• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Why do we hold amplifiers to such high standards?

If a manufacturer is not at least 0.005% THD today, what are they doing? To me, it's a measure of competency.
Power. Because you are more prone to hearing clipping than distortion.


And if you wonder why you need so much power, look at the 2216nd impedance curve.

1709616341172.png
 
What I'm asking is, I suppose, is that 0.1% figure just a conservative engineering target (based on the real world expectation that your listening levels are going to be around 60dB over the background noise, and that if you have a perfect speaker system, you need to achieve this figure)?
Yes.

And because it is so easy to make something that can be much better than that 0.01% is fine.

@DVDdoug has the same experience as me: I've never heard distortion from ANYTHING that wasn't broken or overdriven.
For instance a faulty output relay can make the sound terrible at low signal levels but when cranking it up sounds (and measures) great.
Amps being pushed to clipping levels can react and sound different but when staying below clipping levels of the lowest power output of the amps compared they will sound the same/similar.

That 0.1%, 0.01% or 1% is only a number though and does not specify the harmonic spread and type of distortion and at which levels.
Numbers like that, just as SINAD, is just a number that is easily comparable. And just one (not that important) aspect of signal fidelity.
 
Power. Because you are more prone to hearing clipping than distortion.

granted. if you want a KW amp, it will be higher THD. That said, there are low THD high wattage amps, for a premium price.

But how much is needed for home hifi vs PA system?

I have a McIntosh 2205 from 1975 - 1981 rated at 0.1% THD for 400w (mono) into 0.5 - 4R or 200wpc into 1-8R. Their current lineup of 450 or 600 wpc amps are 0.005%. The MC1.25KW is 0.005% THD. Sure, the current lineup is much more $ than crown.

But for what I would consider suitable for an average home room hifi, does one need much more than 300wpc? 400 wpc? That's should cover a decent sized room with reasonable sensitivity speakers.
 
Hmmm...

Well first I would like to see power ratings measured 20Hz to 20khz, rather than at 1kHz... (the 1khz measure seems to mostly be used for pro/commercial/pa amps?)

Typically, when you measure the same amp at 20-20k, you drop off about 25 to 30% of the power...

So in my head, I estimate that my Crown XLS2500, rated at 440W@8ohm (1khz, 0.5% THD) probably achieves a real 350W @8ohm 20-20khz, 0.01%THD.... However AmirM conveniently tested the 2nd generation version of this amp the XLS2502, and measured it at 310W @8ohm 20-20k, 75db SINAD / 0.018% THD+N (so my rule of thumb is not far off the mark)

My vintage 1980's Quad 606 is rated 140W@8ohm 0.01% THD (20-20k) ie: SINAD 80db

Both of these are excellent sounding amps... I really see no need for an amp specced higher than circa 0.05%THD or SINAD = 66db

Especially given that SOTA speakers struggle to achieve 0.5% THD (!)

So yeah - clearly the THD/SINAD can be used as one measure of engineering excellence (or even engineering hygiene)

But when it comes to the value equation... what is the worth to you of an additional 5db SINAD, above and beyond the threshold of audibility (which is somewhere around 60db / 0.1% THD)...

I picked up the Crown XLS2500 used for US$250.... (around 10 years ago)
The Quad 606 cost me about the same 30 or so years ago

Yes, the Benchmark AHD1 is a SOTA amp with very impressive specifications... but I seriously doubt it would be audibly different.

The value equation just isn't there for me.

In SINAD terms, I require my amps to provide 65db SINAD or more - and that is providing a 5db margin above the threshold of audibility .

Having said that, it is always worthwhile looking at the chart of thd to power.... 99% of the time, your amp is likely to be idling along at 1W and below... or for those who like it loud, in big rooms, 4W and below.... and with a heck of a lot of amps, the power rating provides almost a best case measure whereas the distortion produced at normal operating power levels is often substantially higher (for the Crown, 75db @ 310W, but only 66db @ 1W... and it gets worse with 60db @ 50mW
So when looked at this way, the Crown squeaks into the category I have self-defined, based on threshold of audibility.

I don't have an equivalent graph for the Quad.... I expect it would do better... on the other hand, in actual listening tests in my setup, I could not differentiate between them! (which sort of makes the point about threshold of audibility)

Also although SINAD is important, relatively high THD is inaudible, whereas noise at the same level, is highly audible.
So I prefer specs that break out distortion from noise... SINAD by blending the two, obfuscates the difference between them.
 
Well it's a complex question. If the THD+N of the amplifier is order of magnitude lower than that of the speakers it won't add meaningful difference. So roughly we are talking about 0.01 THD+N or about 75 dB SINAD. But we have different sensitivity for THD across the spectrum and it's lowest in bass and highest in uper mids and lower highs. Luckily speakers also behave similar. I like to push that little higher to 80 dB SINAD across the board and 90 in uper mids and low highs just to be certain even with really great speakers, why not when achievable from technical perspective. Everything above that you won't be able to hear.
Hire things get interesting. Low damping factor of amplifier and load dependance can cause the cuple % shift in speakers response with onse with impedance rise and in such areas and that is something you will be able to hear. Same goes with headphones that are load dependent but other way around if headphones amplifier has high self impedance and as they are low distorsion it's even easier to hear.
 
1kHz is chosen because these distortion products are in our ears most sensitive area . 2nd order is 2kHz for example .

at 10k the 2nd order distortion product is at 20kHz marginally if it's audible for anyone ? especially if you consider the level of it compared to the rest of the music ? given that some electronic music has really huge levels at frequencies extremes most " natural" music does not ?

IF i got this correctly our ability to hear distortion seems to follow the equal loudness contour .

It seems like 1kHZ is chosen to make life easy for amp manufacturers as most amps show rising distortion with frequency . But I thinks its mostly a bit of a coincidence ?

I think the distortion from 100-200Hz to <10kHz in a curve is the most telling .

For example a 250 Hz tone would have higher order distortion products in the middle of our most sensitive hearing ouch ?

Thanks for the IMD articles :)

It provides some tidbits to take home for us that are not electronics engineers .

"an amplifier has no 'understanding' of the waveform at its input, only the instantaneous voltage at any point in time." so goes that arguments that test tones are useless and cant replaces real music for testing ;)

The IMD and THD are interlinked and you cant have one without the other , so the THD is good ( with most 2nd order ) and IMD is bad argument does hold either .
 
IF i got this correctly our ability to hear distortion seems to follow the equal loudness contour .
Yes it reasonableness to it (bottom of the chart) and luckily same goes with analog reproduction devices (drivers, speakers...).
I can't find the graph now, Amir used it cuple times in measurements to show objective performance.
 
Last edited:
It seems like 1kHZ is chosen to make life easy for amp manufacturers as most amps show rising distortion with frequency . But I thinks its mostly a bit of a coincidence ? .
1709627764178.png
 
Why do we hold amplifiers to such high standards?

To me »high standards« not only mean the measurements thing but even so non-audio aspects like this:

• Do I like the visual presence of a given amp in my home? Its looks, its shape, its size proportions? Also its weight or even its smell?

• Do I enjoy touching it? Do I love turning the volume knob (if there is one)? Do I like the click sounds when switching the amp on or off (dang – this is audio after all)?

• Do I feel comfortable with the brand image which I sense the amp is carrying with it, via logo and visual appearance etc?

• And so on and on ...
 
Power. Because you are more prone to hearing clipping than distortion.


And if you wonder why you need so much power, look at the 2216nd impedance curve.

View attachment 354248
Impedance curves are often being wrongly accused... I can replace that curve with a 5 ohm resistor and it will have a "tougher to drive" impedance curve, but if that resistor has a 80dB/V sensitivity it will still be easy to drive like a typical speaker.
 
To me »high standards« not only mean the measurements thing but even so non-audio aspects like this:

• Do I like the visual presence of a given amp in my home? Its looks, its shape, its size proportions? Also its weight or even its smell?

• Do I enjoy touching it? Do I love turning the volume knob (if there is one)? Do I like the click sounds when switching the amp on or off (dang – this is audio after all)?

• Do I feel comfortable with the brand image which I sense the amp is carrying with it, via logo and visual appearance etc?

• And so on and on ...
I think there's a VERY valid place for sensory aspects like this and the best marketers out there in HiFi land understand that. The very best products look and feel delightful, but have the engineering chops internally as well.
 
Impedance curves are often being wrongly accused... I can replace that curve with a 5 ohm resistor and it will have a "tougher to drive" impedance curve, but if that resistor has a 80dB/V sensitivity it will still be easy to drive like a typical speaker.

1709650253694.png


If I look at this, and take something like a Topping LA90 that does 40 W into 8 and 70W into 4, and I look at the 80 Hz bass with 24 ohm impedance, my impression is that you might only have ~15W available there?
 
i think because amps are largely solved technology

we have had good class a/b for as long as many people have been alive here

we have been told what is good, what figures are good and most of us understand what engineering goes into it

and so we sort of know in our guts (partly) what power we think we need + what thd +what ohm... we have LONG expectations that are largely met by many companies at a reasonable price

i also feel that we are less enamoured with playback devices and most of us know DACs are also a known quantity

and so really, it just comes down to amps and speakers... with the last one being the 'final frontier' as with RC
 
I don't think we've seen measurements of the amps inside any of the well-regarded actives yet, but from photos most of them now appear to be custom implementations of class D chip amps. It would be informative to see how the amps in something like a KH-80 or KH-150 stack up against the separates we see measured.
 
My impression is that amplifiers are so good these days that improvements in them would be inaudible. It's pretty much a solved problem. As for tube amplifiers, they can be made to have distortion low enough to be inaudible as well. Thing is why bother with tube amps if they don't add something deemed euphonic to the sound? Some listeners may actually prefer that over near perfect reproduction. It's in the ear of the beholder.
 
If I look at this, and take something like a Topping LA90 that does 40 W into 8 and 70W into 4, and I look at the 80 Hz bass with 24 ohm impedance, my impression is that you might only have ~15W available there?

Speakers are voltage-driven, not power-driven.

When we send 1V of flat frequency response to the speaker (assuming it's own frequency response is flat for this case), 1V at 80Hz will produce as much SPL as 1V into 1kHz in that same speaker, regardless of how much power or current is actually being used at 80Hz.
 
For me it's just about value, once you have a fully transparent amplifier or DAC. For example, we can get the Hypex NCOREx NCx 500 for under $1k and this is about as good as it gets for power amp performance. It generally performs better than it needs to, but if I were to pay around that much or more, it better perform the same or better.

i think because amps are largely solved technology

we have had good class a/b for as long as many people have been alive here

we have been told what is good, what figures are good and most of us understand what engineering goes into it

and so we sort of know in our guts (partly) what power we think we need + what thd +what ohm... we have LONG expectations that are largely met by many companies at a reasonable price

i also feel that we are less enamoured with playback devices and most of us know DACs are also a known quantity

and so really, it just comes down to amps and speakers... with the last one being the 'final frontier' as with RC

I am glad that Class D is also a solved problem these days, that was an important step. The efficiency gains are worth it.
 
Speakers are voltage-driven, not power-driven.

When we send 1V of flat frequency response to the speaker (assuming it's own frequency response is flat for this case), 1V at 80Hz will produce as much SPL as 1V into 1kHz in that same speaker, regardless of how much power or current is actually being used at 80Hz.
And that 1V into 24ohm will require 0.25 amps @ 4ohm, 0.125 amps @8ohm and 0.042amps @ 24ohm (and in power terms the same values as the Voltage multiplier is 1... it will take a massive 0.042W of power @ 24ohm)

For the same output V, as you raise the impedance, the effective "power" halves as the impedance doubles... so 1W@8ohm is equivalent to 0.5W@16ohm.... or 2W@4ohm

Basically raising the impedance causes no issues with any current amp designs that I know of... dropping the impedance below 4ohm (more common than many think) - is where amps get themselves into trouble... it requires more current, more load handling from output transistors, more heat dissipation, and a larger power supply to feed it.
 
For me it's just about value, once you have a fully transparent amplifier or DAC. For example, we can get the Hypex NCOREx NCx 500 for under $1k and this is about as good as it gets for power amp performance. It generally performs better than it needs to, but if I were to pay around that much or more, it better perform the same or better.



I am glad that Class D is also a solved problem these days, that was an important step. The efficiency gains are worth it.
Actual idle measurements on many AB amps don't support the contention of efficiency gains.... It often depends on how heavily A biased the amp is... and techniques of keeping the bias current low while maintaining sound quality (keeping distortion constrained) are well developed and understood.

My (AB) Quad 606 idles at 14W with no signal... whereas my ClassD Crown XLS idles at 35W.

The 606 is a 1980's vintage design, the Crown was released around 2010 I think...

So there is a lot of mythology going on here....

But there are also a bunch of audiophile "heavy Iron" ClassAB amps, that run heavily biased into ClassA.... they typically weigh a ton, have humongous heat sinks, cost a bomb, and heat the room up just as much as most ClassA designs do.... because by default they are designed to run as a Class A amp within "normal" loudness levels, and then to switch to AB/B when large peaks are needed - if you put that switch over point high enough (as some do) - the amp behaves 99.9% of the time like an inefficient power guzzling ClassA amp, even though it is in fact AB.

The Benchmark AHB2 so beloved of many on here, idles at 12W... and is a relative of the 30 years older Quad 606 I have (both use current dumping methods)... 30 years of progress, the idle power is much the same! (but the specs on the AHB2 are current SOTA, the Quad 606 is no longer SOTA....)
 
Back
Top Bottom