Everything a crossover has to do ...


Gang,


In my continuing efforts to encourage informed dialogue, I want to talk about how much a speaker crossover has to achieve, sometimes simultaneously. A crossover is a component which makes sure that the right frequencies go to the right drivers. That is, tweets should tweet and woofers should boom. Otherwise no sound comes out the woofer, and the tweeter becomes ballistic.


So, with those basics out of the way, here is a not-exhaustive list:


  • Filter. The simplest crossovers are just collections of low and high pass filters. Sometimes just one. Filtering is described above: Getting the right signal to the right component.
  • Level match. Drivers all have different voltage sensitivities, so making sure the drivers are playing uniformly is important. Imagine only seeing red on your TV for instance.
  • Equalize. Most crossovers have some sort of basic tilt built into them. Sometimes through dedicated components, sometimes just by picking the poles judiciously.
  • Impedance correct. Either to make speakers more tube friendly, or to make the filters more effective designers may put in components to solve the impedance problem alone. This alone is not something you find in an active crossover.
  • Temper driver resonances. As an example, lots of mid-woofer designs, even very good ones, have a resonant peak above their useful range. A notch filter can take it out and make the low pass better behaved.
  • Phase match. Below, at and above the crossover point drivers need to work seamlessly. Poor phase matching can cause lobeing, notches and peaks you don’t want. This is also very important in active crossovers, but you can often use digital delays to enhance the ease of the design.
  • Baffle Step compensation. A type of EQ that is the result of the driver/baffle size and arrangement. Yes, active crossovers should do this as well.


The point is, designing a crossover, whether active or passive, is not as simple as often assumed. To design any crossover, active with analog devices, DSP based or passive takes tools and effort to do well.


I hope some of you find this informative, and encourages you to learn more, if not experiment on your own.

Best,

E

erik_squires
Amateur application of bi-amping with DSP for example misses several critical functions that a good passive crossover performs.

The bad pattern of DSP crossovers usually goes like this:

1 - Set arbitrary crossover points, with extremely high slopes (4th or higher)

2 - Use global EQ to hammer the design into place.
I wouldn’t say amateur, I would say naive. I encourage everyone who wants to learn more, get the tools and ask for help from experienced builders.


Erik you are spot on ,that is  why I Always recommend upgrading or modifying the Loudspeaker Xover for 
over 85% of speakers have average at Best quality a Xover parts, What is the heart of the Speakers and can make a Dramatic sonic improvement if voiced correctly ,Quality most certainly count.
Hello!
At age 81 I say I've been doing speaker crossovers for about 60 years. Began in college in late 1950's. By late 1990's I thought I was finished, holding patents licensed to Joseph Audio's lineof "infinite-slope" loudspeakers. License expired in 2005 and I went into blessed retirement atage 67. Thought I was done but......
Forget Joseph's record winning "best-sound" at audio shows and well-regarded reviews in audio press.  Three pesky audiophile friends all owning Joseph's products demanded better sound from me and pushed me at age 79 in 2017 to perform research into possible improved crossover performance.  Retrieved my notes now housed in local tech museum which led me to try idea: Merge my infinite-slope theory (based on high-selectivity filters in radio circuits) with "constant-resistance" technology.
Modeled crossovers in computer merging both concepts. Looking for a way to optimize frequency-response and time-response in midrange frequencies simultaneously, something I have only been able to attain nearly optimum in either frequency or time but not in both.I needed to minimize errors in both. By late 2018 I had models which in cyberspace appeared to have flat frequency response and linear-phase (flat time response) through midrange!

Built and tested physical crossovers, finished 2-way prototype speaker system.Prototype worked well in frequency-time measurements in my home instrumentation (FFT analyzer, 1/3 octave pink noise, etc.) showing  flat frequency response, seamless "join" at 2KHz crossover! Repeat tests in anechoic chamber at Binghamton University which corroborated measurements. Box even had decent square-wave response 150Hz-2.5 KHz.
Listen test with 3 Joseph owners (myself included) at home lab and in audio showrooms and 2-way prototype had easily discernible "best-sound" against 3-pairs of Joseph Pearls, as well as against other $33,000+ loudspeaker systems. Word came out and I was invited to demonstrate speakers before audience of 300 at a symposium in August, with great success! 

Installed 3-way invention crossover in three sets of Pearls, with astonishing results. Happy pesky audiophile friends rejoiced.! With performance validated, filed patent on July 2019.
RIMO




















Hello!

I saw my recent post just finished, and noticed "dentdog's" post right above mine. He said to "do without a crossover"  He's right and correct. Install a very-good 4" or 5" driver without crossover in a "transmission-line" box (lossy bass-reflex) and acheive near-perfect reproduction over much of the audible frequencty range limited only by driver performance. Try a coaxial driver, can be larger up to 8" for more bass (and larger box). System will have nearly flat frequency response and linear phase over much of audible frequency range in midrange but with limited sound power. There will be no sweet spot (limited only by driver polar response). If you live in apartment and cannot play loud, and can do without low bass, this kind of system will do well on small-combo jazz and classical chamber music.  Observe if a coaxial it may have built-in crossover which may not work as well as no crossover.
Mechanical crossover ("wizzer cone") might be OK.

RIMO