How to test audio amplifier?

[stmdude]

So, I've designed and built a Class-D audio amp, and it seems to work and sound just excellent.

However, how can I take my ears (and speakers) out of the equation and verify that they're actually linear across the entire frequency range, if it has rolloff (and if so, how much and where), etc, etc.

I was thinking of just hooking my scope up to the outputs, but I have a feeling that the inductance in the speakers would influence the result.
Could I use a power-resistor as a dummy-load when doing this, playing a few frequency-sweeps and letting the FFT in my scope tell me the "truth" ?

Or, is there a more standard way of doing this?  App-notes would be helpful as well.

[dzseki]

Historically the load on the amplifier (for catalogue measurements) is considered as purely resistive, otherwise there would be needed a widely accepted and standardized non-resistive load, which is non existent. For own purposes it is indeed wise to try different load configurations however (be careful with capacitive loads though).
The easy way would be to use simple function generator and oscilloscope, a little more advanced approach would use a half decent PC soundcard and some software like RMAA, professionals usualy settle with instruments from Audio Precision Inc (aka. AP) or Brüel & Kjaer.

[diyaudio]

Whats the correlation between measurement and what you hearing ? 

Siegfried Linkwitz passed away on September 11, one of hes last words of wisdom was,
"what is important to the eye doesn't mean its necessarily important to the ear."

class-d ? you focus on the effectiveness of its modulator, (how well it can handle loads under different conditions) effectiveness of amplifier protection, effectiveness of its clipping detector, OVP, OCP, OTP, dc detect effectiveness. check you with your scope under heavy load, can it deal with sustained abuse? can you recover by gently reducing gain (agc), or terminate until the fault is removed, saving it and whats connected to it. 

now you have a good amplifier, the other stuff is ground for audiophools subjective responses.

       

[JoeB83]

You may have some better luck posting at www.diyaudio.com . There are some really knowledgeable people over there when it comes to audio electronics. There’s a lot less audiophoolery going on there than most audio forums.

[arno]

Hi,
there are several parameters you can measure. Frequency response, distortion, SNR, just to name a few. I mostly used Audio Precision equipment for measuring/testing audio parameters, maybe you have the chance to get access to one.
For a "rough" measurement I would recommend to use your sound card and REW [1] to test. You can use a simple (power) resistor as load, but be careful not to overload the input of the sound card.

[1] https://www.roomeqwizard.com/

BR
Arno

[drussell]

While "normal" testing like maximum power output are being tested, yes, a resistive load is what you want but speakers do not look like pure resistances.  The industry norm is to test to 60 degrees reactive in both directions to simulate likely real-world speaker loads.  You short circuit protection, etc. should allow loads up to 60 degrees either way without limiting.

For stability testing to check for oscillations and other bad behavior I usually go all the way to pure inductance and pure capacitance at the load impedance.  For a typical test of 8 ohms at 1 kHz I use about a 1.3 mH for the inductive test and a 20 or 22 uF (good film cap!) for capacitive to be right around the 8 ohms reactive that we want to test.

[schmitt trigger]

An often overlooked parameter: temperature rise on the power components.

Although Class D amps are more efficient than Class AB, they still produce heat.

With traditional class AB amps, one would monitor the output and perhaps the driver transistors only.
With a class D, the temperature rise of the filter inductors and capacitors is also required. Those components are subjected to very large amounts of ripple current.

[SiliconWizard]

there are several parameters you can measure. Frequency response, distortion, SNR, just to name a few. I mostly used Audio Precision equipment for measuring/testing audio parameters, maybe you have the chance to get access to one.
For a "rough" measurement I would recommend to use your sound card and REW [1] to test. You can use a simple (power) resistor as load, but be careful not to overload the input of the sound card.

Yes. Don't forget to measure that at different power levels.
Also, even though it would then be related to a particular pair of speakers and the room itself to a certain extent, you can make the measurements with the speakers instead of a dummy load. Not something that could be published since it would be non-standard, but for a diy amplifier, that may give you a better idea of its performance in the conditions you're using it.

[calexanian]

There is a lot of literature about this floating about. The first general tests as mentioned before are power tests into a restive load at various frequencies. Next would be a THD test at various frequencies and power levels. After that signal to noise and cross over if a multi channel amp. Then you get into phase and all of that, but the speaker/drivers involved would also be taken into consideration for those tests.

Like i said, there is a lot of literature out there for that., but your basic tests can be performed with a scope, load, and a function generator

[Conrad Hoffman]

This is ancient and they use a spectrum analyzer, but it gives you a good idea of setups, what to look for and how to present it:
http://lcweb2.loc.gov/master/mbrs/recording_preservation/manuals/Tektronix%20Cookbook%20of%20Standard%20Audio%20Tests.pdf