Can we get some DATA about how much power is in each channel PLEASE?!?

[Head_Unit]

Somewhere in some thread on some forum posted actual measured data (using Audacity or such) showing the LCR peak power ran about 10 dB higher than the peaks in the surrounds. Could those who are set up for it do some more measuring like that and post the results? It has a huge implication for recommendations about adding amplifier channels or adding amplifiers to replace internal channels.

 

[Kal Rubinson]

It's quite dependent on the program content and the mastering.

 

[tifune]

Since you're not getting any takers, I grabbed some quick and dirty outputs from my HTP1. The only value on these is the relative output between channels, and even that should be considered a guesstimate when compared to your own situation.

This shows peak output of 3 very recent releases I'm into this week, and then 2 that are far more popular/common. All songs from Apple TV 4K; the Chat Pile song is stereo, up mixed with Auro3D (small=10) just for a different perspective. Rest are atmos. This disclosure song stands out to me because I don't think I've ever seen a song with so much surround L/R emphasis. Again, this is just what I'm into - YMMV.

My system has a global 80Hz crossover. I would expect the output to be a little higher in any channel that's running full range w loudness compensation enabled. I'd probably ignore the sub output altogether (which is fine per your original Q because home-use subs almost always have their own amps)

With my own content I would agree that front L/R channels tend to be loudest by 10dB, if not more, for music. If TV/movies are a regular part of your consumption, center channel should be taken as seriously as front L/R.

 

[-Matt-]

Somewhere in some thread on some forum posted actual measured data (using Audacity or such) showing the LCR peak power ran about 10 dB higher than the peaks in the surrounds. Could those who are set up for it do some more measuring like that and post the results? It has a huge implication for recommendations about adding amplifier channels or adding amplifiers to replace internal channels.

Are you perhaps asking if AVRs have different power ratings for each of their channels? (I.e. Do the LCR channels have higher power rating than the surrounds and if so perhaps only the LCR get measured and not the lower power surround and height channels).

I think there was something about this on Audioholics a while back (was it related to eco nannies on Yamaha AVRs?). I'll see if I can find a link...

In answer to your request for data, @amirm does generally provide power ratings with varying number of channels being driven.

Like this one:

 

[Qbd]

While not explicitly showing the peak values. this thread over on avsforum.com shows the waveforms for many movies.

Judging by the squinting at the waveforms, obviously the total energy is mostly in LCR and LFE, but it looks like there are fairly high peaks in the surrounds too, so the peak power required may not be lower, especially since the surround speakers often are less sensitive than the LCRs too.

 

[voodooless]

Also note that the distance to your speakers have quite a big impact. If your surrounds are further away than your fronts, then you will need to make up for that extra distance in power. So there is not a single answer here.

I suspect that on average, even with the peaks included, the surrounds will be less loud, especially in masters for home audio. On average surround speakers are simply smaller and less powerful, so reliance on them will be lower. The quoted 10 dB on average wouldn’t be a strange figure. To your ears that will sound half as loud. Given that the action takes place on screen, it makes sense to direct most of the sound to it. But as always, YMMV…

 

[Head_Unit]

Are you perhaps asking if AVRs have different power ratings for each of their channels? ...@amirm does generally provide power ratings with varying number of channels being driven.

No, really what my question is getting at is "do you really need to buy a 5- or 7- or etc number of channels external power amp, or is a three channel for the LCR enough?

The data you posted, I'm going to look at more of those tests. If my just-woke-up brain is calculating correctly, that Denon only loses about 1 dB of output power compared to if it was "perfect" which illustrates a different contention I make: buying a stereo power amp of about the same power as an AVR (Emotiva BasX in particular!) is NOT going to help your total output power significantly. You need an amp of significantly MORE power rated, and the purpose of this thread is to get data supporting my contention that is must be THREE channels and that adding just two channels is also shooting yourself in the foot.

 

[Head_Unit]

If your surrounds are further away than your fronts, then you will need to make up for that extra distance in power.

That is an excellent point! If the surrounds were twice as far that would be 6 dB, still less than the hypothetical 10 dB but that's a big difference.

As for the surrounds being smaller, the total power they would get would be reduced if highpassed. However someone else once posted waveforms showing that highpass filtering did not reduce the PEAK power due to the nature of filters. So the amp of a highpassed channel might not run as hot but it would still clip as soon. More data to gather! I gotta figure out how to set this up so I can do it myself...maybe when I retire ha ha.

 

[Blumlein 88]

The center channel averages about 6 db more than left or right. That's 4x the power. Surround channels are 10 to 12 db less than center. So center vs surrounds could be a 16x difference in power.

 

[Chr1]

That is an excellent point! If the surrounds were twice as far that would be 6 dB, still less than the hypothetical 10 dB but that's a big difference.

As for the surrounds being smaller, the total power they would get would be reduced if highpassed. However someone else once posted waveforms showing that highpass filtering did not reduce the PEAK power due to the nature of filters. So the amp of a highpassed channel might not run as hot but it would still clip as soon. More data to gather! I gotta figure out how to set this up so I can do it myself...maybe when I retire ha ha.

Could someone please explain to me why a high-passed amplifier channel would clip as soon as one that is not high-passed? This does not seem to make sense to me. Thanks!

 

[NTK]

Could someone please explain to me why a high-passed amplifier channel would clip as soon as one that is not high-passed? This does not seem to make sense to me. Thanks!

Using Audacity, generated a 1 second white noise (sampling frequency = 48000 Hz), max amplitude 0.5, and duplicated it:

Apply a high pass filter, cutoff at 1 kHz, 12 dB/octave slope to the duplicated track:

Result:

Note: I used white noise in this illustration to show that filtering a signal can increase its max amplitude, but it is sort of a synthetic worst case scenario. If you use pink noise instead (a signal much closer in spectral power distribution to audio), you'll get the typical "expected" response.

 

[Chr1]

Many thanks for the reply, but I still don't really understand why?
Surely a channel that doesn't play the lower frequencies requires less power, and therefore will not clip as readily?

 

[NTK]

Many thanks for the reply, but I still don't really understand why?
Surely a channel that doesn't play the lower frequencies requires less power, and therefore will not clip as readily?

It is just how some signals with the "right" kind of spectral power distribution works out. I'll use square wave as a second example. When you expand a square wave into a series of sinusoids, it is (Wikipedia reference):

Here is how the series computes to when the first sine term is removed (removing the k=1 term is equivalent to applying a brick-wall low pass filter at f ≈ 1.5). The series solution shown is computed using the terms k from 2 to 4000. You can see from the plot that the low passed square wave have spikes higher than the original square wave amplitude of 1.

[Edit] The filtered square wave does have lower average power, but higher peaks! The relationship between peak power and average power is not direct.

 

[Chr1]

Thanks. Very interesting. So I am presuming that this is the case with all amps/channels, not just those in multi-channel amps/AVRs...

 

[NTK]

Thanks. Very interesting. So I am presuming that this is the case with all amps/channels, not just those in multi-channel amps/AVRs...

Yes.

 

[Chr1]

Thanks again. This kind of contradicts the assumption that I have had for years re amplifier power required when high passed. Good to know.

 

[NTK]

Thanks again. This kind of contradicts the assumption that I have had for years re amplifier power required when high passed. Good to know.

But please keep in mind what I showed were corner cases. You should get the typically expected results the vast majority of the time with normal use cases.

 

[danadam]

You can see from the plot that the low passed square wave have spikes higher than the original square wave amplitude of 1.

Any band-limited square wave will also have spikes. AFAICT the spikes in the original wave will be higher, though with 4000 terms the difference should be minuscule. Here's with 110 terms:

 

[dlaloum]

All amps have constraints in their design, typically the power supply is undersized for the maximum power that the amplifier circuits can put out, hence we have a difference between rated continuous and peak power output. (Heat dissipation capability will also impact on the continuous to peak difference depending on design)

When looking at AVR's where there may be anywhere from 5 to 11 powered channels it gets worse (often a LOT worse) - and the full power output is typically only available when driving 2 channels, as soon as the driven channel count increases beyond that, you start to hit the constraints of the power supply in most cases. (yes there are edge case flagship models that are less constrained)

So in my case, using an Integra DRX3.4 ... rated at 100W@8ohm/channel - it was needed to drive my 5.1.4 setup, the front L, C, R speakers are very hard to drive Gallo speakers with an impedance that drops down to 1.6ohm - surrounds are B&O Penta's with a much higher SPL efficiency as well as a benign 8ohm impedance rating, heights are Gallo Nucles Micro's, not overl SPL efficient but again, with a benign 8ohm nominal impedance rating.

Calculating MLP requirements for my preferred listening level (72db continuous, 92db peaks) shows a requirement around 3.5W continuous and no more than 16W peak (@8ohm... double that at 4ohm, 4x at 2ohm etc...) .... which is to say, contrary to many people's beliefs, the power requirements are relatively modest.

However my experience was, that when setting up my system and trying to drive the setup using the AVR alone, the midrange frequencies, vocal area, sounded confused, imaging collapsed.... it wasn't right! (my previous flagship Integra DTR70.4 did not have this problem - but it also had a massive power supply!) - so I plugged in my external amp to handle the front L/R speakers - leaving the critical centre on the AVR as well as surrounds and heights.

The improvement was immediately noticeable.

Removing that load, left sufficient spare power supply capacity for it to handle the just as difficult center channel (also 1.6ohm) - while also handling surrounds and heights.... the clarity was back, vocals instantly became much more understandable, and the imaging and soundstage were back the way they should be.

In the end, I played safe, and provided myself with additional headroom, by running all 3 fronts L/C/R off external power amps appropriately capable of driving very difficult loads - and I have left all the other speakers on the AVR's internal amps.

My personal analysis is that the power supply is far too current constrained to handle those speakers... and once it starts to run out of current, you get a rise in distortion and immediate audible impact. (impact will differ depending on the specifics of the individual models design)

The problem with most AVR's is that they all have this issue. - But if your speakers present the AVR with a benign 8ohm load and reasonable efficiency - then the issue may never be encountered! (and if this is the case for a particular setup, then adding external power amps, will have no audible impact! - other than potential placebo effects)

Current AVR's that have "beefy" power supplies include the Onkyo/Integra RZ70/DRX8.4, Denon A1H... but all the models below those flagships in those brand ranges, do suffer from that type of constraint.

Hence in many cases (speaker dependent) there are benefits to be gained from moving the L/C/R onto external power amps, and value to be gained, by NOT doing the same for the surrounds and heights (assuming that they are an easier load, which is usually the case).

None of this relates to the reality that the surround and height channels are, on average, the most lightly used, on most material - that merely compounds the issue!

 

[FrantzM]

It is just how some signals with the "right" kind of spectral power distribution works out. I'll use square wave as a second example. When you expand a square wave into a series of sinusoids, it is (Wikipedia reference):

View attachment 410702

Here is how the series computes to when the first sine term is removed (removing the k=1 term is equivalent to applying a brick-wall low pass filter at f ≈ 1.5). The series solution shown is computed using the terms k from 2 to 4000. You can see from the plot that the low passed square wave have spikes higher than the original square wave amplitude of 1.
View attachment 410703

[Edit] The filtered square wave does have lower average power, but higher peaks! The relationship between peak power and average power is not direct.

Peace.