Low distortion Class D amplifiers? What to expect?

[abraxalito]

Are you referring to the THD? Its a standard way to indicate an amp's maximum power.

 

[cheater]

What I mean is: the number you posted is 1% THD. The numbers I posted that I'm trying to improve upon are around 0.001% THD. So a thousand times too high. (not 100 like I posted originally)

 

[abraxalito]

Yes but that 1% number isn't indicative at all of the performance below clipping. Rather its an indication of the degree of clipping for the 400W output power.

 

[SemperFi]

What I mean is: the number you posted is 1% THD. The numbers I posted that I'm trying to improve upon are around 0.001% THD. So a thousand times too high. (not 100 like I posted originally)

If you believe an amplifier has 0.0015% THD at the limit of what it can put out in power, then either you believe in santa or someone is cheating.

 

[cheater]

That's the test results of the numbers at full rated power. The power at clipping is about 10-15% more than rated power. So it's not THD at "the limit of what it can put out", there's some headroom. Either way, I don't think Bryston would be a standard in mastering studios around the world if they made their business out of lying to their customers. We can completely bypass any claims as to the validity of those numbers - I have all trust in them I need to have.

 

[cheater]

Yes but that 1% number isn't indicative at all of the performance below clipping. Rather its an indication of the degree of clipping for the 400W output power.

Right, in that case I'll need to figure out the best-case THD numbers, i.e. at what power the numbers are the best. For example, for Bruno's designs, the THD goes down gradually as power increases, and then shoots up by about 60dB at the limit.

However, either way the amplifier you are talking about is going to have good THD at a power that's less than 600W - so what can I do?

 

[abraxalito]

Are you sure you mean THD and not THD+N? Pretty well all the plots of distortion vs output level are THD+N not THD. The 'N' component is why you see the number decreasing with output level.

 

[cheater]

Yes, definitely. The certificate has a separate table of THD numbers for 20, 200, 2k and 20k Hz, and then a single number for noise 20-20k. There's also separate numbers for IMD and output power at clipping.

 

[cheater]

By the way, the noise is at 115 dB below rated output, so it wouldn't really factor that much into THD+N anyways.

 

[abraxalito]

Yes, definitely.

Then I'm curious to know how you're able to claim :

for Bruno's designs, the THD goes down gradually as power increases

Which measurements did you inspect which led you to say this?

 

[cheater]

1ET400A - PURIFI

scroll down to the graph.

 

[abraxalito]

Yes the graphs are of THD+N as I said earlier. But your claim concerns THD, not THD+N. Hence my question.

To put the question another way :

When the THD+N measurements are dominated by the 'N' component, how are you able to deduce the THD contribution?

 

[cheater]

The noise is rated at -117 dB. This was calculated from 20*log((100* ((11.5/1000000)/sqrt(425 * 4)))/20).

The lowest point on the THD+N graph is at 0.0000015, i.e. -114 dB. Therefore the distortion is going to be by far the main component almost everywhere except for the dip where it's merely a little larger than the noise.

It doesn't matter. What I was saying that Class D designs seem to have a sweet spot somewhere close to, but not at max rated power. You mentioned the THD at max rated power. So it would make sense to find out the best case THD, and what power that is at. But either way it won't be 600W.. so I'm not sure this is the right project to be looking at.

 

[abraxalito]

The lowest point on the THD+N graph is at 0.0000015, i.e. -114 dB. Therefore the distortion is going to be by far the main component almost everywhere except for the dip where it's merely a little larger than the noise.

You misinterpret the graph - the opposite is the case. Noise dominates everywhere except just prior to clipping. As you'd expect in a SoTA amplifier designed with no holds barred to minimize THD.

 

[cheater]

The output noise is specified at that level without specifying what power output it's at - I assume it's at all power outputs.

 

[abraxalito]

When the output noise is expressed in dB that's a ratio. If that ratio is the same at all power outputs then the noise would need to reduce as the output signal does. Can you think of a mechanism for that?

 

[cheater]

It's not, it's specified as uV. Look at the page

 

[abraxalito]

I used the word 'when' and was looking at your post #33.

When the noise is expressed as a voltage then it would stay largely constant with signal level. Which means the ratio of 'N' increases as the signal level goes down.

 

[elviukai]

I read the thread, did not understand what thread author want to achieve. 300W is a bit
too much on the hot side for closed vanues(or studio) especialy if this power directly conected to conventional voice coil dynamics speakers. if for some (unknown) reason a high SPL is needed its much wiser to change efcienty of electro acoustical converter-speakers(like going compression drivers based speakers) amp dicortion of 0.5% at 300W output are only small fraction of discortion of complete system( amp, speakers, energized room). also output of EAC are not proportional to input power.
if low discortion are main plan, changing speaker type and checking studio walls with accelerometer whould be my first plans. thread starter barking on wrong tree(amp) todays amps (since ~1975) have astonishing low dicrotions which are not measurable by "analog scopes"- human ear

just my 2cents.

 

[bucks bunny]

With output power below the sweet spot, THD is buried in noise. Which explains not only rising THD+N plots but rising THD plots as well with decreasing power.