Anyone have X-Over Pro? I need a XO designed?

[22mountaineer 10+ year member]

Where do you guys learn this stuff? I remember a little about it from college but nowhere near learning enough to build something like this,

 

[BrianChia 10+ year member]

Where do you guys learn this stuff? I remember a little about it from college but nowhere near learning enough to build something like this,

Learning to use a computer simulation program helps a lot. There's no way I could design it from scratch and expect it to sound good.

 

[joetama 5,000+ posts]

Learning to use a computer simulation program helps a lot. There's no way I could design it from scratch and expect it to sound good.

Oh indeed...

Computers make it much easier...

 

[thylantyr 10+ year member]

You can do standard crossover design, text book, and get very good results. Don't be surprised if high end manufacturers use this method, because they do.

Now that computer software does more, ordinary people can learn the software

and have the software optimize the design to make it technically perfect. But there

is not guarantee that a technically perfect crossover will sound better than something

else that isn't optimized. You can measure and collect data to see which one is more

politically correct, but you can't guarantee that a human will prefer that crossover 100%.

Lets say you want to make a simple off the shelf crossover.

Calculator example;

http://ccs.exl.info/calc_cr.html

They show Linkwitz-Riley, Butterworth, Bessel filter types. You can read about

what each one does. After you mess around with these, you migh have favorites

that you fall back on. For example, if I want to use a Butterworth filter, I like the sound

of 3rd order. If Linkwitz-Riley, I would use 4th or 8th order. For non-optimized crossovers,

people may use the same order throughout the design to ensure the phasing is good,

but sometimes you can cheat. If you want 2nd order on the mids but also want 3rd - 4th

order for the tweeters, then try it out and compare the difference. Usually you can

do a simple test and compare two different crossover designs.

This is the way I see it.

1. You use software to optimize and make a perfect crossover, no guarantees on SQ.

2. You use a generic crossover and be done with it, or do few tweaks as best you can,

no guarantees on SQ.

From what I have seen in cyber, this is how people approach this. They spend a lot of

time using software to make the cool graphs. They are satisfied that the crossover is

sweet. They build the system and accept the performance. They may even measure the

speaker and do more tweaks. These types of people are really anal about getting it

to measure perfect, I sometimes wonder if they even listen to the speakers as they

seem so focused on the technicals.

Then you have the type of person that does listen to the speaker and makes tweaks

as needed, some of these tweaks may sound good, but not measure well, but they

have a better speaker because they like the sound better.

Tony Gee has a good example on this method. Read this article.

http://www.humblehomemadehifi.com/Phlea.html

He added a special filter to the PHL midrange because it had some anomolies, but

in doing so he felt that it robbed the speaker of a sound character that he liked,

so in the end, he removed that circuit.

There is another way, IMO, a very simple way that anyone can make great speakers.

You setup a small test bench with a cheap amplifier, DCX2496 digital crossover and

cheap source. The DCX is your ally, use it wisely.

http://www.behringer.com/DCX2496/index.cfm?lang=eng

filter types (Butterworth, Bessel and Linkwitz-Riley) with selectable roll-off characteristics from 6 to 48 dB/octave.

Surprise! -> Butterworth, Bessel and Linkwitz-Riley

Does that ring a bell ?

Connect a midrange up to the test bench and set a band pass in the DCX.

Whatever you want. Try high pass 60hz - 300hz to keep the bass out. Then

sweep the low pass from full range all the way down to where the tweeter may

take over. Lets say 1khz - 5khz range. Listen to that 1khz - 5khz range and determine

what is the highest you can go where it still sounds good. Use this data to help

you figure out your tweeter crossover. Connect the tweeter up and see if you can

get the midrange and tweeter to have good synergy. You also need to adjust the

gain/levels of the DCX or amplifiers to get the SPL levels to match well.

You can also make a test box if you are really serious about this. All you are doing is

getting data and listening to music to understand what the drivers can do for you.

Use that data and if you want, try to make a generic passive crossover.

Lets say you realized that you like Linkwitz-Riley 4th order on the mids and tweeters

at 2.7khz because that is what the test bench revealed to you as being the best sound.

Now you have data, plug the data into the calculator and make the passive crossover.

This is better than doing it blindly, you now have zeroed in pretty well on what you like.

Add the Zobel to the mids if the impedance curve is wacko. Add the LPAD if you

need to match levels. etc. BSC is an issue that can be tolerated or you can design the

speaker with an extra midwoofer to compensate, etc.

But if you do mess around with a DCX, you may not want to use a passive crossover,

you may love the idea of tweaking the speaker and using all the other features it has.

If you did want to use the DCX, fully active setup, there is no need for a zobel,

no need for an LPAD and the BSC issue is so easy to implement in the DCX in the

form of an EQ setting, a shelving filter that you can adjust and dial in very quickly by ear.

 

[BrianChia 10+ year member]

Great outline on xover design. I've used most of the approaches that you've listed, except I've never had a DCX. I usually start with textbook, optimize on software. Then listen, measure, tweak, and listen.

Textbook designs will work if planned carefully but you can usually maximize the potential of the drivers by optimizing. When I optimize, I'm typically less concerned with obtaining a perfect flat graph but I try to simplify layouts and save money in components.

I agree with Tony Gee's philosophy of how it sounds vs. how it measures. FR plots/RTA doesn't tell you everything. But I also do not build speakers that are near the caliber of what he does. Most of his projects are really anything but humble. //content.invisioncic.com/y282845/emoticons/smile.gif.1ebc41e1811405b213edfc4622c41e27.gif

 

[mavster 10+ year member]

hey Brian, what was that other crossover design about?

thanks

 

[maldecido33 10+ year member]

i use a pen paper and calculator to design my xovers

really quite easy