advertisement


Active Crossover Frequencies

RyanSoundLab

RyanSoundLab

Trade: Ryan Sound Lab
I'm in the process of developing an active crossover with the built-in ability to change the frequencies by the user. It would be both 2 and 3-way capable. I'm interested in feedback on what the most useful frequencies are, particularly for the woofer-to-midrange of a 3-way system.


My thinking is that tweeter crossover points would pretty much stay within the range of, say, 2.5kHz to 4.0kHz (in maybe 500 Hz increments). Does that sound right?


For a 2-way system, that's where the woofer would also cross.


For a three way system, I'm thinking the woofer would cross at something around 250-500Hz. My Isobariks, for example, are at 340Hz or so. Perhaps increments of around 75Hz up and down from there?


Any input would be appreciated.


Kit
 
My objective is to make it a no-cost change that the user can do. Changing the number of poles (roll-off slope) is much more complicated and not practical without replacing boards, which is costly. Most of the active crossovers are either 3 or 4 pole filters and do a good job of rolling off the response outside the band. I'm settling on a 3-pole Bessel filter which can be designed to have constant power output (as opposed to constant voltage) so there are no dips in the frequency response. The actual design of these filters fills books and is not really what this thread is all about but if you want to delve into that aspect of design, please start another thread and let me know. I'll join!
 
RyanSoundLab said:
My thinking is that tweeter crossover points would pretty much stay within the range of, say, 2.5kHz to 4.0kHz (in maybe 500 Hz increments). Does that sound right?
Click to expand...

I’d try and have more range than that, e.g. vintage Tannoys crossover at 1kHz. I’d expect Altecs and most other large two-way horns to be similar.

Again if you have any interest in selling to the horn market the three-way crossover may need more range, e.g. a Klipschorn or La Scala crosses over at 400Hz and 6kHz.
 
Thanks much Tony! That's exactly the kind of input I need. Bigger ranges are possible if I can squeeze enough resistors onto the boards! I forgot about the big horns (often paired with 15" woofers, too). With the price of good quality Class D power amps coming down, more people may consider going active.
 
Hey Kit, I'm sure you have already looked but it might be good to cover the Naim IXO/NAXO/SNAXO filter shapes and frequencies? They quote cut offs and roll offs in the manual though not type, they cover all the Naim speakers but I believe some were made for Linn speakers too. Might be something you would want to cover? 18dB/octave 2.7 kHz and 2.8 kHz, 250 and 350 Hz seems to cover it for the Naim speakers.
 
My thinking exactly. Their 3-pole 18db/octave Bessel filters are just what I plan to use. I have an early prototype of that on my briks now and it sounds smooth across the whole audible frequency range.

Agree that it needs to provide for SBLs, Isobariks, and the rest of their two speaker lines but also much wider. Appreciate getting confirmation of their values. I may not get everyone's preference exactly in the stock crossover (i.e. probably won't have both 2.7kHz and 2.8kHz), but I can always offer to do a custom resistor pattern that covers a narrower range with more resolution for no extra cost. The goal is to have the overall design be flexible enough to cover just about everything.
 
An active crossover must be designed from the outset for a specific speaker. Universal ones such as this are, IMHO, as much use as a chocolate fireguard and I always turn down commissions to build active crossovers for precisely this reason (although I do have one going on at present but it is using Linn's own modules).
 
I've never met a chocolate anything that I didn't like.

Seriously, I've been doing a lot of modeling of the crossover circuits and believe that with careful design (e.g. mostly resistor and capacitor values) a constant-power output can be produced for a set of frequencies by just selecting different resistors. I'll be doing more modeling and then building to confirm this but the fall back would have to be sets of plug-in boards for specific frequencies to match a speaker - still a better option than an integrated motherboard with one design point as the NAXO's I've seen are. At least the unit can be easily changed by the user. The proof will ultimately have to be in the listening, of course.
 
Not sure I can see any point in building something hobbled(in terms of frequencies and slopes) from the outset when compared to stuff already widely available. Bespoke/aftermarket on the other hand is something that makes a lot of sense.
 
RyanSoundLab said:
I've never met a chocolate anything that I didn't like.

Seriously, I've been doing a lot of modeling of the crossover circuits and believe that with careful design (e.g. mostly resistor and capacitor values) a constant-power output can be produced for a set of frequencies by just selecting different resistors. I'll be doing more modeling and then building to confirm this but the fall back would have to be sets of plug-in boards for specific frequencies to match a speaker - still a better option than an integrated motherboard with one design point as the NAXO's I've seen are. At least the unit can be easily changed by the user. The proof will ultimately have to be in the listening, of course.
Click to expand...

IME once we're dealing with real speaker drive units all the theory goes out the window... What are the natural roll off characteristics of the drive unit? What peaks and troughs in the response of the unit need taking into account? What baffle step correction is needed? What phase issues due to driver positioning need addressing?
Designing crossovers that take all this and more into account and give good subjective results is MUCH more difficult than it looks and almost as much an art as a science IME...
For something such as you are suggesting the likes of Rane already have it sown up... then there's minidsp etc..
 
What Rane has sown up is the Linkwitz-Really design, 4-pole flat response that has big dips in it at the crossover points and not-so-great transient response. There is still room for different approaches with different sound qualities and that's what I'm looking at. I'm doing this mostly for my own satisfaction but I suspect others will want to listen and compare for themselves. I had nothing else to do this month anyway (haha).....
 
Out of interest, are general programmable crossovers generally analog or digital domain? I'm asking as it seems like a perfect fit for DSP based algorithms, especially as you can build filters with the desired time domain response, and be able to also provide delay compensation per driver to align them properly. Is it a question of not having the (rather different) digital domain skills to design such a product?
 
Re: digital vs analog approach. There's no doubt that the miniDSP (digital crossover) has amazing capabilities. It really seems to be the future of hifi (almost) here today. The design team for that must have a good number of software experts that can draw on the libraries of routines for filters, etc that have been developing over the decades. There's no way a single individual can put such a device together in any reasonable amount of time. I'm going to have to buy one and compare it to my prototype analog crossover to see what it sounds like. They are so cheap that, if it sounds great, I'll have to abandon the analog version. The analog dinosaurs like me will start dying out!
 
Interesting project ! Don't forget to make it suit Linn Kan (with the higher roll off.. ).It would be very interesting with active Naim N-sats also....
And the human ears are analogue for many years to come...analogue rules OK.
 
Linn Kans will be no problem. I was told they were originally composed of the mid and tweet from the Isobariks, so a frequency around 3kHz is probably about right. I couldn't find a reference to a xover spec on-line just now to confirm that. My friend is using an RSL prototype with N-Sats + homemade isobarik subwoofers(!). He used 80Hz/3kHz breakpoints. Sounds great!
 
You must log in or register to reply here.


advertisement


Back
Top