MUSINGS: How much amplifier power do you really "need"? Why not test for yourself?

Much has been discussed about amplifier power over the years; here for a short primer, here for a contemporary take with advice from Crown, and here for an oldskool audiophile take on the matter when dinosaurs and tube amps ruled the land (notice the positive sentiment towards even low power transistor amps compared to the low-wattage tube jobs in this article!).

We often hear the question: "How much power do I need?" A very reasonable query of course when we're thinking of buying a new amplifier.

The answer to that question of course really depends on your needs. The bottom line is that it really is about your context and preferences, the quantity and quality of the amplifier you need can vary based on all kinds of factors. When it comes to quantity, it will of course depend on the efficiency of the speakers you listen to along with size of your room, distance to the listening position, placement of the speakers, and absorptive/reflective surfaces in the space.

For those who have not seen it, I heartily recommend popping over to DIYaudio and check out the thread "A Test. How much Voltage (power) do your speakers need?" What an awesome, straight forward test provided by Pano in that thread!

Basically, here's the procedure:

1. Download one of the test tones which consists of a sine wave at -12dB RMS. I usually just use the 220Hz tone (a good choice if you have a subwoofer and potential high pass in place) but the 120Hz tone is good if you have full-range speakers. Make sure that you can play back that test tone with whatever computer/streamer/DAC you use to listen to music. I have the track on my music server ready for playback through my Raspberry Pi 3 "Touch" device from Logitech Media Server.

2. Play some dynamic music that's not recorded loud nor unnaturally compressed. This allows you to set the amplitude of your preamp+amp(s) to reflect approximately the loudest volume setting you'd use when listening to more "naturally" recorded music. Maybe 2 or 3 dB above a louder "reference" level you would use which will satisfy the volume demands of essentially all of the recordings in my music library. Remember to turn it up but don't go deaf!

For me, I'll use a classical album like Eiji Oue & Minnesota Orchestra's Stravinsky: Firebird, Song of the Nightingale, Rite of Spring (Reference Recordings, 1996, DR19, average RMS amplitude around -25dB). Maybe John McLaughlin, Al Di Meola, and Paco De Lucia's Passion, Grace & Fire (1983, DR15, about -20dB average RMS amplitude). For pop/rock lovers, how about good 'ol Dire Straits' Brothers In Arms (1985 first pressing, DR16, average RMS level around -24dB).

Remember that when we listen to music, typically only a handful of watts are needed. It is in the dynamic transients that demand brief bursts of power from our amplifiers and it is during these times that you need the extra horsepower so as to avoid transient clipping for high-fidelity playback.

Leave your volume control at the louder reference level after listening to your dynamic album of choice (with maybe 2 more dB's for "good measure").

Here's a view of my room the other day for context:

3. Now take your voltmeter, set it to AC scale and apply the probes to the speaker terminals. Typically for most dynamic speakers, we don't expect a huge voltage across the terminals, use something like a 20-50V range on the meter. Don't use too large a scale or you might run into issues with low-voltage accuracy.

4. Play the test tone (220Hz or 120Hz) over the speakers now. Depending on how loud you set the volume level at, this can be rather unpleasant - use ear plugs if needed. And while it's playing, measure the voltage across the speaker terminals. Take care not to short out the terminals!

So, what voltage do you get?

8.35V across the speaker terminals; note that I'm running a bi-wire Canare 4S11 speaker cable (not that I feel this makes an audible difference) and using the surprisingly accurate, versatile, and inexpensive Tacklife DM01M multimeter.

5. With that voltage reading, estimate the power you need by just squaring that number... For me it's 8.35 x 8.35 = ~70 Watts RMS max into an 8-ohm load.

If you're wondering why simply squaring the voltage will do the trick of estimating peak RMS power needed, it's just Ohm's Law...

V = IR, P = IV
I = V/R, so P = (V/R)V = V^2/R

Since the maximum RMS power for a sine wave is -3dB, and the test tone is -12dB, that means we need to gain +9dB to get the maximum amplitude. This represents a voltage ratio gain of 2.818383.

Therefore, to estimate peak power:
P = [(measured V at -12dB) x 2.818383]^2 / R
or 
P = (measured V at -12dB)^2 x 7.94328 / 8-ohms impedance

In other words, estimated peak RMS wattage across an 8-ohm load = (measured V at -12dB)^2.

So, what I can say from this measurement is that in my 15' x 20' x 8' listening room, with the couch, sound panels, speaker layout, and the Paradigm Signature S8v3 speakers rated as 92dB/2.83V/m sensitivity as pictured above where my sweet spot is about 10' from the speakers, I'd want approximately 70W of RMS amplifier power rated at 8-ohms per channel to achieve the upper listening levels I would likely ever use.

This is of course based on listening to albums with minimum dynamic range compression, obviously also recorded at lower average amplitude. This estimate would easily cover loud, highly compressed pop/rock music with minimum dynamic range. For example, Mick Jagger's recent Gotta Get A Grip/England Lost (2017, DR5, average RMS volume of only -7dB) sounds extremely loud even with the volume control set at the equivalent of 3W RMS peak. Remember, the loss of dynamic range results in the "wimpy loud sound" Mitch Barnett wrote about a few months back. It doesn't take much to make something sound loud on average - but "loud" without much dynamic fluctuation isn't particularly exciting for hi-fi lovers - the tragedy for so many poorly mastered music out there!

Estimating that I would "need" 70W, considering that I am powering my main speakers with 250W (into 8-ohms) Emotiva XPA-1L monoblock amps, I obviously have quite a bit of overhead power to draw from if ever needed.

Remember that the calculations are idealized. Speaker impedance is much more nuanced than a straight 8 or 4-ohm load. If you run a DSP or are EQ'ing, average signal amplitude will typically be attenuated when done right to prevent clipping and you might need amps with a higher overhead (in fact, I set the volume of my system using room correction DSP and turned it off when measuring). At least this test will get us in the general ballpark for an idea of our power requirements.

If we look at the poll result on the thread as of late August 2017, >5 years since inception:

We see that of 588 samples, 70% of those who tried this were happy with <25W (into 8-ohms) in their systems, and 84% would be happy with <100Wpc amplifiers. Those are certainly not unexpected and very reasonable results for home users.

Remember that we're only discussing quantity of power here and not quality. For "quality", well, one would have to consider distortion, signal-to-noise ratio, ability to sustain continuous demands, efficiency, etc. Hard to talk about this unless we focus on specific devices and look at measurements of course. One implication for amplifier measurements is that even with behemoth amplifiers, one must focus on the "quality" of those first few watts and up to 25W which is where most people will be listening with using typical dynamic speakers. I think this reasonably captures the "First Watt" philosophy. Of course folks with speakers using other technologies like ribbon and electrostatic panels may need more current from their amplifiers just as on the other hand there are those with >100dB/W speakers who can get by with just a handful of good quality watts.

Kudos to Pano over at DIYaudio for the test! Certainly a fun, easy and informative exercise for audio lovers...

---------------------------

I saw this at the local shopping center the other day:

"Timeless" vinyl LP as the prop at the local Swarovski store. Laugh now, but I would not be surprised if in 20 years, the currently much maligned CD (especially by analogue-format-lovers) will serve the same purpose as a nostalgic reminder of a time when everyone spun shiny disks. From CD Digital Audio to SACD, CD-V to DVD to Blu-Ray... Remember that the 4.7" (120mm) disk is the most successful physical A/V medium on Earth and no doubt it has had an impact on the human psyche in a way that will be "mined" for nostalgic purposes in the decades to come by marketers.

Sure, while there has been a level of interest in the LP with some younger folks in the last few years, I doubt the emotional bond will remain strong beyond the effect of the Baby Boomer consumer - this easily will be on the decline in the next 5-10 years when it comes to consumer electronics. Note that this has nothing to do with my opinion that vinyl isn't truly "high-fidelity" by today's resolution standards. Rather, it's just an observation about the fact that technology moves on through the generations, fashions come and go, and ultimately it is the human psyche that decides to dig out anachronistic trends that may flourish for a time...

Speaking of flourishing only for a certain amount of time, it's September and the trees are anticipating the fall. This means that there's only one week left to submit your results for the MQA Core Decoding vs. Standard Hi-Res Audio Internet Blind Test. I'll close off the submission page on September 8th.

Remember that MQA has been reluctant over the years to even run A/B comparisons in audio shows. IMO, the likelihood of ever seeing an attempt at studying MQA's effect will be few... I would not be surprised if this blind test will be the largest set of independent results you'll ever see about MQA's sound. If you have the requisite gear, make your entry as "subjects" and send me your results and express your listening experience in the data set.

Hope you're all enjoying the music! Back to school time here in North America...