How much amplifier power do you really need?

[David A]

@Stefan - Thanks for posting the measurements, good stuff.  As I mentioned in another thread, I often consider myself to be a frustrated objectivist.  That is, I rely on subjective observations simply because there is not enough objective data around that corelates with what I am hearing, in many areas there is no decent data at all.  

As a general point, it is always good to see some measurements.  It might be useful if between us we collect all the available Devialet test data relating to power.  It would be good to have it together in one thread, so if anyone knows of any power related measurements, please post!

@David A - Lots of detailed information as always.  With respect to your post #66, please note that I was not complaining about the "how much power does a Devialet produce" posts, quite the opposite, I fully agree that these are related topics.  There is some interesting reading there on many related topics.  My take from it is that what you have stated is more or less in line with my estimate that the SPL in the video was maybe 90 to 95dB(a), and certainly less than 100dB(a).  You make no view on this, I am guessing you position is that it is impossible to say with any certainty?  (you strike me as a cautious man who prefers not to speculate)  OK, there is some speculation here, but from about the 6:00 min mark in the video you can just about make out the voices, not what they are saying, when the camera and microphone are about three to four metres away from those talking.  After 7:00 min when the camera closes up on one attendees smart phone to show the track name, you can clearly hear his voice, it does not sound like he is shouting, and the camera is maybe half a metre away.  So based on your comments, would you be happy to estimate SPL's below 100dB(a)?  I would be interested in you view here.

Regarding your post #67, you make this statement:

Let's say you had 85 dB sensitive speakers and a 4 metre listening distance. We still need to get 101 dB at 1 metre for 92 dB at 4 metres so that's an extra 16 ddB. You still need less than 100 W. 100 W would deliver a 20 dB increase in level so a pair of 85 dB sensitive speakers would deliver a 95 dB level at 4 metres with 100 W.

I fully agree with this in terms of your calculations.  If my earlier assumptions re SPL are about right (if), then what you describe above is more or less in line with what we find in the video, in terms of the SPL's, the sensitivity of the speakers and the listening distance.  Yet, the amps are not showing 100 w, nothing close, they are showing 700 w.  OK, there are assumptions here and a fair margin or error, but the end result is a factor of 7 different.  I think there are some "real world" complexities here that the basic calculations do not fully address.

As am open question to all.  I have been thinking a bit about audio reproduction and the dB "A weighted" scale.  I have attached a dB weighting graph here for easy reference.  What I am thinking about is the fact that human hearing is far less sensitive to sound at lower frequencies, so with equal energy, a sound at 1000Hz will sound to a human a lot louder than a sound at 100Hz.  This is well known.  To me, the obvious consequence of this is that to produce a sound at very low frequency that sounds reasonably loud to a human will take considerably more power than to produce something equally loud at around mid range frequencies.

This is relevant to the Pan Sonic track.  It has a lot of low bass content, right down to frequencies more felt that heard, so not just below 100Hz, but dipping below 20Hz.  (maybe someone has software that can analyse this aspect the track?)  Looking at the a weighted scale, these low frequencies sound -30dB down to human hearing at say 40Hz, even more as you drop to 30 or 20Hz.  -30dB is a lot!  So if you take the power needed to reproduce 1000Hz at 90dB(a), then you would need fully 1000 times the amplifier power for an equally loud sounding sound at 40Hz.  I am thinking aloud a little bit here, but could this be a factor in what we are seeing in the video?

I haven't watched the video. I'll try to do so and comment.

Re the dBA scale: Don't use it!!! The dBA weighting scale was designed specifically for assessing the risk of hearing damage. The bass rolloff is there because our ears become less sensitive at lower frequencies and noise at those frequencies provides less risk of hearing damage than the higher frequencies where the curve is flat. If you want to know whether you're likely to get hearing damage at your listening levels use dBA weighting with an option that averages the levels being measured and calculates a noise dose based on 8 hours continuous exposure to those levels. That's the way dBA weighting is intended to be used.

If you want to know about how much power your amplifier is putting out then ideally you'd use a Flat measurement, ie no weighting curve at all, because low frequencies are important in terms of their impact on power needs so you don't want to downplay their contribution by using a weighting scale that rolls the bass off. A lot of the meters that get sold for use by non-professionals only offer 2 options, dBA or dBC weighting scales. If they are the only options you have, use dBC. It still has a roll off in the bass but it's a smaller roll off than the dBA weighting uses so it gives a more accurate result for the purposes we've been discussing here but it's still neither accurate nor the best option for our sort of purpose.