4 Way Crossover

Thanks JBL. I've heard of two versions with this regard. One said add a tweeter range to the system and take it as a super-tweeter. Others said add a bandpass as you have mentioned.

I've built a prototype of the first type. It sounded well with extended range (for the particular driver set) that's comparatively superior to the three way system I used to have. And,music is there when played loud. The reason for the choice(of the first) is such that three way mathematical models are available whereby four way system is not.

I'll build the other type for comparison later.Thanks again!

Peter
 
Dear KCC

Designing a x-over, no matter if it's 2, 3 or 4 way, requires the same principles - all depends on your drivers- how good they are, how smooth their roll-off if and so on. Usually good drivers (Dynaudio, Vifa, Scanspeak etc.) have very smooth rol-offs at both ends of their bandwidth, which means they are will work very well with low order filters - e.g. check Dynaudio's site and you'll see that they recommend 1st order filters for all of their drivers. Another major point - the better the drivers, the easier is to design a crossover (of course you will need to use Zobel networks for the woofer and maybe the midrange, L-pads for most of the drivers, except the woofer, and possibly notch filters for some of the drivers if there's a need for it).

I always go for 1st order x-overs (considering your drivers can cope with it). They are easier to implement, create no ripple in the pass-band, phase linear, so the speakers could be time-aligned if it's a design goal, in general many, many advantages, but only if you drivers can tollerate all this.

Some of the most sophisticated and highly acclaimed speakers (Duntech, Thiel, Sonus Faber, Sonique, Dunlavy - ex-Duntech creator, Dynaudio) all use 1st order filters.

In short (Ha-Ha-Ha) the advise is: choose your x-over freqs carefully:
woofer - up to 300 - 400 Hz
midwoofer - 400 - 1000Hz
midrange - 1000 - 3000 Hz
tweeter - 3000 and up

These are only recommendations of course, everything depends on your drivers, your design ideas and your taste. Don't forget to check the impedance of your speakers regularly while trying different options, cause the impedance curve can tell you many, many things about the behaviour of your design

Regards

Stan

P.S. If you have any questions do not hesitate to ask
 
Hi Stan,

Thanks for your kind advise. I liked to use first order crossover as well. I have a pair of SEAS speakers employing 6 dB crossover with impedance equalization and filters.They sound fine. I also have speakers of higher order crossover to listen to. Admittedly,I did not find myself prefer one crossover type to the other. They have their characteristics to be appreciated,however.

I placed the question of 4 way crossover for the fact that I have a set of older Philips drivers that does not sound well in mid
range in three way system. Although it's impedance curve(of the 2 inches dome midrange) is flat to 400 hz but sounds "honky" at music level. And,no amount of filters and equalizations can help. I decided to move the crossover point to 2100 hz and place a midbass in between.This raise the question of four way crossover topology. Unfortunately we do not have studys done at this area this point.

Nevertheless,I've received many advises from many kind people.In fact,it's Mr.R.Bullock who suggested the higher crossover point treating the tweeter as super tweeter.

At its presently form,the crossover point for woofer is 350 hz,midbass 2100 hz,midrange to 4900 hz. The presentation is dynamic with details and space not found previously. I listen to music - see the picture - not speakers.

I'll continue improving the system. And,as you know,never is there an end to this. Thanks again Stan!


Peter
 
Or spare yourself the pain and go active using a line level active crossover (for example, from ESP in Australia) or digital (for example. miniDSP or the unit from Tranquility Bass on this forum). Ultimately, you might find it easier, espaecially teamed with a Class D amp, such as a 4 x 100W from Sure.
 
Your best bet might be to grab a copy of XSIM and add 4 of the same drivers to the schematic. Then do simple crossover for all 4 drivers and look at the changes in frequency response and impedance. I think that would answer a lot of questions.
Thanks mlee for your advise, i have tried Xsim but the number of frd files is very limited. Would you pls let me know whêre i can get them? I have some files from Impulse Audio only, but you know, it's very few and many of them are not the drivers i need.
 
You need to measure the drivers on the baffle & save the frequency response into frd file format & the impedance into zma. The response will be wildly different to the manufacturer's published responses due to baffle step & diffraction effects. The alternative (nothing like as good) is to trace the manufacturer response graphs providing you know they are accurate / representative & the conditions they were taken under (if you don't, they're worthless), then reprocess the data to simulate & add in step loss, diffraction etc. & extract minimum phase etc. with appropriate software. Note that the crossover design software you subsequently use will then need to be able to account for the driver positions itself.
 
You need to measure the drivers on the baffle & save the frequency response into frd file format & the impedance into zma. The response will be wildly different to the manufacturer's published responses due to baffle step & diffraction effects. The alternative (nothing like as good) is to trace the manufacturer response graphs providing you know they are accurate / representative & the conditions they were taken under (if you don't, they're worthless), then reprocess the data to simulate & add in step loss, diffraction etc. & extract minimum phase etc. with appropriate software. Note that the crossover design software you subsequently use will then need to be able to account for the driver positions itself.
Thanks for your advíce, I have a omnimic device from Dayton Audio. From what i understand I wil lneed to hung the driver up and measure (1 meter away) their frequency range again using that device and save them in frd files, but for zma files, would you pls let me knoww how i can get it?