Watts are watts - so what's up?

[maxg]

I am now confused as to what everyone means by crossover distortion (actually it is the first time I have heard the term). I take it we are not refering to the crossover in the speaker??

Anyone care to explain to a complete dummy what on earth you are all talking about.

Also Leo,

If I understood correctly you were talking about Harmonic distortion as low as 0.03% for the SET amp at low power - correct?

Whilst I appreciate that for SS amps the distortion will rise as the power goes down I have seen some quoting THD figures of 0.0001 or some such rediculous number. Would it really climb by a factor of 1000 to be more than that of SET at low power?

and finally:

Is distortion of 0.03 Verses distortion of 0.09 (for example) really detectable to the human ear? How about 0.3? Or 0.003?

What are detectable changes in this area?

[leok]

maxg,

SS pp amp crossover distortion, in a general sense, is loss of linearity in the region at the center of the wave (zero Volts crossing region) where both positive and negative pulling transistors are partially turned on and biased where they are least linear. Sometimes the deviation from linearity is so bad there is a little notch in what would otherwise be a smooth waveform. Unfortunately, low power occupies this xero crossing region and so suffers most from this type of distortion. There is another type of ss distortion that also affects small signals at any point in a transistor's bias that is due to the transistor's exponential response to base-emitter voltage. So small signals have a tough time in a ss environment.

The .03% reference to the SET is very academic. I would never claim a SET amp has such low distortion. But, at a low enough signal level, it probably does.

As to what distortion levels become noticeable, that will vary depending on the technology used. People tolerate much higher distortion from tube amps than they do from ss amps, so you can't directly compare the values. The ss amp that has THD figures of 0.0001 (I guess that is %?) might be great at 0.1 Watt where it may be THD figure of 0.01. You have to try it. The problem with pp ss amps is, because distortion increases at lower power eventually it becomes an empirical problem: will the amp eventually get into trouble and sound harsh? With other technologies, if it sounds good at moderate power, it is less likely to be a mess at low power.

Leo

[Parrot]

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

On 6/30/2005 7:31:09 PM leok wrote:

Paul,

I just realized you might have thought I was saying my SET is always below .03% distortion. No, the inverse. Possibly low level harmonics and hall reverb. approach that, and that is important for what I want out of some recordings.

Leo

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

Thanks for clarifying this, Leo. I did indeed take what you said to mean that it never got *above* .03% distortion, and that just struck me as totally different than anything I've ever seen. It makes a lot more sense that it is *always* above .03% distortion, with the caveat that you are just guessing on the figure anyway.

At clipping, would an amp be at 100% distortion? Or how is the percentage determined? Is the measurement made over a specific time duration?

Is crossover distortion a part of THD figures?

[leok]

Paul,

I'm sure others are better at this than me, but I'll try to describe one type of distortion measurement: THD + noise.

1) apply a pure sine to the amp's input at 1 frequency.

2) adjust for some power output, say 1W

3) using a spectrum analyzer sweep the entire audio band and sum, or actually integrate, the total power of all frequencies that are not the input frequency.

4) divide the summed power by the single frequency power and turn the result into a % (basically it's everything you don't want divided by the one thing you do want).

Very severe clipping would be a square wave and there are formulas for the harmonic series that makes a square wave .. it's a series of odd harmonics. I would guess the sum of those harmonics would be greater than the fundamental so distortion would be >100%. At lower clipping levels, the upper harmonic content would drop accordingly, so % distortion due to clipping could be anything from 0 to whatever the max is for a square wave.

Leo

note: the DHD + noise measurement is particularly applicable to switching amps and d/a converters in general, because as they loose resolution they tend to add non harmonic or white or pink noise instead of the harmonic (frequencies that are multiples of the test sine frequency) type. The impact of such noise is background detail becomes obscured. You can hear this on very quiet passages on CD where white noise accompanies the quiet sounds and you can't really hear timbral detail of the instrument, even though there is black silence when the instrument is not playing.