(LONG!)Crossover Design - 2nd Order Butterworth with attenuator


Yup - thats me.. the mad man. From knowing nothing about loudspeaker design to trying a 2nd-order crossover. Anyhow, Ill explain what I have thus far, and what I understand of it. Please correct me where Im wrong, or help me out.

Using lalena.com I have come up with a 2nd Order Butterworth crossover, and Driver Attenuation circuit for the tweeter. For the main crossover, it will handle an 8ohm tweeter with 8ohm driver and crossover at 2500Hz.

In series with the tweeter is a 5.63uF cap, and in parallel with the tweeter, after the cap is a 0.72mH inductor. Similar for the low-pass section, with the inductor being in series, and the cap in parallel after the inductor (same inductor and cap as for the tweeter tho).

My understanding is that the cap for the tweeter will 'roll-up' from essentially nothing to letting the higher frequencies pass with no resistance. The inductor in parallel then lets some of the energy (or current I suppose) from the cap escape an 'easier' route than going through the tweeter. Thus with the cap and inductor combined, I think I see how a faster rolloff is acheived (12db/oct or such for a 2nd-order?). Similar concept for the driver, but in reverse since we're wanting a low-pass filter.

I then have an L-Pad attenuator which has one resistor in series, and one in parallel with the tweeter. The way I understand this to work is that the resistors will offer resistance across the whole frequency range, thus making less current flow thru the tweeter - ie: 3db attenuation with the correct resister values. I do notice that the little L-Pad as a whole gives 8ohm total resistance... which is what the aforementioned crossover was designed to expect. Thus I surmise that the 1st resistor in series is there to help balance this circuit back to 8ohm as the resistor in parallel is essentially bringing the total resistance down when in parallel with the tweeter.

This attenuator I will place where the tweeter would be in the crossover as the only frequencies I want toned down are the ones that make it thru the high-pass and into the tweeter.

I hope I am making sense so far. I think I am. Now I want to move on to complex resistance (ie: impedance) and phase.

Limited talks with a guy at work who did an electrical engineering degree many years back has given me some ideas on this area. Impedance is (to my mind) resistance that varies with regard to frequency (ie: complex resistance).

Now first off, a driver would have complex resistance - and if wired straight to the amp, it makes the amp work harder, and easier depending on the resistance it is offering at the frequency being played. Add into this mix a crossover before the driver, and I reckon theres an issue. Can someone explain what happens when at a given frequency the driver well exceeds its 8ohm's (around which the crossover was designed)? To my mind the current may well decide that the alternate routes in the crossover are easier than going thru the driver and as such the crossover makes the situation worse than if there were no crossover!

Furthermore, the components in the crossover offer phase issues. Ill from here on Ill reference phase as the approximate phase differences between driver and tweeter. Ill reference the phase distortion within frequencies produced by the same driver as time smearing. (How correct this is I dont know, but I recall these terms online somewhere).

With crossovers offering varying degrees of phase distortion, and timesmearing - as well as each driver/tweeter offering its own time smearing across the reproduced frequency spectrum, how does one compensate for this? I assume since crossover generators do not show them, that parts do not exist in reality which adjust phase only (no resistance).

Thus I am lead to believe that if one is to remove as best one can phase distortion, one needs to bring the overall sensitivity down due to all the increased resistance needed to compensate (thus I ask how highly sensitive speaker designs can be phase coherent?)

Futhermore if one intends to counter time-smearing (ie: phase distortion within one driver), essentially that driver (with phase correcting circuitry) will increase quite some from its initial resistance estimate if it were on its own.

I find the whole scenario of working with simple resistance an simple formulae an unusually simplified view of what seems to be quite complex (pun intended!).

If you have read this far, I congratulate you (and you have my thanks). If youre wondering what this guy is smoking, do be aware that 4 weeks ago I could not get my mind around what the difference was between volts and amps. So, as with any area where a person tries to attain a lot of knowledge in a short space of time, it gets a little garbled. Your comments welcome. Site recommendations welcome (although I have read a great many sites on crossovers).

Try not to recommend books. If I had the money to buy them, I would rather invest in powertools (since I dont have any :).

PS: If you could also throw in a word for my buying parts -- 10v (5W?) resistors for the attenuator... good enough?


 
i will answer to the easiest questions...

GeniX said:
Can someone explain what happens when at a given frequency the driver well exceeds its 8ohm's (around which the crossover was designed)?

This will change the crossover point computed. Complex or very cheap drivers need a impedence compensation to ensure the accuracy of the crossover network.

If you have read this far, I congratulate you (and you have my thanks).

Thanks :)

PS: If you could also throw in a word for my buying parts -- 10v (5W?) resistors for the attenuator... good enough?


[/COLOR] [/B]

These values depends on the power handling of your loudspeaker.


As a global conclusion, I always recommend simple networks because complex networks imply a lot of side effects that are hard to control and design around.

Try first a simple network, test and listen, and add components (attenuator, LPad,etc) to do some comparison. This is the best way to learn.

When the passive crossover network becomes too complex, it is easier to use an active crossover :)

Good luck