Class D: Better with Low Efficiency Speakers?

[watchnerd]

I was looking at this charts for the distortion vs output of the Devialet Expert Pro 130 (source credit due to the SoundStage Hi-Fi review):

Red line = 1W, Magenta = 10W, Blue = 30W, Cyan = 60W, Green = 80W

Typical of the class, distortion is not lowest at the lowest output, but in the midband to upper quartile of its rated power output.

Coupled with highly efficient (say >93 dB) speakers, this would mean that the amp is operating in a higher distortion zone at normal listening levels (75-85 dBA) at a typical distance of 10-12 feet, than it would be if partnered with a lower efficiency speaker (<84 dB).

Does this mean lower efficiency speakers are actually a better match for Class D amps than high efficiency ones?

 

[mitchco]

Dunno... but I wonder how audible .005% distortion is... or even .1%...
My Pass ACA's Class A amps have orders of magnitude more distortion. I wonder if it is audible... http://www.firstwatt.com/pdf/art_amp_camp_1.pdf

 

[amirm]

The distortion is not higher but rather noise is. THD+N measurements are dominated by the latter component (Noise) at low volumes.

I have listened to class D amps with audible buzz at zero volume. Same with some other linear amps actually.

 

[March Audio]

I was looking at this charts for the distortion vs output of the Devialet Expert Pro 130 (source credit due to the SoundStage Hi-Fi review):

Red line = 1W, Magenta = 10W, Blue = 30W, Cyan = 60W, Green = 80W

Typical of the class, distortion is not lowest at the lowest output, but in the midband to upper quartile of its rated power output.

Coupled with highly efficient (say >93 dB) speakers, this would mean that the amp is operating in a higher distortion zone at normal listening levels (75-85 dBA) at a typical distance of 10-12 feet, than it would be if partnered with a lower efficiency speaker (<84 dB).

Does this mean lower efficiency speakers are actually a better match for Class D amps than high efficiency ones?

I would say no, simply because the speaker itself is the dominant source of distortion. You are talking about probably a minimum of 0.1 or 0.2 % midrange and much higher at low frquencies.

The hypex amps I just put in my speakers are lower distortion than the devs, 0.01% at 100mW, falling to about 0.0008% at about 100w, and silent.

https://www.hypex.nl/img/upload/doc/ncore_mp/nc252mp/Documentation/NC252MP_02xx.pdf

 

[Blumlein 88]

Think of the EIN spec of microphone pres. It is true of all circuits with gain. Resistances set some minimum noise. Any gain applied amplifies the noise as well as signal. Class D is no worse and no different than any other low distortion device in that noise plus gain limits the THD+N results. High gain with high efficiency speakers is not a good combination. Class D is actually so good you can often get away with a ridiculously overpowered amp on some very efficient speakers because little noise will result. The same thing with a regular class AB amp of huge power usually will cause objectionable noise. It is the N part of THD+N which is making it look like the distortion goes up at very low levels. In reality the device distortion is as low or possibly even lower as mid power levels. The noise covers up this fact.

Depending upon particulars of the gear in use, sometimes it pays to have a reduction of the signal with resistors at the input of the amp. This lowers both signal and prior noise so what noise gets raised by gain is less to start with. Again depends upon particulars and resistances involved.

 

[watchnerd]

I would say no, simply because the speaker itself is the dominant source of distortion. You are talking about probably a minimum of 0.1 or 0.2 % midrange and much higher at low frquencies.

I was thinking the same.

Here is a plot of the JBL 4367 at 100dB at 1 meter at various frequencies, which does quite a bit better than than .1-.2% as you approach the center of the midrange:

 

[watchnerd]

I would say no, simply because the speaker itself is the dominant source of distortion. You are talking about probably a minimum of 0.1 or 0.2 % midrange and much higher at low frquencies.

The hypex amps I just put in my speakers are lower distortion than the devs, 0.01% at 100mW, falling to about 0.0008% at about 100w, and silent.

https://www.hypex.nl/img/upload/doc/ncore_mp/nc252mp/Documentation/NC252MP_02xx.pdf

The new NAD 338 is using Hypex output stages, I don't know if it's the same version, though:

https://nadelectronics.com/product/c-338-classic-digital-dac-amplifier/

 

[watchnerd]

Think of the EIN spec of microphone pres. It is true of all circuits with gain. Resistances set some minimum noise. Any gain applied amplifies the noise as well as signal. Class D is no worse and no different than any other low distortion device in that noise plus gain limits the THD+N results. High gain with high efficiency speakers is not a good combination. Class D is actually so good you can often get away with a ridiculously overpowered amp on some very efficient speakers because little noise will result. The same thing with a regular class AB amp of huge power usually will cause objectionable noise. It is the N part of THD+N which is making it look like the distortion goes up at very low levels. In reality the device distortion is as low or possibly even lower as mid power levels. The noise covers up this fact.

But but but....

Vacuum tubes have waaaaay more noise and people use them with super efficient speakers on a regular basis.

The SET crowd even seems to prefer it.

 

[mitchco]

Running a Crown XLS 1502 Class D amp on a pair of these high efficiency bass bins: http://www.jblproservice.com/pdf/Theatre Series/4739.pdf
I have to put my ear right next to the center of the cone to hear the faintest of noise...

 

[watchnerd]

The distortion is not higher but rather noise is. THD+N measurements are dominated by the latter component (Noise) at low volumes.

I have listened to class D amps with audible buzz at zero volume. Same with some other linear amps actually.

In the Devialet case, do you think that is "digital noise" as the digital volume control starts to approach the digital noise floor or other more general electrical circuit / PS noise?

 

[March Audio]

I was thinking the same.

Here is a plot of the JBL 4367 at 100dB at 1 meter at various frequencies, which does quite a bit better than than .1-.2% as you approach the center of the midrange:

Thats good, just for interest here are the results I got for my speakers

https://audiosciencereview.com/forum/index.php?threads/3-way-dsp-xo.991/page-2#post-45991

 

[amirm]

In the Devialet case, you think thats "digital noise" as the digital volume control starts to approach the digital noise floor or other more general electrical circuit / PS noise?

No it is not related to volume control. That kind of measurement is made by varying the input signal to the amp, not using its volume control.

 

[March Audio]

In the Devialet case, do you think that is "digital noise" as the digital volume control starts to approach the digital noise floor or other more general electrical circuit / PS noise?

Is the buzz mains or more of a digital noise. I would say the volume control itself is not the problem. Its just as blumlein said above, the gain staging is crucial. A power amp might typically have 26 to 30dB of gain, so what was a respectible noise floor straight out of the dac suddenly doesnt look so good.

Hence sometimes the use of an attenuator before the power amp to reduce signal input and noise input to the amp. You then compensate by turning the dac output volume up. The noise level however stays at the same reduced level

 

[watchnerd]

Thats good, just for interest here are the results I got for my speakers

https://audiosciencereview.com/forum/index.php?threads/3-way-dsp-xo.991/page-2#post-45991

Nice, what are your speakers?

 

[March Audio]

Nice, what are your speakers?

A home brew project that has evolved over the last year. Start at the top of that thread

 

[Blumlein 88]

But but but....

Vacuum tubes have waaaaay more noise and people use them with super efficient speakers on a regular basis.

The SET crowd even seems to prefer it.

Well some SETs have very little gain and still have a bit of noise. Most push pull tube amps have lots of gain. Using a preamp turned way down will also turn down the noise though not as effectively as passive reduction with resistors at the input. Years back many tube pre's and power amps were foolishly designed. High gain amp, very high gain pre, and both were less quiet than SS gear. With CD or DAC inputs to such systems the volume ended up turned well down. But it is all down to gain staging.

Even with efficient speakers, excluding things like 100 db and above horns, you are likely only using 3 volts or less. A .1 % level of that signal is -60 db down. You likely can't hear that as distortion. You likely could hear that as a little bit of faint noise with no signal. The Devialet graph shows more like a -80db level at what would be that voltage (a touch over 1 watt). So depends upon the particulars of the speaker if you could hear it or not. For instance you could attenuate the signal at the amp input by 20 db, which kills any upstream noise by 20 db and the level left in the amp isn't enough to matter which would result in a noise free result I do believe were you to use the JBL 4367. The attenuation lets you turn up the source getting the signal out of the noise floor of the preceding gear and amp.

 

[Soniclife]

Is the buzz mains or more of a digital noise. I would say the volume control itself is not the problem. Its just as blumlein said above, the gain staging is crucial. A power amp might typically have 26 to 30dB of gain, so what was a respectible noise floor straight out of the dac suddenly doesnt look so good.

Hence sometimes the use of an attenuator before the power amp to reduce signal input and noise input to the amp. You then compensate by turning the dac output volume up. The noise level however stays at the same reduced level

I've done a simple test with my Dev 440, of taking a quiet music file, making it much quieter in audacity, then playing it back using a player with a digital volume control turned way down, then turning the vol on the amp all the way up.
Using the optical input from a Chromecast audio, or the Ethernet input there is very little noise with my ear almost touching the tweeter, if you pause the music some sort of mute kicks in for total silence after a very short time. The usb input is slightly noisier, and varies with what is driving it. My speakers aren't sensitive, 85db I think, and the noise isn't audible from a few feet away, high sensitivity horns would be worse, but you would never have the vol up that far, at sensible vol settings they are silent.

 

[Frank Dernie]

I was looking at this charts for the distortion vs output of the Devialet Expert Pro 130 (source credit due to the SoundStage Hi-Fi review):

Red line = 1W, Magenta = 10W, Blue = 30W, Cyan = 60W, Green = 80W

Typical of the class, distortion is not lowest at the lowest output, but in the midband to upper quartile of its rated power output.

Coupled with highly efficient (say >93 dB) speakers, this would mean that the amp is operating in a higher distortion zone at normal listening levels (75-85 dBA) at a typical distance of 10-12 feet, than it would be if partnered with a lower efficiency speaker (<84 dB).

Does this mean lower efficiency speakers are actually a better match for Class D amps than high efficiency ones?

Those graphs are, as is common, distortion + noise, typically very low distortion amps have graphs of this shape since the noise contribution is much more than the distortion contribution at low powers.

Also the Devialet is not class D, there used to be a white paper about it on the original web site, which was informative and clear, but since they sold shares to a marketing bloke the thing is almost impossible to navigate and full of screen candy and links to sales
Basically there is a class A amp which controls the output voltage and 4 class D amps per channel transformer coupled to the output and digitally controlled to supply whatever current is necessary to precisely maintain that voltage. In principle the quality is that of a class A amplifier but much more efficient. The class A amp does run quite hot though.
The hybrid circuit is unique to them and patented, I heard Bruno Putzeys was very impressed by the technology but felt a sufficiently good result could be obtained much cheaper and simpler

 

[Frank Dernie]

I've done a simple test with my Dev 440, of taking a quiet music file, making it much quieter in audacity, then playing it back using a player with a digital volume control turned way down, then turning the vol on the amp all the way up.
Using the optical input from a Chromecast audio, or the Ethernet input there is very little noise with my ear almost touching the tweeter, if you pause the music some sort of mute kicks in for total silence after a very short time. The usb input is slightly noisier, and varies with what is driving it. My speakers aren't sensitive, 85db I think, and the noise isn't audible from a few feet away, high sensitivity horns would be worse, but you would never have the vol up that far, at sensible vol settings they are silent.

I use my Devialet with Goldmund Epilog 1&2 which seem more efficient than the 91db/watt spec but a punishing load (a SET barely raised a murmur from them) or with Tune Audio Anima horns which are 109dB/watt - all I do is configure them to a lower maximum output power when I connect the horns.
On any input but phono they are silent when music isn't playing, ear in treble horn gives a just audible hiss, 1 foot away nowt.

 

[watchnerd]

, but since they sold shares to a marketing bloke the thing is almost impossible to navigate and full of screen candy and links to sales

No kidding -- I can't even find a link to the SAM database any longer, nor a link to a downloadable manual.