Amplifier headroom matters
Join Our Community Subscribe to Paul's PostsFact or fiction: an amplifier with 10 times more power than the speaker needs sounds better than one more closely matched?
The short answer is often yes, but it’s worth looking into the why of this statement.
First, let’s get some facts sorted out. Most listeners have no clue how much power their speakers are actually consuming. And this is true even for those that on occasion have had their amplifier shut down. Relying on the speaker manufacturer’s estimate of power handling is pretty much useless information. The fact that a speaker can handle 300 watts doesn’t mean it’s ever going to see that much power.
In fact, most amplifier power spent driving speakers under musical conditions is typically below 50 watts, with an average of around half that. Where peak power gets big is on low frequencies and large transients. Which is why equipment manufacturers once attempted to rate amplifiers with Musical Power rather than their RMS power. (This happened at a time when we were obsessed with big watts).
When music demands power the last thing you want is to run out, or even get close to the system’s limitations. On those peaks and crescendos, we hope for unfettered performance whose sonic qualities match those of softer passages. Unfortunately, this is a rare achievement because of too little headroom.
Headroom is the secret sauce for reproducing effortless music. Headroom maintains an amplifier (or speaker’s) linear performance region—an area we hope to keep sacred. Once we exceed an instrument’s linear region, its sweet spot, sound changes and not for the better. The closer we get to a device’s limits the less free and open the music will sound.
Here’s a good way to calculate this. Most amplifiers and speakers are comfortable at about 20% of their rated output. Exceed that and you venture into areas of strain, struggle, and compression.
Headroom matters when music matters.
and…
I am getting closer to actually publishing my memoir—two years in the making—which has now turned out to be nearly 500 pages! It’s called 99% True and is chock full of adventures, debauchery, struggles, heartwarming stories, triumphs and failures, great belly laughs, and a peek inside the high-end audio industry you’ve never known before.
When it launches I plan a few surprises for early adopters, so go here to add your name to the list of interested readers. There’s an entire gallery of never before seen photos of people you know but haven’t seen like this.
That’s a simple one – here’s a demo where bass and transient heavy music consistently demanding over 500w.
https://www.youtube.com/watch?v=bRMR9JZ1m0s
(Thanks to Soundmind – this was near his link yesterday).
The guys in the video are still happily talking to each other during the demo, showing that you need plenty of headroom to hear this type of music properly at any loudness level.
Thanks to its SMPS, my particular unit has absurdly high peak power, which is a good reason for using the speaker-profiled DSP, as it limits the output to the maximum physical capability of the speaker. I must admit it takes a bit of time getting used to the speed of the sound.
I found that video astounding. My speakers are 89db/1 volt/1 metre. My listening chair is 3 metres from the speakers and the absolute maximum average volume I listen at is 80db (more like 70 at normal listening levels). The calculations indicate that I need about 0.5Wpc to generate 80db at the chair in anechoic conditions. Since I listen in a real room with reflections the actual measured power for that volume level is more like 0.3Wpc. The hundreds of watts shown in the video were astonishing. For the material I listen to the instantaneous peak SPL is around 13db above the average. Allowing 20db to be on the safe side I only need 30Wpc from the amp. This is slightly above the sweet region of my 80Wpc Class G amp (which gives a little over 20Wpc as pure class A). However 80db is ‘party’ volume; at the normal 70db I only need 3Wpc, and this is comfortably in the class A region.
The big question seems to be how much headroom you require. I suppose if one of the second violins flipped out and shot the conductor I might not get the full dynamics of the gunshot. I can live with this. In fact, I can live with this very comfortably indeed.
The “music” in that video is way less intense than normal music and consists mainly of percussive sounds. When just producing a few clicks and pops, it’s easy to use up 500 watts and still be able to talk over it. I’d say that piece was probably deliberately chosen to produce the results you saw.
More traditional listening goes as you’ve described …
It was selected for that purpose. They said so at the start of the video. They also said that peak power is not relevant for “softer” musick like a string quartet.
However, I listen to music by the likes of Kuniko, Philip Glass and Steve Reich that requires that level of peak power on a regular basis.
Even more extreme is Thom Willems (In the Middle Somewhat Elevated), mentioned a few weeks ago. When I first heard it “live” (it’s electronic) at Covent Garden in 1989 I almost had a heart attack. I’ve heard it several times since and these days they play it at a lower volume. Health & Safety.
My listening is far more mundane. During the day my setup is pretty much a digital radio as I put stuff on to keep me entertained as I work. Serious listening sessions tend toward Jazz or light Classical. Heck I’ve even been known to listen to “oldies” rock and roll on occasion, none of which is particularly demanding in terms of high peak values.
But I do get your point, some music is like that. But then you hit the limits of your sources. How much overhead does your CD player or Turntable provide you? How much is actually on the pressing itself?
Very often that’s only 10db or so and that’s only a 10 to 1 power ratio… so unless you are listening at 50 watts average (which would be deafeningly loud most of the time) you would have a hard time justifying a 500 watt amp.
Certainly I could never justify it given my usual sources….
I don’t have a CD player (sold it in 2010) and mostly stream at 24/96. The music I mentioned is mostly purchased downloads from Linn at 24/192.
I do listen to vinyl, usually when the same music is not available in a quality digital source, but sometimes just because I like vinyl.
As I’ve mentioned before, my system has absurdly high peak power that it limited by DSP to protect the speakers. (The system is programmed for whatever speakers are in use.) It makes my sub seem a bit slow and I’m thinking of upgrading that to a more powerful unit to sharpen it up. It shows up in jazz.
Interesting how different our experiences are.
As I mentioned I almost never get past 20 watts per channel and even that is sometimes too loud … but then, I do have neighbours to consider.
Most of my stuff is 24/96 or 24/48 which seems just fine to me.
It’s not about loudness. I had a 22w SET that could deafen me, but it was loose and flabby low down and I had to get rid of it. For that you need lots of power, even at 80db.
For that reason alone Class D powered bass is the most cost-effective solution, whether in hybrid speakers or a separate sub. It’s cheap, efficient and effective.
The power in a sub relies upon having a driver which can move a lot of air very quickly. I seem to remember that you were using a 10″ REL. Going up to a 12″ or larger speaker would probably be beneficial.
I am thinking of going from a 250w to a 550w REL from the same range. I shall go and have a listen. I think they are all 12″ drivers in the same box, but the newer ones with more power have stronger internal bracing.
Steven, I know how much you advocate for your all-in-one, and I certainly like mine as well. My Yamaha R-N803 just got the EISA (Expert Imaging and Sound Association) award for “best stereo receiver 2018-2019” which I certainly would second!
This does make me wonder, however, about how many of these they actually sell (or any manufacturer actually) regardless of awards. My network receiver has been out now a year, and still I am the only review of the product on Yamaha’s site, although there are a few more on Amazon. It would be interesting to see sales figures for various types of products but I know most manufacturers would never want to divulge this info.
Only a very small proportion of audio buyers go on forums or leave feedback. I have no idea what sales are like.
Was discussing all-in-one’s elsewhere and I noted that many of them can be used with external streamers, pre-amps and amplifiers. For example, the Hegel units have both fixed and variable pre-outputs. So they are designed so that they CAN be used alone, but have the flexibility to be used with any number of external sources and power devices. Devialet is unusual in that it can be bridged for dual mono. It’s an expensive option, but cheaper than selling and buying the next model up.
“does not apply to soft music like string quartets” NOT TRUE. Or, only true for the lame recordings of string quartets in symphonic halls from 10 or more meters out. I refuse to see quartets in 1000+ seats. Fortunately, we have some excellent chamber halls in New York with 250, 400 and 600 seats, and we get seats in the first seven rows.
I recorded many string quartets in what amounts to a personal library, a 1,000ft2 room with 1500 books and 2,000 CDs shelved on the walls, with some secondary acoustic treatments. From 2 meters away (front row), the peaks are >110db!
This was measured on the ultra-fast peak reading meters on my DSD HDR.
Probably a bad example. Is John Denver still alive? He’d suit a valve amplifier.
Going to hear the Takacs Quartet next week. Seat C8. Wigmore Hall. Should be fun.
The active speakers from Devialet are said to deliver up to some kW power down to 14 Hz. Thus switching to class D power limitations shouldn’t be an issue anymore!
I heard the new one in Paris last week. It weighed less than 5kg, you could pick it up in one hand, it is rated at 900w, and standing 11m away (the other side of the space) it still gave a really big sound and enjoyable sound.
https://flic.kr/p/2aW5cr3
I think it’s half way between mid-fi and a PA system. There is merit in a unit that you can use to listen to for personal pleasure at modest volumes and then use for a headbanging party. It’s a step up from those things JBL make.
I think they will bring one out later that has a slightly better upper-mid and treble, but the key thing it seems to me is that they wanted to be able to market it at under €1,000. The 600w unit is €999 and the 900w unit €1,249.
Paul is talking about lots of power for unlimited hifi performance, with which I agree entirely. These little Devialet things are all about maximum sound pressure from a small enclosure. They are designed to be used at maximum output, rather than provide headroom, and the remarkable thing is that at 95% volume level I could not hear any distortion. I would happily use one to replace my Sonos Soundbar. It has many uses, hifi is probably not yet one of them. The larger ones are proper hifi, but are something like $5,000 for a pair (it is a fully active hifi system as well).
Wattage ratings of powered speakers are not subject to “truth in marketing” testing. Some subwoofers have been proven off by a factor of ten.
Yes, headroom does matter.
From a service standpoint the biggest problem is clipping. Clipping is the result of asking an amplifier for more output than it can deliver. It literally lops the tops and bottoms off a waveform producing something akin to square waves that have very high harmonic content… at maximum power. This can and will overheat and burn out tweeters and, rarely, even mid-range drivers.
My electronics teacher, way back in the late 1960s, once told me that “If you’re blowing tweeters, you need a bigger amplifier” and given the number of blown tweeters I’ve replaced over the years I am forced to agree. An under-powered amplifier is more dangerous than an one with extra reserve.
Paul is correct at pointing out that most listening averages less than 50 watts. Consider that a speaker rated for 80db/w does that with 1 watt of power. 80db is not quiet, it’s about the sound level of a blender or a vacuum cleaner. Unless you have no neighbours within ear shot, you’re not likely to be listening at levels much beyond that and given the dynamics of most music it’s likely you won’t be hitting more than 10 or 15 watts on the “spikey” parts. Less efficient speakers require more power, of course.
All that said, I’m not so sure that we really need the 1200 watt behemoth amplifiers so many of us seem to favour. When I still had the old Marantz amp, the power meters rarely got over 20 watts per channel… even on my loudest days.
I oftentimes used to remark about my amplifier that “it needs more horsepower” if I felt it lacking (however as I get older become more tolerant of that, choosing to spend brain power on other pursuits). I am a firm advocate of headroom in an amp but seldom tax my current equipment in any way. To use the car analogy, my current car is supposed to do 0-60 in about 5.5 seconds. Do I ever test that, no, but when pulling out into traffic it’s nice to know if I get on it the car would respond. Plenty of headroom in that case. Now, where the car most lacks is in 50-70 acceleration which constantly needs passing gear to achieve satisfactory levels. For you amp aficionados what is the amplifier equivalent of that (handling all the midrange or bass)? Just curious on that part.
Someone please talk about watts versus current.
Power = current squared multiplied by resistance. So one amp through 8 ohms will generate 8 watts, and two amps will generate 32 watts. Since with music we are talking about varying AC signals power, and current, will often be expressed as averages rather than instantaneous values. RMS (root mean square) is commonly used.
In a practical example….when discussing Magnepans, the conversation usually goes something like “Maggies needs lots of power” (assuming they mean watts)…and someone will say ” No…Maggies need lots of CURRENT”. So…what gives? Is current related to what’s needed for sudden loud passages?
I would guess, and this is only a guess, that Magnepans have an impedance which varies widely with frequency. This would mean that at some frequencies where the impedance is low you need a lot more current for a given power than you would expect from the nominal impedance of the speaker. An amp which could not supply this current would be unsatisfactory. Hopefully a Magnepan expert will respond and enlighten us both.
A number of us had a long discussion about Watts many weeks ago. I preferred Overend, many others Charlie. If you are in to beatbox there is Reggie, or the blues there is Ernie! Sorry, off topic ha ha.
Always delighted to be persuaded off-topic. Do not forget James Watt, even if he rather boringly gave his name to the unit.
I did a Google search for “watts” “musician” and was surprised there were a couple dozen results dating back a few hundred years. Little did I know. Methinks a Tidal search of them might be in order just to expand my musical horizons a bit. Secretguy might not be amused that we have veered 🙂
Current doesn’t happen without voltage.
To give the plumbing analogy, voltage == pressure, current == flow, power == stream impact.
Impedance would be a smaller pipe that resists flow and dissipates pressure.
When you apply pressure (voltage) water begins to flow (current) according to how much pressure you apply… at the far end of the pipe the impact (power) can do some work for you.
Current is the result of applying voltage. It cannot be manipulated independently.
So we have the two basic laws of electricity …
Ohm’s Law gives us voltage (V), current (I) and impedance (R) calculation … V = I x R
Watt’s Law gives us power (P) from voltage and current … P = V x I
A more detailed explanation is here …
https://www.electronics-tutorials.ws/dccircuits/dcp_2.html
For a fixed impedance, like in a loud speaker there is a linear relationship between voltage and current. Increasing the voltage increases the current. Decreasing the voltage decreases the current.
Power being a function of voltage and current tends to be an exponential increase where doubling the voltage will also double the current and quadruple the wattage.
So taking formulas from the link above, for a 100 watt amp and an 8 ohm speaker…
V = sqroot(P *R) == sqroot(100 x 8) == sqroot(800) == 28.28 volts.
I = V / R = 28.28 / 8 = 3.53 amps.
To check … P = V * I = 28.28 * 3.53 = 99.82 = 100 allowing for rounding errors.
So calculating current from power, while not exactly head math is not all that hard to do.
Keep in mind that music or sound in general is not a steady state. It is a wild dance of changing voltages and currents so the only accurate way to measure amplifier power is on a steady test tone.
Does that help you?
What might be helpful is that current is measured in amperage. In the US you use mostly 120v, all wall outlets for lamps, etc is 120 volts, protected at the circuit breaker box by 15 amp circuit breakers. Electric dryers and ranges use what is commonly referred to as 220v, a hold over from the days of outlets being 110v. Your standard house has two 120v hot lines with what is known as a common between them, and the box has a ground wire, which is a metal pole driven into the actual ground.
Using the plumbing analogy, current is flow, and flow in electricity is measured in amps, short for amperage.
In the forum there was a discussion, someone wanted to put in dedicated circuits for their audio system. That would require a new circuit breaker and a new outlet, with either Romex or conduit with wires. The gauge of the wire and the outlet will determine the size of the circuit breaker. For lamps, TVs, etc you would use a 15amp circuit breaker, and might have more than one outlet on one circuit breaker. For an audio system with big amplifiers, and power regenerators, you would want 20 amp circuit breakers and 20 amp outlets. Between them you would want 10 gauge wire, although I believe 12 gauge would be acceptable both for flow and building code. Wire gauge, for reasons I never referenced seem backwards, a 10 gauge wire is larger than a 20 gauge wire. A larger wire will allow more flow, measured in amperage.
When someone says you need a power amplifier with lots of current, they mean that it needs to be able to supply more amperage when needed. Amplifiers that can do that often require 20 amp circuits. If you look at the back of the amp, the socket for the power cord will have a different layout of the blades, as will the plug, and the wall outlet.
And just for fun, when you double the voltage, the amperage needed is half as much. That has to do with Ohms law.
FWIW … I was describing the current on the Speaker Leads.
Once upon a time when I was an audiophile I believed exactly the way Paul does. Then one day long after I became an engineer I gave it some more thought. Amplifier output power is a very complex subject, a few simple numbers are not sufficient to explain it let alone how it interacts with loudspeakers and cables. There are many variables. One thing you should know is that an increase of 10 times as much power, that is 10 dbw is subjectively only twice as loud according to audiologists. My experience also has led me to conclude that when a lot of sound comes from only one or two directions at a perceived short distance away it creates a very unpleasant blaring blasting sound that is not at all like music. Measuring power in complex reactive loads is not as easy as putting a VU meter on the output and calibrating it for watts into an 8 ohm resistive load. You can disconnect your speakers so that no output power is being produced and see those meters swing wildly back an forth so much you’d think the amplifier was being driven as hard as it could be. The subject is too complex to discuss here but I will say that high fidelity amplifiers should never be driven to clipping and even one watt available more than you ever need is a waste. Nevertheless it is instinctive among electrical engineers and most other engineers to build in a safety factor so that whatever they design or build is never pushed close to its limits let alone beyond them. This is not always easy as it sounds. I’ve spent much time analyzing and exchanging ideas with structural engineers about why the FIA walkway bridge failed catastrophically. These engineering failure modes interest me, especially in electrical power distribution systems of various types. In complex systems especially with multiple interactive loops, steady state analysis is not nearly sufficient to determine whether or not a safety margin exists let alone how great it is.
All I can add is that I very well pleased with the Sprout100, and only by coincidence does the Sprout mate so well with the Harry Kloss designed Advent Loudspeakers.
I am looking into a pair of ZuAudio Omens if I decide to modernize my speakers.
Sprout100 should be effortless in driving these high-efficiency speakers if what I am reading is correct.
Very very interesting and an eye opener. Just proves how wrong we can be when we form opinions based on scant information. My mental picture of you was correct only fifty percent at best. I wonder how much more I will learn when I read your book. It certainly will be most entertaining. Looking forward to the book. As for the post I was well aware that one must have a lot of power available in reserve to cope with sudden short term demands put on amp which can be many times the average listening level. That amps and speakers do best up to 20% of their capacity is something new I have learnt. I thought that short term peaks could be up to ten times the average level but here I concede to you because I have no measured evidence to support my thinking. Your posts are both controversial and very educational and make one think and have one’s Gods challenged. Keep up the good work. Regards.
“Fact or fiction: an amplifier with 10 times more power than the speaker needs sounds better than one more closely matched?”
Not always … I think the idea of 10x overkill is more of a sales tactic than a genuine benefit.
My current speakers are rated “30 to 100 watts” and I wouldn’t even dream of hooking them up to a Stellar or BHK, both of which have enough power to destroy them if things get out of hand. The best candidate for them is the Sprout.
It’s only been about 15 years since we were happily driving 12 and 15 inch woofers on 30 watts per channel …
“It’s only been about 15 years since we were happily driving 12 and 15 inch woofers on 30 watts per channel”
If you think it was only 15 years, you must of been in a coma for about 20 years. I remember powerful amps becoming quite common in the late ’70s. In 1973 I bought a Crown D150 and my friend bought a DC300A.
More people damage speakers with under powered amps, than with high powered amps.
I suppose an idiot with a high powered amp, could try turning it up with the speakers outdoors, or until his ears were bleeding, but that is rare.
Most damage happens when clipping underpowered solid state amps.
Depending on your speakers, what brand and model are they?
I would think a Stellar S300 would be ideal.
My new friend… Please enhance your calm and read my other comments.
Sorry, I missed your post on clipping. My speakers current model have a recommended amp power of 30-750 watts.
I use an amp that is rated at 400 rms at 4 ohms. If I could afford to I would buy a pair of monoblocks rated at 1000 watts at 4 ohms.
I prefer having more power, so I never have to push my amps anywhere near their limits.
No worries … I understand the logic and I agree with it. If your speakers can tolerate that kind of power, no harm no foul. Heck… go for it!
But in my case there are other factors to be considered. First I live in an apartment with 8 neighbours just inches from my outside walls so structural sound penetration is a major concern for me. This same living arrangement limits how much space I can give to my audio gear so size becomes a consideration… and so on.
Within my constraints I’ve been looking to build the best system possible, just like everyone else here does. What I came up with would likely be laughed at by many but I do have to tell you that the sound quality is breathtaking… I went the “budget audio” route and got a 50w/channel “Chip Amp” that never gets over 1/3 volume so I’ve got tons of headroom. I went with a pair of largish bookshelf speakers that can easily handle the amp and did a few mods for better sound; tweaked the crossover, acoustic treatments in the boxes and ports, better wiring, soldered connections etc. I run the whole thing from a custom built HTPC and I would put my current system up against most audiophile rigs on all but raw horsepower.
Isn’t that the audiophile thing? Creative solutions to problems and pushing the boundaries of technology. (Grin)
I think I’ll just get a Bose wave radio. LOL
LOL … or a Sprout with a pair of nice bookshelf speakers.
In my own recent experiments I would have gone Sprout and may end up there yet. The immediate problem is that PS Audio is not easily available in Canada … Something Paul might want to work on.
https://meyersound.com/video/m-noise/
CREST FACTOR is the ratio between signal peak and average levels.
Meyer Sound, on the forefront of audio metrics for decades, did a survey of music files and found that although average level decreased with frequency at roughly the same slope as pink noise, the peak values did not. This means the crest factor for the top octave is typically 18dB. You need 70 times as much peak power as average power, and 1″ dome tweeters have inadequate surface area, volume displacement and speed to reproduce realistic musical transients.
My measurements based on mid distance, unprocessed live recordings of chamber music using orchestral instruments with nearby acoustic boundaries show peak to average ratios over 20dB for full bandwidth signals. These peaks are not present in commercial releases which use compression and limiting during capture, mixing and mastering. Even “audiophile’ recordings with no electronic dynamic reductions often have “acoustic compression”. By this I mean that the microphones are so distant that the high frequencies are rolled off and phase shifted to reduce the peaks by 3-10dB, and they are outside the Schroeder limit where the reverberations are louder than the direct sound, which shaves another 3-6dB off the crest factor by increasing the average level.
In Geffen Hall, for example, the third row is outside the direct sound envelope! Mushy, “warm” 19th Century sound suitable for Wagner, Berlioz and Mahler but neither Baroque, Classical nor late 20th Century composers.
With typical 86dB efficiency of “high end” direct radiator speakers, you will need >500 Watts per channel to reach orchestral levels, so 5x or 10x headroom is a moot question because the average power will melt your voice coils. You really need pro level efficiency, >96dB/Watt; but even then for the peaks of live music, I use 60-100WPC tri-amped, or 360-600 Watts total. I have not tried 300WPC x6 to hear if it sounds better.