Amplifier Power Ratings - Inflated power specs (Audioholics)

[pedrocols]

I really wish we could limit engineers' participation. Is there a session in this forum just for engineers?

[audiosyndrome]

Wow. Someone has either a very short attention span or a preference for audiophile nonsense. 😉

Russ

[DYohn]

I say the obsession with measuring into a reactive load is moot to the discussion about measuring amplifier power ratings, not that it does not matter at all.

A standard sets out a method for testing. Any two products tested using the same standard creates a valid method for comparing their performance (on paper.) There is nothing wrong with that statement. That's what standards are for. Everyone uses standards, and the best tools comply to published standards.

What audio analyzer does Emotiva use? Does it comply to CTA-490? If not, what standard does it use as the basis for its tests? How do you calibrate it?

[KeithL]

Not "none" by any means... but it is true that a typical switching supply may have less "total joules of stored power reserve" than a linear power supply of similar power.

But, then, because it operates at a higher frequency, a switcher can also "refill its reserves" much more quickly than a linear supply operating at a mere 60 Hz.

"Power reserves" must be considered at a variety of time scales....

At one scale "big capacitors" may enable an amplifier to continue to play for several seconds if the lights go out.

(But, to be fair, if the lights go out, a few more seconds of music aren't going to matter much.)

But, from the opposite perspective, a linear power supply only has the opportunity to recharge those big capacitors 120 times per second (at the peak of each half-wave cycle).

Between those energizing pulses they have a lot of time to drain down - which is why they need to be so large.

And, when those capacitors are charged, they are always charged to some percentage of the peak line voltage, whatever that happens to be at the time.

This offers the opportunity for more unnecessary and unwanted variations in operating voltage.

In contrast, our SMPS also uses the line voltage to fill a set of primary storage capacitors.

However, the power it delivers at the output is taken from a separate set of secondary storage capacitors.

The regulation circuitry carefully monitors the output voltage, compares it to a reference standard, and moves more power to the output as needed, "to make sure the output voltage remains what it should be".

Unlike the linear supply, rather than simply deliver "some percentage of the line voltage", our SMPS actually delivers a consistent REGULATED output voltage.

And, unlike the linear supply, it has the opportunity to correct this voltage hundreds of thousands of times a second if necessary to ensure that it remains precisely what it should be.

(The voltage in the primary storage can be allowed to vary considerably; rather than being used by the amplifier itself that power is simply used as "raw materials" for producing the regulated output voltage.)

Under normal operating conditions, the REGULATED power provided by our SMPS provides cleaner and more consistent performance than the UNREGULATED power provided by a normal linear power supply.
(We don't normally "term drop"..... but see how many of our competitors' amplifiers power their main supply rails from a REGULATED power supply....)

Jun 7, 2019 7:51:25 GMT -5 mgbpuff said:

Jun 6, 2019 22:03:14 GMT -5 DYohn said:

Look up CTA-490. OrCTA-2000 for mobile amps. And your obsession with reactive loads is moot. A standard is a standard.

In one fell swoop you have managed to diss and dispel an entire branch of mathematics involving imaginary number systems and therefore the entire science of explaining AC power and circuitry. The power cube method, although a much more thorough test than a purely resistive one, still does not address the issue of stored power reserve (which a switching power supply has none)!

[mgbpuff]

Unfortunately, the voltages that occur in a highly inductive circuit are not 'regulated' in that they respond to L (di/dt) and may be many times above the nominal power supply. Therefore a 'over designed' but non regulated power supply may be more desirable than a regulated one.

[KeithL]

We use Audio Precision test sets for all of our automated tests....which are the industry standard in the audio industry.

We have quite a few of them, but most of the ones we have are variations of the AP-58x series (AP-585's mostly).

(Calibration is as per the manufacturer's recommendations.... although I'm not involved in the details.)

CTA-490 is a standard for how to perform a particular set of tests.

The AP-585 is a general purpose test set.

It measures whatever you tell it - however you tell it to do so - from a large library of available functions.

It can be programmed to perform pretty well any test, from any standard, that you choose to program in.

The AP does come with a long list of preprogrammed "projects" - and I see that "CTA-490 " is included on that list.

(However, it's not one of the ones we normally use, and I'm not familiar with it.)

We actually do our power testing using load banks assembled from really big 1 kW non-inductive ceramic resistors.

And, of course, when designing new models, we actually listen to them a lot with a variety of actual speakers.

We've considered testing using some sort of accurate model of a loudspeaker.

However, to be quite blunt, there wouldn't be much purpose in doing so, and publishing the results, because there's nothing much to compare them to anyway.

We comply with the current version of the FTC Power Amplifier Testing Regulations (whatever the latest revision of the the original 1972 version is) - where appropriate.

This the only "legally official" standard for testing audio power amps - and only specifies a very few ratings - under very specific circumstances.

(That standard started out being very concise... and very narrow... but has become much less precise, and much more convoluted, with every revision and proposed revision.)

It's also worth mentioning that modern test equipment is rather different than test equipment from the old days.

For example, on a vintage THD test set...

I would connect my amplifier to a load resistor...

I would then set my test oscillator to put out a clean sine wave...

Connect a probe from my meter to that load resistor...

Configure some sort of "nulling filter" in my test set to block out the test frequency sine wave... (so I can look at "everything else")

And anything that shows on the meter after the original signal has been removed would be "THD + noise".

(And there would be calibrations for the test oscillator, the input amplifiers, the meter, etc.)

On an AP, the first part is the same...

I would connect my amplifier to a load resistor...

I would then set my test oscillator to put out a clean sine wave...

(The test oscillator is built in, and crystal controlled, so it's inherently accurate to some small number of parts per million.)

Connect a probe from my test set to that load resistor...

However, on the AP, there is no nulling process...

The AP will simply analyze the entire signal and provide whatever analysis I configure it to...

I can then ask it to display the sum of all the harmonics (THD)...

or to show me just the 7th harmonic...

Or to give me a graph showing the relative level of each of the first ten harmonics or so...

Or to give me a spectral analysis of just the noise (not counting the harmonics)...

And, if I'm testing something other than an amplifier, which has the potential to alter the frequency of the signal...

I can ask it to display the frequency... accurate to quite a few decimal places...

Or to generate a graph showing how that frequency varies... over the next second... or the next minute... or the next day...

And I can have it show me the results in real-time, or average several readings over time, to get a more accurate reading...

(I could even get a graph of, on average, how much the individual readings are varying from the average.)

In the old days, when equipment did one specific thing, the trick was to decide what test equipment you really needed...

Nowadays, when an AP can do seventy or eighty "canned tests", and an infinite number of custom programmed ones, the trick is deciding what you WANT to measure or look at, and how.

Jun 7, 2019 9:22:51 GMT -5 DYohn said:

I say the obsession with measuring into a reactive load is moot to the discussion about measuring amplifier power ratings, not that it does not matter at all.

A standard sets out a method for testing. Any two products tested using the same standard creates a valid method for comparing their performance (on paper.) There is nothing wrong with that statement. That's what standards are for. Everyone uses standards, and the best tools comply to published standards.

What audio analyzer does Emotiva use? Does it comply to CTA-490? If not, what standard does it use as the basis for its tests? How do you calibrate it?

[DYohn]

Perfect.

And by the way I only mentioned the CTA standards because as a former member I have read them thoroughly. Like I said, there are many valid standards.

www.cta.tech/

[AudioHTIT]

I think there’s actually some useful information given by ALSO including the single and dual channel ratings. Some may use their multichannel amps to listen to stereo recordings and knowing the performance under those conditions is useful — my XPA-7 G2 rated 200Wx7 ACD, is rated 520W 2CD.

Now it’s not likely someone would use a multi-channel amp as a monoblock (though it’s possible), but a loud sound effect moving from channel to channel could take advantage of power supply reserves and single channel capabilities (what was referred to as headroom). Which is to say, as long as the ACD ratings are included, more lower channel ratings could help a buyer make a decision when choosing an amp.

As to where the ACD rating should be placed, the Monk* in me wants to see them in order, so 1, 2, 7 works for me; but 7, 2, 1 would be fine as well, and might be more appropriate for how most would use the amp, but being called out as ‘slipping’ for not putting ACD first seems a bit petty (and I’m a big believer in standards).

* For those who don’t know the reference to TV’s Detective Monk (and his sometimes paralyzing obsessive compulsive disorder), I thought OCD in this context could be confusing.

[KeithL]

The output of an amplifier is, by definition, "regulated". (The technical term is "load invariant".)

(We are talking about the voltage that exists at the output terminals of the amplifier.)

A "normal audio amplifier" is a "voltage amplifier" and is expected to deliver to the speaker a specified multiple of the voltage of the input signal.

(In other words, the output voltage of the amplifier is specified and is not supposed to change regardless of the load.)

The only "issue" with a reactive load is that, in order to maintain that relationship, the amplifier may need to source or sink more current than it would if the load was simply a resistor.

So, for example, maintaining 1V across an 8 Ohm pure resistor requires a certain amount of current - which will be the same at 20 Hz as at 20 kHz.

However, if you put a 1 uF capacitor in parallel with the resistor, you now have a reactive load.

And, with that reactive load, the amplifier will have to deliver a LOT more current at 20 kHz than it will at 20 Hz to continue to deliver that same 1V.

(And it's quite possible that a certain amplifier model may be unable to do so... which might cause it to distort, or even fail, if it tries.)

In fact, as long as the amplifier maintains this relationship properly, how it does so, and what sort of power supply it has, is totally irrelevant.

In general, when you design audio equipment, at the first level of approximation, you assume a regulated supply.

For example, you say "the amplifier runs on +/- 40V rails"... as such you are describing a regulated power supply.

For a non-regulated supply, running on 60 Hz, those rails will be: "between +/- 35V and +/- 44V, varying according to a 120 Hz pseudo-sawtooth wave, whose shape will vary depending on the load".

(For those who prefer frequency terminology that would be: "+/- 40 vDC, with a variety of frequency components added, mostly harmonics of 120 Hz, varying with load".)

(Luckily most modern amplifier designs are very tolerant of this sort of variation - so it doesn't cause too much of a compromise in their performance.)

At BEST, an unregulated supply may be "close enough to the ideal to avoid causing problems"....

And, under certain circumstances, it may have other virtues....

However, in practice, all of those virtues are variations on a theme: "overloading more gracefully".

Comparing two power supplies, both designed to have the correct capabilities, a regulated supply will always perform better.

However, comparing two UNDERDESIGNED power supplies, a non-regulated power supply may have fewer problems if you overtax it.

(And, since an "overdesigned" power supply will never be called upon to reach its limits, any type should work equally well.)

(We actually did design the SMPS in our Gen3 amps to act somewhat more gracefully than many other SMPS would if you do manage to exceed its capabilities.

However, since it can deliver nearly 3 kW of actual power under most circumstances, this doesn't seem to happen very often.   )

Jun 7, 2019 9:50:42 GMT -5 mgbpuff said:

Unfortunately, the voltages that occur in a highly inductive circuit are not 'regulated' in that they respond to L (di/dt) and may be many times above the nominal power supply. Therefore a 'over designed' but non regulated power supply may be more desirable than a regulated one.

[KeithL]

Indeed......

And the marketing department likes to put "the most impressive specs first"...

(Or, perhaps, the specs people are most likely to look at and compare first.)

Jun 7, 2019 10:48:20 GMT -5 AudioHTIT said:

I think there’s actually some useful information given by ALSO including the single and dual channel ratings. Some may use their multichannel amps to listen to two stereo recordings and knowing the performance under those conditions is useful — my XPA-7 G2 rated 200Wx7 ACD, is rated 520W 2CD.

Now it’s not likely someone would use a multi-channel amp as a monoblock (though it’s possible), but a loud sound effect moving from channel to channel could take advantage of power supply reserves and single channel capabilities (what was referred to as headroom). Which is to say, as long as the ACD ratings are included, more lower channel ratings could help a buyer make a decision when choosing an amp.

As to where the ACD rating should be placed, the Monk* in me wants to see them in order, so 1, 2, 7 works for me; but 7, 2, 1 would be fine as well, and might be more appropriate for how most would use the amp, but being called out as ‘slipping’ for not putting ACD first seems a bit petty (and I’m a big believer in standards).

* For those who don’t know the reference to TV’s Detective Monk (and his sometimes paralyzing obsessive compulsive disorder), I thought OCD in this context could be confusing.

[Deleted]

Most of the amplifier standards were agreed upon when stereo was the norm.
Power supply design has changes so much in recent years & become a science in itself.
As Keith mention, and is a big point that may have been missed, today we want multichannel (5, 7, 11) to perform consistently as a 2 channel. It's just not fair to expect it.
"SMPS actually delivers a consistent REGULATED output voltage." - that is the secret to make multichans work & sound good.
Having a single power supply to supply multiple amps is not as easy as some may think;. Ask yourself why monobocks are so loved.

[mgbpuff]

Jun 7, 2019 10:54:45 GMT -5 KeithL said:

The output of an amplifier is, by definition, "regulated". (The technical term is "load invariant".)

(We are talking about the voltage that exists at the output terminals of the amplifier.)

A "normal audio amplifier" is a "voltage amplifier" and is expected to deliver to the speaker a specified multiple of the voltage of the input signal.

(In other words, the output voltage of the amplifier is specified and is not supposed to change regardless of the load.)

The only "issue" with a reactive load is that, in order to maintain that relationship, the amplifier may need to source or sink more current than it would if the load was simply a resistor.

So, for example, maintaining 1V across an 8 Ohm pure resistor requires a certain amount of current - which will be the same at 20 Hz as at 20 kHz.

However, if you put a 1 uF capacitor in parallel with the resistor, you now have a reactive load.

And, with that reactive load, the amplifier will have to deliver a LOT more current at 20 kHz than it will at 20 Hz to continue to deliver that same 1V.

(And it's quite possible that a certain amplifier model may be unable to do so... which might cause it to distort, or even fail, if it tries.)

In fact, as long as the amplifier maintains this relationship properly, how it does so, and what sort of power supply it has, is totally irrelevant.

In general, when you design audio equipment, at the first level of approximation, you assume a regulated supply.

For example, you say "the amplifier runs on +/- 40V rails"... as such you are describing a regulated power supply.

For a non-regulated supply, running on 60 Hz, those rails will be: "between +/- 35V and +/- 44V, varying according to a 120 Hz pseudo-sawtooth wave, whose shape will vary depending on the load".

(For those who prefer frequency terminology that would be: "+/- 40 vDC, with a variety of frequency components added, mostly harmonics of 120 Hz, varying with load".)

(Luckily most modern amplifier designs are very tolerant of this sort of variation - so it doesn't cause too much of a compromise in their performance.)

At BEST, an unregulated supply may be "close enough to the ideal to avoid causing problems"....

And, under certain circumstances, it may have other virtues....

However, in practice, all of those virtues are variations on a theme: "overloading more gracefully".

Comparing two power supplies, both designed to have the correct capabilities, a regulated supply will always perform better.

However, comparing two UNDERDESIGNED power supplies, a non-regulated power supply may have fewer problems if you overtax it.

(And, since an "overdesigned" power supply will never be called upon to reach its limits, any type should work equally well.)

(We actually did design the SMPS in our Gen3 amps to act somewhat more gracefully than many other SMPS would if you do manage to exceed its capabilities.

However, since it can deliver nearly 3 kW of actual power under most circumstances, this doesn't seem to happen very often.   )

Jun 7, 2019 9:50:42 GMT -5 mgbpuff said:

Unfortunately, the voltages that occur in a highly inductive circuit are not 'regulated' in that they respond to L (di/dt) and may be many times above the nominal power supply. Therefore a 'over designed' but non regulated power supply may be more desirable than a regulated one.

As long as you are talking about a voltage controlled wave form, and most amplifiers are voltage controllers, you are correct about the regulated voltage power supply being superior (provided it is adequate). But magneto restrictive transducers such as dynamic cone speakers are intrinsically current responsive devices rather than voltage responsive. Therefore a power supply approximating an ideal current source would be more appropriate than one approximating an ideal voltage source. If the transducer is electrostatic in nature, then voltage control is the more proper approach. Why aren't there more current controlled amplifiers like some of the ones that Nelson Pass has designed?

[audiobill]

Blah, blah, so what does it sound like or does that not matter ?

[KeithL]

Let me fill you in on a few more details...

First off, it sort of goes beyond"most".

Show me a single commercially produced audio amplifier that truly operates in "current output mode".

(You'll see it used in other applications.... probably ranging from motor controllers to electric jackhammers...)

Yes, even an ordinary loudspeaker is really "intrinsically current drive in nature"

However the currently existing examples have been optimized to work with voltage drive amplifiers.

The main reason you don't see current drive amplifiers is that they wouldn't work well at all with virtually all currently available speakers.

(I wouldn't be surprised if some of the dedicated amplifiers inside some subwoofers may turn out to operate in current mode or some sort of hybrid mode.... but I don't offhand know of any.)

For example, with a typical loudspeaker, there is an impedance peak at resonance, as well as a peak in mechanical output efficiency.
With a voltage drive amplifier, this produces a drop in current at the impedance peak, which partially cancels out the peak in efficiency.

(At resonance, the efficiency goes up, but the impedance goes up, while the voltage stays the same, so the current goes down, and those two partially cancel each other out.)

However, with a current drive amplifier, the impedance peak does not cause a drop in current (current drive delivers the same current regardless of impedance).

(At resonance, the efficiency goes up, the impedance goes up, but the current stays the same, so the acoustic output goes WAY up, and you get a huge bump at resonance.)

Preventing this should surely be possible - but it would require significant changes to the design.) 

Another potential issue is damping.

A current source has a very high output impedance...

So the damping factor on any current source amplifier is basically zero.

(It's not like with tube amps, where the damping factor is limited to a low value; a theoretically perfect current source has an infinite source impedance and so a damping factor of zero.)

Also, without going into a lot of detail, a current drive AMPLIFIER in no way suggests a current source power supply.

There's nothing especially difficult about designing a "current amplifier" - where the output current, rather than the output voltage, is proportional to the input signal.

For example, you could convert one of our XPA amps to be a "current source amplifier" by adding two or three parts and moving two or three connections.

(It would NOT require any changes to the power supply at all for a similar output power.)

(However, it would probably sound awful with speakers designed to work with a voltage drive amplifier.)

A current source power supply is simply not especially appropriate for the amplifier topologies that anybody is currently using for audio amplifiers.

(It's a great topology for constant-output LED light bulbs.)

It's also worth mentioning that a "current source power supply" is simply another name for "a current REGULATED power supply".

So you're simply comparing one type of regulated power supply to another.

I've always though it might be interesting to design a current drive headphone amplifier.

(Headphones are generally a simpler load than speakers - with fewer and less extreme impedance variations - and the power and voltage levels involved are a lot safer.)

The real short answer to your question about Nelson Pass' designs is this.....

"Each method has some advantages and some disadvantages."

"His method doesn't have any obvious and clear overall advantages over doing it the other way."

"And, on top of that, everyone is just used to doing it the other way."

Jun 7, 2019 11:28:59 GMT -5 mgbpuff said:

Jun 7, 2019 10:54:45 GMT -5 KeithL said:

The output of an amplifier is, by definition, "regulated". (The technical term is "load invariant".)

(We are talking about the voltage that exists at the output terminals of the amplifier.)

A "normal audio amplifier" is a "voltage amplifier" and is expected to deliver to the speaker a specified multiple of the voltage of the input signal.

(In other words, the output voltage of the amplifier is specified and is not supposed to change regardless of the load.)

The only "issue" with a reactive load is that, in order to maintain that relationship, the amplifier may need to source or sink more current than it would if the load was simply a resistor.

So, for example, maintaining 1V across an 8 Ohm pure resistor requires a certain amount of current - which will be the same at 20 Hz as at 20 kHz.

However, if you put a 1 uF capacitor in parallel with the resistor, you now have a reactive load.

And, with that reactive load, the amplifier will have to deliver a LOT more current at 20 kHz than it will at 20 Hz to continue to deliver that same 1V.

(And it's quite possible that a certain amplifier model may be unable to do so... which might cause it to distort, or even fail, if it tries.)

In fact, as long as the amplifier maintains this relationship properly, how it does so, and what sort of power supply it has, is totally irrelevant.

In general, when you design audio equipment, at the first level of approximation, you assume a regulated supply.

For example, you say "the amplifier runs on +/- 40V rails"... as such you are describing a regulated power supply.

For a non-regulated supply, running on 60 Hz, those rails will be: "between +/- 35V and +/- 44V, varying according to a 120 Hz pseudo-sawtooth wave, whose shape will vary depending on the load".

(For those who prefer frequency terminology that would be: "+/- 40 vDC, with a variety of frequency components added, mostly harmonics of 120 Hz, varying with load".)

(Luckily most modern amplifier designs are very tolerant of this sort of variation - so it doesn't cause too much of a compromise in their performance.)

At BEST, an unregulated supply may be "close enough to the ideal to avoid causing problems"....

And, under certain circumstances, it may have other virtues....

However, in practice, all of those virtues are variations on a theme: "overloading more gracefully".

Comparing two power supplies, both designed to have the correct capabilities, a regulated supply will always perform better.

However, comparing two UNDERDESIGNED power supplies, a non-regulated power supply may have fewer problems if you overtax it.

(And, since an "overdesigned" power supply will never be called upon to reach its limits, any type should work equally well.)

(We actually did design the SMPS in our Gen3 amps to act somewhat more gracefully than many other SMPS would if you do manage to exceed its capabilities.

However, since it can deliver nearly 3 kW of actual power under most circumstances, this doesn't seem to happen very often.   )

As long as you are talking about a voltage controlled wave form, and most amplifiers are voltage controllers, you are correct about the regulated voltage power supply being superior (provided it is adequate). But magneto restrictive transducers such as dynamic cone speakers are intrinsically current responsive devices rather than voltage responsive. Therefore a power supply approximating an ideal current source would be more appropriate than one approximating an ideal voltage source. If the transducer is electrostatic in nature, then voltage control is the more proper approach. Why aren't there more current controlled amplifiers like some of the ones that Nelson Pass has designed?

[Lsc]

Jun 7, 2019 13:24:37 GMT -5 audiobill said:

Blah, blah, so what does it sound like or does that not matter ?

Let’s let the enginerds have their day in the sun 🌞.

I agree, I only care about how it sounds and how heavy the amp is lol.

[DYohn]

The thread is about power ratings, not about sound quality.

[AudioHTIT]

Jun 7, 2019 13:24:37 GMT -5 audiobill said:

Blah, blah, so what does it sound like or does that not matter ?

Well, this thread is about specs ...

[pedrocols]

Jun 7, 2019 16:34:42 GMT -5 AudioHTIT said:

Jun 7, 2019 13:24:37 GMT -5 audiobill said:

Blah, blah, so what does it sound like or does that not matter ?

Well, this thread is about specs ...

Aren't the specs in the products' specs section of the product? AKA "Product Specs."🤔

[AudioHTIT]

Jun 7, 2019 16:49:57 GMT -5 pedrocols said:

Jun 7, 2019 16:34:42 GMT -5 AudioHTIT said:

Well, this thread is about specs ...

Aren't the specs in the products' specs section of the product? AKA "Product Specs."🤔

I’m just saying that this particular thread is about (power) specs, how they’re arrived at by manufacturers, and how they’re interpreted by consumers and reviewers. It’s a valid discussion. How specs do or don’t relate to the sound of an amplifier may also be an interesting topic (for another thread).

[creimes]

Jun 6, 2019 16:40:48 GMT -5 Hair Nick said: