Why do we hold amplifiers to such high standards?

[Duke]

Now, the golden standard for a hi-fi amplifier worthy of its name is that it is capable of x watts at @0.1% THD+N (60dBs down)...

I would like to understand the reasoning behind certain criteria.

THD is a failed metric when it comes to predicting perception of distortion. Better metrics exist but they are much more difficult to derive, so THD persists.

how do you notice 0.1% THD if even the best speaker is going to produce, realistically speaking, 0.5-1% THD (that is, 14-20dB higher distortion) at normal listening volumes? 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)?

Loudspeaker harmonic distortion is always very low order and is therefore perceptually relatively benign.

Earl Geddes and Lydia Lee conducted an investigation into distortion perception that resulted in the GedLee Metric, which correlates well with distortion perception. Here is an interview with Earl Geddes cued up to where he begins to discuss correlations between metrics and perception:

 

[GXAlan]

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)

Maybe I don’t truly understand this the way that I think I understand it.

If the Topping LA90 can produce 15V max, how loud could that Polk floorstander play a 1 kHz test tone (at 6 ohms impedance) versus a 70 Hz test tone (at 24 ohms impedance)?

 

[RayDunzl]

Maybe I don’t truly understand this the way that I think I understand it.

I'd view it as the speaker being more efficient (higher impedance = lower power draw) at some frequencies, yet giving approximately the same sound output level.

 

[NTK]

Maybe I don’t truly understand this the way that I think I understand it.

If the Topping LA90 can produce 15V max, how loud could that Polk floorstander play a 1 kHz test tone (at 6 ohms impedance) versus a 70 Hz test tone (at 24 ohms impedance)?

We need to get away from our fixation on power. Loudspeakers (and headphones) SPL output is proportional to voltage, not power. That's why loudspeaker sensitivities are now specified at 2.83 Vrms (which corresponds to the old superseded convention of 1 W into 8 ohm, which also means loudspeaker efficiency is not constant with frequency). The requirement for the amp is for it to be able to deliver the voltage as determined by the input voltage and gain setting, and the current as demanded by Ohm's law.

ANSI/CTA-2034 is also making the attempt to switch the focus on specifying amplifier power to amplifier output voltage. Below is from the "Example report for a Passive Loudspeaker System".

 

[GXAlan]

The requirement for the amp is for it to be able to deliver the voltage as determined by the input voltage and gain setting, and the current as demanded by Ohm's law.

That’s how I understand it. As you have a desired voltage and your amp runs out of current (because the impedance is too low), the amp cannot supply the required current.

On the other hand, if the amp is hitting a max voltage, then the amp is not struggling, but if that max voltage is in the setting of high impedance, my understanding is that the actual SPL at the listening level is poor if I am playing a 70Hz test tone versus a 1 kHz one?



The other way to think about

 

[NTK]

On the other hand, if the amp is hitting a max voltage, then the amp is not struggling, but if that max voltage is in the setting of high impedance, my understanding is that the actual SPL at the listening level is poor if I am playing a 70Hz test tone versus a 1 kHz one?

I can't see why amp driving a loudspeaker at a frequency when it has a high impedance will produce poorer sound than when it has a low impedance. The high impedance load is easier for the amp as it demands less current.

 

[GXAlan]

I can't see why amp driving a loudspeaker at a frequency when it has a high impedance will produce poorer sound than when it has a low impedance. The high impedance load is easier for the amp as it demands less current.

Not about poorer or better sound. My question is strictly about the SPL.

 

[NTK]

Not about poorer or better sound. My question is strictly about the SPL.

SPL is related to voltage. It is how the frequency response plots are made.
[Edit] The Spinorama standard for passive speakers is to measure at a constant 2.83 Vrms from 20 to 20 kHz.

 

[radix]

Maybe Im missing something. I thought it was the current that creates magnetism and thus SPL. Speakers have a dB/W/m rating. An AC signal at 10V will produce different SPLs depending on the impedance. Or is that all just wrong?

 

[NTK]

Maybe Im missing something. I thought it was the current that creates magnetism and thus SPL. Speakers have a dB/W/m rating. An AC signal at 10V will produce different SPLs depending on the impedance. Or is that all just wrong?

It somehow all works out. See this example simulation of a loudspeaker driver using basic physics by COMSOL.

https://doc.comsol.com/5.6/doc/com.comsol.help.models.aco.loudspeaker_driver/models.aco.loudspeaker_driver.pdf

 

[ZolaIII]

Not about poorer or better sound. My question is strictly about the SPL.

Simply put 2xV=+6 dB, 2xW=+3 dB and you can calculate in either but it will be rough calculation as impedance is not constant and low end eats a lot.

 

[wwenze]

Maybe Im missing something. I thought it was the current that creates magnetism and thus SPL. Speakers have a dB/W/m rating. An AC signal at 10V will produce different SPLs depending on the impedance. Or is that all just wrong?

Yea but you need voltage to push that current.

Our speakers are designed such that 1V at 70Hz will create the same SPL as 1V at 1kHz, regardless of the underlying impedance. So further calculation is not needed. And our measurement results confirm this.

 

[GXAlan]

It somehow all works out.

So if we say that the sweep is at 2.83V, then at 15V output, there is +14.5 dB of gain. So the Topping can hit 99 dB at 70 Hz and 104.5 dB at 1 kHz.

At 1 kHz, it is pulling 56W (2.5A) and at 70 Hz, it is pulling 9.4W (0.625A).

If you “could” deliver 56W into 70 Hz, then your amp would need to push 36.6V, and that would be a 22.1 dB gain over 2.83V or 106.6 dB. Even though that is only 1.525A, the LA90 cannot hit 36.6V because this sweep shows that the max voltage is ~15V.

So the impact of the high impedance “somehow works out” is that if you have the FR response curve and maximum voltage of the amp, you can do the math.

The other way we can do it (since I based it off the 4 ohm measurements) is that 24 ohms should be 1/8th of 4 ohm and 1/4 of 8 ohm, or 7W-9W give or take, which matches my 9.5W calculation earlier because it might be easier for the amp to drive at high impedance.

The impact is the “90 dB @ 1 watt” sensitivity. Here, if you assumed the LA90 hit 46W into 6 ohms, you would see 106.6 dB predicted but you don’t really get that.

If you wanted 36.6V into 70 Hz, your amp would need to deliver 335W into 1 kHz, which is how the impedance affects the “power” of the amplifier you need when shopping, even though that is to get to 56W at 24 ohm.

So now, if you wanted to hit 100 dB at 3 meters / 10 ft….

At 15V and 10 ft, your 1 kHz drops to 95 dB. To get to 100 dB, you need 26.7V or an amp that could deliver “180W” into 4 ohms.

However, your 70 Hz also drops to 89.5 dB at 15V but now 10 ft away. To hit 100 dB, I would need 50.2V.

Which means my amplifier would need to be rated at 4 ohms at 630W of power, when shopping, if I wanted to hit 100 dB at 70 Hz at 10 ft, even though it would only be pulling “100W” from the unit.

Again, all assuming voice coils don’t overheat, etc. These numbers are also skewed by the 1 kHz bump in the FR of the speaker I brought up.

So the trick is to look at the FR plot, figure out dB gain needed for desired SPL and distance and then use the impedance to estimate the actual marketing amplifier power needed…

 

[dlaloum]

Maybe I don’t truly understand this the way that I think I understand it.

If the Topping LA90 can produce 15V max, how loud could that Polk floorstander play a 1 kHz test tone (at 6 ohms impedance) versus a 70 Hz test tone (at 24 ohms impedance)?

Assuming it can put 15V into 6ohm... that would be circa 37W @ 6ohm (and 2.5amps which the power supply must provide)

I have seen their "efficiency" quoted as 92db - assuming that this is at 2.83V (1W@8ohm)

then the SPL for a single speaker at 1m would be circa 106db for 15V - the loudness is directly related to the V... and it assumes that the amp can provide the required power and current to be able to keep the V clean

 

[ZolaIII]

@GXAlan that's why there is sustainable max hold, recommend max hold and shoot peak burst (usually differently given in ms). If given (at all) and how accurately milage will depend of (making reputation of it) manufacturer.
Edit:

It clips when it gets stuck and it fals (hard clip) when it can't hold voltage anymore. There are complex factors of generated heat and sustainable leakage (environmental and generally how good cooled, needless to say any pasive is really bad on blowing it off). Heat is the worst enemy no matter how you look at it (life spin and getting more out).

 

[GXAlan]

I have seen their "efficiency" quoted as 92db - assuming that this is at 2.83V (1W@8ohm)

We have this though. SPL for a given frequency at 2.83V. So we should be able to use dB gain relative to voltage to do your calculations. I don’t see where I am wrong, but genuinely had to work things through to get to my conclusion about how everyone is right. You need to worry about voltage if something was perfectly flat and it works out as well as the experience you need amplifiers with a huge amount of rated power for bass even if the simple sensitivity measurement is forgiving.

And

 

[dlaloum]

We have this though. SPL for a given frequency at 2.83V. So we should be able to use dB gain relative to voltage to do your calculations. I don’t see where I am wrong, but genuinely had to work things through to get to my conclusion about how everyone is right. You need to worry about voltage if something was perfectly flat and it works out as well as the experience you need amplifiers with a huge amount of rated power for bass even if the simple sensitivity measurement is forgiving.

View attachment 354544

And
View attachment 354545

If you assume the line on that F/R chart - 85db as the SPL / efficiency spec - then the 106db @15V would drop to circa 99db @15V

The impedance isn't the main issue, some of those peak phase angles might be the real issue ... I wonder if anyone has done an EPDR calc/graph for it?

Either way, it needs a robust amp...

 

[Joseph Crowe]

One thing I’ll point out is that loudspeaker distortion is much lower than most people realize.
The issue surrounds cheap or poor measurement mics and preamps which have a bottleneck for distortion. A good tweeter for example can easily hit 0.01% (-80dB) for upper treble.

 

[wwenze]

Pretty sure a good mic like umik1 can measure 70dB below fundamental with mo distortion

Which also proves we have loudspeakers with no distortion at 70dB below fundamental... At certain frequency ranges. Our concern is usually the bass as well as any non-bass region where it screws up.

Yea but good point raised nonetheless. Instead of simplifying it as Loudspeaker = 0.1% distortion we should consider the entire frequency range which would mean we should have at least -60 to -80 dB (or 0.1% to 0.01%) for the amp to stop being a bottleneck.

 

[ZolaIII]

Even mention of THD and uper highs in same sentence makes me laugh. Second harmonic is A*2 so eventually up to 10 KHz it theoretically has some sense and only theoretically. In practice to about 6 or 6.25 KHz.