Northcreek 2nd order crossover magic

[Klimon]

Upon designing a crossover for a pair of northcreek D28's I've stumbled upon something very intriguing; they recommend a 2nd order filter using a different ratio between C and L as the well-known 12db slopes: C*L = 0,350 *10(-9). George Best courteously answered my email saying that they determined this to be the best sounding ratio. The actual choice of C and R is a function of net tweeter impedance and L-pad; higher impedance = smaller C & bigger L // lower impedance = bigger C & smaller L....

Using the 0,350 ratio George says for the D28 a 1651hz crossover point requires 11µF / 0,315mH filter. This would be a suitable crossover point for the D28 in my speakers also, but I don't know how to calculate the formula to determine low-pass crossover point at 1651hz for my woofer, with 0,350 ratio. The extra variable of driver impedance makes it all the more difficult to recover the magic formula, the tweets are 3.6ohms, the woofers 5.6.

I hope someone gets a clue of what I'm saying here, it would be great to combine crossover point, driver impedance and C & R values (with C*R = 0,350 *10 (-9) in one formula to make the northcreek ratio as workable as the classic 2nd order filters.

Simon

[Zaph]

To be clear, that's not 2nd order... it's 4th order. The number of electrical components is not what determines the order, it's the resulting combined acoustic and electrical rolloff.

The resulting values are not (or should not be) determined with standard "cookbook" formulas. No calculator, pencil and paper were (or should be) used. They are modeled with software to achieve a target slope. Basic formulas do not consider all the variables such as varying impedance or natural rolloffs while modeling software does.

My advice: throw away those equations. The magic formula is to get Soundeasy, LSPcad or Speaker Workshop. Then get a microphone, preamp and construct a impedance measuring jig.

You'll have more questions, which I'll let others answer.

[Klimon]

That's a good starting point, Zaph. I've done some reading meanwhile and I think the slope of the tweeter roll-off (at around 1,6k with second order filter) is rather 18db/ octave. My woofer is a 6,5" alu-cone unit so I'm aiming for a third order slope at 1,6k to get the cone-breakup silent.

Is it possible to obtain a 18db/octave slope with a second order electrical filter and without 'using' the natural roll-off of the woofer (e.g. using different L*C ratio)? I've found only this reference that points to this direction (wikipedia): "Third order acoustic crossovers are often achieved with a first or second order filter circuit." but am not sure how to achieve this, if it is possible at all....

Simon