hypex power ratings

[March Audio]

Are we going to again get a new capacitor like joust where both of you are correct from a different point of view?
For what purpose exactly?

No. John is on ignore and wrong for the reasons pointed out over and over again. I can't be bothered to listen to his FUD

He doesn't understand the technical reasons why the things he thinks are important are of no relevance.

It's all pretty basic but he just doesn't get that music is not continuous sine waves.

He doesn't understand that music power content reduces dramatically with increasing frequency.

Until he takes these simple facts on board and acknowledge the flaws of the blunt instrument the FTC test is, its clear that he will continue to make this uninformed noise and threadcrap any subject to do with class d.

It's just best in this case to not indulge and just ignore him.

Just curious, has anyone ever heard any complaint from anyone that Hypex amps don't live up to expectations with respect to their power output? Anyone herd any reports of them shutting down due to temp? I haven't.

 

[mhardy6647]

The FTC says the amplifier has to run continuously at its full power rating with a single tone sine wave. I agree that standardizing the test conditions is necessary, but those conditions need to be meaningful. IMO, designing to the FTC specs results in a poorer product because the specs do not represent real life conditions.

... and the 1974 FTC regulations included a preconditioning period at 1/3 of rated power.
In the real world, it wasn't terribly uncommon for a component to fail the preconditioning, based on reviews published in the popular hifi rags of the era* (Audio, High Fidelity, Stereo Review).

Browse 'em all at https://worldradiohistory.com/

Power is an interesting topic, isn't it?
Seems to me it comes down to a debate between a straightforward, reproducible, easily replicated standardized set of conditions (which the FTC regs could be argued to be an example) vs. a more "meaningful" real world approach (which could be more abstruse, perhaps much more abstruse).

In either case -- but perhaps even moreso in the latter case -- it would be straightforward for an unscrupulous manufacturer to game the system (or, as they say in the Ed. biz, to teach to the test).

Corollary: It's probably a good thing that Volkawagen doesn't build and sell audio amplifiers.

_________________
* admittedly not a statistically meaningful sample set.

 

[March Audio]

Hypex in thermal limited, it can "only" output a 100 watts for a sustained period (something like a couple of minutes). However you will never run into that situation in real life, your amplifier will never output a constant 100+ watts. Class A/B amplifiers hit their distortions limits long before their thermal limits. In practice that means that Hypex can easily provide more power to your speakers than your current amplifier.

Not quite right, it is thermally limited like any amp but the full power duration (its not just 100 watts) depends on the heatsinking. See my video above.

 

[March Audio]

... and the 1974 FTC regulations included a preconditioning period at 1/3 of rated power.
In the real world, it wasn't terribly uncommon for a component to fail the preconditioning, based on reviews published in the popular hifi rags of the era* (Audio, High Fidelity, Stereo Review).

Browse 'em all at https://worldradiohistory.com/

Power is an interesting topic, isn't it?
Seems to me it comes down to a debate between a straightforward, reproducible, easily replicated standardized set of conditions (which the FTC regs could be argued to be an example) vs. a more "meaningful" real world approach (which could be more abstruse, perhaps much more abstruse).

In either case -- but perhaps even moreso in the latter case -- it would be straightforward for an unscrupulous manufacturer to game the system (or, as they say in the Ed. biz, to teach to the test).

Corollary: It's probably a good thing that Volkawagen doesn't build and sell audio amplifiers. View attachment 107104
_________________
* admittedly not a statistically meaningful sample set.

Yes that is the debate. It's not about getting rid of a test to provide comparitive data to consumers. It's about having tests that are relevant and meaningful.

Just for info that 1/3 power conditioning period is far more of an issue for class A/B than it is for class D (total non issue for class d) due to the fundamental differences in efficiencies.

 

[tmtomh]

1. To ensure low roll-off (flat frequency response), low distortion, and stability margin for signals below that frequency.
2. Hypex specs the power bandwidth to 35 kHz, that is from their own datasheet, so it should meet that spec. It does, but only for <30 seconds. That does not meet the FTC spec (as says John) but is probably good enough for the real world (as says Alan). That is based on the output coupling capacitor, not thermal limits, something I did not notice before.

Say you have a simple two-pole (second-order) roll-off someplace just above the audio band, and you want to spec 20 kHz to 1 dB. The two-pole response could be the output LC filter (which many amps have, not just class D, for RFI immunity). The simple transfer function is H = 1 / sqrt[1 + (f/fp)^2n] where fp is the pole frequency and n = 2 for a second-order response. If I churned the equation rightly (no promises!) then you need fp = f / rt2n[(1/H^2) - 1]. At -1 dB, H = 0.89, and choose f = 20 kHz for our rolloff. Then you need about 28 kHz bandwidth to get -1 dB at 20 kHz. Spec a little tighter, or provide a little more design margin, and 35 kHz is not unreasonable for full-power bandwidth.

HTH - Don

Thanks Don, for that detailed explanation!

It seems to me that @restorer-john is using the FTC rules as a cudgel: he makes the unstated - and unsupported - assumption that the FTC rules are the only proper and honest standard, and then based on that, claims that Class D amps are "toys" (direct quotation) and that their specs are dishonest and manipulative because they don't have the same ultrasonic power bandwidth as Class AB amps.

He knows full well, of course, that Class D requires filtering of a kind not contemplated in the FTC specs, and he also knows full well that such filtering is in no way, shape, or form a design feature that can be shown to result in any measurable performance deficiency - and he knows full well that bandwidth beyond 20kHz plus a reasonable buffer (as Don H describes above) is irrelevant to the actual sonic performance of an amplifier. But he won't admit it. @pma has gone further by asserting that the difference between resistive loads and actual speakers make Class D amps' real-world frequency response nonlinear. He is certainly correct that this can happen under certain design conditions. But @March Audio and others have repeatedly shown - with both technical explanation and graphs of actual measured performance under the conditions in question - that the Hypex and Purifi designs do not actually have this problem. So pma's hypothesis is certainly reasonable, but it has been disproven in the case of the Hypex and Purifi designs, and he refuses to admit it.

This stubborn clinging to disproven hypotheses, this intractable insistence on flawed conclusions based on reasonable facts taken in isolation or without sufficient context, seems to me to be anathema to the goals and ethos of this forum.

 

[PierreV]

One thing I've always wondered is if there's a price point were amps don't really improve that much beyond. I guess I'm going to find that out. There certainly does seem to be a lot of folks online that think their hypex or purifi sounds better than their previous x5 or x10 priced amp

Count me in that camp. My diy NC400 are currently driving a pair of Focal Scala Utopias V2 better than a five times more expensive A/B amplifier did. The difference in the bass is obvious. The semi-rational analysis of the issue that led me to try the Hypex amps was that the Focals seemed to go quite low in terms of impedance in that FR. I believe that the high instantaneous current delivery ability of the Hypex made a difference there. I have other speakers where I don't hear a difference though but those speakers (per measurements available on the web) don't seem to go as low as the Focals.

The difference is so evident to my ears that I have been planning to take FR measurements to see if I can correlate my experience objectively but I have been lazy as it involves moving heavy stuff around in a tangle of cables. Then, even if I managed to show a difference, it may simply indicate that the A/B amp is past its prime and needs a recap. Who knows.

If I were you, I'd try to borrow an Hypex/Purifi amp and see what the result is. Or even buy it and resell it if it disappoints.

 

[March Audio]

1. To ensure low roll-off (flat frequency response), low distortion, and stability margin for signals below that frequency.
2. Hypex specs the power bandwidth to 35 kHz, that is from their own datasheet, so it should meet that spec. It does, but only for <30 seconds. That does not meet the FTC spec (as says John) but is probably good enough for the real world (as says Alan). That is based on the output coupling capacitor, not thermal limits, something I did not notice before.

Say you have a simple two-pole (second-order) roll-off someplace just above the audio band, and you want to spec 20 kHz to 1 dB. The two-pole response could be the output LC filter (which many amps have, not just class D, for RFI immunity). The simple transfer function is H = 1 / sqrt[1 + (f/fp)^2n] where fp is the pole frequency and n = 2 for a second-order response. If I churned the equation rightly (no promises!) then you need fp = f / rt2n[(1/H^2) - 1]. At -1 dB, H = 0.89, and choose f = 20 kHz for our rolloff. Then you need about 28 kHz bandwidth to get -1 dB at 20 kHz. Spec a little tighter, or provide a little more design margin, and 35 kHz is not unreasonable for full-power bandwidth.

HTH - Don

Just for info Hypex amps are typically 0.5dB down at 20kHz and 3dB at 50kHz.

Purifi are Flat to 20kHz and 3dB down at 60KHz.

 

[DonH56]

Hypex amps are typically 0.5dB down at 20kHz and 3dB at 50kHz.

Purifi are Flat to 20kHz and 3dB down at 60KHz.

Yah, when I ran the numbers it was clear their target was better than -1 dB, but I was too lazy to do the math again.

 

[Koeitje]

Not quite right, it is thermally limited like any amp but the full power duration (its not just 100 watts) depends on the heatsinking. See my video above.

I think I could have been more clear, but my main point was that NCore (and Purifi) can output their rated peak power longer than A/B amplifiers.

For me, I didn't hear a difference between my recapped Marantz PM-16 (€2000 in 1996, 20kg) class A/B and NC400's on Revel M106's. But the Marantz was simply too big to find a spot for, because it doesn't adhere to standard width and depth for audio equipment. Plus I wanted to just have a pre + power amplifier setup for more versatility.

 

[March Audio]

.

This stubborn clinging to disproven hypotheses, this intractable insistence on flawed conclusions based on reasonable facts taken in isolation or without sufficient context, seems to me to be anathema to the goals and ethos of this forum.

Sums it up

 

[March Audio]

I think I could have been more clear, but my main point was that NCore (and Purifi) can output their rated peak power longer than A/B amplifiers.

Well I think that depends, it is possibly/probably a bit too much of a generalisation. I think it would depend on design and implementation.

However with reference to the OP and his existing amp, an NC500 based amp would be significantly more powerful. On paper, in bench tests and in real world use.

 

[Deleted member 65]

Interesting, thank you. Assumed the power supply can deliver, I should then buy some big heat sinks for optimized continuous power? Not so easy to attach them to a plate amp, though.

Or you can mimic my Purifi build

 

[DonH56]

Thanks Don, for that detailed explanation!

It seems to me that @restorer-john is using the FTC rules as a cudgel: he makes the unstated - and unsupported - assumption that the FTC rules are the only proper and honest standard, and then based on that, claims that Class D amps are "toys" (direct quotation) and that their specs are dishonest and manipulative because they don't have the same ultrasonic power bandwidth as Class AB amps.

He knows full well, of course, that Class D requires filtering of a kind not contemplated in the FTC specs, and he also knows full well that such filtering is in no way, shape, or form a design feature that can be shown to result in any measurable performance deficiency - and he knows full well that bandwidth beyond 20kHz plus a reasonable buffer (as Don H describes above) is irrelevant to the actual sonic performance of an amplifier. But he won't admit it. @pma has gone further by asserting that the difference between resistive loads and actual speakers make Class D amps' real-world frequency response nonlinear. He is certainly correct that this can happen under certain design conditions. But @March Audio and others have repeatedly shown - with both technical explanation and graphs of actual measured performance under the conditions in question - that the Hypex and Purifi designs do not actually have this problem. So pma's hypothesis is certainly reasonable, but it has been disproven in the case of the Hypex and Purifi designs, and he refuses to admit it.

This stubborn clinging to disproven hypotheses, this intractable insistence on flawed conclusions based on reasonable facts taken in isolation or without sufficient context, seems to me to be anathema to the goals and ethos of this forum.

I think that's a bit harsh. I have avoided stepping into this; I respect both Alan and John, and they are arguing from valid but opposing sides. The issue is not the filtering, except that Hypex's output cap appears underrated for the FTC test (they appear to leave thermal management to the end manufacturer, not atypical as they cannot control that aspect of the final design). I do not recall seeing a five-minute 20 kHz full-power test and Hypex's datasheet says pretty clearly the capacitor is not rated for that. Note many amps have included output coupling caps that may have the same problem (I expect myself, Alan, and John have dealt with a few over the decades); in a class D design, the capacitor is actually more of a decoupling capacitor since it is not in series with the output. The FTC also required 1/3 power pre-conditioning (I think it's 1/8 now but don't quote me) for an hour; that is actually pretty stringent for a class AB amplifier but not a big deal for a typical class D amp.

From my point of view there are a couple of (OK, three) things going on. The following is my opinion and I apologize in advance for misrepresenting Alan, John, Hypex, or any other parties past, present, and future (hopefully that's CYA enough).

1. Hypex states you cannot (should not) exceed 30 seconds for a full-power test over 5 kHz due to the output capacitor's ESR. That parasitic resistance heats up and can cause capacitor failure.
2. Thermal management is left to the final manufacturer and is another bound on long-term power dissipation. Fair enough, Hypex provides modules, so it is up to the assembler to design the heat sinks and chassis to support the thermal load.
3. Then the big debate about the FTC specs for the US market. The FTC says it should provide full rated power over the rated full-power bandwidth for five minutes (though there are some caveats on that now). The final amp (any amp, not specifically talking about March Audio) may or may not meet that spec based on the thermal design; but it seems like the module itself is limited to <30 s over 5 kHz.

Does it matter? There's following the FTC to the letter of the law, and real-world usage. John is arguing for the former and Alan for the latter, both from a sense of "right" and adherence to best principles for the consumer. I can read the data sheet (though frankly have not except a few snippets being debated here) and understand the specs so from that point of view Hypex is not doing anything wrong. It is up to the manufactures to decide how to rate the final product. That's where Alan and John get crosswise; until John sees a full-power FTC power test, in his eyes it is not FTC compliant (mine too). Alan feels the FTC test is not reasonable and provide lots of evidence to back that up (ditto). Knowing the dynamic nature of music, I can accept that many (probably most) Hypex (etc.) amps do not met the FTC spec, but as long as the actual specs are provided and the amps are not claimed to meet the FTC tests, I can make an informed decision.

There is risk either way; risk there is less margin in the class D design for thermal and sustained HF output; and, risk in cost, weight and power by purchasing an amp that meets the FTC spec but may be overdesigned for the real-world task. Up to me to choose. Alan and John have both made good cases for their point of view IMO.

Hopefully my last word as I really have nothing to add to this discussion. - Don

 

[March Audio]

I think that's a bit harsh. I have avoided stepping into this; I respect both Alan and John, and they are arguing from valid but opposing sides. The issue is not the filtering, except that Hypex's output cap appears underrated for the FTC test (they appear to leave thermal management to the end manufacturer, not atypical as they cannot control that aspect of the final design). I do not recall seeing a five-minute 20 kHz full-power test and Hypex's datasheet says pretty clearly the capacitor is not rated for that. Note many amps have included output coupling caps that may have the same problem (I expect myself, Alan, and John have dealt with a few over the decades); in a class D design, the capacitor is actually more of a decoupling capacitor since it is not in series with the output. The FTC also required 1/3 power pre-conditioning (I think it's 1/8 now but don't quote me) for an hour; that is actually pretty stringent for a class AB amplifier but not a big deal for a typical class D amp.

From my point of view there are a couple of (OK, three) things going on. The following is my opinion and I apologize in advance for misrepresenting Alan, John, Hypex, or any other parties past, present, and future (hopefully that's CYA enough).

1. Hypex states you cannot (should not) exceed 30 seconds for a full-power test over 5 kHz due to the output capacitor's ESR. That parasitic resistance heats up and can cause capacitor failure.
2. Thermal management is left to the final manufacturer and is another bound on long-term power dissipation. Fair enough, Hypex provides modules, so it is up to the assembler to design the heat sinks and chassis to support the thermal load.
3. Then the big debate about the FTC specs for the US market. The FTC says it should provide full rated power over the rated full-power bandwidth for five minutes (though there are some caveats on that now). The final amp (any amp, not specifically talking about March Audio) may or may not meet that spec based on the thermal design; but it seems like the module itself is limited to <30 s over 5 kHz.

Does it matter? There's following the FTC to the letter of the law, and real-world usage. John is arguing for the former and Alan for the latter, both from a sense of "right" and adherence to best principles for the consumer. I can read the data sheet (though frankly have not except a few snippets being debated here) and understand the specs so from that point of view Hypex is not doing anything wrong. It is up to the manufactures to decide how to rate the final product. That's where Alan and John get crosswise; until John sees a full-power FTC power test, in his eyes it is not FTC compliant (mine too). Alan feels the FTC test is not reasonable and provide lots of evidence to back that up (ditto). Knowing the dynamic nature of music, I can accept that many (probably most) Hypex (etc.) amps do not met the FTC spec, but as long as the actual specs are provided and the amps are not claimed to meet the FTC tests, I can make an informed decision.

There is risk either way; risk there is less margin in the class D design for thermal and sustained HF output; and, risk in cost, weight and power by purchasing an amp that meets the FTC spec but may be overdesigned for the real-world task. Up to me to choose. Alan and John have both made good cases for their point of view IMO.

Hopefully my last word as I really have nothing to add to this discussion. - Don

Excellent summary.

The FTC tests goals, assisting the domestic consumer to make informed choices, are valid. However the tests simply aren't up with current knowledge, technology and measurement capabilities. They are a blunt and simplistic instrument still firmly stuck in the 70s.

The "HF margin" was never required, not in the 70s and not now, yet due to the test design it's implied that it is. This is wrong and John is wrong to infer that it is.

Just take apart an active speaker and check the power rating of the tweeter amp V the woofer amp the tweeter amp is usually much lower powered because its simply not required.

As a quick example

KEF LS50 Wireless
Frequency response 45 – 47,000 Hz ±3 dB (depending on settings)

Power output 200 W low frequencies,
30 W high frequencies
Maximum SPL106 dB

 

[mhardy6647]

Just for info that 1/3 power conditioning period is far more of an issue for class A/B than it is for class D (total non issue for class d) due to the fundamental differences in efficiencies.

My "preconditioning" comment was indeed very much a retrospective one.
That is good to know -- I would hope that to be the case, but one cannot beat empiricism!

 

[pma]

Just for info that 1/3 power conditioning period is far more of an issue for class A/B than it is for class D (total non issue for class d) due to the fundamental differences in efficiencies.

This depends on heatsinks used, for both class AB or D. For example, I can run this amplifier
http://pmacura.cz/pmacfa_en.htm
at any power between 0 - 100% of the specified power continuously, no problem with the 1 hour test at any power.

When I had the Hypex NC400 modules to test here (they were sold by Hypex? few years ago, see below)

they were running too hot in that 1/3 power test.

And the class D AIYIMA A07 (that I am testing now) specified at 2x300W/4ohm switches off itself after 1 minute of 1x100W/4ohm/1kHz test.

So it is definitely not true that is a "(total non issue for class d)", as you are saying. No, it is a question of design for any class of amplifiers. Inadequately small heatsinks will make the amp to fail in this test, regardless its class of operation.

As you know, Hypex NC500 OEM datasheet states that continuous power is only 1/5 of the "Peak Output Power" as per their specs. I hope I do not need to paste this datasheet again. It is the old game of "Peak Power", which has been with us for at least 50 years and has not changed and the main reason is marketing.

 

[Blumlein 88]

My "preconditioning" comment was indeed very much a retrospective one.
That is good to know -- I would hope that to be the case, but one cannot beat empiricism!

As I recall the 1/3 power is the point that would cause close to Max dissipation across the output transistors in class AB designs. For class A or D it doesn't relate to anything.

 

[mhardy6647]

As I recall the 1/3 power is the point that would cause close to Max dissipation across the output transistors in class AB designs...

I have the same recollection (albeit vaguely) -- FWIW.

I do agree, again FWIW, with your comment (which I excised above, no offense!) vis-a-vis Class A and I presume you're also correct for Class D, although I really dont' know about the latter. I once read a not-entirely-facetious rumination about Class A, suggesting that the efficiency is actually a wee bit higher for a Class A amp when it is operating at its full audio output power, since not all of its output, under those conditions, is in the form of heat.

 

[pma]

As I recall the 1/3 power is the point that would cause close to Max dissipation across the output transistors in class AB designs.

That's correct. Based on voltage drop and power (Vce x Ic) on output transistors and load.

 

[Inner Space]

The FTC tests goals, assisting the domestic consumer to make informed choices, are valid. However the tests simply aren't up with current knowledge, technology and measurement capabilities. They are a blunt and simplistic instrument still firmly stuck in the 70s.

I'm an eight-month member here and have sincerely enjoyed the many and various contributions from both @March Audio and @restorer-john over that period. Both seem like fine and fascinating fellows with much of value to learn from.

I have no dog in their fight (thank goodness), so as a neutral bystander would merely observe that their dispute seems minimally technical and maximally emotional. Clearly John has enormous emotional investment in the big Japanese iron of a vanished era (hence bolding above) and is profoundly unhappy that his old certainties are slipping away.

This kind of thing manifests everywhere, in every arena of life. The way it used to be is a powerful feeling. Fortunately audio disputes are a trivial outlet in comparison with others (he said, in America, in January 2021).