[msmith]

Lycan,

As you know, most amplifier bench tests are done with steady state supply voltage, steady state signals and purely resistive loads. It is possible for two amplifiers to test identically (theoretically) under the above conditions, yet test differently when measured with dynamic signals, reactive loads and a "squishy" supply voltage.

This, of course, does not negate your basic premise... it simply points out that the typical "bench test" may not be complete and conclusive. In simple terms, to paraphrase Paul Klipsch: "If it measures the same and sounds different, you're not measuring the right thing."

 

[SoundChaser]

some people buy corvettes and Ferraris

Yes, amplifiers can & do sound different. But it's perfectly reasonable to understand that, given the state of electronic circuitry, it's not hard (like it was 50 years ago) to build an amplifier whose errors in gain, power, frequency response, noise & distortion are all well below the limits of human hearing (this, i suspect, is behind the "competently designed" phrase).

So, if nothing else comes of this discussion, consider this when comparing ANY two amplifiers : there's a well-known list of things that can CAUSE one amplifer to sound different than another. These are :

1. Gain difference by 0.25dB or more2. Frequency response errors of 0.25dB or more
3. Power
4. Noise & distortion greater than, say, -80dB.
5. The psychology of knowing which brand name you're listening to

So consider these KNOWN things that cause amps to "sound" different. Wouldn't you like to know which one is causing the sonic difference you "heard"? Most importantly, how much are you willing to PAY for a sonic difference due to one of these parameters?

Trust me ... audio salesmen know these effects VERY well. Crank the gain on the brand you want to move by a quarter dB or so, crank the brightness on the TV brand you want to move ... all well known tricks.

You think you hear a difference? ask WHY !!!!! Hell, MEASURE the outputs of the two amps with an AC voltmeter at a few different frequencies Fear not, you won't "spoil" the amplifier magic by putting a voltmeter across its output terminals.

Then, ask how much you're willing to pay for that tweak of the gain knob, or that slight bump in midrange frequency response ...

I don’t follow how the gain difference would skew the comparative results of 2 amplifiers.

1 amp reaches its full potential @ ½ volume off the head unit. The other amp reaches the same SPL level @ ¾ volume off the head unit.

What’s the difference?

 

[lycan]

I don’t follow how the gain difference would skew the comparative results of 2 amplifiers.

1 amp reaches its full potential @ ½ volume off the head unit. The other amp reaches the same SPL level @ ¾ volume off the head unit.

What’s the difference?

small gain differences (fraction of a dB) are often "perceived" by human hearing as differences in "quality", rather than differences in "volume". Same can be said for small differences in brightness in TV screens

Oh dear, how oh how can we possibly eliminate that difference when comparing two amplifiers? All hope is lost! We have no choice but to accept the mysteries of SQ ...

How much will YOU pay for that small tweak of the gain knob? How can one possibly determine if that's the cause of the difference heard ????????

 

[lycan]

I'd like to believe that someone who has been listening to say, a Milbert tube amp, could easily pick his amp out of a line-up, but since reading this thread I'm not so sure.

If you had read and understood this thread then you would readily accept that a person could easily pick out his Milbert tube amp in an amplifier comparison ... but the reason WHY is because of the tube amplifier's gain, power, frequency response (probably), noise or distortion (equally probable)

I'll repeat it for the miliionth time : All amps do NOT sound the same. I can pull two amps from the same damn production line, and set their gains or crossovers different, and i promise they WILL sound different. It's astonishing to me that nobody understands this.

Seems to me, everyone wants to put themselves into one of two EXTREME categories : either all amps sound mysteriously different, for reasons beyond all logic and understanding .... or else, at the other extreme end of the spectrum, any two amps under any circumstances (wildly different gains, completely different crossover settings, etc) will sound exactly the same.

Are these really the only two categories? All amps are EXACTLY the same no matter what, OR it's impossible for any two amps to be sonically indistinguishable no matter what?

That's what this debate always degrades into

 

[DS-21]

To be honest, I don't know too many people who would agree to have their Zuki/Genesis/Audison/Sinfoni/Luxman/Milbert/McIntosh/F1 Alpine/ODR/XES etc., amplifier tested against a Boss or Legacy in a double blind study.

Why not?

A now-dead Miami audio dealer once agreed to blindly compare his own Pass Labs amp, which costs a helluva lot more than any piddly car-fi boutique crap, to a cheap Yamaha integrated.

I've gotten people to compare Meridian gear (again, more expensive than pretty much anything available for car-fi) against mass-market (two-figure pricetag) gear. I've compared my own Classe gear (see a trend) against much cheaper gear.

So, are you saying that people who buy boutique car-fi gear on the illusory grounds that it "sounds better" are exceptionally deluded even by the moronic standards of "high end" home audio?

Or just that that people who buy boutique car-fi gear on the illusory grounds that it "sounds better" are too poor for their little brains to withstand knowing that they might have eaten peanut butter and ramen for a month for no sonic reason?

Seems to me, everyone wants to put themselves into one of two EXTREME categories : either all amps sound mysteriously different, for reasons beyond all logic and understanding .... or else, at the other extreme end of the spectrum, any two amps under any circumstances (wildly different gains, completely different crossover settings, etc) will sound exactly the same.

Sorry, but that sounds a little bit like Republicans screaming that Obama is a "socialist." As in that debate, there is one extreme position that is apparently still widely-held despite being easy to disprove. (Well, I suppose there are still people who think the earth is 6000 years old, too...) And there is a second position that only exists in the collective imagination of those aforementioned extremists, but not in real life. The only things that claiming said second position exists do is demonstrate to sane, reasonable people that the holders of the former position are
a) utterly decoupled from reality.
b) too intellectually unsophisticated to understand the difference between questions of fact (does x sound different from y?) and matters of opinion/preference.

Nobody here's articulated an argument anything remotely like the your illusory "other extreme end." And, in fact, I've never heard anyone say things like level and FR are inaudible in controlled listening. Obviously, were someone to articulate that position, rational people who deride the advocate as being just as brainless and/or deaf as the "magic voodoo sonic differences" crowd.

 

[lycan]

Nobody here's articulated an argument anything remotely like the your illusory "other extreme end." And, in fact, I've never heard anyone say things like level and FR are inaudible in controlled listening. Obviously, were someone to articulate that position, rational people who deride the advocate as being just as brainless and/or deaf as the "magic voodoo sonic differences" crowd.

I agree ... the extreme position has not been articulated, but it certainly has been implied. Here's a "version" of the same position :

I'll take two amplifiers from the same production line. I'll make sure they are level matched at 1kHz. But the crossover settings in one amp are wildly different than the other amp : one amp is high-passed at 50Hz, the other is high-passed at 200Hz (and yes, the crossover circuitry is, strictly speaking, in the pre-amp section ... exactly where the gain circuitry is, in most car amps). Surely, they "sound" different. Which one is the incompetently designed amp?

A stupid, trivial, meaningless example? Not to this crowd! Because there's a thought process that many will NOT follow. "Well, duh lycan ... of course we mean the crossovers must be the same too. Or else the frequency responses will be wildly different !!". And i'll respond by saying : How CLOSE must they be? If you were to put MONEY on the line ... which you do, when you make a purchase decision .... how will you determine if the frequency responses are "close enough" for a valid comparison? Will you look at the approximate position of the detent on the crossover knob, or ... will you MEASURE the frequency responses, according to a target spec of, say, a fraction of a dB?

Disable the crossovers, you say? And that's enough to convince you that the frequency responses are within a fraction of a dB ... if there's real money on the line?

I promise you ... most of the audience still doesn't get that the fundamental proposition here is NOT that all amps sound the same, but rather that all amps which measure the same, will sound the same ... and that we can use a small set of well-established electrical parameters or specs to completely & thoroughly characterize any & all sonic attributes, just like every manufacturer does when they design & test these appliances in large volume.

Most don't even understand the difference

EDIT : and we haven't even begun to explore the point i tried to make earlier, and that Manville started to elaborate on ... how do we comprehend a loudspeaker's reactive impedance in an amplifier comparison, when most tests & specs are made with a simpler resistive load? Two amplifiers that spec, and therefore sound, identical with a 4 ohm load that's almost purely resistive, may in fact not spec, nor sound, identical with a load that's "nominally" 4 ohms, but has substantial reactive variance. Such a proposition would surely confound the "all amps sound the same" thesis, but not necessarily the "all amps that measure the same, will sound the same" proposition.

 

[DS-21]

I'll take two amplifiers from the same production line. I'll make sure they are level matched at 1kHz. But the crossover settings in one amp are wildly different than the other amp : one amp is high-passed at 50Hz, the other is high-passed at 200Hz (and yes, the crossover circuitry is, strictly speaking, in the pre-amp section ... exactly where the gain circuitry is, in most car amps). Surely, they "sound" different. Which one is the incompetently designed amp?

I'm confused why you're bringing crossover settings into the argument. If they're defeatable, then it's not relevant to a discussion of sonic differences.

But let's assume the settings are made at the factory and non-defeatable. Then the answer is:

Both, if they're both impliedly intended for full-range use.

The latter, if they're both expressly intended just for use driving mains in a "typical" system (60-150 Hz highpass).

Neither, if they're expressly designed only to drive tweeters (1KHz and up, let's say).

And i'll respond by saying : How CLOSE must they be? If you were to put MONEY on the line ... which you do, when you make a purchase decision .... how will you determine if the frequency responses are "close enough" for a valid comparison? Will you look at the approximate position of the detent on the crossover knob, or ... will you MEASURE the frequency responses, according to a target spec of, say, a fraction of a dB?

Disable the crossovers, you say? And that's enough to convince you that the frequency responses are within a fraction of a dB ... if there's real money on the line?

If there's purchase money on the line, a reasonable person isn't too worried about it. Even if the level variations are slightly audible with some program material, a well-designed system will use signal processing that can fix such issues anyway.

If it's an "amp challenge" I would spot-measure voltage at a few frequencies. In one I recently did (which, for the record, netted me an Arc KS125.4 Mini and a couple hundred bucks for Pakistani flood relief in the person's own lorry with his own speakers and own choice of program material he felt had revealed differences in sighted listening) I just measured voltage off of since waves at 50, 1k, and 10k Hz and matched gains. That was sufficient for the amp to be sonically indistinguishable.

EDIT : and we haven't even begun to explore the point i tried to make earlier, and that Manville started to elaborate on ... how do we comprehend a loudspeaker's reactive impedance in an amplifier comparison, when most tests & specs are made with a simpler resistive load? Two amplifiers that spec, and therefore sound, identical with a 4 ohm load that's almost purely resistive, may in fact not spec, nor sound, identical with a load that's "nominally" 4 ohms, but has substantial reactive variance. Such a proposition would surely confound the "all amps sound the same" thesis, but not necessarily the "all amps that measure the same, will sound the same" proposition.

I dismiss the point out of hand, because it has never been shown to be meritorious in a controlled listening test.

It's all just hand-waving until/unless someone can point to a positive identification in a controlled same/different subjective listening evaluation.

I think you would agree that a Yamaha integrated amp with a standard AB amp circuit probably acts somewhat differently into reactive loads than a tweaky single-ended solid state amp, right? Such as, say, the old Pass Aleph series. Well, we know how that went!

 

[rommelrommel]

EDIT : and we haven't even begun to explore the point i tried to make earlier, and that Manville started to elaborate on ... how do we comprehend a loudspeaker's reactive impedance in an amplifier comparison, when most tests & specs are made with a simpler resistive load? Two amplifiers that spec, and therefore sound, identical with a 4 ohm load that's almost purely resistive, may in fact not spec, nor sound, identical with a load that's "nominally" 4 ohms, but has substantial reactive variance. Such a proposition would surely confound the "all amps sound the same" thesis, but not necessarily the "all amps that measure the same, will sound the same" proposition.

This is what I really want to hear about. I believe that any differences are measurable, but that there are likely differences. If there is an impendance spike in the passband and two amplifiers handle impedance spikes differently, could it not produce an audible difference? Or does competent design account for this?

 

[lycan]

This is what I really want to hear about. I believe that any differences are measurable, but that there are likely differences. If there is an impendance spike in the passband and two amplifiers handle impedance spikes differently, could it not produce an audible difference? Or does competent design account for this?

More likely, the problem would be an impedance "dip" ... amps must work harder into lower impedance loads.

I'll allow for the possibility, as does Manville (i think). Whether or not it's ever been uncovered in a finite sample size of controlled listening tests is irrelevant ... it's certainly possible, and logical, based on well-established electronic theory and known listening thresholds. An amp which shuts down, goes unstable, or current-limits, into a 2-ohm load (while being perfectly happy with a 4-ohm load), would be expected to behave "differently" while driving a speaker whose reactive impedance dips to 2-ohms ... as compared to an amplifier whose power simply doubles into 2-ohms. To expect otherwise, is irrational. Therefore, i'll allow for the possibility ... even if i can't point to a specific listening test that revealed the principle.

The key point for me, as it has always been, is this : if a difference in the "sound" of electronic equipment can be heard, then that difference can be electrically measured. And it follows, then, that if two pieces of electronic equipment are electrically indistinguishable (into the same load!), they must be sonically indistinguishable. This is simple "if p, then q" logic ... nothing more.

 

[DS-21]

I'll allow for the possibility, as does Manville (i think). Whether or not it's ever been uncovered in a finite sample size of controlled listening tests is irrelevant ...

Why? Lots of things are theoretically possible, but don't actually happen.

When it comes to audibility, the only thing that's important is, well, establishing audibility. There can be no question of preference, relative fidelity to the source material, etc., unless the threshold determination that there is an audible difference has been affirmatively made.

The only way to do that is to isolate the variable in question and listen to see if a difference can be actually established to a reasonable degree of statistical significance.

Unless we're talking about a measured property that has already been shown to be audible in such circumstances, anything less is just mental masturbation.

The key point for me, as it has always been, is this : if a difference in the "sound" of electronic equipment can be heard, then that difference can be electrically measured.

We're in agreement on one crucial point: the priority of sound (is there an audible difference in controlled listening?) over all other factors. Where we disagree is on what follows. I say that it follows logically to say that if there's no audible difference established, there's no need to measure esoteric things such as output into weird loads and so on. Everything must stem from that threshold determination of audibility.

 

[lycan]

Why? Lots of things are theoretically possible, but don't actually happen.

When it comes to audibility, the only thing that's important is, well, establishing audibility. There can be no question of preference, relative fidelity to the source material, etc., unless the threshold determination that there is an audible difference has been affirmatively made.

The only way to do that is to isolate the variable in question and listen to see if a difference can be actually established to a reasonable degree of statistical significance.

Unless we're talking about a measured property that has already been shown to be audible in such circumstances, anything less is just mental masturbation.



We're in agreement on one crucial point: the priority of sound (is there an audible difference in controlled listening?) over all other factors. Where we disagree is on what follows. I say that it follows logically to say that if there's no audible difference established, there's no need to measure esoteric things such as output into weird loads and so on. Everything must stem from that threshold determination of audibility.

I know what you're saying, but you can have a theory that allows for, even expects, and "event". Now perhaps, at this point in time, the "event" has not yet been observed. What does that mean? Not much, really. It doesn't disprove the theory ... the theory simply remains open to the possibility. That's all i'm really saying. Well, i'm actually saying something more ... i'm saying that if a difference is really heard in a controlled test, i'm hoping to *point to* some simple electrical tests that can help uncover the simple question : why?

There are some extreme examples that come to mind in other realms of science, but i simply won't state them in yet another damn thread about sonix of electronix (cuz they will be taken WAY out of context).

I think it is important for the general audience to know, that we're not in any real disagreement. We may argue over semantics, or the possibility of extreme "outliers" in a statistical distribution ... but i don't think we're saying anything different of substance.

It's just that one of us is an engineer, and the other is a lawyer

 

[MarkZ]

It's certainly possible for an amplifier's performance to be well within spec (however we define it) driving resistive loads, and then tank under reactive loads -- either by shutting down or producing some sort of audible...um...artifact.

But I'll reiterate something I said earlier, to which you (lycan) echoed (or maybe you said it first...I don't remember ): it's pretty easy and relatively cheap to design an amplifier that will behave "perfectly" under a varied assortment of (reasonable) loads... given that the design's minimum impedance isn't exceeded.

Why? Because amps with fairly high gain negative feedback usually have an extremely low output impedance, often dominated by the output zobel/protection circuitry (if it's got 'em). I wouldn't expect the FR or distortion performance to necessarily change appreciably (again, assuming "competent" design), and the noise characteristics shouldn't be affected. So I think you're being a bit conservative to entertain the notion that competently designed amps will change their characteristics to an audible level when put under (reasonable) reactive loads.

What might happen though (and this also goes for Manville's "squishy supply voltage" comment) is that the output capabilities of the amp decrease more for one amp than the other. But this isn't really a "sonic signature" issue. It's a "where does my amp clip?" issue, which is irrelevant to the conversation, IMO.

Anyway, I think one of the biggest problems in this thread is that nobody's defined "competent" yet.

 

[lycan]

It's certainly possible for an amplifier's performance to be well within spec (however we define it) driving resistive loads, and then tank under reactive loads -- either by shutting down or producing some sort of audible...um...artifact.

But I'll reiterate something I said earlier, to which you (lycan) echoed (or maybe you said it first...I don't remember ): it's pretty easy and relatively cheap to design an amplifier that will behave "perfectly" under a varied assortment of (reasonable) loads... given that the design's minimum impedance isn't exceeded.

Why? Because amps with fairly high gain negative feedback usually have an extremely low output impedance, often dominated by the output zobel/protection circuitry (if it's got 'em). I wouldn't expect the FR or distortion performance to necessarily change appreciably (again, assuming "competent" design), and the noise characteristics shouldn't be affected. So I think you're being a bit conservative to entertain the notion that competently designed amps will change their characteristics to an audible level when put under (reasonable) reactive loads.

What might happen though (and this also goes for Manville's "squishy supply voltage" comment) is that the output capabilities of the amp decrease more for one amp than the other. But this isn't really a "sonic signature" issue. It's a "where does my amp clip?" issue, which is irrelevant to the conversation, IMO.

Anyway, I think one of the biggest problems in this thread is that nobody's defined "competent" yet.

Again, no real disagreement ... but i don't think i'm being unreasonable in allowing for load variation (beyond resistive) to upset certain amplifiers.

Yes, it's very well known how to design otherwise (i think you said it first ). But it *may* come down to cost. It's not a "small signal" issue; i absolutely agree that a negative feedback amplifier's output impedance will be positively tiny compared to any reasonable load.

Instead, it's a question of large-signal heat, and stress. The lower impedance load will cause the output stage to be more thermally stressed ... when is a danger threshold crossed, for a given amplifier designed to a price point (and one that may measure fine at a higher impedance load)? And potentially worse is a very reactive load, where simple impedance magnitude doesn't tell the whole story. A capacitive load is particularly nasty for any Class AB amp, because the output stage will be delivering it's HIGHEST current when the voltage across the output transistors are also at their HIGHEST ... unlike a resistive load ... meaning worst-case thermal stress of output devices under reactive, rather than resistive, load conditions.

Again, no disagreement on fundamental substance of the discussion. I just don't think it's very unreasonable to expect performance variation ... that may differ from one amp to another ... under reactive, versus tested-resistive, loads.

 

[MarkZ]

I think the reactive/resistive issue falls under the "measures the same" umbrella, regardless.

No doubt. I think Manville's point was that it's another variable that needs to be accounted for, and that your measurements need to address it when trying to extrapolate sonics from a set of test parameters.

Which is sort of why I like DS-21's approach of forgetting what's under the hood, so to speak, and instead directly measuring outcomes.

 

[lycan]

i think we all agree, that none of this load discussion refutes the validity, and comprehensive nature, of electrical measurement.

Amps have to be specified somehow, with some kind of representative load. Surely small variations, at least, can't be expected to upset a "competently designed" amp

And yes ... understanding the "stressful" nature of reactive, rather than purely resistive, loads is one of the factors (maybe the main one) that lead to the "power cube" representation of loads for amplifiers.

 

[lycan]

i think we all agree, that none of this load discussion refutes the validity, and comprehensive nature, of electrical measurement.

Amps have to be specified somehow, with some kind of representative load. Surely small variations, at least, can't be expected to upset a "competently designed" amp

And yes ... understanding the "stressful" nature of reactive, rather than purely resistive, loads is one of the factors (maybe the main one) that lead to the "power cube" representation of loads for amplifiers.

EDIT : but i don't know that Rockford gets the credit for developing the "power cube" load representation ...

 

[14642]

http://www.audiograph.se/

Here's the powercube guys. I do believe Rockford may have been the first company to use this thing to sell amps. I remember sitting through a training when I was a retail installer where our rep attempted to explain the measurement. Basically, the arguement was, "we have this book with all these measurements and that makes our amps sound better." Thank God for the Autosound 2000 guys who taught us all to make sure we didn't believe anything that any company ever told us about anything so we'd spend 10 bucks on their little newsletter from their company, which was the only one that ever told the truth so long as they weren't in the IASCA judging lanes.

Anyway, I digress. Carry on, mates.

 

[14642]

Oh, and of course all amps and head units don't sound the same.

 

[thehatedguy]

Easy fix or the gain knob variations...get rid of it with those coupling caps.

Why does Milbert get all of the tube love? HSS is much nicer amp.

But what can I say, I love even order distortion profiles.

And I really respect JL amps for increasing power into reactive loads. Any amp can drive a resistor...but some are better than others when you throw in some back EMF etc to the mix. Most will decrease power output, but the JLs increase over resistive loads.

 

[14642]

I am still at a loss for what happens in a Rockford amp that occurs a few minutes after warming up. I read somewhere that you shouldn't judge a Rockford amp on it's sonics until this period of time has passed, and I'll be darned if I don't hear a clearly audible difference when it happens.

Is it frequency response shaping? Is it a distortion reduction? Is it parts of the spectral response on a hold until some temperature differential is achieved? I thought it might have something to do with that, one of their trade-marked circuits is responsible, don't remember which one.

Who knows, but many amplifier manufacturers in the quest for ever sleeker and smaller heat sinks without fins use limiters that are activated when the heatsink temperature rises. Some of them have pretty nice transitions that make them difficult to hear and others are pretty abrupt.