No announcement yet.

LR or butterworth filter, which one and why??

  • Filter
  • Time
  • Show
Clear All
new posts

  • LR or butterworth filter, which one and why??

    Provided Link: AJ crossover equations

    I'm building a 5.1 HT system with the front mount dayton neo and the tang band W4-654S. I found a crossover calculator on line (see link below) and it give the option of Linkwitz-Riley, butterworth, bessel or chebychev crossovers. I know most people just use the first two mentioned so I'll stick with that, but which one? what's the difference?

    still new at this, sorry.
    thanks, Aaron Garmon

  • #2
    Some people are addicted to Vicodin. I'm addicted to speaker building.

    The Chorales - Usher 8945A/Vifa XT25TG Build
    ESP Project 101 Lateral MOSFET Amplifier
    LM4780 Parallel Chipamp
    Sonata Soundbar Project
    The Renditions - Active/Passive Towers


    • #3
      In layman's terms . . .

      First, everything said above and here assumes that the response of both drivers is basically flat for an octave or so beyond the cutoff frequency. If that's not true, the summation of your electronic filter (what you build) and the natural rolloff of your driver(s) will be an *acoustic* rolloff very different from the ideal rolloff predicted by the equations.

      Okay. For you, the basic difference between between the filters you mentioned is the way the intersecting curves will add together. The LR adds flat. Very easy to visualize this. Under the same conditions, the BW adds a 3 dB hump with its peak at the XO frequency. Very bad if your curves are flat, but perhaps useful if the unfiltered curves have dips at f(xo) or have already started to rolloff (hopefully to about -3dB at f(xo) ). You can also adjust the humps and dips by choosing slightly different cutoff frequencies for your two drivers, but this is a little harder to visualize. Not impossible, but you'll need to stare a lot.

      There are also phase issues involved in your choice, but without serious software or a penchant for parabolic and logarithmic equations, you'd do best to ignore them.

      Fact is, once you get hold of that serious software, you'll say goodbye to all these dudes and calculate component values based on what yields the best combination of a flat summation and phase integration. The resulting curves might look like a BW or LR or whatever, but the values you got there will have nothing to do with textbook equations.

      Best advice for a quick, fun DIY project. Select drivers that will cross with room to spare somewhere in the low 2000s. Pad the tweeter as needed so the average SPL for both drivers is the same or so. Zobel your woofer. Calculate a LR XO using the textbook calculator. Your f(xo) is a function of L * C, so when you select your actual components and have to adjust to the closest available values, if one must go up, the other must go down.