Teddy Ray
pfm Member
bet you couldn't id them blind. im certain of it, actually.
Do amplifiers really sound the same?
- Thread starter avole
- Start date
- Status
bet you couldn't id them blind. im certain of it, actually.
Tigerjones
Bagpuss
When someone can combine cars, wine and amplifiers in the same analogy I think we'll have made some real progress.
I think that's a fair enough challenge which deserves a reply. If two amps can be shown to be audibly different in blind ABX testing by being distinguishable in a significant number of cases I would expect one or all of the following measurements to show a difference, all other things being equal: output signal demonstrably different for a given input music (not test) signal - perhaps detected by differencing the outputs; output signal different for pink noise, ditto; difference of input to output detectably different; amplitude and phase spectral response detectably different; impulse response detectably different; harmonic distortion detectably different; clipping distortion detectably different.
Tony L
Administrator
To my mind what is becoming known as the strict "objectivist" stance breaks down even before that as there isn't even agreement as to the proposed operating environment, let alone the measurement criteria. An amplifier is a tool to drive a loudspeaker, and there is simply no standard here. A Quad 57, Linn Isobarik, Apogee Scintilla, Lowther and Klipschhorn are all very fine loudspeakers, some of the very finest in fact, yet so different from one another as to have different and often contrary input requirements. Differences in efficiency, loading, current demand and mechanical damping of these speakers is so extreme as to require entirely different amplification solutions. To my mind amplifiers can not be considered in isolation from their intended operating environment - they are a tool one selects for a very specific role, so any attempt to generalise with a few simple specs is largely an irrelevance. The real world is very different.
I think that's fair - obviously completely ignoring the load in the claim that all competent amplifiers sound the same is not necessarily right. The claim, as I understand it, does include the proviso that they are working within their capability - ie not clipping, for example, which mitigates some of your criticism, but in general it's a good point.
Robert
Tapehead
The situations where those additional considerations arise are however few and far between, and generally covered by the usual rider that amplifiers should not be driven into clipping.
In the vast majority of proposed combinations the amplifier can be considered in isolation, and if all you want is transparency, a few headline specifications will guarantee that's exactly what you get.
There are only two real worries when selecting amplifiers for unusual drive conditions:
- Driving very low impedance and/or very reactive loads. In such cases the amplifier needs to have sufficient drive and very low output impedance to be transparent, others things being equal.
- Vintage loudspeakers designed when high amplifier output impedance was common. You can fake this condition perfectly with a couple of resistors.
But whichever way you cut it, the measurements dictate precisely the result achieved in these odd situations. In this particular area, we know everyone there is to know.
James
Lord of the Erg\o/s
Power supply stiffness, noise and rejection seem to matter too, but I'm not sure how these materially affect the transfer function of an amplifier, pre or power.
Sue Pertwee-Tyr
Accuphase all the way down
Thanks. Are these measurements routinely published by the manufacturers, perhaps available on their websites? I can't say I've seen them, as a rule.
If not, do those people who advocate measurement (I'll exclude Serge, as I think I know the answer) measure these parameters for themselves? If so, before or after purchase?
You did, by the way, forget to answer the bit of the question about by how much these parameters would have to change, to be audible, or to produce the sort of musically relevant effect I described.
My point, in case it has passed anybody by, is that reading the numbers on a makers bumf might make it possible to conclude that two amplifiers are different, if their numbers are different. But being able to read those numbers and draw anything other than general conclusions about them seems either unlikely, or (perhaps) beyond the technical capabilities of some of those who espouse these views.
I'd hazard a guess that those on here who can genuinely interpret the measurements, decide which ones are critical and how (and by how much) they would need to change to achieve an audibly different result, are in a very small minority. What's more, I have my doubts that those are the ones (Serge excepted, perhaps) who would advocate such an approach.
If so, I think that choosing your amp by the numbers without being able to defend the decision by explaining what those numbers mean, compared to an amp you disregarded, starts to look less like science and more like numerology. In other words, it's a subjective impression of whether you like the look of the numbers not a rational scientific one.
Tony L
Administrator
I don't agree at all, especially given the current fashion for very small ported two way stand-mounts which makes for some remarkably inefficient and hard to drive loudspeakers e.g. low 80s sensitivity and sweeping reactive loads with severe impedance dips and spikes. A lot of very highly regarded amplifiers will respond to that in audibly differing ways. If all speakers were say 90db with a perfectly flat 8 ohm load you'd have a point, but they are not. Even quite cheap and compromised speakers can prove quite challenging these days.
James
Lord of the Erg\o/s
Perhaps with valve amps, but a "proper" solid state amplifier should approximate a true voltage source, and double its power into halving impedances - is what I expect to be one parameter that serge would look for. Krells can do that, but I rarely hear them espoused for "musicality", whatever that means.
It's rare for that to happen, as the power supply and current driving capability isn't possible for most amplifiers. The Quad 405 and 303 were quite good, and have been used to test the wire with gain/ amps sound the same hypothesis, but Quad produced a graph showing that the amplifiers were no where near your ideal.
sergeauckland
pfm Member
In answer to your questions:-
A) Manufacturers seldom put a pull set of specs on their literature or web sites, but sometimes a full set of measurements is available if a Magazine does a full review. Paul Miller at HFN does this, but clearly, only a very small number of products get this full treatment. Nevertheless, with modern electronics, and if one has reasonable confidence in the manufacturer's quality, one can make some assumptions before purchase.
B) Yes, I do measure any purchases myself, and compare them with the spec. If the numbers indicate a fault, or seriously out of spec, then the product goes back for replacement or refund. However, as was the case with my Behringer A500s, they didn't meet spec, but were still perfectly acceptable as they were. I then modified the power supply as I needed less noise than the spec, and as it was a trivial modification, I didn't mind doing it. Unless using the A500 for active 'speakers with a driver of 92dB sensitivity as my mid-range driver is, it wouldn't have been a problem.
C) How much do parameters have to change to be audible? Well, that all depends.....As above with noise, normally the noise of an A500 would be perfectly acceptable, but not in my case due to the particular nature. Distortion is generally accepted to be inaudible if 0.1% at all powers and frequencies. However, this doesn't mean that 0.2% is audible. Depending on the order of the distortion and whether Harmonic or intermodulation, a level of 1%THD, with low orders predominating as is normal, is generally considered inaudible by most people. Consequently, a figure of 0.1% was chosen as a number that would give a blameless performance for virtually everybody under all circumstances. It should be recognised that studio quality Reel-Reel machines normally had 3% distortion on peaks as that was how peak level for these machines was established. Considering how many fine recordings were made on these machines, even 3% isn't obviously bad, all other things considered. So really, there's no one answer to your question. Distortion can vary by large amounts and still be inaudible, frequency response variations can be quite large (say 2-3dB, possibly more) and still be inaudible if of high Q whilst a large dip of 2dB over an octave range in mid frequencies would easily stand out in an AB comparison. However, for frequency response variations, a roll-off of 1-2dB at the frequency extremes would generally be inaudible, even 3dB at 20kHz if more than single order, as a 6dB/octave roll-off at 20kHz is still 1dB at 10kHz, possibly just audible.
D) Choosing an amplifier by numbers makes complete sense to me as most modern amplifiers are so way beyond transparency, that unless faulty, even if they don't quite meet spec, they would still be transparent. Comparing modern amplifier specs for mainstream amplifiers like Quad, Naim, Cyrus, Arcam and so on leads me to expectno audible difference whatsoever unless the amplifiers are being used outside of their design envelopes.
S.
sergeauckland
pfm Member
One way of ensuring an amplifier doubles it's power into half the load is to design it for the worse case. For example an amplifier that's designed to provide 1kW into 1 ohm will do 500 watts into 2 ohm, 250 watts into 4 ohms and 125 watts into 8 ohm. In practice, the higher impedances may well produce even more power as if the amplifier has an unstabilised supply, then the power supply voltage will rise. Clearly, an amplifier that can produce 1kW into 1 ohm continuous would be a huge and expensive beast, and commercially could only be sold as a 125w power amp using conventional specs, so would be at a commercial disadvantage. Hence Krells are sold for large amounts of money by specialist dealers whereas one can buy a 125w amplifier from Behringer (The A500) for £150. However, the Behringer can't do 1kW into 1 ohm!
As to musicality, I have no idea what that means. All I know is that a Krell technically is about as good as it gets, and well over what anyone needs. (Unless driving 1 ohm!)
S.
Tony L
Administrator
In reality only a handful of exceptionally large and expensive amps achieve this ideal, e.g. Krell, Mark Levinson etc. Most fall very far from it, e.g. my second system Quad 303 actually delivers more watts into 8 Ohms than it can into 4 (45 vs 28 IIRC)! It *hates* low impedances! I don't think any UK amp manufacturer has a product that correctly doubles output as impedance halves, especially to Krell level where the amp is able to do it right down to 1 Ohm IIRC. If you want accurate that is what it looks like, anything else is compromise.
PS I was very aware of this fact when writing my earlier post ;-)
James
Lord of the Erg\o/s
I don't think being able to go down to 1-ohm whilst maintaining full drive voltage is necessary unless driving Apogee Scintillas. Any amp that gets close to doubling their 8-ohm power into 4-ohms is good enough for most nominal 8-ohm loudspeakers, which shouldn't dip much below 5-ohms at minima. They can get into trouble with 4-ohm designs though, which is favoured by American designers and hence the need for arc-welding amperage.
Mescalito
pfm Member
Mescalito
pfm Member
Mescalito
pfm Member
I am a person who believes that any two transparent amplifier, operating within their design parameters and driving identical speakers, sound essentially identical.
Yet I use a £8000 pre-amp & active speakers costing £28,000.
Why? for the same reason I have a rolex watch, a Du Pont lighter, etc, etc.
First off, the kit looks good, 2nd, pride of ownership, bling, call it what you may.
Chris
Generally not I would say, although most or all of them would have been measured by the manufacturer in development and some used as routine end-of-line tests.
I have no idea, but I suspect not. I don't.
I don't know. As I've stated a couple of times, in many cases we don't know at what level different measurable results become reliably audible for different classes of listener, and the question is complicated by the fact that you can't generally give a single number for any measurable. It often depends. However there are some generally accepted limits as Serge has pointed out and many real world amplifiers measure way below them. Having said all that, it is not my view that we can say with confidence what measurable effect will be audible and what will not, particularly near to the threshold. My position is complementary to your question: which is that if two amplifiers can be shown to be audibly diferent then they will measure different to an extent that plausibly explains the audible difference - I do not accept the idea that we can hear things that result from unmeasurable differences.
But what people are claiming isn't that you choose your amp on the basis of the maker's bumf. They are saying that, assuming you don't have some exotic speaker which presents a bizarre load, any competent SS amplifier working within its operating range will be indistinguishable - then all you need to decide is what the efficiency of your speakers is and how loud do you want to go. Not necessarily my view, although I do think that the amplifier in most systems has the least effect on the performance of the system as a whole, compared to sources and speakers, to the point of no difference to my cloth ears. I do agree however that choosing your amplifier by comparing THD+N figures in manufacturers' glossies makes no sense.
Paul R
pfm Member
An amp suitable for the Scintilla will be ideal with the others, but a gratuitous waste of resources.Tony L said:
There's a misconception here. There are two conditions, the boundary of the output of the amp, so the maximum power an amp will produce into various loads, you will seldom find a true voltage source here because of PSU limitations, unless the amp is nobbled into higher impedances, which I'm sure happens in dark corners of the high end market. The other is the output impedance of the amp which applies all the time and will never be 0, although nowadays it will be low compared to the wiring.James said:
For example Tony's 303 is nominally a 45W amp, because into 8 Ohms it will produce about 45W at low distortion. But it will make more than 25W into 16 Ohms, and less than 25 into 4.
The important spec however, when you're not operating the amp into clipping regardless of load, is the output impedance which is given as 0.3 Ohms, which is low but not very.
Paul
- Status