A Modular System for Line-Level Loudspeaker Active Crossover
XEN Audio
November 2018
Forward
We wish to dedicate this project to the late Siegfried Linkwitz, from whom many of us have learnt a
great deal about loudspeaker design and sound reproduction.
Background
A few years ago we published a line-level active crossover for our Avalon Clone [1] using the latest
Accuton Cell series chasses. A dedicated PCB was made for that specific circuit / purpose after
testing with a prototyping board. And we were very pleased with the results.
Lately a friend of ours who is into speaker building finished a project called the DUO-DXT [2], a two-
way bookshelf with SEAS chasses and a somewhat complex passive crossover. Just for fun, we took
on the challenge to come up with an active crossover with essential the same (driving voltage)
frequency response at the chassis terminals.
At the same time, we also have a few other speakers around for which we had long wanted to try out
active crossovers with. Among them a pair of second-hand B&W CDM1s that is worth very cent we
spent on them.
We were looking for something that was totally flexible and modular, so that we could make up any
filter system by putting different circuit elements in series and/or parallel. Also we do not want to have
any DIP switches to select different values, and prefer soldering instead. Unfortunately nothing readily
available appealed to us, so we ended up designing and building our own.
Design
The basic design consists of a number of small PCBs stackable in a series. Each is approximately
50 x 25mm in size, and can support one filter function for 2 single-ended or one balanced channel(s).
Examples of such basic filter functions include [3].
Input Buffer with Passive Attenuator
Sallen Key Low & High Pass (Bessel, Butterworth, Chebeshev, Linkwitz-Riley)
All Pass for Delay Correction, positive and negative
Shelving Filter with Notch, Low & High Pass
Notch filter with optional Gyrator (work in progress)
As much as possible, we only want to use passive filter elements (resistors and capacitors) and unity-
gain buffers. The latter can of course be dual JFET opamp’s such as OPA2134 in unity gain, but can
also be discrete buffers. In some cases, such as the all-pass delay, opamp’s cannot be avoided
unfortunately.
The dual buffer or opamp is to be mounted on DIP8 IC sockets. That means one can also use
single / dual SOIC8 opamp’s (such as 2x OPA627) on adaptor PCBs, or in case of unity gain, dual
discrete JFET buffers with a quad of matched low-noise N-JFETs (2SK170, 2SK369, 2SK117, …..).
Each filter module has its own local power-supply-decoupling capacitor. While MKS-02 1μF were
used in ours, you can also freely pick any electrolytic caps with 2.5mm pitch and 5mm diameter with
>30V rating.
Figure 1 Dual Source Follower with TO92 N-JFETs on DIP8 header
To allow for a wide choice of components, capacitors are on 5mm pin pitch with additional 0805 SMD
pads, and resistors on 5 & 10mm pitches. For our prototype build, we used WIMA MKP2, FKP2 and
Panasonic ECHU SMD extensively for capacitors. For resistors, anything from Vishay S102, Caddock
MK132, Nikkohm RP44, to Dale RN55, PRP9372, Welwyn RC55Y will all fit.
The connections at both input and output edges of the boards are all on 2.54mm pitch. One can then
choose between hard wiring (as we did) or DIL header / sockets. Each set of connections (common
on all boards) consists of Signal left, Gnd, -Vs, +Vs, Gnd, & Signal right (from left to right).
Application Examples
1) Linkwitz approximated Duelund 3-way (see front page photo)
In one of his webpages [4], Siegfried Linkwitz discussed the Duelund 3-Way crossover [5] in some
detail, and proposed an approximated version at the end. This differs from the traditional LR-4 3-way
in that the mid-range HP and LP are only 2nd order, instead of 4th. Not only does the latter see less
phase shift around the crossover frequencies, but there is also less active devices (3 buffers instead of
5) in the signal chain. So we decided to build one, just for fun. Note that both the tweeter and the
bass driver has to be connected in reverse polarity.
The circuits are as follows.
Figure 2 Schematics of the Linkwitz Approximated Duelund 3-Way Active Crossover
The simulated frequency response is shown below :
Figure 3 Linkwitz Approximated Duelund 3-Way Active Crossover - Frequency Response
2) B&W CDM1
The B&W CDM1 is Loudspeaker of the Year 1995 [6]. They are still popular on the second hand
market, and prices have been holding steady for the last 10 years. For a bit more than 300 USD, they
are a real bargain.
The components used in the original B&W crossover are not particularly high hand. So an active
crossover can be a good upgrade. In order to determine the require driving frequency response of
each chassis, the simplest way is to power the loudspeaker with a sweep sine of constant amplitude
via a power amplifier of sufficient bandwidth, and measure the AC voltage of each chassis as a
function of frequency. Once this is known, one can try to emulate this with line level filters.
The circuit is shown below. As in the original passive crossover, the main driver gets a 1 st order low-
pass, whereas that for the tweeter is 2 nd order. The beauty with the modular system is that one can
use the same breadboard to make up different combinations, as shown in the photos.
Figure 4 Schematics of B&W CDM1 Active Crossover
Figure 5 B&W CDM1 Active Crossover - Frequency Response
3) DUO-DXT Reference Bookshelf
The original passive crossover of the DUO-DXT is rather complex. Thus, the same approach as in the
CDM1 was used to determine the desired filter characteristics. After some iterations, the active filter
circuits shown here were arrived at, which resembles the original passive counterpart to within 2dB
from 10 to 25kHz.
Fig. 6 Schematics of Active Crossover for the DUO DXT
Figure 7 DUO-DXT Active Crossover - Frequency Response
Out1 = passive Mid-bass Out1a = active Mid-bass
Out2 = passive tweeter Out2a = active tweeter
But why Analogue and not DSP ?
People will say, why bother these days with analogue crossover when you can do everything (and
more) digitally. Which has of course been made easy by nice, inexpensive products from the like of
mini-DSP. But our own experience, and perhaps also that of others, suggests that analogue might still
be subjectively the preferred solution in the end [7, 8].
References
1. http://xen-audio.com/documents/accuton/130701%20Accuton%20CELL%20Public.pdf
http://xen-audio.com/documents/accuton/130818%20Accuton%20CELL%20Public%202.pdf
2. https://www.hifi-selbstbau.de/bauvorschl-mainmenu-36/2-wege-lautsprecher-mainmenu-75/416-
duo-dxt
3. http://www.linkwitzlab.com/filters.htm
4. http://www.linkwitzlab.com/crossovers.htm
5. http://www.musicanddesign.com/Duelund_and_Beyond.html
6. https://www.stereophile.com/standloudspeakers/630/index.html
7. http://www.diyaudio.com/forums/the-lounge/146693-john-curls-blowtorch-preamplifier-ii-
9870.html#post5285486
8. http://www.diyaudio.com/forums/the-lounge/146693-john-curls-blowtorch-preamplifier-ii-
9870.html#post5285509