How much amplifier power do you really need?

[RebelMan]

@RebelMan

The points you quoted from Roger Sanders' white paper are his conclusions, not assumptions.

They follow pretty much from this simple maths:

Let's say you need around 5 W RMS to achieve normal listening levels (whatever they are) with a low-ish efficiency speaker in a normal-sized room.  Given that -- from unbiased sources -- we know some music can have a crest factor of over 20 dB, that means the amplifier has to deliver over 500 W peak power to reproduce that music without clipping.

That seems very straightforward and uncontroversial to me.  Which of the two premises do you disagree with, or what is the fault in the logic?

Actually, it is an assumption Roger is making and I will explain. 

I quoted him directly from his paper so I don't see how you saw a conclusion at that point early in his paper.  It looks to me like you extrapolated what he is saying to make your math work around what he is saying and I do agree with your math.  However, I have never seen 20dB crests at anytime in my many years in this hobby.  I have seen 10+dB but that is not to say that maybe someone somewhere has actually seen crests go that high.  It seems to me it was most likely an exaggeration.  In any case, using your figures a 500W draw doesn't say much and this is where the assumption comes in.  He does not say anything about the load.  Have you ever seen an impedance-phase curve?  If Roger superimposed such a curve over the power curve then he would have some good news.  He didn't.

I bet my little 140 could muster 500W on it's way down to 1 ohm.  That 4KVA power supply it has makes it a BEAST.

[RebelMan]

After reading my post to RebelMan, I'm embarrassed for display such an inappropriate rant.  My apologies to the members.  It supplied nothing to the ongoing discussion.
Jim

Jim,  I was not offended. In fact I could see how you saw the humor in my remarks as well.  No worries.

[RebelMan]

Just to note; in case of our Devialets the sheer number of variables is even larger and makes this discussion even more difficult as their internal microprocessors in the power supply and output stages run code determining the voltage and current limits short and long term. It’s certainly possible these code imposed limits (said to be there for thermal management and accounting for the differing heat dissipation capabilities of the different cases) could affect SQ e.g. when comparing a 170 vs. 240 or any model. The smaller models like the 130 have a different class D output stage.

You are correct, there are some limits that were imposed and across the board but it is unlikely that they impact sound quality.  The limits are mostly thermal and functional (for dual-mono output) and I also suspect for some cost savings too.  The 130 and 140 are using different Class-D amplifiers that limit their output current to 1.5x the power factor but they are not of an inferior quality.  You could liken it to different types of batteries, say D and C.

What does concern me are the limits placed on the 220.  It is not as linear as the 130/140.  Output at 2 ohms is limited to 330W.  I have been trying to confirm this but if it is in fact true then the 130 and 140 are more powerful than the 220.  Surprise!

[RebelMan]

Joking apart (sorry, I couldn’t resist), I think Antoine’s post #20 is the most likely reason there was an obvious difference between the D170 and 240, with the Blades at least.  In other words, the reason the amplifiers performed differently was due to the known technical differences between the amplifiers.  No need for Sherlock Homes.  In terms of level matching, we pushed the 240 as loud as I would ever want to listen, so irrelevant in this case.

That was too funny @Confused, nice catch.  Hopefully you'll catch my response to @Antoine post as well.  Here's one for you... https://www.youtube.com/watch?v=x0ow4X8tiMI

[Jean-Marie]

Because de Devialet is fully digital it is a bit different and in particular you can know exactly what is the peak power.

You will never exceed 0dBFS (if you except the pathological situations where the reconstruction filter moves above).

If you look at the configuration file of your Devialet, you find a section like this (in my case this is a D220):

"ADH_CONFIG":{
"VOLTAGE_0DB":51,
"PUISSANCE_0DB":220
}

It tells your you that the peak voltage is 51volt which corresponds exactly to 440W under 6ohms or a pure sinus tone with a RMS power of 220W. (The crest factor of a sinus is 3dB in power).

Jean-Marie

[Antoine]

I vaguely remember hearing a while back that the volume control was arrange so that 0 dB was full rated power when the input levels were at max - for a digital input at least I suppose that's pretty well defined.  (I don't think I'm making that up...)  Maybe it has some additional gain available to handle lower-level inputs?  Another question to add to the growing list for Mathieu, perhaps.

Yes, I also usually listen with the volume control in the range -40 dB to -20 dB, with rare exceptions usually for orchestral music, and I don't think I've ever ventured above -10 dB.  My speakers are meant to be 87 dB at 1 m for 2.83 V input and have nominal 4-ohm impedance.  If I've got the maths right that means 84 dB at 1 W.

Interesting thought about whether/how the dynamic range indication in Roon might relate to crest factor. It makes me wonder what the DR value would be for the PanSonic track used in the video.

IIRC from 0dB and up on volume control the Devialet DSP applies digital amplification so mathematically changes the input signal to make it ‘louder’. Of course the chances of the input signal clipping over 0dBFS become larger here and thus one’ll see the clipping indicator more often.

[RebelMan]

What you have defined is energy.  Power is the capacity for work per unit time.  Never mind, it doesn't have that much bearing on the discussion.

Incorrect.  Energy is stored work.  Work is the amount of energy transferred to or from a system by a force.  Power is the rate at which work is performed. Implicit in my explanation was the matter of time.  My mistake was assuming you would grasp this.

One of the interesting points of the video was that it did indeed measure instantaneous power delivered to the load.  To be really explicit, the meters on the amplifiers showed the amount of power needed to drive the speakers with a faithful (undistorted) signal at the set volume level.

I don't disagree but it still only shows half the picture which makes their point all the more meaningless because it is taken out of context.  I can Dyno test my car to 600HP but that doesn't tell you a thing about how fast it really is.

GOOD, power that is sufficient in supply at any given time for any given load.

CLEAN, power that is free from artifacts inherent of the architecture's design and components.

LINEAR, as the demands for power increase (instantly or continuously) the supplies of power also increase in lock step.

None of those qualities are inherent to power in itself (where do they appear in the definition of power?).  Of course I agree that they are desirable characteristics of an amplifier.

Are you trying to be obtuse?  You asked for a "distinction" between terms and I provided it to you.   A symbiotic relationship exists between power and amplifier that can not be divided.  If you found a way then apply for a Noble Prize.

The video shows that the amplifiers are delivering up to 600 W per channel peak power into the load at some points.  Assuming the amplifier knows how to properly measure the peak power it's delivering, the power is just what the meter shows it to be whether it's into a 1-ohm, 3-ohm, 6-ohm or any-other-ohm load.  You don't have to make any assumptions whatever about the load -- or any other "conditions", for that matter -- if you believe that the amplifier is designed so that the peak power meters are accurate.

Correction, they said 750W peak, but that's beside the point.  The amplifier is not telling us anything more than what the power was at a given time.  750W into an 8 ohm load is vastly different than 750W into an 2 ohm load.  So which is it?

How can you conclude that an amplifier needs 100 times more power available for peaks when we don't even know what those peaks are?  That's purely conjecture.  You cannot solve what you do not know.  Again how do you know?  You need to measure.

It's not my conclusion.  It follows directly from the definition of crest factor, as explained in the article I linked to, without any interpretation or assumptions on my part.  Neither is it conjecture, unless you disagree with the values of crest factor that are quoted by that article and other independent sources.  I am not trying to "solve" anything, in your words.

I never implied it was your conclusion.  However, you adopted the conclusion to buttress your position.  Although, if you were more open minded about the case of power vs sound quality "you" as I used it would have been interpreted as defined by Merriam-Webster... —used to refer to any person or to people in general

I suspect our conversation has gone beyond the point of being interesting or useful.  You're welcome to continue if you wish but I will bow out here.

Not so fast.  Talk is cheap.  Time to put up or shut up.  I made you an offer to put this issue to the test.  Will you take it?

[RebelMan]

Because de Devialet is fully digital it is a bit different and in particular you can know exactly what is the peak power.

You will never exceed 0dBFS (if you except the pathological situations where the reconstruction filter moves above).

If you look at the configuration file of your Devialet, you find a section like this (in my case this is a D220):

"ADH_CONFIG":{
    "VOLTAGE_0DB":51,
    "PUISSANCE_0DB":220
  }

It tells your you that the peak voltage is 51volt which corresponds exactly to 440W under 6ohms or a pure sinus tone with a RMS power of 220W. (The crest factor of a sinus is 3dB in power).

Jean-Marie

I don't think the meaning of this code has ever been clarified.  @Antoine asked for clarification back when the Pro series debut.  How are we suppose to interpret what this is saying?

I have never seen an amplifier produce more than about 4dB of dynamic headroom to serve signal peaks. Many amplifiers are capable of producing between 1-2dB. The Devialets (all of them) have virtually no dynamic headroom (0dB) according to the tests I have seen.  So how is it the 220 would have a crest factor of 3dB?

I suspect the "VOLTAGE_0DB":51 statement only pertains to the 220 when it is in bridged mode (dual-mono configuration) and that the actual peak voltage it artificial limited somewhere else in the code that is not evident in the snipet above.  Would you agree?

[David A]

Because de Devialet is fully digital it is a bit different and in particular you can know exactly what is the peak power.

You will never exceed 0dBFS (if you except the pathological situations where the reconstruction filter moves above).

If you look at the configuration file of your Devialet, you find a section like this (in my case this is a D220):

"ADH_CONFIG":{
    "VOLTAGE_0DB":51,
    "PUISSANCE_0DB":220
  }

It tells your you that the peak voltage is 51volt which corresponds exactly to 440W under 6ohms or a pure sinus tone with a RMS power of 220W. (The crest factor of a sinus is 3dB in power).

Jean-Marie

I don't think the meaning of this code has ever been clarified.  @Antoine asked for clarification back when the Pro series debut.  How are we suppose to interpret what this is saying?

I have never seen an amplifier produce more than about 4dB of dynamic headroom to serve signal peaks. Many amplifiers are capable of producing between 1-2dB. The Devialets (all of them) have virtually no dynamic headroom (0dB) according to the tests I have seen.  So how is it the 220 would have a crest factor of 3dB?

I suspect the "VOLTAGE_0DB":51 statement only pertains to the 220 when it is in bridged mode (dual-mono configuration) and that the actual peak voltage it artificial limited somewhere else in the code that is not evident in the snipet above.  Would you agree?

I'm stupid but I'm going to make a guess at what is being set in those lines for Power and Puissance.

Since they occur in the section for ADH config, I'm going to guess that the line for Voltage defines the setting for the class A amp which supplies the voltage and the 51 V that it shows for Jean Marie's 220 equates to the voltage for 220 W output per channel for a 220 in stereo mode. 

"Puissance" seems to translate to power. There's a setting in the configurator's speaker section where you can set maximum power output and the maximum you can set is your amp's specified power (that's also the default setting) but you can limit to to a lower output. I suspect that the 220 shown in Jean Marie's file indicates that he has chosen to leave that setting at the maximum for his 220. If you have a dual amp setup I suspect the maximum/default will be the mono output so for a 440 which is comprised of 2 220s the figure shown would be 440 which would be the power output from 51V for bridged mode for the amp.

If you limit the power output I think you'd see a lower number for Puissance reflecting the lower wattage limit you've set and that the lower output would be reached at a 0 dB volume setting. Maximum power output, whatever you set it to, will always be reached at a volume setting of 0 dB. I don't know whether the voltage value will also change if you limit maximum volume but it may also.

Basically, taken together, I think those 2 values control the settings for the Class A and Class D amps respectively given the maximum power output you choose in the speaker settings in the Configurator.

Just a guess as I said but there''s no reason to specify ADH configuration if you don't give the user the option to vary maximum power output and given the fact that the Devialet is a Class A/Class D hybrid it would seem to make sense to have separate values for the Class A and Class D stages if you are going to be able to vary output.

I've got a 140. That section of my config file shows 41 V and a Puissance of 140. I think if I added a companion to make a 210 the file would show 41 V and Puissance 210.

[RebelMan]

Since they occur in the section for ADH config, I'm going to guess that the line for Voltage defines the setting for the class A amp which supplies the voltage and the 51 V that it shows for Jean Marie's 220 equates to the voltage for 220 W output per channel for a 220 in stereo mode. 

"Puissance" seems to translate to power. There's a setting in the configurator's speaker section where you can set maximum power output and the maximum you can set is your amp's specified power (that's also the default setting) but you can limit to to a lower output. I suspect that the 220 shown in Jean Marie's file indicates that he has chosen to leave that setting at the maximum for his 220. If you have a dual amp setup I suspect the maximum/default will be the mono output so for a 440 which is comprised of 2 220s the figure shown would be 440 which would be the power output from 51V for bridged mode for the amp.

If you limit the power output I think you'd see a lower number for Puissance reflecting the lower wattage limit you've set and that the lower output would be reached at a 0 dB volume setting. Maximum power output, whatever you set it to, will always be reached at a volume setting of 0 dB. I don't know whether the voltage value will also change if you limit maximum volume but it may also.

Basically, taken together, I think those 2 values control the settings for the Class A and Class D amps respectively given the maximum power output you choose in the speaker settings in the Configurator.

Just a guess as I said but there''s no reason to specify ADH configuration if you don't give the user the option to vary maximum power output and given the fact that the Devialet is a Class A/Class D hybrid it would seem to make sense to have separate values for the Class A and Class D stages if you are going to be able to vary output.

I've got a 140. That section of my config file shows 41 V and a Puissance of 140. I think if I added a companion to make a 210 the file would show 41 V and Puissance 210.

The logic behind your hypothesis follows but some of the math is still elusive.  Looking at the 220, max voltage is set to 51V at 0dBFS and the corresponding impedance at this voltage is 6 ohms.  Power in this case computes as 434W which is almost twice what the 220 is rated for.  @Jean-Marie suggests this to be peak power.  If we apply the math to the 140 using your figures power computes to 280W peak which is exactly twice what the 140 is rated for continuously.  If power from the Devilet 140 and 220 are not artificially limited (somewhere in the code) then where did the peak power go?

According to HiFi World the Expert 220 Pro reached a peak power of 190W into 6 ohms which is nowhere near the 434W computed from above.  @Jean-Marie suspects this is due to the evaluation unit being tested with SAM enabled. However, Audio was unable to measure the peaks of the Expert 140 Pro because the sophisticated power protection circuits prevented it...   

"It is worth mentioning the presence of a special protection circuit, which, after exceeding a certain input voltage (sensitivity), does not allow distortion of the terminals, so distortions do not increase, but stop at a level even lower than the standard THD + N = 1%. Hence, our results concern such conditions, because it is the highest power that can be obtained."

It would seem peak power may never be reached regardless of the amplifier's ability to produce it because it's trying to play it safe.  If this is true then the 220 has a major handicap.  I confirmed with Devilalet that continuous power of the 140 and 220 is as follows...

                 Expert 140 Pro            Expert 220 Pro
8 Ohms      105 W                        165 W     
6 Ohms      140 W                        220 W
4 Ohms      210 W                        330 W
2 Ohms      420 W                       330 W

If peak output power is limited to continuous output power (as multiple tests have shown) because of the protection circuits then that can be a problem for people with speaker impedances that dip below 3 ohms and mine do.  For the people that still believe more power sounds better than less power, have some crow.