Those of you who believe measurements aren't the whole story, do you have a hypothesis why that is?

[steve59]

I'm agreeing with you while adding there are some speaker designs that can bring even the best designed amps to their knees.

 

[mdunjic]

I'm agreeing with you while adding there are some speaker designs that can bring even the best designed amps to their knees.

IMO, any amplifier that is brought to its knees by a speaker (unless speaker impedance drops down to zero ohm, i.e. shorted outputs) may not be considered ‘best designed’ … even then protection should kick in and prevent damage

Any ‘best designed’ power amp shall be able to double its power into halved speaker impedance … all the way at least to 2 ohm and have wide enough flat bandwidth to not distort the signal in any way.

 

[ahofer]

IMO, any amplifier that is brought to its knees by a speaker (unless speaker impedance drops down to zero ohm, i.e. shorted outputs) may not be considered ‘best designed’ … even then protection should kick in and prevent damage

Any ‘best designed’ power amp shall be able to double its power into halved speaker impedance … all the way at least to 2 ohm and have wide enough flat bandwidth to not distort the signal in any way.

As has been widely discussed here, the "doubling" of power is typically a misdirection achieved by lowballing the 8 ohm figure in the first place, a practice that dates back to the 1970s. My old Bryston 3B (1987) allegedly doubled 100/200 but an 8 ohm load tested at nearly 140 watts.

At some point, perhaps 2 ohms as you suggest, but maybe more, speakers that 'bring amps to their knees' are a design fail in and of themselves.

 

[mdunjic]

As has been widely discussed here, the "doubling" of power is typically a misdirection achieved by lowballing the 8 ohm figure in the first place, a practice that dates back to the 1970s. My old Bryston 3B (1987) allegedly doubled 100/200 but an 8 ohm load tested at nearly 140 watts.

At some point, perhaps 2 ohms as you suggest, but maybe more, speakers that 'bring amps to their knees' are a design fail in and of themselves.

That's exactly why we have Amir ... to prove or call that out !

 

[mdunjic]

As has been widely discussed here, the "doubling" of power is typically a misdirection achieved by lowballing the 8 ohm figure in the first place, a practice that dates back to the 1970s. My old Bryston 3B (1987) allegedly doubled 100/200 but an 8 ohm load tested at nearly 140 watts.

At some point, perhaps 2 ohms as you suggest, but maybe more, speakers that 'bring amps to their knees' are a design fail in and of themselves.

Yeah agree, speakers falling to close to 2 ohm aren't well designed probably either ... good point

 

[LTig]

As has been widely discussed here, the "doubling" of power is typically a misdirection achieved by lowballing the 8 ohm figure in the first place, a practice that dates back to the 1970s. My old Bryston 3B (1987) allegedly doubled 100/200 but an 8 ohm load tested at nearly 140 watts.

Yep. A doubling of power when halving the load is technically impossible, at least using bipolar transistors or MOSFET in a class A or AB power stage, even with a regulated power supply. I'm not familiar enough with class D power amps but the limits of the transistors should apply as well.

At some point, perhaps 2 ohms as you suggest, but maybe more, speakers that 'bring amps to their knees' are a design fail in and of themselves.

I agree when we talk about multi way speakers using dynamic drivers. There is no excuse for designs going down to 1 Ohm or less. Some exotic drivers though suffer from very low impedances, e. g. full range electrostatic or ribbon drivers. For those you may need matching amplifiers.

 

[mdunjic]

Yep. A doubling of power when halving the load is technically impossible, at least using bipolar transistors or MOSFET in a class A or AB power stage, even with a regulated power supply. I'm not familiar enough with class D power amps but the limits of the transistors should apply as well.

I agree when we talk about multi way speakers using dynamic drivers. There is no excuse for designs going down to 1 Ohm or less. Some exotic drivers though suffer from very low impedances, e. g. full range electrostatic or ribbon drivers. For those you may need matching amplifiers.

Agree ! What I was describing is an aspiration, but due to the losses it will probably never double its power … but it should come as close as possible … or at least be honest and transparent in the published specs

 

[phoenixdogfan]

Yep. A doubling of power when halving the load is technically impossible, at least using bipolar transistors or MOSFET in a class A or AB power stage, even with a regulated power supply. I'm not familiar enough with class D power amps but the limits of the transistors should apply as well.

I agree when we talk about multi way speakers using dynamic drivers. There is no excuse for designs going down to 1 Ohm or less. Some exotic drivers though suffer from very low impedances, e. g. full range electrostatic or ribbon drivers. For those you may need matching amplifiers.

Probably the biggest design failure of the Original Apogees and Scintillas was the failure to make them actives with matching amplifiers and electronic crossovers.

 

[atmasphere]

Any ‘best designed’ power amp shall be able to double its power into halved speaker impedance … all the way at least to 2 ohm and have wide enough flat bandwidth to not distort the signal in any way.

One problem you are up against right away is the simple fact that every amp made has more distortion into lower impedances (people have challenged me on this statement but when you dig into the specs you see that this is true). By this metric there is no 'best designed' amp. Quite often I've seen that distortion described as 'negligible' but IME it isn't. This is because the ear uses higher ordered harmonics to sense sound pressure and in addition assigns them a tonality of 'bright and harsh'. For this reason you can make any amp sound smoother and more detailed simply by having it drive a speaker of higher impedance.

Also IMO/IME its not important that the amp double its full output power as impedance is halved. It is important instead for it to behave as a voltage source at any volume with which the amp will be played on a given speaker.

 

[mdunjic]

One problem you are up against right away is the simple fact that every amp made has more distortion into lower impedances (people have challenged me on this statement but when you dig into the specs you see that this is true). By this metric there is no 'best designed' amp. Quite often I've seen that distortion described as 'negligible' but IME it isn't. This is because the ear uses higher ordered harmonics to sense sound pressure and in addition assigns them a tonality of 'bright and harsh'. For this reason you can make any amp sound smoother and more detailed simply by having it drive a speaker of higher impedance.

Also IMO/IME its not important that the amp double its full output power as impedance is halved. It is important instead for it to behave as a voltage source at any volume with which the amp will be played on a given speaker.

sure, you are right … amp should aspire and be designed to maintain its voltage gain basically more or less unchanged throughout the whole frequency bandwidth and through the impedance variations of the speakers it is driving

 

[Cbdb2]

Also IMO/IME its not important that the amp double its full output power as impedance is halved. It is important instead for it to behave as a voltage source at any volume with which the amp will be played on a given speaker.

Same thing. If its a perfect voltage source it will double the power. And amps are (if they use feedback) close until they start to current limit. Those numbers are at full power. If you try a 100watt amp at 1watt into 4ohms it will put 2watts into 2ohms.

 

[escksu]

Also IMO/IME its not important that the amp double its full output power as impedance is halved. It is important instead for it to behave as a voltage source at any volume with which the amp will be played on a given speaker.

Its pretty much the same thing. If your voltage remains the same and your resistance is 1/2, your current doubles. High current is what is hard to achieve. You need bigger transformers, more power transistors..... Whats even worse is that BJT current output drops as their temperature increase. Your cost goes up, multiple transistors in parallel has it own design challenges too. You need output resistors.

This is why not that many amps out there (esp. budget ones) are able to drive lower impedance speakers.

 

[escksu]

I'm agreeing with you while adding there are some speaker designs that can bring even the best designed amps to their knees.

Yeah, this is why monster amps from brands like Krell and MArk Levinson exist!!

 

[solderdude]

Also IMO/IME its not important that the amp double its full output power as impedance is halved. It is important instead for it to behave as a voltage source at any volume with which the amp will be played on a given speaker.

A true voltage source will double the output power when impedance is halved. When it does not it is current limited and thus not a voltage source.
This will only happen when current capabilities are reached (often also sagging max output voltage).
So what you wrote (well it is your opinion at least) doesn't jive with what your requirements.

This is because the ear uses higher ordered harmonics to sense sound pressure and in addition assigns them a tonality of 'bright and harsh'.

This too is indeed your opinion and not universally considered a truth. Higher order harmonics would have to be really high (> 0.05%) and not be masked. To affect tonality the distortion would have to be substantial >1% or so.
Only tube amps or amplifiers 'designed by ear' would be able to reach such levels.
Can you point to any research showing what type of distortion and if low levels (say well below 0.1%) would affect tonality and make an amp sound bright or harsh ? Preferably with blind level matched tests as provided evidence ?

 

[mdunjic]

Same thing. If its a perfect voltage source it will double the power. And amps are (if they use feedback) close until they start to current limit. Those numbers are at full power. If you try a 100watt amp at 1watt into 4ohms it will put 2watts into 2ohms.

Exactly

 

[atmasphere]

This too is indeed your opinion and not universally considered a truth. Higher order harmonics would have to be really high (> 0.05%) and not be masked. To affect tonality the distortion would have to be substantial >1% or so.
Only tube amps or amplifiers 'designed by ear' would be able to reach such levels.
Can you point to any research showing what type of distortion and if low levels (say well below 0.1%) would affect tonality and make an amp sound bright or harsh ? Preferably with blind level matched tests as provided evidence ?

That the ear uses higher ordered harmonics to sense sound pressure is well known (take a look at the Radiotron Designer's Guide 3rd edition from the 1930s) and easy to demonstrate with simple test equipment.

What you are saying in your 2nd sentence is what I'm talking about so we are in agreement in this regard.

I think the issue is that of masking as you rightly point out. But it does not stop at some arbitrary point such as 1% THD. It simply depends on how well the lower orders are able to mask higher orders. To give you an example, you may have noticed that some class D amps are remarkably smooth- I know of at least one that I regard to be just as smooth as any tube amplifier (which IMO/IME rely on masking to be as smooth as they are because they actually generate more higher ordered harmonics than most solid state amps). In a class D amp, some of the non-linearities you run into (dead time, issues with the encoding scheme) cause lower ordered harmonic distortion. Even though the amp might be two orders of magnitude lower distortion than a tube amp, its actual distortion signature can look very similar (depending on the design) to that of a tube amp- and so have similar smoothness, although IME greater detail simply on account of it being lower distortion.

In many amplifier designs the feedback is able to suppress the lower orders. But if you don't use enough feedback what happens is you get (thru bifurcation) additional higher orders generated by the feedback itself (see Norman Crowhurst; this is not new information and should not be controversial). The other issue is how feedback falls off at higher frequencies (read: distortion increases above a certain frequency, which is why showing distortion vs frequency is so important) due to a lack of gain bandwidth product; see the link below.

It is arguable that the ear is more sensitive to the higher ordered harmonics than anything else, simply because it does use them to sense sound pressure, and that covers a pretty wide range.

If you've not read this Bruno does a good job putting it in perspective:
https://linearaudio.net/sites/linearaudio.net/files/volume1bp.pdf

Its pretty much the same thing. If your voltage remains the same and your resistance is 1/2, your current doubles. High current is what is hard to achieve. You need bigger transformers, more power transistors..... Whats even worse is that BJT current output drops as their temperature increase. Your cost goes up, multiple transistors in parallel has it own design challenges too. You need output resistors.

This is why not that many amps out there (esp. budget ones) are able to drive lower impedance speakers.

I probably could have worded that better- my point was simply that a design can be a proper voltage source and the fact that it may not be able to double power from its 8 Ohm full power level to 4 Ohms does not affect that. Heck, there are tube amps that act as voltage sources too, but instead of doubling their power from 8 Ohms going down, whatever their 4 Ohm power is they cut that in half as the load impedance is doubled. Expensive way to do it IMO.

 

[solderdude]

That the ear uses higher ordered harmonics to sense sound pressure is well known

It wasn't for me and I suspect that for that to happen the harmonics would need to be very high in level. Quite common in 1930 today most higher harmonics (on any decent design) is > 70dB below the signal level so not an issue.
You can use the program 'distort' to find out your audibility levels.

There has been done quite some research on perception. Just recently there was an interview with @Sean Olive where he mentioned distortion is not a deciding factor in sound quality.
Only very poor (mostly cheap) implementations of switching amps have higher than usual harmonics but near 19kHz the IM distortion increases which can fold back in the audible band.

I don't think class D is 'tube like'. The harmonic spread is different, the levels at which HD (and IM) occur are different and the reaction to load impedances as well as output R differs a lot.

But I agree with @Sean Olive that distortion isn't a very determining factor, certainly not with modern designs. For designs predating 1930 ... sure.

 

[atmasphere]

Just recently there was an interview with @Sean Olive where he mentioned distortion is not a deciding factor in sound quality.

To me this comment seems really suspect! Otherwise why is there an ongoing effort to keep it down? This comment must be out of context.

 

[artburda]

Otherwise why is there an ongoing effort to keep it down?

To be able to market such specs? Why are street cars able to drive faster than 200 km/h?

 

[Momotaro]

As has been widely discussed here, the "doubling" of power is typically a misdirection achieved by lowballing the 8 ohm figure in the first place, a practice that dates back to the 1970s. My old Bryston 3B (1987) allegedly doubled 100/200 but an 8 ohm load tested at nearly 140 watts.

Per John Atkinson "the [Krell] Evolution 600 was a powerhouse, clipping with continuous drive at 610W into 8 ohms (27.85dBW), 1190W into 4 ohms (27.75dBW), and 2.2kW into 2 ohms (27.4dBW). I don't hold my AC wall voltage constant for these tests; at 125.8V with the amplifier quiescent, it had dropped to 121.8V with the amplifier clipping into 4 ohms, 119V with it clipping into 2 ohms." Good enough for me.

At some point, perhaps 2 ohms as you suggest, but maybe more, speakers that 'bring amps to their knees' are a design fail in and of themselves.

You can certainly make that argument (as the departed Mr March did in extremis and contra Magico) but if the amp/speaker combination works for you (or me) it's moot.