Jay said:
Where did I say that THD is my measure for hi-fi?
People will form a judgement about how much they trust your ears.
If CFP sound different from others, and different from each other, then maybe those CFP amps are not hi-fi. This does not mean that CFP cannot be hi-fi; merely that those particular samples are not hi-fi.
I know of no fundamental reason why a CFP output should have higher output impedance than a Darlington. It is a while since I looked at this, but it could be that CFP has lower output impedance. However, depending on exactly how it is driven, it is conceivable that a CFP may have a rather non-linear output impedance.
Why should using the appropriate quiescent current for an output architecture cause nonlinearity? If people building CFP outputs think like you then they may all have the wrong quiescent current and so have higher crossover distortion than necessary.
No, too much current is more likely to give odd order distortion.
I'm not sure about 1%, but anything much less than this is likely to be indistinguishable. However, it would depend on the order of the distortion and the accompanying IM. 0.5% 2nd order may be indistinguishable from 0.05% 2nd, but I suspect that 0.5% 7th order may sound worse than 0.05% 7th. That is the weakness of THD, which is why I don't think I mentioned THD.
Sufficiently genuine to satisfy me. I am not in a position to judge how well they would satisfy others, including you. However, I am not one of those who pursues 'tube sound' - I want my valves to be inaudible.
As headphones unavoidably modify the acoustic properties of the ear it is not surprising that some compensation for this is needed. That is not an excuse for some modern expensive headphones which make no attempt at serious sound reproduction. Some people may prefer them, but for sound reproduction they are not 'better'.rdf said:
A couple of peevy pets of mine:
(sorry, I'm probably just crotchety this morning... I found yesterday that a driver I'm designing into a speaker system isn't going to be easily obtainable anymore)
Why do people talk about phase like it's a mysterious unknowable effect? Phase response is TRIVIALLY measurable these days, there's no sense in implying that maybe some obscure effect is causing a "phase problem", phase is as easy as magnitude to measure. Didn't use to be in the 70's maybe, but that was a long time ago.
This kind of statement (which is not at all rare) makes little sense except when listening to pure sine waves. THD (harmonic distortion) is a function of signal level of a sine wave. Music waveforms cover quite large ranges of level and are not simple sine waves, so I don't know how you could even arrange to listen to music "at 1% distortion". You can listen to music on an amp that is rated at 1% distortion at a specified power level (with a sine wave , and nothing but, again) but the actual distortion you might measure to get there will be different at different power levels. If you listen to an amp rated at 5% at what the manufacturer calls it's max output, and another rated at 0.1% at its declared power, there is nothing to say which will have the closer waveform accuracy at the much lower average power levels that you'd be listening to music on. The rated distortion is usually assumed to be the highest a sinewave would cause up to the declared power so you could think that things will be clean below that peak level. But rating a higher distortion to a higher declared power (for example) could even be with the exact same amp!
(sorry, I'm probably just crotchety this morning... I found yesterday that a driver I'm designing into a speaker system isn't going to be easily obtainable anymore)
Last edited:
FWIW, I used to count myself in the "lower than .05% (pick your favorite number) hardly matters camp, but recently had a chance to listen in my own system to a pair of SS amps using the same chip ( LM3886) by one of our forum members - Tom Christiansen - that disabused me of a couple of my steadfast "facts".
One design has a rather elaborate feedback topology, while the other is dare I say, a bit more "pedestrian"?. Both have distortion and noise levels that by standard tests are excellent, and my previous "conventional wisdom" was that after 3 decimal places of distortion, -85dB noise etc., it'd be academic - but there was no question as to which I preferred the sound of. My audiophile random analogy generator is currently dysfunctional, so I won't bother with the purple prose, but I guess that made one of them "better" ?
One design has a rather elaborate feedback topology, while the other is dare I say, a bit more "pedestrian"?. Both have distortion and noise levels that by standard tests are excellent, and my previous "conventional wisdom" was that after 3 decimal places of distortion, -85dB noise etc., it'd be academic - but there was no question as to which I preferred the sound of. My audiophile random analogy generator is currently dysfunctional, so I won't bother with the purple prose, but I guess that made one of them "better" ?
- Status
- Not open for further replies.