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

Section B

Using the “Canvas” project as my example in this case for a complex Series Xover
Emulation, this will probably change the way you look at crossovers in general as a whole.

Before reading this, I recommend you try and not think about Parallel Xovers and how they are modeled. Series Xovers represent a different form of thinking, and the design process is quite different.

A brief primer:

In a Parallel Xover, you increase resistance and therefore the impedance where you want to attenuate the drivers’ issues. In a Series Xover, you try to minimize resistance where problems lie. This is because everything is in series in a Series Xover, and ‘shorting around a problem’ is how you fix stuff. Generally speaking, smoother drivers are used in Series filters to avoid having to fix stuff, as more series reactance in the circuit is generally frowned upon. This also means that lower DCR coils can have a big favorable quality in this topology. Tweeters used should be taken well care of, as Series xovers are typically harder on tweeters as a whole with less overall protection. As a general rule relating to this fact, I have not employed a planar or ribbon tweeter in any form of series xover to avoid a fatality. A lot of users of this type either exemplify the ‘lots of parts’ or the ‘minimalistic’ mentality, so these are not that different from the typical Parallel xover designer’s attitude. You can go right down the middle just as you did for the other flavor of design. The tweeter and woofer do both connect at the same node in the xover and are in series, so that is not a piece that is misdrawn below. You cannot bi-amp a Series xover. The possibility exists that you can get better acoustic phase correlation with a series xover with less parts, and I have seen that happen a few times.

One common misconception:
You don't add the impedance of the drivers to get the net impedance. It's just like parallel xovers in that the driver's range relates to the impedance in that range.

Case in point of above, I collaborated with Charles Laub and some others to produce a series xover for the “NorCal Collaberation 2-way” with the DQ25 and a Peerless buyout. The resultant network was minimalistic in parts count, and most were of small value.




There was concern for bleed-through of high-frequencies in the woofer’s range, but strapping a 1.5 uF across the woofer remedied that issue. Acoustic phase correlated between the drivers from 100 Hz to 8 kHz in the initial simulation, and beyond that a smidge with the added capacitor. 8 parts yielded a tanked BW3 acoustic xover, and very good phase alignment.

Referencing an older article from Andy G or Clayton Oxendine, you will read that deciphering the Series Xover schematic is like traveling the path of least resistance for any frequency you want to try. Highs go through caps of large enough value and tweeters, while lows pass through the minimal coil value and woofers. If you can trace and visualize what the frequencies do, you will see what the circuit is doing to the frequency response.

Oh- and one last thing before we move on; If you change ONE part in the circuit, the responses of BOTH drivers change. Everything is in series, so everything affects everything else!!

Now- To break down the circuit into its parts of what it actually does, it’s kind of the opposite of parallel duties. In the “NorCal” schematic above for the simplistic approach, the 0.25 mH coil acts as the first part of the high-pass filter for the tweeter, and the 5.6 uF makes it second order. The 6.8 ohm is the series padding resistor for the tweeter, and the 4 ohm is the parallel factor. The 7.5 uF is the first leg of the woofer’s xover, and the 1.8 mH coil is both the second leg and the point where BSC starts to take affect. The 0.22 uF steepens the woofer rolloff like the tank does in a parallel network, and the would be 1.5 uF cap (mentioned but not pictured) would make the 3rd leg for a 3rd order filter on the woofer.

That is how the filter for a Series xover is generally looked at, but I’ve been known to be on the odd side of the fence at times. Let’s see how I look at it….

Now- from the ‘parallel-xover-minded’ designer, let’s look at this a bit differently….
The 0.25mH can also be seen as a 6dB Lowpass (LP) for the woofer, and the 7.5 uF seen as the 6dB Highpass (HP) for the tweeter. The ‘bridge-connection wire’ is then attached to the drivers. It depends on which input leg you start from to visualize it this way. You will effectively halve the circuit as a means to an end from a visual standpoint.
From the negative input, since the 7.5 uF is followed by a 6.8 ohm resistor to pad the tweeter (negative T connection), the 4 ohm shunts for further attenuation, and the 5.6uF is on the other side of the tweeter to finish the dual-cap setup, while the 0.25 mH actually is the shunt in between to the other input side to make it a 3rd order electrical rolloff.

Make sense??

Now the woofer for the other direction…
From the positive input, you have a 0.25 mH LP coil, and the 7.5 uF and 6.8 ohm resistor make it a damped second order, while the 1.5 uF cap across the woofer and the 1.8 mH BSC coil make it an overall 4th order electrical layout.

Also of note- the resistor placed between the main coil and cap lets the amplifier see a very constant impedance load. Try to keep this above 4 ohms or so to avoid impedance issues.

If you have made it this far and you are nodding your head in agreement, then you can probably design a Series Xover via simulation in PCD. Select “Series 2-way” or “Series 2.5 way” to begin your exercise.

Your parts go in different places in modeling a Series Xover, so be sure to keep your schematic picture open for constant reference. I have to keep it open myself to make sure I put things in the right places. The toggles become very important, and the ‘textbook-value calculator’ can give you a really good head start if you don’t know where to begin. Now that I’ve given you a way to look at Series Xovers that might be understandable and relatively easy to model in PCD, here is the schematic and the layout for the “Canvas” project, followed by the FR/TF/Impedance simulations:









In regular SXO terms, this is a 4th electrical on the tweeter, and 3rd-tanked on the woofer, to yield a BW6 acoustic xover at just over 2 kHz.

Let me know if you have any questions, and I hope this helped you to be able to perceive how to model a Series Xover!!

---------------------------------------------
An experience from greywarden:

I tried to model a series crossover the other day for the speakers I had recently done a parallel with success and decided that you have to start with 1 component, make it flat-ish, add another, stare at the transfer functions, eventually get it flat (but not enough), add another component to each driver, stare again, fubar the whole thing and quit. haha.

Joking aside- they are not easy the first time, just like gw found out.

I still hope you give them a try!