How true? Class D amps can't sustain their maximum spec'd power

[levimax]

In any amplifier, if a switching supply and a linear supply both have exactly the same output voltage under no-signal conditions, the linear supply will sag to a lower voltage than the switching supply under heavy load. How much the linear supply will sag has to do with the amount of capacitance on its output. So in this instance, the linear supply is at a disadvantage and will supply fewer watts under sustained heavy output than the switching supply.

If the linear supply has a higher output voltage than the switching supply but sags to the same voltage as the switching supply under sustained heavy load conditions, then the linear supply will have the ability to deliver higher output wattage for a short period (in other words transient output), and will have the same ability to output sustained wattage as the switching supply. This is typically how a linear supply is implemented (at least in our amplifiers) so that there is transient ability above what a switching supply can offer.

Looking at @amirm amp tests it looks like most of the class D amps do not have a lot of "headroom" for transients. However I watched the QSC video linked a couple posts above this and it sounds like they have plenty of headroom on their new class D amps. I would love to see one of these amps tested. The DPA 4.3 is less than $2,000 for 4 Channels of 625 watts continuous power with almost 3 dB headroom. Looking at the specs it looks like the high SINAD fans might be disappointed but power / dynamics fan may be happy. I know I should buy one and send it in but I am not going to unfortunately.

 

[DonH56]

I am not sure why "headroom for transients" is such a big deal. Buy an amplifier rated for the power you need to deliver the transient peaks you anticipate in your system. Headroom over spec is pretty nebulous and seems to built into some sort of mindset that says "if it is rated for xxx W, then I need to have xxx+yyy W or it is not a good amp". A class AB, or D, or whatever amplifier with a well regulated power supply will not have much headroom (unless they intentionally under-spec the power to provide a higher "headroom" rating). The difference is that many class D amplifiers use switch-mode power supplies to improve their overall efficiency and reduce heat loss, and those are almost always regulated, so now a 400 W amp really is a 400 W amp. Or you can buy a 300 W class AB amp with 1.25 dB of "dynamic headroom" and get the same power, for a short time (transient). Of course, distortion is never specified for that "dynamic power" rating, nor how long the amp can sustain that power level...

Headroom these days is more about marketing than engineering IMO. Maybe since most AB amplifiers cannot compete on actual steady-state power ratings, they rely on "dynamic headroom" as a feature to distinguish them from class D (or AB amps with stiff power supplies)? At the same time they turn it into a feature: "My AB 100 W amp has 3 dB of dynamic headroom! Your 200 W class D amp does not have any headroom so it's not nearly as good!" The only time dynamic headroom would play into my decision is if the base power rating is not enough to play the loudest peaks when I am listening. And then I would look for an amp with higher base specs. Class of operation would not really enter into it; if I need 100 W, I would look at other specs and features, with class of operation being low in consideration, and dynamic headroom weighted essentially zero in my buying decision.

All high dynamic headroom tells me is that the amp's power supply and/or thermal system is inadequate...

Caveat: As @MakeMineVinyl noted in an earlier posting, multichannel amplifiers are frequently designed with higher headroom to create a multichannel amp in a reasonable box with reasonable heatsinking and power supply in the space (and cost target) available. Since music and movies typically have anywhere from 17 dB (50:1 power ratio) to 20 dB (100:1 power ratio) or more peak-to-average ratio, the trade in long-term output for a small'ish box makes sense to me. But still I bought my multichannel amps to provide the power needed for my expected loudest transients based upon their rated power output and not their dynamic headroom.

FWIWFM/IME/IMO/etc. - Don

 

[JustJones]

The class D amps I use are in my speakers. I assume they sustain whatever power they need as I've never had them shut down using the built in protection. I guess I could wear some ear protection to see how far they would go.

 

[levimax]

I am not sure why "headroom for transients" is such a big deal. Buy an amplifier rated for the power you need to deliver the transient peaks you anticipate in your system. Headroom over spec is pretty nebulous and seems to built into some sort of mindset that says "if it is rated for xxx W, then I need to have xxx+yyy W or it is not a good amp". A class AB, or D, or whatever amplifier with a well regulated power supply will not have much headroom (unless they intentionally under-spec the power to provide a higher "headroom" rating). The difference is that many class D amplifiers use switch-mode power supplies to improve their overall efficiency and reduce heat loss, and those are almost always regulated, so now a 400 W amp really is a 400 W amp. Or you can buy a 300 W class AB amp with 1.25 dB of "dynamic headroom" and get the same power, for a short time (transient). Of course, distortion is never specified for that "dynamic power" rating, nor how long the amp can sustain that power level...

Headroom these days is more about marketing than engineering IMO. Maybe since most AB amplifiers cannot compete on actual steady-state power ratings, they rely on "dynamic headroom" as a feature to distinguish them from class D (or AB amps with stiff power supplies)? At the same time they turn it into a feature: "My AB 100 W amp has 3 dB of dynamic headroom! Your 200 W class D amp does not have any headroom so it's not nearly as good!" The only time dynamic headroom would play into my decision is if the base power rating is not enough to play the loudest peaks when I am listening. And then I would look for an amp with higher base specs. Class of operation would not really enter into it; if I need 100 W, I would look at other specs and features, with class of operation being low in consideration, and dynamic headroom weighted essentially zero in my buying decision.

All high dynamic headroom tells me is that the amp's power supply and/or thermal system is inadequate...

Caveat: As @MakeMineVinyl noted in an earlier posting, multichannel amplifiers are frequently designed with higher headroom to create a multichannel amp in a reasonable box with reasonable heatsinking and power supply in the space (and cost target) available. Since music and movies typically have anywhere from 17 dB (50:1 power ratio) to 20 dB (100:1 power ratio) or more peak-to-average ratio, the trade in long-term output for a small'ish box makes sense to me. But still I bought my multichannel amps to provide the power needed for my expected loudest transients based upon their rated power output and not their dynamic headroom.

FWIWFM/IME/IMO/etc. - Don

I look at it differently, I don't see what the big deal with continuous power is. Continuous power has little to do with playing back music. Why pay extra for a physically larger amp designed to deliver continuous power at a level only needed for peaks.

 

[Doodski]

I look at it differently, I don't see what the big deal with continuous power is. Continuous power has little to do with playing back music. Why pay extra for a physically larger amp designed to deliver continuous power at a level only needed for peaks.

So when turning up the bass and cranking the volume @ parties and solo events the amp does not fry or clip. I don't see why peeps are scared of continuous power output @ 1KHz. Perhaps the new surround amps can't do it but all the amps that I tested up to ~2002 took it just fine. Any decently newly made amp can handle it.

 

[Tks]

If they can't. I'd love to see examples where the PSU or the thermal solution employed within the unit isn't somehow involved in that equation for why the Class D amp is failing to sustain the load.

I feel this goes for any device. It's almost always related to how well heat dissipation is being handled, or the power supply being used is not up to the job. I guess these days you can have sloppy designers lying about the spec, and simply using parts that aren't up to spec for their intended goal.

 

[AdamG]

I am not sure why "headroom for transients" is such a big deal. Buy an amplifier rated for the power you need to deliver the transient peaks you anticipate in your system. Headroom over spec is pretty nebulous and seems to built into some sort of mindset that says "if it is rated for xxx W, then I need to have xxx+yyy W or it is not a good amp". A class AB, or D, or whatever amplifier with a well regulated power supply will not have much headroom (unless they intentionally under-spec the power to provide a higher "headroom" rating). The difference is that many class D amplifiers use switch-mode power supplies to improve their overall efficiency and reduce heat loss, and those are almost always regulated, so now a 400 W amp really is a 400 W amp. Or you can buy a 300 W class AB amp with 1.25 dB of "dynamic headroom" and get the same power, for a short time (transient). Of course, distortion is never specified for that "dynamic power" rating, nor how long the amp can sustain that power level...

Headroom these days is more about marketing than engineering IMO. Maybe since most AB amplifiers cannot compete on actual steady-state power ratings, they rely on "dynamic headroom" as a feature to distinguish them from class D (or AB amps with stiff power supplies)? At the same time they turn it into a feature: "My AB 100 W amp has 3 dB of dynamic headroom! Your 200 W class D amp does not have any headroom so it's not nearly as good!" The only time dynamic headroom would play into my decision is if the base power rating is not enough to play the loudest peaks when I am listening. And then I would look for an amp with higher base specs. Class of operation would not really enter into it; if I need 100 W, I would look at other specs and features, with class of operation being low in consideration, and dynamic headroom weighted essentially zero in my buying decision.

All high dynamic headroom tells me is that the amp's power supply and/or thermal system is inadequate...

Caveat: As @MakeMineVinyl noted in an earlier posting, multichannel amplifiers are frequently designed with higher headroom to create a multichannel amp in a reasonable box with reasonable heatsinking and power supply in the space (and cost target) available. Since music and movies typically have anywhere from 17 dB (50:1 power ratio) to 20 dB (100:1 power ratio) or more peak-to-average ratio, the trade in long-term output for a small'ish box makes sense to me. But still I bought my multichannel amps to provide the power needed for my expected loudest transients based upon their rated power output and not their dynamic headroom.

FWIWFM/IME/IMO/etc. - Don

Well said Don!

 

[Doodski]

I'd love to see examples where the PSU or the thermal solution employed within the unit isn't somehow involved in that equation for why the Class D amp is failing to sustain the load.

Class A/AB amps can have thermal protection too. Usually a PCB ~85C trip temp and that means reallly hot heatsinks. Those rarely trip unless the amp is driven into a low impedance for extended periods of time and so with class D efficiency I doubt the thermal protection will be used much where it is @75C or 85C as I've read is common in various models of class D amps.

 

[levimax]

I don't see why peeps are scared of continuous power output @ 1KHz.

I dont see it as being afraid of continuous power rather just good design for playing music. For what ever power level an amp can play continuosly at it is not hard or expensive to design it to deliver 2 to 3 dB more peak power. Why leave all that power on the table which many class D amps seem to.

 

[DonH56]

I dont see it as being afraid of continuous power rather just good design for playing music. For what ever power level an amp can play continuosly at it is not hard or expensive to design it to deliver 2 to 3 dB more peak power. Why leave all that power on the table which many class D amps seem to.

It is not leaving it on the table, it is part of the design. 3 dB peak power is twice the power, meaning everything around the amp has to be able to handle that much additional power even for a short time. Voltage rails must be higher (about 40% higher for 3 dB headroom), so all the components and devices inside must support the higher voltage, current capacity must be higher, capacitors need higher voltage ratings, transistors need higher breakdown voltage, you'll need more gain to provide those peaks with a reasonable input voltage (which means more noise at the output), etc. And if you have a regulated supply rail, as many if not most class D amps do, using a SMPS to be more efficient and reduce heat, then the only way to get more "peak power" is to underspec the output. It is not something they are "leaving on the table", it is how the design works.

 

[mocenigo]

I am not sure why "headroom for transients" is such a big deal. Buy an amplifier rated for the power you need to deliver the transient peaks you anticipate in your system.

Exactly. My Purifi/Neurochrome build will deliver 165W over 8ohm, 330W over 4Ohm. My speakers are 6Ohm min impedance (and mostly nearly resistive over the whole range – designing the crossover with this goal instead of "perfect constant 7 ohm" or "kill all peaks" or "smooth everything within 0.1Db" has been an... interesting challenge) and my speakers have a sensitivity close to 98 Db/W/m. So I know I have quite a headroom...

 

[DonH56]

p.s. Don't forget to look at some of @amirm or other reviews and note how close to clipping most amps are at rated power output. Now double that (3 dB in power) and take a gander at the distortion.

 

[mocenigo]

So glad we have our cheeses properly characterized now.

Good cheese is better than SINAD ;-)

 

[levimax]

It is not leaving it on the table, it is part of the design. 3 dB peak power is twice the power, meaning everything around the amp has to be able to handle that much additional power even for a short time. Voltage rails must be higher (about 40% higher for 3 dB headroom), so all the components and devices inside must support the higher voltage, current capacity must be higher, capacitors need higher voltage ratings, transistors need higher breakdown voltage, you'll need more gain to provide those peaks with a reasonable input voltage (which means more noise at the output), etc. And if you have a regulated supply rail, as many if not most class D amps do, using a SMPS to be more efficient and reduce heat, then the only way to get more "peak power" is to underspec the output. It is not something they are "leaving on the table", it is how the design works.

QSC class D amps have 3 dB of headroom and are not expensive so it can be done. I understand what you are saying but what I am saying is it is way cheaper to design in 2to 3 db of headroom as compared to 2 to 3 db more continuos power.

 

[pjug]

It is not leaving it on the table, it is part of the design. 3 dB peak power is twice the power, meaning everything around the amp has to be able to handle that much additional power even for a short time. Voltage rails must be higher (about 40% higher for 3 dB headroom), so all the components and devices inside must support the higher voltage, current capacity must be higher, capacitors need higher voltage ratings, transistors need higher breakdown voltage, you'll need more gain to provide those peaks with a reasonable input voltage (which means more noise at the output), etc. And if you have a regulated supply rail, as many if not most class D amps do, using a SMPS to be more efficient and reduce heat, then the only way to get more "peak power" is to underspec the output. It is not something they are "leaving on the table", it is how the design works.

NAD has used their rail switching scheme for dynamic power with Class D. I'm not advocating for that but it shows that Class D doesn't preclude taking this approach.

 

[MakeMineVinyl]

QSC class D amps have 3 dB of headroom and are not expensive so it can be done. I understand what you are saying but what I am saying is it is way cheaper to design in 2to 3 db of headroom as compared to 2 to 3 db more continuos power.

Continuous power is more expensive. The cost to provide a measure of transient headroom above this is relatively less expensive. It comes down to capacitors, transistors and other assorted parts which can handle the extra voltage/current on a transient basis and a transformer which provides a somewhat higher rail voltage. That's relatively cheaper to do than build with bigger heatsinks and gobs of extra output transistors to be able to supply higher continuous power output. Of course its possible to have both - a high continuous power output with some transient margin above that.

I just looked at the website for our amplifiers, and we don't even mention transient power. As a matter of fact, I've only even seen it being measured in the lab on one instance, and that was out of curiosity with one of our more powerful amps which had over 2000 watts transient power against 750 watts continuous if I remember correctly.

 

[Doodski]

I dont see it as being afraid of continuous power rather just good design for playing music. For what ever power level an amp can play continuosly at it is not hard or expensive to design it to deliver 2 to 3 dB more peak power. Why leave all that power on the table which many class D amps seem to.

Class D amps have so much power for the dollar that the big wattage numbers solve the headroom stuff anyway.

 

[MakeMineVinyl]

I look at it differently, I don't see what the big deal with continuous power is. Continuous power has little to do with playing back music. Why pay extra for a physically larger amp designed to deliver continuous power at a level only needed for peaks.

You can certainly make an amplifier less expensive that way, but the fly in the ointment is that you can't know before hand how long your transients are going to be relative to how long the amplifier can hold out supporting those transients.

 

[RayDunzl]

In practice it doesn't matter, because no content uses full power all the time.

I don't understand the fascination with "maximum power continuously."

Yeah.

Here's some Daft Punk.

This section looks dense and demanding of power:

Zoom in there...

Get a demand for "max power" for a couple of cycles at intervals of 1/2 second.

Waveform view:

Decibel view, with time bar:

 

[DonH56]

NAD has used their rail switching scheme for dynamic power with Class D. I'm not advocating for that but it shows that Class D doesn't preclude taking this approach.

Thanks, I was not aware of that. There are a few other class G (and one H that I have read about) designs using class D rather than class AB cores, but the ones I have seen before target low-power applications.