Crossover symmetry

[bvbellomo]

Most 2 way designs I've seen are electronically symmetric, i.e. a 2nd order low pass on the wofer and 2nd order high pass on the tweeter. Or 4th order, but usually both the same.

This leads to an acoustically higher order slope on the tweeter, as impedence rises closer to fs, and we may have a Zobel or notch at FS, so sound level drops as we get closer to FS. On the woofer side, spl is typically the same as we go up to crossover frequency.

So how do I pick target curves to design a crossover? If I pick acoustically symmetric curves, I have to use an electronically asymmetrical crossover to hit them. If I pick acoustically asymmetric curves, I get a bump or dip somewhere near crossover.

 

[eriksquires]

Hi BV!

There's no canonical correct answer that says either of these answers are correct for most designs.

Acoustic offsets and therefore phase matching across the crossover point are also very important. This may force your hand. When I'm working on a design I have to take into account all of it. Frequency, phase and impedance. I use XSim to do all the math for me.

You are correct in thinking that it's not just the crossover response, but the speaker's as well that must line up correctly. I've even seen experienced designers like Troels say something like "I tried to make this a symmetrical (or 4th order) crossover, but I couldn't make it work..."

Of the 3 speakers I designed last year, only one was symmetric, while the other two ended up being 2nd order LP, 3rd order HP.

For most of us, the order of decisions is Drivers -> Cabinet -> crossover but in rare, masochistic cases, designers want to achieve time-conicidence, in which case the order may be Cabinet -> Crossover -> Drivers. That is, they have to find drivers that will tolerate the crossovers needed.

Best,


E

 

[eriksquires]

I should also say, I cheat, a lot!

That is, my crossovers may have "hidden" EQ. So the values of the parts is not what's needed for a bend necessarily but also to adjust the response of the drivers. 🙂 This is sometimes a little easier with a 3rd order than a 2nd.

Best,

E

 

[giralfino]

Most 2 way designs I've seen are electronically symmetric, i.e. a 2nd order low pass on the wofer and 2nd order high pass on the tweeter. Or 4th order, but usually both the same.

Maybe you've seen not well designed crossovers. A lot of people think only at electrical slopes, and forget about natural and baffle induced driver behavior, and acoustic offset. See how a reputable designer describes one of his design:

This crossover is relatively simple with a low component count. It is functionally 4th order Linkwitz-Riley with a slightly shallower slope on the woofer to help with the acoustic center offset between the woofer and the tweeter. We have third order electrical on the tweeter and 2nd order electrical on the woofer to reach our target acoustic slopes. The tweeter uses a single padding resistor and also requires a conjugate impedance flattening notch across the terminals. A traditional L-pad would not have been enough to damp the tweeter's impedance peak and a LCR circuit will be needed with almost all 6600 applications to avoid ringing.

See here: Zaph|Audio - ZRT - Revelator Tower

If you want an example of a crossover with 2nd order electrical high and low pass (but acoustic LR4):

This crossover is asymmetrical 4th order Linkwitz-Riley at 1750 Hz. The slightly steeper tweeter slope along with a slightly shallow woofer slope helps to align the listening axis with the tweeter.

Electrically, on the woofer we have a simple two component 2nd order network. Also shown is an optional impedance flattening circuit for those with tube amplifiers. (more info on that and other options below) On the tweeter, we have a 2nd order network, a single resistor for level adjustment and a top octave response shaping network. These components combine with the natural rolloffs to hit the asymmetrical 4th order slopes.

See here: Zaph|Audio - ZA-SR71

So there is not a single answer, but I tend to be cautious on every design that doesn't provide measurements.

Ralf

 

[Eldam]

What are saying soft optimiser like SoundEasy most of the time ? Or Xsim ? Do we end with an asymetrical filter ?

 

[eriksquires]

Hey Eldam,

XSim doesn't have a sound optimizer, or filter choices. 🙂 You pick them. And it's still my favorite.

Best,

E

 

[jrh0516]

PCD will show you target slopes which can be helpful when designing circuits. However IMO flat first arrival frequency response should be the goal over symmetrical filters.