Amplifier harmonics and load stability

[A.S.]

There are repeated claims made about the sonic abilities of certain amplifiers, around which folklore develops and legends are born. Scientifically structured blind listening tests conducted over the years suggest that, within their specified operating power ability (which must be somewhat speaker load dependent according to circuit design) when amplifiers are compared like-for-like these sonic differences vanish.

A couple of years ago I measured one channel of three power amplifiers driving a certain speaker load. The amps didn't have any volume controls, so their gain or loudness was fixed by the internal circuitry alone. I can't recall what the speaker was, but if it was one of ours it was certainly a benign, easy load. The graph attached shows variations in voltage at the output terminals of three power amplifiers (blue, black and red) across the 20Hz to 20kHz audio band (horizontal scale). The vertical scale shows the voltage in dB. A perfect amplifier would be a perfectly flat line from left to right raised to some vertical scale marker. For example, the black amp has a fixed gain of about 39dB and it is as flat as you could wish for (or need) except the slightest and utterly inaudible droop at the very highest frequencies.

The black and red traces tell a different story. Both black and red are one channel of two QUAD 405 (or 405-2? can't recall) power amps. The black amp shows a very slight (perhaps +/- 0.5dB) variation from a perfectly speaker-independent flat line and the gain is slightly less than the blue amp at about 37dB. So, if the blue amp was exchanged for the black amp it would sound about 2dB louder and this alone would be more than enough for the listener to attribute some marvellous extra 'punch', 'dynamics' or 'rhythm and pace' to the blue one. All of that would be expected for an amp that is somewhat louder. And if we started off with the blue amp and switched to the quieter black one, we'd probably note that the black amp was 'thinner', 'softer', 'has less guts' ... even 'more detailed, refined'.

The red trace is a warning to us all. First the overall sensitivity is lower than the black sister 405. This one is down at about an average of 35dB. That means it is about 2dB quieter than the other 405 and a whopping 4dB quieter than the blue amp for the same music input signal. For no more than the reasons of the difference in loudness between these three specimens they will definitely sound different. Of that I am certain. But what of the red trace, it's far from flat. Given that it's average loudness is about 35dB (a visual guess) across the frequency band for the very same speaker load it is about +/- 2dB either side of that 35dB line. That's a very big variation indeed. Since the black and blue curves are the very same brand, model and age of amplifier how is this possible? It's possible because the red amp is faulty - it still works and may continue to do so for some years with ever more significant change in performance until it fails. And then it may destroy the speakers going 'DC'.

As the black and red amps are about the same age, how would you know that one was faulty from the serial numbers? You wouldn't. You could cheerfully buy on on Ebay and it would be pot luck. Fortunately QUAD can still service these and presumably could pick up this fault ... but there again, maybe not as it is rather subtle and in the lab, they may well test the amp with a fixed resistor, not a real speaker load.

Conclusion: amplifiers can sound different solely for reasons of ageing. And as I've said before, get your amp checked regularly by a qualified engineer, best of all the makers themselves. Capacitors age horribly and as they are the heart of the amp, when they fail mayhem results. And remember, when comparing power amps (or indeed any audio equipment) unless you equalise the loudness (the gain) any conclusion you draw just won't stand up to proper scrutiny when the gains are equalised.

>

[EricW]

Would it be fair to say that, apart from the issue of gain/loudness differences, one never actually hears an amplifier: one only hears a specific amplifier/loudspeaker combination? Hence, even if valid subjective comparisons between amplifiers were even possible, they'd still only be meaningful if comparison were made on the same speakers?

(Which, of course, hardly anyone seems to do.)

[A.S.]

Would it be fair to say that, apart from the issue of gain/loudness differences, one never actually hears an amplifier: one only hears a specific amplifier/loudspeaker combination? Hence, even if valid subjective comparisons between amplifiers were even possible, they'd still only be meaningful if comparison were made on the same speakers?(Which, of course, hardly anyone seems to do.)

100% correct. That's why I find it so curious that those who believe in the magic of amplifiers feel such a compulsion to ensnare others with their enthusiasm. They completely ignore the fact that, as you say, the speaker + amp load can hugely influence the performance. And as I have shown, as the amp ages and the circuit degrades, dramatic changes in performance result. So, in practice Amp A off the production line today could/may/will sound and perform quite differently to Amp A made fifteen years ago.

Unless amplifiers from different brands (and same brand, different age) are compared after measuring their gain (loudness) across the audio band as shown in my graph using test equipment and then used with the same loudspeaker that was the load-under-test, any subjective conclusions drawn are utterly and totally invalid. This is so blindingly obvious that I shouldn't have to restate it.

The ear is sensitive to loudness. Change the loudness and you change the perception. A violin played at a whisper, normal level and an ear splitting 120dB will have the same (let's assume) measured audio spectrum but will sound totally different for reasons solely of loudness.

Buy an amplifier on durability, styling, size, warranty, after care; without test equipment you are not able to make a rational decision about sound quality. And that's a fact.

[weaver]

In asking the following question I am aware that it is both seemingly straightforward but also one that could prompt needless diversion. I am also aware that I am trying to draw a comparison between two things which in reality are not strictly comparable; one is an electromechanical device the potential shortcomings of which have been set out on many occasions; the other is an electronic device whose task is to faithfully pass a signal with gain.

The question is:

Is there anything within a power amplifier that can 'mask' detail in a way that is analogous to the way that certain cone materials can mask detail to a greater extent than others?

Possible straightforward answers would be:

No

Yes

or

Yes, but the environmental noise present in most listening rooms is such that we needn't worry about it.

[MikeH]

Hi Alan
Did you investigate the red line amp any further? Was it manufactured about the same time as the other? What was its age. I just wonder if factors such power supply capacitors aging was the cause of such a response.

Intresting??

Regards

Mike

{Moderators comment: approx same age. Yes we also suspect psu caps. Quad will service for peanuts.}

[A.S.]

Is there anything within a power amplifier that can 'mask' detail in a way that is analogous to the way that certain cone materials can mask detail to a greater extent than others?

Well, we have to ask ourselves what is actually doing the amplification inside the amp. It is a combination of transistors [or tubes] resistors, capacitors and coils. All behave with extremely completely behaviour in a physical sense so that the entire circuit can be modelled long in advance of picking up the soldering iron.

For example, I'd expect all modern amp design to be conceived, optimised and 100% analysed using SPICE and MathCad to the point of utter perfection. So what factors within the components or circuitry could make a huge but not predictable sonic effect?

[weaver]

So what factors within the components or circuitry could make a huge but not predictable sonic effect?

Thanks Alan, I wasn't asking that they necessarily be huge and I would expect that to a certain extent they would in fact be predictable but would generally come under the heading of 'good enough, within the cost restrictions and expected performance requirements of this product'.

The first one that springs to mind would be the power supply; is there for example electrical noise associated with power supplies that can contribute to the noise floor* of the amplifier or would the whole notion of 'competent design' mean that this issue (if it is one) ceases to be an issue.

*I hope I am using this term in its correct sense in that, as with say a vinyl record there will always be a certain amount of noise present and any signal quieter than that will not be heard.

[EricW]

Buy an amplifier on durability, styling, size, warranty, after care; without test equipment you are not able to make a rational decision about sound quality. And that's a fact.

I recently read on online review that was a rave about a particular pair of loudspeakers from a well-regarded Canadian manufacturer. They sell for $5,000 and the reviewer stated that he considered them competitive with other speakers he had heard at $10,000 and up. So far so good.

The only problem was, that he stated he first tried the speakers with a good 50 wpc amplifier that he had used to good effect with other speakers, but wasn't happy with how they drove these speakers because, as he commented, they were a difficult load particularly at bass frequencies, and the combination gave an unsatisfying sound. It wasn't till he switched to a different, more powerful amplifier that the speakers began to sound good (according to the review).

And it wasn't a matter of efficiency, either - the efficiency spec for the speaker was 90 dB - and yet a good 50 wpc amplifier didn't get the job done, apparently.

So maybe this is really is one of the world's great $5,000 speakers - but apparently only if you have the right amp. If you don't, too bad, I guess. For a well-heeled audiophile who gets good advice from a dealer who knows what the speaker requires, it may not be a problem and the results may be terrific. But I suspect at least some buyers might use an amp that wouldn't drive these speakers properly, and then wonder what the problem is. Is it the amp, or is it the speakers?

[A.S.]

...The first one that springs to mind would be the power supply; is there for example electrical noise associated with power supplies that can contribute to the noise floor* of the amplifier or would the whole notion of 'competent design' mean that this issue (if it is one) ceases to be an issue.

The design of the power supply within the power amplifier and the routing of the cables to and from the psu is, I know from bitter experience, a significant performance variable. Yes, low level noises can and do creep into cables and circuits due to the high currents and the pulsed nature of current waveforms in a push-pull class AB amplifier. I admire anyone who can design a properly quiet amplifier.

But - we're back to our old friend masking. Is it really necessary to chase a level of performance which is difficult to measure even under lab conditions with anything but state of the art test gear? Is a noise floor of -110dB audibly better than one of -100dB or even one of -65dB which is about as good as an analogue recording is capable of? Yes, if we're being offered a super-quiet noise performance in the electronics for free we'd be stupid not to accept it graciously. But in fact, when the music is playing at a normal level in our home hifi system the background noise can be really quite terrible (technically that is) and yet not detract one jot from our musical enjoyment. In other words, fantastic state of the art specifications are nice to have but they help salesemen flog hifi gear: they don't do anything for listening pleasure if the music masks the latent noise (as all vinyl lovers experience every time they play a disk).

But that aside, the function of a power amplifier is about as basic as electronics can get. It's been researched in microscopic detail for 30+ years. It's been shown many times that when the levels (gains) of power amplifiers are carefully matched to remove the subjective variable of loudness, the performance differences with a given speaker load are impossible to identify with certainty.

If only one aspect of human audiology makes an impact on you let it be this: If there is a sound level differences between two pieces of audio equipment that loudness difference alone will give rise to various subjective judgements in your brain. Remove the level differences by making the quieter one as loud as the other and those subjective differences greatly diminish or disappear altogether.

[STHLS5]

Is there any chance of comparing Amplifier A vs Amplifier B and explaining about the myth behind the measurements? What should we take into account by looking at these measurements?

ST

[A.S.]

I'm not sure that we are the right folk to be critiquing amplifiers made by third parties. We are not experts on amps (our views are well known). Are there any specific aspects of the measurements that we could compare here? On the face of it (a ten second peek at the graphs in both links - ignoring all the prose) I'd say that the performance was broadly comparable - just from that brief look there is nothing that would induce me to necessarily buy one over the other. I didn't look at the styling, size, price or any other factor, just the technical graphs.

A more detailed study of the graphs may reveal some small advantages of one over the other, but based on my quick peek, these amps seem to be in the same technical class, and I'd expect price range. Compared with even the very, very, very best loudspeaker ever invented they are likely to be thousands of times better in all and every technical aspect.

[Phil100]

Capacitors age horribly and as they are the heart of the amp, when they fail mayhem results.

Another reason to remove the capacitors from the amplifier design with the exception of those needed for RIAA equalization and power supply filters. See http://www.metaxas.com/pages/technology/four.html

{Moderator's comment: We should have said electrolytic (wet) capacitors age in amps. Dry film such as polyseter, mular, polypropylene and the like are stable for hundreds of years - probably.}

[STHLS5]

...Compared with even the very, very, very best loudspeaker ever invented they are likely to be thousands of times better in all and every technical aspect.

1) Are you suggesting that a well designed amplifier far exceeds the minimum requirements of a loudspeaker to perform at its optimum level?

2) Are you also suggesting that there could not be possibly any improvement in the loudspeakers sonics once a well designed amplifier exceeds the threshold of the minimum requirements of a loudspeakers because the weakest link in the audio chain is the loudspeakers?

That makes sense to me.

ST

[A.S.]

1) Are you suggesting that a well designed amplifier far exceeds the minimum requirements of a loudspeaker to perform at its optimum level?

2) Are you also suggesting that there could not be possibly any improvement in the loudspeakers sonics once a well designed amplifier exceeds the threshold of the minimum requirements of a loudspeakers because the weakest link in the audio chain is the loudspeakers?

No. Taking your points one by one ...

1) Are you suggesting that a well designed amplifier far exceeds the minimum requirements of a loudspeaker to perform at its optimum level?

I'd look at this from a very different perspective. There is no 'optimum level' for a high fidelity transducer other than passing on the signal it receives without corruption. The loudspeaker struggles in measured performance areas in which the amplifier is, for all intents and purposes, perfect. That is, the amplifier is far better than the speaker by a huge margin. The amp has a flat frequency response and its self-generated distortion is so low that state-of-the-art test equipment is needed to measure it accurately. Also, the amp's distortion is not much level dependent, and the nature of its distortion predictable. None of these things can (sadly) be said about a loudspeaker, which as a mechanical device has mass and inertia and simply cannot start and stop instantaneously, hence distortion (of motion, of the signal) is inevitable along with non-flat response, highly level-related non-linearities etc. etc.. Objectively, loudspeakers (and other electro-mechanical devices such as pick-ups) are technically nothing to be proud of and the fact that the technology stagnated thirty years ago is a real shame.

2) Are you also suggesting that there could not be possibly any improvement in the loudspeakers sonics once a well designed amplifier exceeds the threshold of the minimum requirements of a loudspeakers because the weakest link in the audio chain is the loudspeakers?

I think your question is ' ...improvement in the system sonics once a well designed amplifier exceeds the threshold of the minimum requirements of a loudspeaker ...' Right?

It's baffling to me that if a speaker introduces a very typical 1% distortion (that means, at any instant 99% of the music it reproduces is as intended but 1% of what you hear is distorted, objectionably or not) that we should be concerned that the amplifier generates 0.1 or 0.001% distortion. How can we hear and isolate 0.1% distortion in a mix of 1% + 0.1%? True, the sonic nature of the distortion may be such that one contributor adds a 'rasping' hard edge and the other a warm rather pleasant tone so that they could be subjectively identified. But in reality, the speaker is so much better at generating continuous distortion of a complex, broad band type that continually drones along with the music that there must be a degree of amplifier perfection beyond which the speaker will mask the amplifier.

This subject was thoroughly researched years ago in the analogue era. It was concluded that the degree of tolerable distortion - that is, non-audible - depended somewhat on the type of music being reproduced. Pop music (synthetic by nature) could have 10% or more distortion and sound OK - piano revealed distortion in the replay system, depending about several factors incl. listener acuity, at around 0.5%.

Again, when comparing hifi electronics in a scientific objective way, every effort must be made to normalise the replay loudness as distortion as a % figure is level dependent. Example: If a CD player A outputs 10V and 1% of that is distortion it is generating (as a voltage) 100mV of distortion. If CD player B outputs 2v for the same music and has the same 1% distortion, that's 20mV of distortion: still amounting to 1%. But subjectively A may appear to have much more distortion simply because in voltage terms there is much more of it eventually arriving at the speakers .... but as a percentage they are identical.

So a level difference between two audio equipment items has a huge influence on how you rank them subjectively. The 'big con' is surely to ignore this painful fact and just swap wires between two pieces of electronics then draw barmy unsubstantiated subjective comments about their audio characteristics. What you are hearing - and you definitely are hearing sonic differences - is a comment on how your ear/brain works, not necessarily any magic characteristics of the items under test. Is this clearly understood as a fact?

[kittykat]

Is there any chance of comparing Amplifier A vs Amplifier B and explaining about the myth behind the measurements? What should we take into account by looking at these measurements?

ST

Might be easier (and less ambiguous) if we compare the previous amp A compare A with this, Amplifier C ....

http://www.stereophile.com/content/m...r-measurements :-)

[A.S.]

That's a more interesting comparison. May I suggest that you totally ignore (don't even read) any of the words and just concentrate on the graphs. Don't look-up the price or look at the cosmetics. Just look at the graphs.

Disclaimer: I'm sure that both of these amps have a market and satisfy their users. They are both likely to be technically a lot better than even the best loudspeaker. It is not our intention to criticise either brand.

[A.S.]

Again, without looking at any pictures, without reading a single word of the review, trying not to see the brand names or models (not that they would mean anything to me), without being aware of the prices, without knowing anything about the technology (tube? transistor?), solely looking at the published graphs and taking them at face value ..... if I was in the market for an amplifier I would have to select amp C in preference to either amp A or B. That is, unless A or B was much cheaper and I was on a tight budget.

Can you see why from the curves?

I'm ignoring such unknowns as brand after-care back-up, brand image, country of origin etc. etc.. Just considering the narrow technical aspects, which would be a rather crude way of making a selection and certainly doesn't include a listening test which could, conceivably be decisive.

Amp A

Amp B

Amp C

[STHLS5]

....... if I was in the market for an amplifier I would have to select amp C in preference to either amp A or B. That is, unless A or B was much cheaper and I was on a tight budget.

Can you see why from the curves?..

I am not sure if I can make anything out of the graphs.

distortion (%) vs 1kHz continuous output power into (from bottom to top at 100W): 4, 8 ohms.

Vs

distortion (%) vs 1kHz continuous output power into (from bottom to top at 10W): 8, 4, 2 ohms.

Vs

distortion (%) vs 1kHz continuous output power into (from bottom to top at 1W): 8, 4 ohms


And Amp C is a power amplifier whereas Amp A and B are Integrated Amps.

ST

[kittykat]

I am not sure if I can make anything out of the graphs.

Imo, think the graphs you are referring to tell quite a bit, regarding how well an amp is designed, the philosophy and discipline. The graphs show distortion levels as more power is required from it, where it peaks out (the “notch” -normally should correlate relatively closely with rated standard specs). It might also tell how much noise there is to begin with. In other words, we can observe where distortion of an amp. begins on the scale, how it travels and starts to deteriorate (close to rated claims).

Think this thread started on “how load dependent an amp is”. If im not mistaken, the first graphs on any of those 3 pages with the “simulated loudspeaker load” line (the most wavy one) answers the question most well. Some amps have a much more “wavy” line than others.
But Alan does put it into a beautiful perspective. The ugliness of an amplifier is by far more liveable than the slightest blemish of a speaker.

[Haligonian]

The ugliness of an amplifier is by far more liveable than the slightest blemish of a speaker.

I am sorry, I do not understand what you mean by this. Could you (re)paraphrase?