By Irving "Bud" Fried.
In Kalman Robinson's review of the OLS Karma Ceramique CE-2.0 loudspeaker (Stereophile Oct 2000 p209) he refers to the designer's description of the network as a "subtractive" crossover network of the series type.
I have been a longtime proponent of series networks, of which I was apprised over 20 years in my researches in Denmark. (My article on this subject is available from me for $2.50)
I am happy, therefore, to note that the subject of series networks has been brought up: they are generally considered to be "mysterious" and not to be discussed by the profession of speaker designers. I would like to make it several points regarding them and why they are virtually ignored by the profession.
First of all, Kalman recites, they are more difficult to design:, or you have to be able to visualize &endash; ie. have a brain. (My article points out that Peter Walker has often mentioned the dearth of first-class minds in speaker design. Peter, who created the Quad Electrostatic, has a first class mind!)
Second, while there are formulas for designing series crossovers (David Weems gives such in his book), there are no computer programs available, as there are for parallel networks.
Third, they are "self-balancing" as OLS assets, meaning that small variances in parts values does not destroy them (as in sharp parallel filters); rather it merely shifts the transition point up or down. Only if there is a very large error do bad things happen -- ie, presenting a capacitive or inductive load to the amplifier.
Otherwise, series networks present an "resistive" load to the amplifier. That is good, because amplifiers prefer such loads to the highly reactive loads of sharp-filter parallel networks -- the networks that are used all too frequently in supposedly high-end loudspeakers today.
Kalman in his listening, pointed to the merits of series networks, when he commented how the mid, woofer, and the tweeter seemed much more integrated than that to which he was accustomed. This is because series networks, done with the proper drivers and with the proper parts, are much more in-phase at the crossover points. Indeed, when one is accustomed to a properly operating series network loudspeaker, when one goes to another loudspeaker with parallel networks, no matter what the slope rate, the first shock is to notice that the drivers all seem to be separate -- what has been called "woofer/tweeter" sound.
Why is this? Several theories have been evolved regarding this phenomenon. I will list them briefly, and suggest that further elaboration can come from that article of mine.
Irving M. (Bud) Fried