My understanding of the function of a crossover is that at a certain frequency the signal will start dropping at whatever the slope of the crossover is. So with an 80Hz lowpass filter and a 12db/Octave slope if you have 90db from 30-80Hz you would have 78db at 160. This site however states that the crossover point is the 3db point. Meaning that you'd be flat upto some point below 80Hz, 3db down at 80Hz, and 15db down at 160Hz. That doesn't really make sense to me, but it's how I read their doc. http://www.mtxaudio.com/caraudio/edu...ersFilters.cfm Can someone clear this up for me? brianca...

Come on, folks. Surely someone can share a little knowledge about crossover workings and vocab. brianca...

brian, the frequency where the effects of crossover attenuation begin is alternately called the turnover, corner, or crossover frequency, and is typically the frequency at which the signal is attenuated by 3db. i think the 3db gain or loss is a function of the drivers, and NOT a result of the action of the crossover - the crossover starts working actively only AFTER the corner frequency. but i could be wrong. this applies for both high- and low-pass filters. my understanding is that the crossover needs to begin working at some point, and this is just the point usually selected. maybe someone else can shed some more light on this...

According to the article the point where the action starts is the cutoff point, the crossover point is the 3db down point, and the slope starts from there. I was just wondering if that was correct, as it seems to not be the way that crossovers are discussed in most of the posts I have read. brianca..

Remember a crossover has both a low pass filter and a high pass filter. 3dB is a doubling (or halving) of power. The LPF and HPF meet at the crossover point at -3dB. So you have two signals at half power which equalizes the power output.

brian, the description of crossovers on the mtx web-site is the same as the one i learned. it may be helpful if you explained what it is specifically about that particular description that you feel conflicts with corssovers as they have been discussed around here.

Brian, Crossovers have slopes, meaning that there will be some overlap. Imagine if your tweeter went down to 2kHz, and was full up at 2kHz, then started to roll off at 12dB/octave as you go down from there. Now, add to that your woofer which is also full up at 2kHz, then rolls off at 12dB/octave as you go up from there. You're going to get a boost centered at 2kHz, right? This is the reason for stating the crossover points the way they do. There are also some specific kinds of crossovers, such as Butterworth and Linkwitz-Riley. As for your other question, yes, in general a 3dB increase is a doubling of SPL.

I guess what I'm not clear on is if the crossover point that I set on my B&K is the point when the signal will be 3db down or where it will start to slope. My confusion also comes from how the highpass and lowpass meet. Let's assume a crossover point set in the pre/pro at 80Hz. So my mains are -3db at 80Hz due to the lowpass, and my sub is -3db at 80Hz due to the lowpass. That seems to me that I am down 3db at 80Hz. But what you're saying is that two speakers (or in this case 7 speakers and 1 sub) at the same level create twice the output to make up for the drop. Correct? That seems counterintuitive to me. I'm trying to gain a better understanding of how this happens, so I guess I need to spend a little time with the math. brianca...

John is correct 3dB is a doubling of power not sound level. Brain, Keep in mind that the dB scale is logarithmic not linear.

The crossover point is NOT normally where they begin to slope, but where they are 3dB (or possibly 6dB down - see below). They start to barely drop much, much earlier than that, how much is dependent on the crossover filter type, and order. In general, the crossover point IS the 3 dB point, or where the amplitude is 0.707 of the pass band value. There is a phase shift involved at the crossover point, so when you add them together, you can't add them together algebraically, but as vectors. To see this graphically, draw 2 sine waves out of phase with each other. Then add them together at each point. You will get a pure sine wave as a result, but the amplitude will be smaller than you would expect. If I remember correctly, a high pass and a low pass crossover that are each 3dB (0.707) down and 90 degrees out of phase at the crossover point add together to 1 at that point. Just what you want. But not all crossovers are 3dB down at the crossover point: One exception is the Linkwitz Riley crossover. By definition, a 4th order LR is actually 2 cascaded 2nd order Butterworth crossovers. Each of these cascaded sections is 3dB down at the crossover point, so together they are 0.707 x 0.707 = 0.5, or 6dB down at the crossover point. Is this wrong?? No, for these filters, the high pass and low pass sections are exactly IN phase at the crossover point, they simply add together to 0.5 + 0.5 = 1. Again, just what you want. To further complicate matters, some types (Chebyshev is a good example) exhibit some ripple in the pass band (where the gain may be greater than one at times), so the crossover point is defined as where the ripple ends, or where the amplitude is Back to 1. Makes you think a little harder before you design a filter using these, but not to worry - these have pretty horrible phase response, and more ragged pass band response, so they are almost never used in audio. Did that help any? Cheers, Chris

Chris, Some of that doesn't jive since a 2nd order are 180 deg out of phase, so if there is a 2nd order HP & LP, either the 180 deg out of phase would cancel when played together, or one being +180 deg and the other being -180 deg out of phase, resulting in 0 deg relative phase shift, except for it being 1 wavelength behind.