Some measurement of commercial class D

[banana]

A friend of mind lent me his Acoustic Reality eAR202 for some AP measurement.

[banana]

attached the 8ohm 1W FFT

Red line is right channel, blue is the left channel.

1KHz ouput 2.8Vrms, set as 0dBr reference.
K2 = -95dBr = 0.0018%
K3 = -90dBr = 0.0032%
K4 = -103dBr = 0.0007%
Calculated THD (noise not included) = sq rt(K2^2 +K3^2 +K4^2) approximately 0.004%

when using calculation from the AP2 THD+N is 0.015%

Figure from AP2 seems higher than the FFT have shown. I suspect this is due to out of band energy. The AP2 I used doesn't contain the AES17 filter, so I used the build in 20KHz filter instead.

Maybe I should try adjusting the analog analyzer's bandwidth to see if I can get better figure.

Nevertheless, -90dBr K3 and <110dB noise floor are impressive.

[IVX]

Originally posted by banana
attached the 8ohm 1W FFT..

..Nevertheless, -90dBr K3 and <110dB noise floor are impressive.

If you need only several watt's, in other words <50W, then you can be much more impressed by using some output stage chip from TI, Philips etc in the selfoscillating configuration. E.g.: http://www.diyaudio.com/forums/showt...168#post740168

[koolkid731]

It would be interesting to run a THD vs frequency at different power levels. I did that on a module from the same OEM company and found it rather disappointing, so I'd be curious if any improvement has been made.

For the record I'm not at present affiliated with any manufacturer, lest some accuse me of being biased.

[IVX]

Originally posted by koolkid731
It would be interesting to run a THD vs frequency at different power levels. I did that on a module from the same OEM company and found it rather disappointing, so I'd be curious if any improvement has been made.

For the record I'm not at present affiliated with any manufacturer, lest some accuse me of being biased.

I see noise floor rise according frequency (at the AP screenshot from banana), so seems THD/Freq isn't flat at all.

[banana]

Now THD+N vs Freq.

loading is 8ohm, with 20KHz filter.
Red line 1W, Magenta 20W, blue 50W, cyan 120W

THD rise with higher frequency is expected, probably due to running out of loop gain.

[banana]

THD vs output power

Blue line is 8ohm load, red line is 4ohm.

I'm becoming more and more convinced that the 0.015% THD+N was quite real, most of it being inband noise.

I've tried different filter, with 20KHz filter the residue noise level is about -77dBr (reference to 2.83V). But even with A-weighting, or 400Hz highpass filter, it is still only -78dBr.

[banana]

CCIFF Intermodulation

Dual tone 19KHz and 20KHz, -3dBW each (0dBr is 1W).

Note that the whole noise floor rised when dual tone is applied. Also -75dBr side band is not as good as expected.

[koolkid731]

Thank you Banana for your posts.

I usually make measurements with AES17 40kHz filter to capture 2nd harmonics of 20kHz signal, and publish graphes with AES17 30kHz filter.

It's also good idea to document the curves of each plot - which is which at what conditions? - and I use abbreviations such as f3 for 30kHz filter, f4 for 40kHz filter, 8R for 8-ohm load etc...They make later analyses easier.

[classd4sure]

Sorry to hear you're not with what's their names anymore.... I'd have quit too, we'll not speak of this again


Capturing 40kHz with a 40kHz cut off seems.. well... what's the point?