Master Thread: Are Measurements Everything or Nothing?

[John_Siau]

Wow John, that looks like a ton of distortion on the competitor's amp! Are you saying there are no other options outside of your amps and Class A?

Seriously, I am trying to visualize the scale on your scope trace, is that thing just broken? Is it even close to typical?

Just asking since there are many ways to make an amp sonically differentiate, and this is one that will stick out especially as volume is reduced, right?

The distortion produced by the other amplifier is typical of many class AB amplifiers. It is rated at 0.02% THD+N with a SNR of 105 dB. Rated output power is 150 Watts into 8 ohms. 0.02% THD+N is -74 dB near full output.

This means that the THD is -74 dB relative to a high-level signal while the noise is 105 dB relative to maximum output. Like many class AB amplifiers, the actual amplitude of the THD does not change much with signal level. At 1 Watt, 0.1 Watt, and 0.01 Watt, the THD is still about -74 dB relative to the rated output of the amplifier. In other words, the amplifier distortion was always produced at about the same level (due to the fixed-level push-pull crossover distortion of the output stage). This distortion level was high enough to produce a distortion SPL of 14 dB at the listening position when using speakers with an 87 dB sensitivity. This is not unusual in any way. This test clearly exposes the shortcomings of traditional class AB amplifiers.

See more information about these measurements.

The distortion residual waveforms for both amplifiers have exactly the same scale factor. The trace for the distortion residual is boosted by 60 dB relative to the trace for the output waveform (top trace). There is a 35 dB difference in THD between the AHB2 and a typical class AB amplifier when operating at 0.01 dB. This is a huge difference and the difference is clearly audible at low test levels.

The THD from the traditional class AB amplifier was 14 dBSPL at the listening position and, in contrast, the THD from the AHB2 was -21 dBSPL. 14 dB above the threshold of hearing vs. 21 dB below the threshold of hearing.

Keep in mind that these differences we clearly audible through a set of speakers!

 

[Titurel]

Here is the evidence:

At Benchmark we ran an ABX test using a 1 kHz low-level tone. We compared a class AB amplifier to the Benchmark AHB2. The test level was 0.01 watt, producing an SPL of 67 dB at the listening position. The measured distortion produced by the class AB amplifier was reproduced at a level of 14 dB SPL. In contrast, the measured distortion produced by the AHB2 was well below 0 dBSPL (actual level was calculated at -21 dBSPL). Input voltages to the speaker terminals were set to 0.2828 Vrms using a precision volt meter.

Given these well-controlled conditions it was very easy to hear the difference between the two amplifiers in a fully-blind random ABX test. I scored 100% correct on every ABX test sequence I ran. Other listeners achieved similar results.

You can read the entire writeup here: "Power Amplifiers - A "First Watt" ABX Test"

Given the results of this test, there is no question that power amplifiers can sound different even when they have respectable specifications at high output levels. This specific test shows that we should pay more attention to the 0.01 watt THD.

Well done. I think this could very well nail the most significant area under dispute - how an amp behaves at its lowest levels (0- 0.1 watt), representing power demands of most home systems most of the time.

In a nutshell, we shouldn't assume a given suite of tests will yield meaningful results just because they are the bog standard suite. One should not just evaluate the object under test, but the test assumptions as well. Again, well done.

 

[hwest]

You clearly lack the capacity to listen (to statements of fact about audibility thresholds) and learn (that your listening impressions are influenced by non-sonic factors).

Not sure how much longer the mods are going to tolerate your continued statement of impossibilities as fact.

We are old school science here, not old school snake oil.

So you're one of those that believe the Marantz sounds as good or better than the AVM 90 due to how the numbers look is what you're saying and now I'm snake oil because I disagree with that as many many folks do? I see.

You clearly lack the capacity to listen (to statements of fact about audibility thresholds) and learn (that your listening impressions are influenced by non-sonic factors).

Not sure how much longer the mods are going to tolerate your continued statement of impossibilities as fact.

We are old school science here, not old school snake oil.

So you're one of those that believe the Marantz sounds as good or better than the AVM 90 due to how the numbers look is what you're saying and now I'm snake oil because I disagree with that as many many folks do? I see.

 

[hwest]

Well, you know what they say about opinions and certain bodily organs

Luckily, we have objective metrics.

Again, throw all the numbers you like the Anthem 90 beats the AV 10 easily in sound quality it's not up for debate everyone agrees that has tested both not just me so something is off if you're suggesting otherwise.

 

[hwest]

Well, you know what they say about opinions and certain bodily organs

Luckily, we have objective metrics.

You're sticking to numbers that aren't giving you great sound for whatever reason so you trust your numbers I'll trust what most people would chose myself included, AVM 90 over any Marantz or Denon period.

 

[hwest]

That's what we are all paying attention to. Anecdote vs data/evidence debates tend to reach a dead end pretty quickly around here.

No evidence? Not much more to say.

Tons of numbers out there. The numbers are close but the 90 is just a much better sounding unit and the popular opinion is just that so that's real world. Perhaps those big box vendors can tell you why just slightly better numbers result in a worse sounding unit because that's real world stuff.

 

[Chagall]

@hwest I think you have made your "point" - you don't like numbers. Anything else or is it just spam at this point?

 

[solderdude]

Perhaps those big box vendors can tell you why just slightly better numbers result in a worse sounding unit because that's real world stuff.

Vendors are only interested in selling items. Usually the more expensive they are or the higher margin they get is what they recommend.
They resort to flowery language because they NEED a sale so talk along with the (potential) customer towards the best possible sale.
Why listen to or trust recommendations from them ?

 

[John_Siau]

Well done. I think this could very well nail the most significant area under dispute - how an amp behaves at its lowest levels (0- 0.1 watt), representing power demands of most home systems most of the time.

In a nutshell, we shouldn't assume a given suite of tests will yield meaningful results just because they are the bog standard suite. One should not just evaluate the object under test, but the test assumptions as well. Again, well done.

Given the measured 1W, 0.1W and 0.01W performance of the class AB amplifier under test, we could predict that the THD would exceed 0 dBSPL at all of these power levels. Given our speakers and listening position we predicted a 14 dB SPL level.

The ABX listening tests proved that a distortion playback level of just 14 dBSPL was easily detected in the presence of a 67 dBSPL 1 kHz sine tone.

At these low levels, the THD (measured in SPL) produced by the class AB amplifier probably exceeded the THD (measured in SPL) produced by the speakers, although we don't have any measurement data to confirm this. At the very least, the two sources of distortion were spectrally different.

The ABX listening test proved that the THD produced by the speakers was not high enough to mask the THD produced by the amplifier. It also proves that the 67 dBSPL tone did not mask the 14 dBSPL push-pull crossover distortion of the class AB amplifer.

The crossover distortion contains many high-order harmonics and I suspect that this makes it more audible than we might expect otherwise.

 

[MAB]

The distortion produced by the other amplifier is typical of many class AB amplifiers. It is rated at 0.02% THD+N with a SNR of 105 dB. Rated output power is 150 Watts into 8 ohms. 0.02% THD+N is -74 dB near full output.

This means that the THD is -74 dB relative to a high-level signal while the noise is 105 dB relative to maximum output. Like many class AB amplifiers, the actual amplitude of the THD does not change much with signal level. At 1 Watt, 0.1 Watt, and 0.01 Watt, the THD is still about -74 dB relative to the rated output of the amplifier. In other words, the amplifier distortion was always produced at about the same level (due to the fixed-level push-pull crossover distortion of the output stage). This distortion level was high enough to produce a distortion SPL of 14 dB at the listening position when using speakers with an 87 dB sensitivity. This is not unusual in any way. This test clearly exposes the shortcomings of traditional class AB amplifiers.

View attachment 383981
See more information about these measurements.

The distortion residual waveforms for both amplifiers have exactly the same scale factor. The trace for the distortion residual is boosted by 60 dB relative to the trace for the output waveform (top trace). There is a 35 dB difference in THD between the AHB2 and a typical class AB amplifier when operating at 0.01 dB. This is a huge difference and the difference is clearly audible at low test levels.

The THD from the traditional class AB amplifier was 14 dBSPL at the listening position and, in contrast, the THD from the AHB2 was -21 dBSPL. 14 dB above the threshold of hearing vs. 21 dB below the threshold of hearing.

Keep in mind that these differences we clearly audible through a set of speakers!

The amp doesn't look properly biased.
As you said, this distortion will be more audible as volume is reduced.

 

[John_Siau]

The amp doesn't look properly biased.
As you said, this distortion will be more audible as volume is reduced.

You are looking at a highly amplified distortion residual. Biasing never fully removes crossover distortion. This amplifier was meeting or exceeding all of its published specifications.

To eliminate audible effects, the crossover distortion needs to be about -90 dB (depending on speaker efficiency) relative to 1 watt and this in not achieved by most class AB amplifiers. This fault gets worse if you examine the effects of temperature changes, load impedance, and load phase angle.

 

[BDWoody]

...the 90 is just a much better sounding unit and the popular opinion is just that...

So you keep claiming.

When you have evidence that supports these claims let me know. Until then, I think we're done here.

 

[pma]

At Benchmark we ran an ABX test using a 1 kHz low-level tone. We compared a class AB amplifier to the Benchmark AHB2. The test level was 0.01 watt, producing an SPL of 67 dB at the listening position. The measured distortion produced by the class AB amplifier was reproduced at a level of 14 dB SPL. In contrast, the measured distortion produced by the AHB2 was well below 0 dBSPL (actual level was calculated at -21 dBSPL). Input voltages to the speaker terminals were set to 0.2828 Vrms using a precision volt meter.

At these low levels, the THD (measured in SPL) produced by the class AB amplifier probably exceeded the THD (measured in SPL) produced by the speakers, although we don't have any measurement data to confirm this. At the very least, the two sources of distortion were spectrally different.

Would you mind mentioning which class AB amplifier had such high distortion at low power? I am used to get typical results of THD vs. level like this.

 

[MAB]

You are looking at a highly amplified distortion residual.

You already said this.

Biasing never fully removes crossover distortion.

Agreed.

This amplifier was meeting or exceeding all of its published specifications.

Which one is it?

To eliminate audible effects, the crossover distortion needs to be about -90 dB (depending on speaker efficiency) relative to 1 watt and this in not achieved by most class AB amplifiers. This fault gets worse if you examine the effects of temperature changes, load impedance, and load phase angle.

I am surprised by your competitor's amplifier.

 

[antcollinet]

but the 90 is just a much better sounding unit

Another totally unevidenced claim. What are you not understanding about:

Sighted listening comparisons (or poorly controlled blind listening) are not reliable as evidence. It doesn't matter how many people do them and think the same as you. We are far into the statement "the plural of anecdote, is not data"

 

[SIY]

We compared a class AB amplifier to the Benchmark AHB2.

Was it the one with misadjusted bias that was used in your marketing blurb?

The distortion produced by the other amplifier is typical of many class AB amplifiers. It is rated at 0.02% THD+N with a SNR of 105 dB. Rated output power is 150 Watts into 8 ohms. 0.02% THD+N is -74 dB near full output.

This means that the THD is -74 dB relative to a high-level signal while the noise is 105 dB relative to maximum output. Like many class AB amplifiers, the actual amplitude of the THD does not change much with signal level. At 1 Watt, 0.1 Watt, and 0.01 Watt, the THD is still about -74 dB relative to the rated output of the amplifier. In other words, the amplifier distortion was always produced at about the same level (due to the fixed-level push-pull crossover distortion of the output stage). This distortion level was high enough to produce a distortion SPL of 14 dB at the listening position when using speakers with an 87 dB sensitivity. This is not unusual in any way. This test clearly exposes the shortcomings of traditional class AB amplifiers.

View attachment 383981
See more information about these measurements.

The distortion residual waveforms for both amplifiers have exactly the same scale factor. The trace for the distortion residual is boosted by 60 dB relative to the trace for the output waveform (top trace). There is a 35 dB difference in THD between the AHB2 and a typical class AB amplifier when operating at 0.01 dB. This is a huge difference and the difference is clearly audible at low test levels.

The THD from the traditional class AB amplifier was 14 dBSPL at the listening position and, in contrast, the THD from the AHB2 was -21 dBSPL. 14 dB above the threshold of hearing vs. 21 dB below the threshold of hearing.

Keep in mind that these differences we clearly audible through a set of speakers!

Apparently so. This is disappointing for an engineer who has created some absolutely excellent products.

 

[SIY]

Just for reference, here's the residuals from the last two engineered AB amps I measured. Parasound is dominated by plain HD, the Cambridge by noise. No signs of crossover distortion because the bias was not misadjusted.

 

[John_Siau]

Would you mind mentioning which class AB amplifier had such high distortion at low power? I am used to get typical results of THD vs. level like this.

View attachment 383993

Lets examine the A250W4R.

Here is a more detailed THD curve for the A250W4R:

From this we can see that the THD is about 0.003% at 1 Watt. Converting % to dB we have -90.5 dB. At this 1 W output level, the THD measures -90.5 dBW. This means that the SPL of the THD would be -0.5 dBSPL at 1 meter from a 90 dB efficiency speaker (and about the same level at a typical listening position). In other words, this very good class AB amplifier just barely achieves the required numbers with resistive loading in ideal conditions. If you do the math, the amplifier we tested had a 1kHz THD that was 17 dB higher (0.02% THD at 1 W, 0.1W and at 0.01W).

Our ABX listening tests showed that 0.02% THD at 0.01 W was clearly audible. The above curves suggest that the A250WR4 achieves about 0.003% THD at 1W, 0.1W, and 0.01W. Note that the noise level is about the same as the THD level, so it will be hard to see the crossover distortion when viewing the residual on a scope. The 150W class AB amplifier we tested had a SNR of 105 dB, and if you do all of the math, you will find that the 1W THD was 10.3 dB higher than the noise (making it easier to see the distortion residual on the scope).

Based on the above curves, the SPL produced by the crossover distortion from the A250W4R (with 90 dB speakers) should be at (or slightly below) the threshold of hearing when the biasing is perfectly adjusted and load conditions are ideal. If you run the same calculations on most class AB amplifiers, you will find that most fail to achieve this level of performance. If you are using high-efficiency horns, the A250W4R may produce audible distortion at low power.

As you make these calculations it is best to do the math with the THD expressed in terms of dBW (dB relative to 1 W). This makes it easy to convert to dBSPL. Add the THD in dBW to the speaker efficiency to calculate the SPL. The single-speaker SPL at 1 meter will be very close to the two-speaker SPL at the listening position in a typical room.

Here is a set of calculators that make these conversions easier.

 

[voodooless]

So what about noise here? You’re about 1.5m from the speaker. Poor SNR will be much more noticeable than the distortion components, and best of all, you can pass an ABX without even needing a signal present if it’s poor enough.

 

[pma]

Lets examine the A250W4R.

Here is a more detailed THD curve for the A250W4R:

View attachment 383994
From this we can see that the THD is about 0.003% at 1 Watt. Converting % to dB we have -90.5 dB. At this 1 W output level, the THD measures -90.5 dBW. This means that the SPL of the THD would be -0.5 dBSPL at 1 meter from a 90 dB efficiency speaker (and about the same level at a typical listening position). In other words, this very good class AB amplifier just barely achieves the required numbers with resistive loading in ideal conditions. If you do the math, the amplifier we tested had a 1kHz THD that was 17 dB higher (0.02% THD at 1 W, 0.1W and at 0.01W).

Our ABX listening tests showed that 0.02% THD at 0.01 W was clearly audible. The above curves suggest that the A250WR4 achieves about 0.003% THD at 1W, 0.1W, and 0.01W. Note that the noise level is about the same as the THD level, so it will be hard to see the crossover distortion when viewing the residual on a scope. The 150W class AB amplifier we tested had a SNR of 105 dB, and if you do all of the math, you will find that the 1W THD was 10.3 dB higher than the noise (making it easier to see the distortion residual on the scope).

Based on the above curves, the SPL produced by the crossover distortion from the A250W4R (with 90 dB speakers) should be at (or slightly below) the threshold of hearing when the biasing is perfectly adjusted and load conditions are ideal. If you run the same calculations on most class AB amplifiers, you will find that most fail to achieve this level of performance. If you are using high-efficiency horns, the A250W4R may produce audible distortion at low power.

As you make these calculations it is best to do the math with the THD expressed in terms of dBW (dB relative to 1 W). This makes it easy to convert to dBSPL. Add the THD in dBW to the speaker efficiency to calculate the SPL. The single-speaker SPL at 1 meter will be very close to the two-speaker SPL at the listening position in a typical room.

Here is a set of calculators that make these conversions easier.

You are quoting an old measurement that is limited by the noise of the older measuring system. Thus, the results below 1W are useless and reflect only system noise. All your deductions from the old graph thus are pointless. Please stick with the new measurement of the same amplifier, otherwise your post is misleading.
Care must be taken if we want to measure at low level and best way is to use an autoranging instrument. Which I do not own.