Originally Posted by russ_777
Power Needs for Salon2's
I've been driving my Salon2's since I got them ~15 months ago with 2 channels of amplification from a Denon AVR-4311CI. I realize I'm in the minority driving 22k speakers with a $2k multichannel, mass market AVR. But I've rationalized it on the following grounds: (1) it's powerful as AVRs go and I'm only using 2 of the 9 PA modules -- predecessor 4310 measured 163/268 wrms @0.1% THD into 8/4 ohms with peak outputs of 192w/310w and the 4311 is rated 10 wpc higher; (2) I have the Salons crossed over to a 15" sub at 60 Hz which pains me to do BTW; (3) my MSP is only ~ 10' from the Salons; (4) I rarely listen at very high SPLs; and (5) the AVR has never gone into thermal fault. The final of these 5 factors had concerned me the most since the Salons do have some significant current demands at some frequencies and are fairly inefficient.
All of that led me to the thought of trying to measure the actual power being delivered to the speakers under actual listening conditions. I'd like to know anyone's thoughts on the methodology I'm using (actually, making it up as I go to a degree). Keep in mind I'm not interested in sub-atomic precision, just some ROM estimates that might shed some light. In a nutshell, I'm measuring power drawn from the wall outlet on the AVR AC power lead using a Kill-a-Watt device. Obviously there are some adjustments that need to be made to that number to get a decent estimate of what's actually going to one Speaker or the other. I assumed that if I measured the average power draw from the AVR when energized but not processing any audio or video and subtracted that number from the power draw I measured as music was playing then I'd get a decent estimate of the power input to the power supply when music is playing. That number will vary with SPL and thus time, and likely some configuration settings on the AVR. I further assumed the average efficiency of the PS/PAs to be 50%. The final assumption I made was that of the power that makes it from the wall to the outputs of the two channels on average it will be split equally between the two channels. Worst case from the standpoint of sizing the amplifier would be the scenario where 100% of the wall current at a given moment goes to one channel, but that doesn't seem very realistic to me in a 2 channel system.
Now for some numbers. I suppose I could have gotten a mic and some software to measure the SPLs but that wasn't important to me since I know how loud the loudest I listen is. That said, I measured the quiescent power when the AVR had no audio or video inputs to be around 95 watts. That seemed high but that's what it was and it didn't vary more than a watt or two. I then played several types of music all at average volume levels that were as loud as I would be able to listen to for more than a few minutes. As the music played, I watched the LCD display on the Kill-a-Watt display the power readings which it updates at one second intervals. The results were highly dependent on the source material as you would expect. Highly compressed tracks kept the power levels ranging from the 100-teens up into the 130s. Again, these were average levels, not the highest peaks. When I listened to some classical music (Rutter Requiem, Reference Recordings) the average level dropped way down, but when they hit peaks they were generally higher than the pop-oriented compressed crap.
The highest reading I saw was 210w. So let's just assume that corresponded with close to if not the highest peak SPL I'd likely encounter in normal listening, and let's further assume that a typical average power draw would be 115w. In each case following the assumptions I made, we would subtract 95 watts off the top, multiply that by 0.5 to account for efficiency of the PS/PA then divide that by two assuming the load was equally divided between the 2 channels. The resulting numbers would be 5 watts average per channel and 28.75 watts peak/channel.
Frankly, I don't believe those numbers, especially the peak number as it's well less than 10 dB above the average number and seems very low in an absolute sense. Some of that is likely attributable to my missing some of the highest peak values though I would think the LCD readout would be running off a peak hold circuit within the 1 second sampling interval which should capture the peaks.
Any ideas on where I may have made significant errors in this analysis?
o Home Loudspeaker Efficiency
Home loudspeakers are typically less than 1% efficient. Speaker efficiency really isn't a useful way to analyze home speakers.
o Peak power usage
of 5 watts is useful number for further analysis. The peak values you developed are likely rather averages over short periods of time.
The crest value of music or movies is often given as 10dB to 20dB. Using the 5 watt average value this would make the peak values somewhere between 50 watts and 500 watts. It is these peak values that cause huge demands on power amplifiers. Having sufficient power to meet these peak demands is required if amplifier clipping is to be avoided.
Of course these calculated peak values are very sensitive to the 5 watt average. For example, if the actual number is 4 watts then the peaks are 40 watts to 400 watts.
o Looking at this another way:
I keep a sound level meter in my media room. I occasionally take readings, in a attempt to avoid going completely deaf. I sometimes use C weighting, which seems relevant here, and the fast reading setting. The meter reading may or not be the real peak, I don't know. There are perhaps 5 - 10 readings per second.
The readings easily vary between 45dB SPL and 95dB SPL, likely more. In power terms this is a variation of 50dB or a factor of 100,000, if I've gotten the decimal in the correct place. Based on the variation in these SPL readings, I would expect the peak power readings, if they were somehow taken for my equipment, to range from 0 watts over the base reading to 200 to 500 watts over the base. This is if the average power is assumed to be 5 watts as measured above.
o Have others here performed more rigorous experiments in this area?