Do Class D Amplifiers Deliver Rated Power?

[Oddball]

I feel that a source of confusion derives from a lack of experience of bench testing power amplifiers.

There are definitely corner cases for certain film tracks and electronic music with difficult crest factors or frequencies which are uncommon in normally recorded music. But none of them are actually similar to a pure sine wave full power test across a wide range of frequencies.

I've run sine wave full power and soak tests on power amplifiers and absolutely nothing in music comes close to the heat generated!

That is helpful to know. So then how would you choose a D class replacement for the AB amp that was proven as adequate? Primarily interested in HT and that is where the most of the amp tool takes place.

For example, my requirement would be to run the bed channels large (LFE+main for additional sub support), and in addition to send the LFE channel to all the bed speakers (sans center) so I can use them as additional supporting subwoofers? It is difficult to say how much toll this takes for the amps, but it certainly makes them real busy.

Currently my amps (see the signature) are adequate for that, but they are regrettably near the end of their useful life. Looks like going 2x up in wattage for the D class might not be enough?

EDIT: just to visualise the current setup:

 

[somebodyelse]

Currently my amps (see the signature) are adequate for that, but they are regrettably near the end of their useful life. Looks like going 2x up in wattage for the D class might not be enough?

Or you go for one of the manufacturers that does spec for continuous full power output, like Vera Audio:
https://www.audiosciencereview.com/...ncore-amplifiers-overheat.49087/#post-1758153

Most hifi amps, whether AB or D, aren't built for continuous output because for most hifi use it's adding unnecessary expense for the extra cooling. Class A is different as they don't get a choice - peak thermal output is at idle.

 

[Oddball]

You might be right.

I am primarily concerned about replacing my amps.

5 Rotels RB 1070 seem to be performing above its AB class and could not choke them with any curveball yet despite their modest 125Wps rating into 8ohm. Despite the low-is SINAD, even Amin was happy to recommend and I am really happy with them doing the surround and Atmos amplification.

Rotel RB-1070 Amplifier Review

This is a review and detailed measurements of the Rotel RB-1070 stereo power amplifier. It was released in early 2000. Can't find the cost then but I see it used for around $400 to $500. I have always found Rotels to be most attractively designed power amplifiers. One look at that front panel...

www.audiosciencereview.com

It will be a very sad day when my Brystons 4B SST fail, but given their status of the junior monsters in AB class, I wonder what would be there to replace them in D class (hopefully at much lower cost). They do so well with LCR that are driven down to 30hz and in addition given the LFE duties to perform in that range.

 

[Waxx]

I think an NC2K based amp like the Apollon NC2K SLM Monoblocks are easely way to powerfull for what you need and will certainly have enough headroom. That is s 1600w peak in 8R (so 5x what is needed in RMS). Even the lesser NC1200 variant could do it. Both are way cheaper than Bryston amps.

 

[Oddball]

That would be a pretty safe bet - inflation adjusted a bargain compared to Bryston 4SST but still 4+K for 2 channels. Not really thinking of AB amp as want to reduce the replacement cost and gain the benefits of the D class.

Thanks to all for their input - I think I got the point. Will likely size up 3-4 times when transitioning to D class.

 

[DVDdoug]

From what I've seen MOST manufactures are fudging the specs no matter the class. And it doesn't seem like the FTC is enforcing the rules.

Crown makes amplifiers rated for 3000W and more and that's more power than you can get from a regular American power outlet!

IMO - For high-power class-D is the way to go.

 

[audioresearch]

Movie soundtracks have a crest factor of 15 to 20 dB. You don’t need full power for the 2 hour duration of the movie. So you’ll need on average 25 to 100 times less than your 300W of max power.

I should add this. I use multiple pairs of Bose 901 speakers in addition to 4 large speakers of my own design each with among other things an 18" woofer. The Bose equalizer boosts the highs quite a bit in addition to what bass I let them output that I do not crossover to my own monster speakers. So a lot of power is needed. Also, my hearing is poor in the high frequency range and so I boost the highs even further to compensate.

All I really care about in regard to whether or not typical class D amps can output rated power is that they do not shut themselves off when I am listening.

It gets really messy and very involved to spec how much power is needed at what frequencies and for how long.

If I stick to a simple spec such as "can deliver X watts continuously over the entire 20Hz to 20KHz range indefinitely long" that simply covers all the bases and I don't have to get involved with all the details of an easier-to-meet spec. If instead I start getting involved with crest factors, peak power levels, how short are their durations, etc, it gets complicated and messy fast.

I'm still researching all of this and so I really like the opinions and facts about it that people here have been giving me. Thanks.

 

[voodooless]

I think you’re over complicating things. We’ve heard little to no feedback from people complaining that their Hypex, Purifi, Pascal or B&O amps have issues with certain types of musical content for prolonged periods.

 

[DVDdoug]

If I stick to a simple spec such as "can deliver X watts continuously over the entire 20Hz to 20KHz range indefinitely long"

That's the ideal but since program material is dynamic it's not really necessary. The FTC rule says 5 minutes, which you'd also never see except with test tones.

To me the bigger question is usually if I trust the specs at all! Can this "100W" amp even deliver 100W briefly without clipping?

I'm not sure what Amir considers "continuous" but he did kill an amplifier recently during testing.

It gets really messy and very involved to spec how much power is needed at what frequencies and for how long.

Right! It's too complicated to make a usable spec. Program material varies too much.

All I really care about in regard to whether or not typical class D amps can output rated power is that they do not shut themselves off when I am listening.

"Typical"??? A badly designed (or poorly ventilated) amplifier of ANY class or design can overheat when stressed. Better to shut down than to burn-out! But it shouldn't happen with any well-designed and properly ventilated amp. (It's probably less-likely with class-A since they basically run at full-power full-time, even when idle.)

Class-D amplifiers run cooler than other designs when "pushed". But because of that, the manufacturer uses smaller heatsinks (and MOSFETs rated for lower power etc.) so that doesn't automatically make it more reliable

Under normal operation, or when idle, there isn't much difference in heat or efficiency compared to class A/B but they are more efficient under worst-case test conditions. (There was a thread about that recently.)

I've never had an amp of any kind shut-down or burn-out from playing it "too loud". But I HAVE heard of it happening to other people.


...Speakers aren't rated for continuous test tones and tweeters can't handle as much power as woofers. If have an amp & speakers both rated at 100W and you run 100W test tones into your speakers, the speakers will usually die before the amp.

 

[Newman]

I listen to movies/tv, not just music, and there could always be cases where I might want a lot of power at any given frequency-depends what is in the sound track. Maybe aliens are killing people with sound in a movie soundtrack??

Then you are best served with a subwoofer. Guess what? They are routinely fitted with Class D amps. Guess what? The Class D amplified subs don't have a generic problem doing their job.

Does anyone own a class D amp that can output, say, around 300 watts per channel or more both channels driven continuously at 20Hz, at 1KHz, and at 20KHz for a very long period of time and that amplifier is lightweight?

See above. Then your speakers, relieved of subwoofer duties, can get by with quite light duty Class D amps. (Subject to usual caveats about difficult speakers.)

cheers

 

[Oddball]

Then you are best served with a subwoofer. Guess what? They are routinely fitted with Class D amps. Guess what? The Class D amplified subs don't have a generic problem doing their job.


See above. Then your speakers, relieved of subwoofer duties, can get by with quite light duty Class D amps. (Subject to usual caveats about difficult speakers.)

cheers

Not so quickly. What you are describing is common, and a bit boring stereotype for HT. Big speakers, big subs and big amps, they can still play together well - but they might need the power on continuous basis.

 

[MaxwellsEq]

If I stick to a simple spec such as "can deliver X watts continuously over the entire 20Hz to 20KHz range indefinitely long"

That is NOT a "simple" spec at all. I've performed power amplifier testing. Please give me some instruction as to how to set up a test for this "simple" spec? Think really hard before you answer.

 

[Newman]

Not so quickly. What you are describing is common, and a bit boring stereotype for HT. Big speakers, big subs and big amps, they can still play together well - but they might need the power on continuous basis.

I considered that, and still think my answer is on the money, and the exception is such an edge case that I let it drift.

I mean, the OP specifically mentioned "alien invasion with sonic death ray" movies, meaning 20 Hz 'earthquake' type continuous LFE channel content at 300W levels.

Try delivering that with any floorstanding main speaker. Just try getting the actual SPLs out into the room, 20 Hz at 115 dB, even for burst, never mind extended periods. So I advised him on what 'best serves' his stated conditions. I'll stand by that.

cheers

 

[Old_School_Brad]

I feel that a source of confusion derives from a lack of experience of bench testing power amplifiers.

There are definitely corner cases for certain film tracks and electronic music with difficult crest factors or frequencies which are uncommon in normally recorded music. But none of them are actually similar to a pure sine wave full power test across a wide range of frequencies.

I don’t think anyone’s claiming they’re the same -certainly not me. My point is simply to emphasize the importance of continuous power capability. The exact wattage or crest factor values are somewhat arbitrary in this context. What really matters is that the amplifier must be able to sustain significant output for extended periods, sometimes minutes, depending on the material. That includes specific sound effects in home theater, or particular music genres.

Calling this type of content a "corner case" doesn’t quite sit right. Maybe it is for some, but for me, it’s not. I have electronic music with sustained low bass as part of my daily playlist. And when I want to listen to it at elevated levels in my living room system, quick 20 ms bursts of xxx watts just aren’t enough for my main speakers. Or, rather, it could be, if the continuous power is high enough, but then I wouldn't care as much to look at the burst power, you know. -It's a point-of-view.

That is helpful to know. So then how would you choose a D class replacement for the AB amp that was proven as adequate? Primarily interested in HT and that is where the most of the amp tool takes place.

For example, my requirement would be to run the bed channels large (LFE+main for additional sub support), and in addition to send the LFE channel to all the bed speakers (sans center) so I can use them as additional supporting subwoofers? It is difficult to say how much toll this takes for the amps, but it certainly makes them real busy.

Currently my amps (see the signature) are adequate for that, but they are regrettably near the end of their useful life. Looks like going 2x up in wattage for the D class might not be enough?

EDIT: just to visualise the current setup:

View attachment 448193

How do you figure these excellent 1070s are nearing end-of-life? They're only around 20 years old. Replacing them with new gear of equivalent quality would cost a small fortune. That said, it might be worth doing a quick check on the idle power consumption of all of them. I’d estimate the whole stack draws a minimum of 380 W. Depending on how many hours per year it’s in use, switching to Class D could actually be a cost-effective option in the long run.

 

[Oddball]

How do you figure these excellent 1070s are nearing end-of-life? They're only around 20 years old. Replacing them with new gear of equivalent quality would cost a small fortune. That said, it might be worth doing a quick check on the idle power consumption of all of them. I’d estimate the whole stack draws a minimum of 380 W. Depending on how many hours per year it’s in use, switching to Class D could actually be a cost-effective option in the long run.

I guess just being realistic in expectations. I do hope that 1070's last for another 20 years, but just expecting the worst. They don't really draw much power in idle - did not check the wattage but they barely get warm.

Brystons on the other hand are handy in the winter as double up as a heaters and need to be AC'd in the summer. More worried about their longevity as one of them already started making some mild trouble - but was able to resolve that myself with a bit of help from excellent Bryston support.

Now we can get back to the D class - sorry for the AB rant.

 

[Oddball]

I considered that, and still think my answer is on the money, and the exception is such an edge case that I let it drift.

I mean, the OP specifically mentioned "alien invasion with sonic death ray" movies, meaning 20 Hz 'earthquake' type continuous LFE channel content at 300W levels.

Try delivering that with any floorstanding main speaker. Just try getting the actual SPLs out into the room, 20 Hz at 115 dB, even for burst, never mind extended periods. So I advised him on what 'best serves' his stated conditions. I'll stand by that.

cheers

Nothing passive will go to 20hz at even 105dB, which is the HT reference for all the channels except LFE that gets +10dB boost. D&M LFE distribution will send the LFE to large bed channels before +10dB boost so that peaks will be limited to 105dB at reference volume.

Depending on the tower speaker capability and placement, 105dB is still a tall task until some 50-60hz (or higher) and does come with significant distortion below that range. IMO the best way to address this is to start reinforcing the large bed channels with subs at eg 60hz (could be some other frequency with eg Trinnov but D&M offer only choice of 40/60/80hz). This LFE+main setup lets you use the speaker capability all the way to its useful output and making up the SPL with subs transitioning in.

Just to be clear, I am not recommending this setup for budget oriented folks as value vs cost equations might not work well. But if one wants reference HT than anyway having as close to full range towers is the requirement so what you do with them is then a question.

 

[Fidji]

to OP question , answer is simple YES, for all the usual suspects - Purifi, Hypex, IcePower etc. For the assemblers - you need to watch for PS used [e.g. 1ET7040 was always a problem, as no compatible PS was available from Hypex]. If you want to be sure to have enough Watts - go for Hypex 1200 or Purifi 1ET9040 [NC2K is not good enough in terms of distortion, IMO]. Subs is different sport - but usually if you enter high performance territory and passive subs, you will get proper recommendation from custom installer - e.g. Ascendo DSP with 4x2500 W or Powersoft [but both for rack install, not living room]

As a sidenote - there are also D Class amps with Linear Power Supply, that offer more of a headroom for peak power, compared to SMPS - e.g from ATI or Trinnov.

But overall - there are usually more differences in 2 HT setups than they have in common. Speaker count, room size, acoustic properties of the room, receiver/processor voltage, power amp gain, speaker sensitivity and impedance and many more parameters. It always helps if one understands it and chooses accordingly.

@audioresearch - for 18in woofer - i would go with IcePower 2000W for bass section, that is about right, unless you have some high sensitivity design.

 

[MaxwellsEq]

I don’t think anyone’s claiming they’re the same -certainly not me. My point is simply to emphasize the importance of continuous power capability. The exact wattage or crest factor values are somewhat arbitrary in this context. What really matters is that the amplifier must be able to sustain significant output for extended periods, sometimes minutes, depending on the material. That includes specific sound effects in home theater, or particular music genres.

Unfortunately, the crest factor is not "somewhat arbitrary". Averaged over a period, and comparing other signals including music, there is nothing tougher than a pure sine wave to test load. Unless you only listen to pure sine waves, the real crest factor of music - including extremes of electronic and film music - means you never load your amplifier or speaker drivers even closely to the same extreme as you do with a sine wave test.

Calling this type of content a "corner case" doesn’t quite sit right.

You are correct, I should not have stated that this music or film soundtrack is a "corner case". The fact it exists and people listen to it means it's "inside the corners" so to speak.

What I was trying to impart was that music most people listen to does not have a lot of energy below 50Hz (i.e. most rock, classical, pop etc.) and has a downward slope of energy from c 1kHz, often being down by at least 20dB at 20kHz. E.g. a piece of electronica I have:

If I performed a wide bandwidth scan with sine waves from 20Hz to 20kHz at 0dbFS, what you would see above is a straight line running along the top of the grid from left to right. It's obvious that even very extreme soundtracks and electronic dance music does not get close to a sine wave at full power at all frequencies.

I'm to to busy to do this - but I had this thought experiment: use a random word generator to select 100 tracks from Spotify and perform a frequency plot of their content. Then average these frequency plots to get "model" of the 100 tracks. Then identify how many tracks are 10dB above the average at 50Hz and 20dB above the average at 20kHz. If these "difficult" tracks were analysed, we would still find their crest factors considerably less challenging than pure sine waves.

 

[Old_School_Brad]

Unfortunately, the crest factor is not "somewhat arbitrary". Averaged over a period, and comparing other signals including music, there is nothing tougher than a pure sine wave to test load. Unless you only listen to pure sine waves, the real crest factor of music - including extremes of electronic and film music - means you never load your amplifier or speaker drivers even closely to the same extreme as you do with a sine wave test.


You are correct, I should not have stated that this music or film soundtrack is a "corner case". The fact it exists and people listen to it means it's "inside the corners" so to speak.

What I was trying to impart was that music most people listen to does not have a lot of energy below 50Hz (i.e. most rock, classical, pop etc.) and has a downward slope of energy from c 1kHz, often being down by at least 20dB at 20kHz. E.g. a piece of electronica I have:

View attachment 448670
If I performed a wide bandwidth scan with sine waves from 20Hz to 20kHz at 0dbFS, what you would see above is a straight line running along the top of the grid from left to right. It's obvious that even very extreme soundtracks and electronic dance music does not get close to a sine wave at full power at all frequencies.

I'm to to busy to do this - but I had this thought experiment: use a random word generator to select 100 tracks from Spotify and perform a frequency plot of their content. Then average these frequency plots to get "model" of the 100 tracks. Then identify how many tracks are 10dB above the average at 50Hz and 20dB above the average at 20kHz. If these "difficult" tracks were analysed, we would still find their crest factors considerably less challenging than pure sine waves.

Sure, I agree with your point about sine waves versus average music -but that wasn’t really the point I was raising. My point is that when you're dealing with a track featuring a continuous bassline, the crest factor becomes largely irrelevant and doesn't tell the whole story. The amplifier still needs to deliver sustained power for the duration of that content, regardless of crest factor. What truly determines the required power is the desired SPL and the speaker's characteristics -those are the real variables.

As for the experiment with 100 randomly selected Spotify tracks -I don't think it tells us much either. Like referencing the average crest factor of electronic music, it only gives you a general picture. It doesn't account for the specific scenario where someone wants to play a track with sustained 30–60 Hz bass at the same average dBA level as, say, a Metallica track. That kind of content places very different demands on an amplifier.

So it's not that we're necessarily disagreeing, I think. -I think we're just coming at the topic from slightly different angles and maybe talking past each other a bit?

 

[MaxwellsEq]

So it's not that we're necessarily disagreeing, I think. -I think we're just coming at the topic from slightly different angles and maybe talking past each other a bit?

Probably. That's what the Internet is for

I also agree that what people experience in the real world - e.g. wanting to play a sustained bass line at 30Hz to 60Hz for 3 minutes at high power - seems probably tougher than using a sine wave to measure clipping point at 30Hz for 20 seconds into a resistive load. All I can add, is that, having done both the sustained bass line and the sine wave testing ( ) , sine wave testing is a great deal more brutal with much quicker temperature rise than the sustained bass line! However, 3 minutes of sustained load is obviously also tougher than 10 seconds. So it's probably horses for courses...