Tips! Volume 3, By Ben “Wolf” Shaffer; 08/01/2010.

Explanations:
The Actual Placement in reference to a xover schematic,
For: Parallel Networks (A) and Series Networks (B) as used in Jeff Bagby's PCD7.

If you follow Jeff’s tutorial for this part of the design, you probably already know how to do most of this. However, in the case that something is still not clicking for you, maybe this will help!?

Section A

Let’s start with the Parallel Xover style…..
Sometimes you have to use notches or ‘tanks’ to get the response you desire, and not use a ton of parts in the long run. Since my “MAX” design is to the Nth degree on parts, I figured it would make the best visual with what you can accomplish in PCD.

If you look at this filter with fresh eyes and/or little experience you’ll probably just pass out from brain implosion, but looking carefully, it isn’t all that hard to comprehend.
Here is the schematic:



Now- before you faint- Look at its layout. It looks essentially like a 3rd order on the woofer and
5th order on the tweeter (plus some additional parts to be more complex). It has your standard
Coil-cap-coil arrangement on the woofer and Cap-coil-cap-coil-cap on the tweeter. Not that difficult to understand yet, so let’s make the other parts come into play.
The LCR that is aft the circuit in the woofer is to tame the upper breakup on the M8n. This particular driver has 2 intense breakups to minimize. One is about 2.1 kHz just above the xover frequency, and the other is about 7 kHz.



The LCR tames the 7 kHz spike alone, and the dual-tanks (0.33uF/2.2uF) and damping resistor (6.0 ohm) between drop the knee* at the xover freq and attenuate harshly above it. Sometimes the single tank and the damping resistor is enough, but it was not this time. The spike at 2.1 kHz is about 8-10 dB above reference level and is definitely going to be audible. Most designers like the resonances in question to be -25 dB or less in audibility than the nominal output level of the system, so that is something to attempt to achieve. You can’t always do it cheaply or easily, but sometimes it makes or breaks the system as a whole.

Here is how this is setup in PCD:



To make the notches and tanks be where they are in the circuit, you have to get a bit creative on placement of parts. To make the resistor infinite, it’s 99999, or short would be 0.00. Since we don’t want to short the maximum effect of the tanks, 99999 is used. DCR is placed where the coils would normally be, but to make them tanks or notches, they need to go before and aft the xover central node. That means to not be redundant, the normal coil positions would be set to 0.00 as well.

The combination of the Acoustic rolloff of the M8, and this electrical filter yields a Linkwitz-Riley 8th order, or “LR8”. LR, Bessel (BS) filters are even-order, and Butterworth (BW) can be either even or odd order.
Since the one-order shift allows for better phase alignment from offset-Z between the tweeter and woofer in virtually all situations; using a 5th order electrical on the tweeter along with its 2nd order inherent acoustic rolloff yields a total 7th order acoustic slope. This is one order of difference from the woofer, and will make phase alignment easier.

Now on the tweeter (DX25)-
Since it has a response that is fairly ‘hilly’, I tried to do the best I could with it. The DCR for the 2 coils might seem a little high because it is! By adding resistance to the coils here, you get a damping effect at the rolloff of a section of the xover (just like you do with 6 ohm above in the woofer or a typical zobel) at the knee* that corresponds to it.

*Maybe I didn’t mention this yet, but a ‘knee’ is the transitional-area of the driver’s rolloff between nominal level and the tangential decline of amplitude the xover provides. It looks visually like a ‘knee’.

I was able to manipulate the transfer function of the tweeter to effectively do the opposite of what the tweeter’s natural FR does. This will flatten out the curve. The resistor (2.5 ohm in schematic by ear, 1.5 ohm in the simulation) attenuates what still needs to be lower in level that the damping resistors before did not already take care of. The LCR after the resistor is to tame the top-end’s unflat response. I will say this- the usage of this filter here is used in bandwidth where the tweeter is operating, and can easily cause some low impedance issues if not watched carefully. I was not able to use a parallel notch instead to do this, so the series LCR got the nod. It was able to minimize the response, but not totally repair it. The series LCR across the tweeter is to remove the problems of the tweeter’s natural resonance (Fs), and is an out of bandwidth affect. It will not affect the impedance in the range it is operating below the xover. Normally, this LCR is not necessary but with shallower xover slopes. I heard buzzing from the tweeter being active so close to resonance, so I nuked it to make it sound cleaner.

Here is how the tweeter is setup in PCD:

EDIT!: Josh caught an error I made, and the tweeter resistor is in a different position in my schematic, even though that is how I have it wired, It does not match the PCD layout. The resistor should be before the 12 uF cap to be accurate.



Notice I had to use the 99999 to emulate the 5th order electrical circuit. The damping resistors are all placed in their respective places, and that’s easy enough to see. Mind you- the DCR of the coils is subtracted from the resistance value to get the actual added resistance in these locations.

And you get this graph, transfer function, and impedance as a result:



Notice the 2 very deep dips in the woofer transfer function (Blue)? Those are the combination of the tanks and the series LCR on the woofer. The dip at about 3.8 kHz, and the peak at 10 kHz in the tweeter’s TF are the result of the damping resistors and the in-band LCR.

Granted- this is not the smoothest result for an FR simulation, but it is the best I was able to get in this case. I hope this is informative!

On to Tips! Volume 3 Section B- SXO’s!!!