Please review the site Rules, Terms of Service, and Privacy Policy at your convenience. Rules, TOS, Privacy
Get familiar with the reaction system: Introducing the Reaction System

crossover tutorial

edited September 15 in DIY
School is winding down allowing time to work on ideas for all the drivers accumulated thru the past few years.  

Yea, I know, most of you guys here are versed enough in crossover topology not to need a layman's guide 101, but I do. 

In particular:
  • When to use rcl in series both before and after
  • When to use them in parallel legs
  • They're variations of both rcl networks, what are they used for? 
  • How to attenuate in band anomalies
  • How to attenuate out of band anomalies so the handoff is smoother.
  • How to widen or narrow the attenuation band     

Been playing around with xsim and pcd for some time and if the drivers are easy to work with things are fine, but a hiccup here or there and I'm in trouble.  Found a few answers by experimenting, but more often than not the sheer number of possible combinations do more to confuse than enlighten.    

First things first.  I have a knee in both a woofer (~3.5k) and tweeter (1.8k) slope that lies ~8-12db below reference that I would like to clean-up.  Where do I start?

I know you guys have your own slang for what is what so you may have to explain it a couple times ... or write really slow. 

  There are many more combinations than this (pcd), but for easy reference:


       

rjj45

Comments

  • Jeez guys, 22 views and no comments.  Are my questions that lame? 
  • No, I just don't have the time or energy to write you a book on crossover design. Analog circuit design was an entire course in my electronics education, not just a simple forum post online..

    Look up terms like "parallel notch filter", "series RLC filter", thing like that aren't "slang" but proper terms for the circuit elements. 

    Since it looks like you are playing with PCD, that's an easy way to learn how different components in different circuit locations affect the results. You don't even need to load any frequency response into PCD, just add components and look at the change it has to the transfer function. That transfer function is then applied to your frequency response that you load in. You may also want to load in an impedance file to see how a non-linear electrical load on the filter affects things as well.
    JasonPBilletrjj45kennyk
    My opinions are 100% factual
  • edited September 15
    Since Craig is doing a larger Peerless build I though about doing a budget build using the recently on sale Tymphany SDS-160F25PR01-08 and an nd25fa.  Not against using another tweeter just wanted to hear this one.  IIRC, Ben has commented about liking it crossed around 3-4k and I wonder if this might be due to a rather large knee that tends to lie ~12db below reference in some of the other designs.   

    More than likely, once finished these speakers will be give-mes' to someone at work, so they need to remain economical.  Keeping that in mind, crossover parts can add up quickly.  The 2.8mh is going to cost as much as either driver.         

    Dinkn' around.  Needs work.     



     

     



  • dcibel said:
    ... Since it looks like you are playing with PCD, that's an easy way to learn how different components in different circuit locations affect the results. You don't even need to load any frequency response into PCD, just add components and look at the change it has to the transfer function. That transfer function is then applied to your frequency response that you load in. You may also want to load in an impedance file to see how a non-linear electrical load on the filter affects things as well.
    100% this. If you want to learn how a crossover filter functions without the electrical college class, this will get you there.
    = Howard Stark: "This is the key to the future. I'm limited by the technology of my time, but one day you'll figure this out."
  • I think the comment is spot on. By looking at the filter response you can see how different components change the filter then the response. I think the filter response graph is necessary for messing with notches/ RCLs.

    I want to add that correcting every blip is a waste of time imo. Start simple and check off axis. If you are using omni mic or other real time FR measurement, pick up the mic and move it around, you will see the issues pretty quickly.  I learned that the hard way rushing through a few designs...
    Billet
     John H, thanks to JP I did get that email
  • @Kornbread ; - no I don't think your original question was "out of line".
    Various experts have dropped hints and tips over the years, but those have never been gathered and written up. For instance, someone said (I could very well have this backward)- use a parallel RLC for in band anomalies, and a series RLC for out of band anomalies.
    Nice response in the sim, but I'm not sure whether you need the RLC traps on both drivers.
    I always strive for simple, minimal component crossovers.
    Don, Donno, or "Hey you" all work for me, But never "Mr Johnson"
  • Is that Xsim simulation done with in box measurements and a measured z-axis offset?  If not, you may be chasing ghosts.  And you can pick up a 3 mH iron core at PE for $5.60.  Not cheap, but not overly expensive either.
  • I had a thought, and even started a while back to make a few writeups on design work, the focus was going to be on taking measurements correctly and with minimal tools but eventually moving into the actual crossover design. The idea was to make some good reference documents that can help answer a lot of the beginner questions that get asked repeatedly, explaining that simple questions can have complex answers and to provide a resource to point people to who want to get started into actual acoustic design instead of the response they usually get of "just build a kit".

    It turns out that making a detailed instruction that has the right flow of information and doesn't leave a lot of questions unanswered is a lot of work, so I never did get back to it. Maybe over this winter I'll get motivated again.

    rjj45
    My opinions are 100% factual
  • Guys, I am trying. 

    And no, those are 'factory' files.  The box is simmed but I still need to look at Edge and better understand combing before settling on given proportions.  At this point just trying to be sure the drivers should play nice with each other before cutting wood. 

    @rjj45 ; Different sim but it is an attempt at removing some knees that lay just below the reference level from the crossover slopes.   

     


    [youtube]



  • @Kornbread ; - yep -- those out of band "knees" have caused me more than a little trouble at times. 
    IME, you can sim a whole lot, but once you put the drivers on a baffle, the baffle diffraction will swamp
    the ripples you are trying to smooth now in your sim. 
    Some guys do a lot of sims. I do measurements in box, then import those into Xsim. 
    More than once, I've gone "down the rabbit hole" in simming and tuning a crossover, thrown out all the
    sims and variations, and started from scratch.
    Hope this helps.
    Don, Donno, or "Hey you" all work for me, But never "Mr Johnson"
  • If we don't work this weekend, I'd like to start cutting wood. 
    rjj45
Sign In or Register to comment.