Passive Crossover Network

[Rich D.]

Hi electrotechs!

I am designing a audio amp/speaker combination. This is to be a 3-way system, bi-amplified. I am using an active crossover filter that will drive a mono subwoofer, band limited to 1 KHz and lower. Two more amplifiers will drive the rest of the audio in stereo. This stereo amp will drive two 2-way speaker systems consisting of a 2X mid and 2X high drivers on each side. I am going to use a passive crossover here because it is not feasible in cost and complexity to add two more amplifier channels and their power requirements.

My tweeters can run down to 1.5 KHz, and the mid-range driver is good up to 10 KHz, so I chose a middle frequency of 4 KHz as the crossover point that should reduce the funny business these drivers can produce near the limits of their frequency range.

So when I use my calculator software to figure out the passive crossover at about 4 KHz, I find I can use a 3.3mH air-coil and a 0.47uF cap. I would normally elect to use a larger cap and smaller inductor but I have found a decent deal for 3.3mH. Besides: inductors may be lossy but caps are less reliable. My calculations show a decent curve with no peaks, but that is with 160 ohms used as a damper in the network. Usually I see designs without a resistor, and I'd rather not have one absorbing too much music power.

Without the damper I get a theoretical 30dB peak at the cutoff frequency. That scares me into wanting to use a resistor in the network. BTW: I have a particular sensitivity to excess peaking in the 3-4 KHz range so I REALLY don't want to hear that.

So my question is: If I don't use a separate resistor in the network, can I rely on the losses of the components & drivers or will I be able to hear this strong peak at the cutoff frequency?

Ideally I would like to make a more complex multi-stage filter network, but my software and I don't know how to calculate those, so I'm stuck with 6dB/oct.

 

[audioguru]

It is impossible to design a passive crossover without knowing that it has a load of a speaker but you forgot about the impedance of the speaker. Without the speaker then at the crossover frequency there is a huge peak but with the speaker (8 ohms for your L and C values) then the response is Butterworth which is flat to the crossover frequency that is down -3dB.

You will have phase cancellation or phase boost problems with your 2nd-order filters. If the midrange and tweeter speakers are connected in-phase then the 2nd-order filters will cause them to be 180 degrees out of phase and cancel at the crossover frequency. If the speakers are connected out of phase then the 180 degrees phase shift of the filters cause their outputs to add +3db at the crossover frequency. A Linkwitz-Riley response (instead of Butterworth) will correct this problem.

 

[Mikebits]

Hmm, never heard of the Linkwitz-Riley dudes, very interesting stuff, here is something they had to say about a passive crossover.

Crossovers may be implemented either as passive RLC networks, as active filters with operational amplifier circuits or with DSP engines and software. The only excuse for passive crossovers is their low cost. Their behavior changes with the signal level dependent dynamics of the drivers. They block the power amplifier from taking maximum control over the voice coil motion. They are a waste of time, if accuracy of reproduction is the goal.

Interesting read, if your into this sorta thing: https://www.linkwitzlab.com/crossovers.htm

 

[bountyhunter]

This article was interesting.

**broken link removed**

 

[dr pepper]

By damper and resistor do you mean an impedance equalizer.
If either your mid or tweeter has a peak near the xover freq then you might need a impedance equalizer resistor/cap.
Otherwise what you say sounds ok to me.