How do you design a crossover??????

[legendaryfrog]

for me, it just about seems impossible. how do you choose all the parts and stuff? could someone "fill me in" on this?

 

[Cal Weldon]

Frog,

Crossovers are complex. It takes years to get good at making them. You can't be just "filled in." It's like asking "can someone fill me in on how to tear down an engine?"

Here's an XO calculator. It also shows you how to do box sizes etc. That should help.

http://www.mhsoft.nl/spk_calc.asp

 

[heater]

This "filling in" could take some time not to mension expense.

I suggest you first read this: http://sound.westhost.com/lr-passive.htm

Then, if you are like me, you might decide that going active might be easier. As for example here http://sound.westhost.com/lr-passive.htm

Then when you have understood all you can you start experimenting and probably come back here to discus why things don't work out as planned

Still, nothing ventured, nothing gained.

 

[Daveis]

Prototype first using multiple amps and an active crossover.
Once you are satisfied with the sound, design the passive crossover using software like Xover Pro, etc. Or pay/commission someone at Madisound or online to build you one.

Multiple amps, a volume control solution, and a Behringer DCX2496 and 15 minutes can result in a passable solution.

Dedicating a PC as crossover can also work. This is expensive.

Main thing is to have the box dimensions right. For the beginner (me included) I stay away from vented boxes. I just use an oversized sealed box. Amazingly, a sealed box that's too small will only make your bass roll off prematurely. It won't necessarily sound bad.

But if you want a sure thing look for Jon Marsh, Zaph, etc and build with a proven winner.

 

[Cloth Ears]

Then, if you are like me, you might decide that going active might be easier. As for example here http://sound.westhost.com/lr-passive.htm

Did you mean http://sound.westhost.com/biamp-vs-passive.htm?

 

[heater]

What I actually meant was the 24db/octave active c/o project itself:http://sound.westhost.com/project09.htm

Your link is Rods excellent introduction.

 

[Geek]

Crossovers are complex. It takes years to get good at making them. You can't be just "filled in."

Indeed!

I ended forgetting the calculator's and spreadsheets and just setup a test mic and sweepgen and breadboarded until I got it right, both on screen and sound wise.

 

[Pano]

Black Magic is best - or Voodoo. (Unlike speaker cables where HUNA is best.)

Before you begin - be sure to lay in a good supply of chicken lips, toad ears and bat feathers. These will greatly aid your calculations.

There are software crossover calculators available, but they work best by the light of the new moon. Burn a green candle, if you don't have new moonlight.

Components made of cryogenically treated unobtainium can really simplify your calculations and curses, as they tend to receive and transmit ethers and ectoplasm extremely well. Your ears will know the difference!

But don't expect results overnight. And of course - YMMV. IMHO.

 

[hermanv]

for me, it just about seems impossible. how do you choose all the parts and stuff? could someone "fill me in" on this?

Designing a crossover is a complex operation. Many do their own design, it can be learned.

The following is very general, it is meant to provide a bird’s eye view of the process rather than being an instruction guide. (Hopefully, I got it right).

Item 1. You need to decide on the crossover order, i.e. 1st order, 2nd order, 3rd order or 4th order. Each order increases the slope of the attenuation past the crossover point by 6 dB per octave. i.e. 1st order is 6 dB/octave, 2nd is 12dB/octave etc.

How is this decision made? The driver behavior out of band mostly determines this. The newer ultra stiff exotic coned drivers tend to have very bad out of band break-up characteristics and usually work best with steeper crossovers. These crossovers steeply reduce any out of band energy reaching the driver. On the other hand gentle slopes are usually preferred if a phase flat or wide dispersion design is your goal. Low order designs use the fewest parts.

Item 2. You need to decide if a Linkwitz-Riley or conventional crossover (usually Butterworth) is preferable. This is important because conventional crossovers have a phase of plus and minus 90 degrees at the crossover frequency which means the crossover point must be 3dB down, whereas Linkwitz-Riley crossovers have a plus and minus phase of 180 degrees at the crossover point which means the crossover frequency point needs to be 6dB down. Likwitz-Riley crossovers have very good phase characteristics and are often preferred for 1st and 2nd order designs, the advantages decline as the order number increases.

At the crossover frequency both the lower and higher frequency driver each contribute ½ of the acoustic output. Summing the two acoustic signals requires the correct phase for each driver. Depending on crossover style and order(slope) chosen, the drivers might be connected in phase (+ to +) or out of phase (+ to -).

Once these decisions are made any number of component value calculators are easy to find on the internet.

The Loudspeaker Design Cookbook by Vance Dickerson describes most of these points in fine detail. Many have learned the craft from this book.

 

[Bigfootpete]

how do you choose all the parts and stuff?

It might help if you have some electronics knowledge...

You can do a search on the web for capacitors and resistors etc.
And there are a few threads on here where people recommend components.

There are also lots of websites with tons of info about crossovers...Search for butterworth or Linkwitz Riley.

 

[Cal Weldon]

Maui Man,

That must be the way they do it on the southern tip of the island. I've noted that in places like Kihei and Lahaina they tend to stick to the more conventional means.

Are you sure you're not up Haleakala where virgins are still thrown into the crater?

Herman,

Good advice but I wonder if that's a little advanced for him.

Pete,

Good advice. I think Frog should spend a little time researching. It will give him a better idea how involved XO building is.

Frog,

You owe it to yourself to Google. It is good to know about resistance, impedance, inductance and capacitance before heading into the abyss.

 

[PeteMcK]

Design Crossover

Choose crossover freq

Choose xover order and type
Choose series or parallel
Choose whether to zobel or not

decide tweeter attenuation
Decide BSC level

Calculate component values

Design notch filter if there are any peaks which need additional filtering

Try, measure, and tweek…..

 

[Cal Weldon]

Pete McK,

Again good advice but I think Frog has to go and find out what all those things mean first.

 

[Cloth Ears]

Hi L-frog,

Cross-overs. Hmmm. Difficult to know where to start. Even the Rod Elliot stuff (from soundwest) and Vance Dickerson's cookbook assume a bit of knowledge.

And I don't know what your knowledge is of electronics (or DIY) is either - although you do seem to be asking in the right place. Have you done a search on this subject within DIYaudio (I found a couple of threads: http://www.diyaudio.com/forums/showthread.php?threadid=88567&highlight= & http://www.diyaudio.com/forums/showthread.php?threadid=80560&highlight= & http://www.diyaudio.com/forums/showthread.php?threadid=40648 & http://www.diyaudio.com/forums/showthread.php?threadid=53031&highlight=)

If you have no knowledge, the the easiest place to learn about what they do would be from a David B Weems book (he's written a few - probably in your local library). They're not neccesarily the best for hi-fi, but they explain it in simple terms. Then move on to Rod Elliot and Vance Dickerson.

There's also some members here who have good sites to learn from. Andy Graddon has a site on crossovers, Tony Gee (Geenius) has a site on DIY speakers, which includes some crossover design and John Kreskovsky has a site also - and these are all for series cross-overs. Parallel cross-over information is available in many places.

 

[hermanv]

Herman,

Good advice but I wonder if that's a little advanced for him.

When I was young I knew everything, as I got older, it turned out I knew very little, now older yet, I am learning.

It's a journey, you got to start somewhere.

Try this resource
http://ldsg.snippets.org/idx.php

 

[e-side]

Re: Re: How do you design a crossover??????

Likwitz-Riley crossovers have very good phase characteristics and are often preferred for 1st and 2nd order designs, the advantages decline as the order number increases.

Only first order Butterworth filters exist.

You should use Linkwitz-Riley for even order filters.
Butterworth will give the best results when an odd order is used.

Have a look at http://www.trueaudio.com/st_index.htm > "An Overview of Crossovers (- Part II)" for basic info about x-overs.

 

[hermanv]

Re: Re: Re: How do you design a crossover??????

Only first order Butterworth filters exist.

You should use Linkwitz-Riley for even order filters.
Butterworth will give the best results when an odd order is used.

Perhaps a language or translation problem?

Butterworth filters can be made for any order. I think you meant all first order filters are Butterworth filters. I did mis-speak, I don't think a first order Linkwitz-Riley is possible.

"Should" is subjective, not everyone agrees about the "best" crossover design. Although L-R crossovers have real advantages they are not automatically the only answer for a given design or application.

Because the drivers once mounted in their enclosure exhibit their own frequency amplitude and phase characteristics, any automatic selection of a crossover style may be flawed. In the end, it is the summed acoustic response we presumably want to be flat that determines the best crossover solution for a speaker design.

I does get worse, many designers intentionally color the speaker response to make it more pleasing. Absolute neutrallity might not sound best for a given design.

 

[e-side]

Re: Re: Re: Re: How do you design a crossover??????

Perhaps a language or translation problem?

Butterworth filters can be made for any order. I think you meant all first order filters are Butterworth filters. I did mis-speak, I don't think a first order Linkwitz-Riley is possible.

"Should" is subjective, not everyone agrees about the "best" crossover design. Although L-R crossovers have real advantages they are not automatically the only answer for a given design or application.

Because the drivers once mounted in their enclosure exhibit their own frequency amplitude and phase characteristics, any automatic selection of a crossover style may be flawed. In the end, it is the summed acoustic response we presumably want to be flat that determines the best crossover solution for a speaker design.

I does get worse, many designers intentionally color the speaker response to make it more pleasing. Absolute neutrallity might not sound best for a given design.

Yes, I meant that butterworth's the only possibility for a first order filter.

I agree that a 'flat response' doesn't have to give the best sound. FR won't be flat at all when listening to your speakers in a room.
However, I think in theory a flat response would be desired. That's why I referred to Linkwitz-Riley for even order filters.

Of course everything has its + and -.

 

[legendaryfrog]

thanks. ill wait till next year when my school's speaker building courses begin. For now ill stick to full range. (no crossover is the best crossover!)

 

[Daveis]

No crossover is better than crossover? I wanted to think that, too.

Most of the music I listen to and due to my hearing range I would need a fullrange driver that could cover the range from 40hz to 16khz. If only such a beast existed...

Even when a speaker can produce on-axis frequencies close to full-range it always seems to suffer off-axis.

My BG RD40 planars are supposed to be good from 250hz to 18khz. The reality is that they sound best from 400 hz to 10khz. I thought I would be able to get away with a 2-way design under 400 hz. The reality is that I discovered that it sounds much nicer with a dynamic ribbon on the top crossed in at 10khz or ever as low as 4khz, depending on the music.

I'd suggest that you compromise and use a fullranger and cross in a standard tweeter at 4khz using a 1st order Butterworth. You should have plenty of overlap in the FR of both drivers to do that.