What's the deal with POWER?

[Joe Bee]

I could not think of a more descriptive title, but I actually have a serious question about the power ratings of non-instrument amplifiers such as PA or home stereo systems.
In the very old days I didn't know much about the dynamic range of music so I assumed that just like the guitar amps I am a lot more familiar with, any amp could continuously supply its rated power.
Later I learned that most music (other than super compressed top-10 pop music) has transient peaks exceeding the average level by about 20dB or a factor of 100, and I assumed that this would explain the shady PMPO ratings that most consumer grade audio equipment comes with. Wouldn't it make sense for a boom box to supply an average power of maybe 5W and pump out transients at 500W?
But then I came across the divine musings (sorry steve ) of Rod Elliott, where I learned that PMPO is complete bogus and the rated RMS power is actually not some thermally limited average power, but the maximum power that can be obtained in a pinch by connecting the power supply rails to an 8Ohm resistor.

OK, so much for the short intro, now my actual question:

Wouldn't it make a lot of sense to build amplifiers that have a lot of "dynamic headroom", i.e. that can supply a lot more REAL power than the rated average power? I think about it this way: MOSFET Semiconductors are usually thermally limited in peak current, and I don't see why it should be different for bipolar devices, so it would not be a problem to ask a 5W amplifier to supply 500W for a few miliseconds. The power supply would also not be a problem, since even a flimsy wall-wart with 10VA can easily supply 100A or more given a reasonably sized filtering cap.

The only "problem" that I see with this scheme is that the rail voltages would have to be increased by an obscene amount. To pump 500W into an 8Ohm load, a 65V supply rail is needed, and that's a lot more than the 12V rails a common 5W boom-box will have. This voltage will have to be dropped in the active devices under normal listening conditions, so the total dissipation at 5W listening level (800mA into 8 Ohm) will go up from
12V * .8A = 10W (of which 5W are waste heat)
to something like
65V * .8A = 52W (of which 47W are waste heat)

That is a lot of power to be sinked, and I reluctantly agree that the average boom box will not have the capacity for this. But still this means that an amp with a 100VA supply and a heatsinking capacity of a 100W amp could produce transients of 500W, something like a 7dB headroom gain!

Seriously, why would any "500W amplifier" ever need a power supply of more than 100VA if no program material with less than 15dB of dynamic range is reproduced and the amp is not driven into clipping?



Oh, and the real killer, but I am not even getting into that, would be some kind of bi-amping / rail switching / Class G arrangement where the "500W amplifier" is only ever fired up when a transient approaches. This way a 500W transient from a 10VA wall-wart could probably be made reality.

[Steve Conner]

Hi Joe,

Power ratings are subject to a lot of marketing spin, as Joe Public knows that more watts = better but has no idea how to measure it.

If an amp's power rating is given as "RMS", then I understand that to mean that the amp can be connected to a sine wave generator and dummy load resistor, and will supply the rated amount of power without clipping, for at least a few minutes before the thermal cutout trips. I think there are IEC and DIN standards for just how long the "few minutes" should be, in what ambient temperature.

If the power rating isn't "RMS" then it's simply the biggest number that the marketing department had the guts to print. The old "music power" was simply the instantaneous peak power, which worked out at roughly 2-3x what the RMS rating would have been, depending on how saggy the power supply was, and again how brave the marketing guys were

Your points about dynamic ratings are quite valid. You can undersize the transformers in a hi-fi amp grossly. One I saw that surprised me was a Mission integrated amp with +/-55V rails and just a single transformer that looked around 150VA. I also built a stereo amp of my own that did 180W/channel into 4 ohms using one 160VA transformer per channel, and has never caused any trouble in the 8-10 years I've had it, even though it would probably burn out its transformers on a prolonged sine wave test.

However, you can't get away with quite as much in PA amps, that may need to thump out compressed bass all night long in clubs.

[Enzo]

Joe, whay market are you trying to serve? They already make multirail commutating amps, Carver for example has been at it for decades. But that means the amp will have redundant parts, multiple supply rails, the extra switching circuits, etc. All that adds to the cost for diminishing returns.

You could probably stand there and hold a 100 pound object, maybe 150 pounds. COuld you hold a 200 pound object? For a few seconds? OK, but for a "while?" How about a 900 pound piano or something? You probably can't just pick the piano up, but if you yanked up on it, you could probably get it off the floor for a moment long enough for a helper to pull something out from under the leg where it had been pinned.

So you have a 150 pound rating with a 900 pound peak rating. WOuld you make the case it would be better if you could always be able to lift 900 pounds? Really, how often does it come up? Would life be any less liveable if that peak were only 600 pounds?

The amp makers make amps with a given nominal power. They can make peaks a little hotter than the steady flow, but as you note, they aer limited by power voltage. Look at that 500w peak on the 5 w signal. WOuld your ear notice if the peak were only 200w? That is only a couple db. Tube amps compress away peaks all the time, and we love them.

For that matter, we often place compressors and limiters in the signal chain before it even gets to the PA.

I don't agree with your assertionthat power supply would not be a problem. To hold up under that peak flow, the caps would have to be a LOT larger. That adds cost and space needs to the amp. And how much better would that make the audience experience?

It sounds like what you want to do is build a 1000 watt amp and call it a 10 watter.