Andrés,
Sorry I missed your question in the original thread.
... a simple method to check and correct any crossover issues.
Yes, the idea is to create two tracks out of one, splitting the original track into high and low frequency ranges so each can be manipulated separately.
Depending on the type of spectrum analyzer you have, either pink noise or white noise will work. Pink noise is appropriate for power spectrum analyzers and white noise is appropriate for amplitude spectrum analyzers. Most built in wave editor spectrum analyzers I have encountered graph the amplitude spectrum and will yield a flat curve with white noise.
Split a noise track into high and low frequencies using your "crossover" filters. Then mix them back together and look at the spectrum. If you see a dip or bump at the crossover frequency, you have a problem. Since your filters are somewhat of a black box, there is no good way to analytically determine how to fix the problem. You'll just have to do it experimentally. There are two "knobs" you can adjust to remedy the situation - the separation in frequencies between the high and low cutoffs, and the relative phase between the two. For example, you could try setting the high and low pass cutoff frequencies +- 5% from the center frequency, or 152Hz and 168Hz respectively, rather that both at 160Hz. Or you could try adding a slight delay (zero to several milliseconds) to one of the tracks. Once again it's impossible to predict which track or exactly how much without knowing the exact nature of the filters. Then finally try some combination of the two.
When you find a good solution for a given filter and crossover frequency, you can probably scale the values so it works for other frequencies. Say for example, at 100Hz a +- 8% frequency spread and a 3ms high pass delay flattens out the response. Then at 50Hz a +- 8% frequency spread and a 6ms high pass delay should work. Likewise, a +- 8% frequency spread and a 1.5 ms high pass delay should work at 200hz. Make sense? Of course, I would verify this with the spectrum analyzer, because who knows how the hell some of these filters are designed....??? Also remember that adding different effects to the two tracks can wreck this adjustment.
... checking to see if an amp power per channel is the same
Accurate spectral power density measurements require some relatively sophisticated test equipment. I think, however, you're just interested in determining whether or not you amplifier channel gain in equivalent to within a few tenths of a dB or not, right? A very simple way to do this is to hook one speaker up across the positive L and R amp outputs. Put an identical signal through both cannels and listen. A 1kHz sine wave is a good test tone because it's in the middle of the spectrum and your ear is very sensitive to it. The signal should completely cancel. It probably wont, so using your software adjust the output of one of the cannels up or down in steps of 0.1 dB until hear the minim loudness. This setting is the difference in gain between the two channels. Set the levels on the computer equal again, and then you can adjust one of the amp channel levels until the signal is minimized. Leave it this way. If you cant seem to make it get any quieter, then your nominal value is probably best.
If you want numbers, then get one of those 8 Ohm 20 Watt power resistors from your local electronics supply store. Hook the resistor up across the positive L and R channels like you did the speaker. This time with a 60 Hz test signal, measure across the resistor with your AC voltmeter and adjust one of the amplifier channel levels until the voltage reads a minimum. However, your ears could be more sensitive than your meter in this case so the 1 kHz listen method might be better.
Hope this helps!
barefoot
