Merit in measuring @ 1 watt (first watt performance)?

[dshreter]

Is there evidence that measuring at 1W would reveal something that 5W does not? Most importantly, are there situations an amp would measure well at 5W but not at 1W?

 

[fpitas]

Is there evidence that measuring at 1W would reveal something that 5W does not? Most importantly, are there situations an amp would measure well at 5W but not at 1W?

All I can think of is noise and crossover distortion. And the latter is only of concern with class AB.

 

[DonH56]

A couple of comments, did not wade through the whole thread, just a quick skim:

1. There are many things that can affect distortion and noise as signal level varies, including biasing (particularly dynamic biasing, which includes things like class G/H designs), crossover distortion (usually a function of bias and feedback), bandwidth (small-signal bandwidth is generally much greater than large-signal bandwidth, leading to more noise but also more potential for stability issues and such), operating temperature and other factors that can change the gain of the devices (again mitigated by feedback), and so forth. Some things like power supply noise, clock noise, and such may be independent of the signal level and will also cause low-level SINAD to drop. But the THD+N sweeps virtually every reviewer does will indicate if there's a problem or not, and FFTs can help pinpoint any problems noticed.
2. Crest factor is very important, but a single sine-wave tone provides a reasonable measure of clean power output that in most cases carries through to multitone distortion. Predicting multitone distortion from a single-tone sweep has been done reliably for ages. There are a couple of caveats:
   • Distortion varies over frequency, from things like falling loop gain (feedback factor) and slew-limiting as frequency increases, so a single frequency measurement will not in general translate directly and accurately to multiple tones spanning a broad bandwidth; and,
   • The crest factor of the source is not in general known for things like music, so we usually go by estimates (typically 17~20+ dB for music) in determining the peak power needs of our system -- it could be measured, but how many of us sit around with o'scopes and impedance meters plumbed to our speakers? (OK, this is ASR, so maybe there are a few, at least now and then... ) That allows us to estimate average and peak power needs and relate them to single-tone measurements.

FWIWFM - Don

 

[mhardy6647]

So profligate. I'm usually around 20mW. Go ahead, burn down the planet!!!

Yeah, I'm a wild man, what can I say‽ Dude, crank it up!

Plus, you gave me an excuse to use an interrobang. ‽

 

[LiquidSound]

A bunch of opinions, albeit not uninformed ones.

1) I think that the first watt is important in real life even for modern, average reproduction chains (i.e., the glut of modern, low-sensitivity loudspeakers).
2) I think the trick is mostly noise down in the low-output
3) I think folks are even more noise-obsessed now than they were 20 years ago (which I find a bit ironic).
4) I think it is really unfortunate that the proliferation of RFI 'hash' from modern technology has led to both real and imagined noise issues in audio reproduction.
5) I think that the obsession with power conditioning and esoteric power and audio cables is driven, to a significant extent, by the added noise of RFI and digital signal processing artifacts (irrespective of the low level of particularly the latter).
6) I think that y'all'll likely disagree relatively vehemently with me on some, perhaps all of these items!
(actually, I am pretty sure about number 6)

bonus opinion

7) For those of us troglodytes (mancave-dwellers, if you will ) using (very) high-sensitivity loudspeakers, the first 100 mW is important (and, perhaps, even the most important)!

At the first watt, I'm pushing close to 106db so yes ! The first 100mw are very important ! So is the actual noise-floor. With very sensitive speakers, the noise floor can hide micro-dynamics...

 

[fpitas]

At the first watt, I'm pushing close to 106db so yes ! The first 100mw are very important ! So is the actual noise-floor. With very sensitive speakers, the noise floor can hide micro-dynamics...

Yes; hopefully you can put your ear to the horn like the RCA dog and not hear anything.

 

[mhardy6647]

At the first watt, I'm pushing close to 106db so yes ! The first 100mw are very important ! So is the actual noise-floor. With very sensitive speakers, the noise floor can hide micro-dynamics...

Roundabout 104 dB here.

While we're on the subject -- I think one of the more salient (if non-dramatic) manifestations of the value of high-sensitivity loudspeakers is the way that fadeouts retain their cohesiveness, even at very low volume levels.

 

[LiquidSound]

Yes; hopefully you can put your ear to the horn like the RCA dog and not hear anything.

Hopefully, the noise floor is well below the 100mw level ! LOL !

 

[fpitas]

Hopefully, the noise floor is well below the 100mw level ! LOL !

At my normal listening level, the horns get about 20mW signal.

 

[Depoxy]

The issue is a little bit more subtle. There is a "problem" with measuring power amp output using single tone sine waves. Music is not a single tone sine wave. The important difference is in the crest factor. Single tone sine waves has a crest factor of 3 dB, which means the V_peak is 1.414 * V_rms. Music typically has a crest factor of >12 dB (see below), which means when outputting the same average power, V_peak is >4 * V_rms.

The implication is that, if you are listening to music at the 80 dB SPL at 1 W average power (as in your earlier example), your amplifier will need to have the voltage output capability to produce an equivalent of >8 W when tested with a single tone sine wave. Add another 10 dB headroom for program peaks, and you will need an amplifier capable of delivery unclipped power of >80 W (as measured with single tone sine waves) when your listening level only requires an average power of 1 W.

View attachment 76423

 

[LiquidSound]

Roundabout 104 dB here.

While we're on the subject -- I think one of the more salient (if non-dramatic) manifestations of the value of high-sensitivity loudspeakers is the way that fadeouts retain their cohesiveness, even at very low volume levels.

Hopefully not fading out to a higher level noise floor !
OK... I normally don't consider all of this when listening. I do consider that a speaker/driver rated at 1-watt of average power produces X-db at 1-meter, I then calculate the loss at the actual listening position at an ~ -10db +3db for stereo speakers yielding ~ -7db loss at the listening position. So, IF my speakers truly produce 106db at 1-meter with 1-watt of power, subtracting the 7db loss at the listening position would yield ~ 99db of sound at 1-watt... If my listening is at an 80db average, then if I have this correct, I'd be using ~ 1/64 watt and adding 12db for crest-factor and 10db for headroom (22db), this would bring me to needing an amplifier that can produce ~ 2 - 4 watts of average power to cover the headroom and crest-factor. I realize there may be other factors in my estimation not considered, however I think I'm in the ballpark on this... I'm certainly open to correction on this estimation !

 

[SeshatCZ]

LiquidSound, thanks for the math. It is very helpful for me. So, for us mere mortals, we can talk about higher-sensitivity speakers e.g. 91 dB, which are still more often Numbers should be for a 15 dB lower base in sensitivity circa threetimes more power needed, i.e 6-12 Watts. These should cover any decent Class A amplifier, primarily in biamp setting for tweeters.

 

[Koeitje]

No, because 1W is not enough for the dynamics for pretty much all speakers in normal listening situations. My Revel M106 would give me a little over 83dB at my listening position with 1W. This includes room gain and using two speakers. Dynamics go far above 83dB under normal situations. At 5W I'd get a little over 90dB, which is pretty much what you can expect in terms of dynamics and moderate listening levels.