Class D Amp Specs

[dpasek]

I haven't seen anything but laughable distortion specs for the class D amplifiers and modules. What use is it to spec 300 W at 10% distortion? (For example, the 2x300W TAS5630 Class-D Amplifier Board.) I'm not driving a servo motor with it, I'm using it for a speaker. Don't the manufacturers have any pride in their products, or are these amps still vastly inferior in THD as compared to class AB? I would appreciate seeing power ratings at 0.05% THD max if these amps are capable of delivering it.

[Face]

Re: Class D Amp Specs

http://www.wyred4sound.com/141528/html/page.html

[Sydney]

Re: Class D Amp Specs

Big wattage numbers sell product.
What you will find is that distortion "knees" upward quickly in amplifiers.
That module produces 240 watts @ 1% distortion into 4 ohms, lowest distortion figure is .07%.
See:
http://www.ti.com/lit/ds/symlink/tas5630b.pdf

[Wushuliu]

Re: Class D Amp Specs

Lack of research is laughable.

Another example, here.

[dpasek]

Re: Class D Amp Specs

Ah, the vendor just regurgitated the tabular specs from TI. I can see from Table 1 on P. 12 of the spec sheet that this is really an approximately 100 W/ch at <.15% THD (8 Ohm load) amp module. Of course, (Sydney) as soon as any amp starts to clip, odd harmonic distortion rises quickly and the amp starts to sound bad. The amp starts to become nonlinear even before that. The 300 W spec is just hype as is the 'audiophile' claim. The .03% at 1W THD spec is also bogus since this is just a local minimum. But 100W/ch at <.15% THD really isn't too bad, especially at the price point, and particularly for the efficiency which is around 90% at 100W. The retailers should provide the realistic specs along with any hype they want to use and that is called truth in advertising. There have been laws passed supporting the concept in some areas of trade. Audio amplifiers for automotive applications and motors for compressors are a couple of exceptions (to the concept of truth) that come to mind. (Wushuliu, the 2x250W IRS2092 module has more appropriate specs.)

[djg]

Re: Class D Amp Specs

I'm happy with the performance of my amps built from kits from


www.classdaudio.com

[fastbike1]

Re: Class D Amp Specs

The point that seems to escape you is that the specs are available, you just need to read a bit further for a given board/amp. You will be able to see a power spec @ a THD value comparable to what is seen w/ AB amps.

The specs game is no different than what has gone on for years.

Ah, the vendor just regurgitated the tabular specs from TI. I can see from Table 1 on P. 12 of the spec sheet that this is really an approximately 100 W/ch at <.15% THD (8 Ohm load) amp module. Of course, (Sydney) as soon as any amp starts to clip, odd harmonic distortion rises quickly and the amp starts to sound bad. The amp starts to become nonlinear even before that. The 300 W spec is just hype as is the 'audiophile' claim. The .03% at 1W THD spec is also bogus since this is just a local minimum. But 100W/ch at <.15% THD really isn't too bad, especially at the price point, and particularly for the efficiency which is around 90% at 100W. The retailers should provide the realistic specs along with any hype they want to use and that is called truth in advertising. There have been laws passed supporting the concept in some areas of trade. Audio amplifiers for automotive applications and motors for compressors are a couple of exceptions (to the concept of truth) that come to mind. (Wushuliu, just because a device spec sheet says something doesn't mean it will be realized in a given implementation.)

[Deward Hastings]

Re: Class D Amp Specs

You are missing several points . . . the first being that the 10% distortion point is a common standard for spec'ing maximum amplifier power, as it clearly marks the onset of clipping. If you dig even a little deeper you will find that most Class D amp manufacturers *also* report distortion at more reasonable power levels, and commonly show distortion+noise/power curves as well. There is no "mystery" to it, and nothing "hidden" or deceitful about it.

Ah, the vendor just regurgitated the tabular specs

Um . . . well . . . I'm not familar with (haven't used) the TI chipset, but have IRS2092 based amps done surface mount and on two different through-hole boards. They all test essentially the same and match the manufacturer results linked by Wushuliu above . . . and sound the same as well. Build the reference circuit, get the reference results. There is no reason it should be otherwise, unless something is done wrong in the build.

The absense of crossover distortion makes the better Class D amps (2092 based included) sound like the best Class A at low power, while maintaining that quality at power levels unattainable in practice with Class A for thermal and efficiency reasons, and difficult even with Class AB. If you're an antiquarian, or just looking for cheap-but-decent amps on the "used" market, then "linear" still makes sense. If you're looking to build or buy new, however, then anything other than Class D is just silly (well, maybe the LM3886 still makes some sense for low-to-moderate power, and it also sounds good).

[dpasek]

Re: Class D Amp Specs

I took a look at the IR reference design. It really looks quite nice.

[dpasek]

Re: Class D Amp Specs

... the 10% distortion point is a common standard for spec'ing maximum amplifier power, as it clearly marks the onset of clipping. If you dig even a little deeper you will find that most Class D amp manufacturers *also* report distortion at more reasonable power levels, and commonly show distortion+noise/power curves as well. ...

The 10% (continuous power) distortion point would seem to be something of an indicator of transducer damage (blown speakers) rather than a useful measure of headroom or transient response. Wouldn't it be a more meaningful indication of transient response to measure the response to some form of impulse stimulus (such as a haversine) in the frequency domain to see where it begins to show a significant change?

As for more reasonable power levels, 1% is probably the limit of useability with <.1% a more appropriate measure for continuous power performance. That, plus a meaningful measure of transient response would seem to provide more reasonable measures of performance.

The minimum distortion spec seems deceptive since it is just a local minimum in the curve and not representative of overall performance. If they want to claim 'audiophile' performance, it seems that the THD spec should be something less than .05% over a broad power and frequency range.

[fastbike1]

Re: Class D Amp Specs

Actually I think you're off by a factor of ten. IIRC there was some research years ago that indicate 10% was the level of distortion that listeners considered the limit of usability and 1% was the threshold for fairly reliable identification (i.e. better than random guess).

[Deward Hastings]

Re: Class D Amp Specs

Wouldn't it be a more meaningful indication of transient response to measure the response to some form of impulse stimulus

No . . . there is no "transient response" past clipping.

[dpasek]

Re: Class D Amp Specs

No . . . there is no "transient response" past clipping.

I'm not talking about clipping in this part of my comment. It seems that I diverged without sufficient context change.

[dpasek]

Re: Class D Amp Specs

Big wattage numbers sell product.

So it seems.
I found a comment in an EE-Times article to that effect:
"Amplifier manufacturers will sometimes specify the output power at a higher level of THD in order to allow them to advertise a higher output power. A clipped sinusoidal output waveform with 10% THD has an average output power 28% higher than an undistorted sine wave output from the same amplifier."

[dpasek]

Re: Class D Amp Specs

Actually I think you're off by a factor of ten. IIRC there was some research years ago that indicate 10% was the level of distortion that listeners considered the limit of usability and 1% was the threshold for fairly reliable identification (i.e. better than random guess).

That's possible. It depends on application, program source, and the tolerance of the individual listener. A boom box user listening to rap probably won't care. It also seems that you can get away with using a class C amp in a bullhorn and the voice will still be understandable for communications purposes. But these are both well outside of 'audiophile' territory.