You sure you are not looking at THD+N? what you have described is a typical curve when you take noise floor into account, as this remains constant, so as signal goes down the noise gets closer.
Look at the distortion graphs here: http://www.hypex.nl/docs/NC400_datasheet.pdf
THD at 20W is significantly lower than at 2W; at 2W THD is masked by noise but as such THD is higher than at higher output levels.
Maybe not such a bad idea to force the NC400 into more power where it has less noise and lower THD?
THD at 20W is significantly lower than at 2W; at 2W THD is masked by noise but as such THD is higher than at higher output levels.
Maybe not such a bad idea to force the NC400 into more power where it has less noise and lower THD?
Last edited:
You are wrong. totally and utterly wrong. You cannot reduce noise by drawing more power from the amplifier.
Noise will not be less at higher power; it is steady state.
However at higher power S/N ratio will improve; that's what matters.
As for distortion: it is well known that class A/B amplifiers don't excel at low power (cross-over); something comparable might occur to class D amplifiers but I am not an expert. Maybe dead time plays a role.
However at higher power S/N ratio will improve; that's what matters.
As for distortion: it is well known that class A/B amplifiers don't excel at low power (cross-over); something comparable might occur to class D amplifiers but I am not an expert. Maybe dead time plays a role.
The SNR of the amp won't change - if you look at the graphs, you see that the noise stays constant (while distortion starts to increase after 50 W), but the SNR caused by input noise will get worse if you burn away power in a resistor (as you need more gain to reach the same power level).
That is a specific problem with A/B amps (crossover distortion). It is not an issue with class D.
Class D may have other issues; they are not distortion free.
When generally comparing distortion graphs between class A, class A/B and class D it seems that class A/B and D have more in common (whatever the cause) than class A.
Class A: rising THD with output as SY mentioned.
Class A/B and D: lowest THD at some point between low and maximum output.
When generally comparing distortion graphs between class A, class A/B and class D it seems that class A/B and D have more in common (whatever the cause) than class A.
Class A: rising THD with output as SY mentioned.
Class A/B and D: lowest THD at some point between low and maximum output.
Pieter, it is rising THD+N. The THD actually decreases going down in frequency but as the S/N gets worse, what you see is a rising towards LF with the THD invisible.
A spectrum at different levels tells the truth. Or a THD graph without the noise.
Jan
No, it doesn't. Remember that the THD graphs measure THD and noise in %, not in dB (as noise is usually measured). That is % of the actual signal, not maximum signal (SNR is normally measured as the ratio of noise to *maximum* signal.
If the noise level stays the same in terms of an absolute value (in mV), a number measured against a 2W signal will look 3 times as high as one measured against 20W (and that is exactly what the hypex graph shows). So the noise level stays exactly the same with a 2W and 20W signal. If you attenuate the output (with a resistor) you need to increase the gain to reach the same power level - increasing the noise.
Thus, attenuating the output to make the amp "work harder" increases noise and decreases SNR.
Pretty much the equivalent of tuning a car by making the brakes bind...
Class D *may* have other issues, yes, so are we now moving into speculation? They definitely are not distortion free, but you are making specific claims about the distortion characteristics that don't hold water.
Please point us to graphs showing that (preferably for the hypex topology).
Speculation is not the right term; I do not aim to "talk" NC400 into distortion territory which would be ridiculous as it sets very high standards.
Up to some 3 W the THD is dominated by noise; I get it is THD+N and THD alone might be about non existant.
The THD+N graph of NC400 and about every class AB amplifier reminded me of Nelson Pass' comparison of class A and class B:
Here we see distortion profiles of class A (red curve) and class B (blue curve) without N; I had the class B graph in mind which shows rising distortion at low level (crossover artifacts), but realize that THD and THD+N graphs are not comparable.
With N involved the class A graph will look much more like class AB and D graphs.
Last edited:
Class B is very different from even class A/B. Yes, class B suffers from increasing (relative) distortion at lower power levels - pretty much a structural problem with class B. But that does not apply to even class A/B, not to mention class A and D.
- Status
- Not open for further replies.