Another question, its obvious that a resistive load is quite different of real loudspeakers, I would like to build a test circuit wich has the same caracteristics of a real load. Where can I get such schematics?
Buy a cheap speaker and use that. It'll cost less and be easier than trying to do it on your own. If you really wanted to have something to test with, you could mimick a speaker pretty well with a coil that has a matched inductance to your speaker.
There's a nice piece of freeware called Speaker Workshop that allows you to make THD measurements using the sound card in your PC. I'm not totally convinced that the figures are accurate but it might give you a guide.
It also measures impedance, passive values, freq resp (with a mic), xover optimisation (rather primitive) and box calculation.
The only way you can entirely accurately test an amp is to use the load you will be listening through. As the load is a complex thing to model why go to all the bother?
If you wanted an accurate "dummy load" it would have to mimic so many varying factors - complex impedance curves created by the interaction of the crossover network and the voice coils, not to mention resonances and their associated harmonics and how this effects the impedance and load characteristics.
I'm not saying a dummy load is a bad idea, the point I was making is that testing should in theory be done as a system as a whole - the speakers and amp are used together and will interact with one another.
Testing this way will allow the amp to be talored to the speaker and vice-verca.
I do agree though, there should be an "industry standard" dummy load to make matters simpler, also I don't see why so many tests are a simple mono-frequency sinewave - when have you ever heard a pure sinewave in music?
In the brave new world maybe testing will be done with a signal with a range of different frequency sinewaves all at once to better simulate REAL music. (distortion and THD testing)
Would you not agree that this philosophy makes more sense than the industry standard 1Khz sinewave?
A standard load I have seen referred to in some magazine articles/books is an 8 ohm resistor in parallel with a 2 uF cap. I don't believe this is a particularly realistic test load, but at least it is simple to reproduce and if your amp has any tendencies to instability then this load will very likely make it show up.
I normally use a pair of 400W, 8 ohm resistors in parallel for testing hi-fi amps, they don't usually get very warm. Since most speaker's impedance curves dip close to 4 ohms at some frequency, I think this is a reasonable test. I also have a 750W, 2 ohm resistor I use for testing PRO amps, but this really needs a fan for extended testing and I hate to use it during the summer.