[get menu]

What's wrong with the stock Klipsch crossover networks ?

What's wrong is VERY difficult to explain to anyone who has no background in passive filter design. This is nearly all of us! I will do my best to explain the problems with the Klipsch networks by the numbers using the Klipsch "AA" network as an example.

First, a quick course in very basic electronics is a must! There are three basic types of electronic components. The inductor, capacitor and resistor. Here is a quick explanation of the three.

Another term you will often hear is impedance. This is the combined effect of all three types of opposition to the flow of current. That is, resistance plus reactance of either type. The phase shift between voltage and current can be anything from +90 to -90 degrees depending on the proportion of inductance to capacitance.

All stereo power amplifiers are measured and specified as delivering power into a RESISTOR. That is, a load having no inductive or capacitive component and the current drawn in perfect step with the voltage applied. This is what a crossover network and loudspeaker should look like to the amp. The Klipsch networks DON'T! Good quality amps can handle goofy loads that look like inductors at one frequency and capacitors at another frequency. Little ones (like SET amps) can't!

Below are impedance plots of the "AA' and ALK Universal networks comparing the impedance seen by the amp at all frequencies. The top two curves show the reactance phase of the impedance. It should be a straight line at ZERO degrees. Below zero indicates the amp is seeing a capacitor and a resistor. When the line is above zero the amp sees an inductor and a resistor.

The bottom two curves show the resistive error. Again the curve should be a straight line at 8 Ohms. Look at the AA at about 2000 Hz (the little red X). It's almost 30 Ohms! That's right in the most important spot in the midrange! All of the ALK networks look like pure resistors of 6 to 8 ohms over the entire audio range. Above and beyond the actual audible effects of a squirrely impedance curve itself is that it is the "measure" of how well designed a network is. Properly designed networks show a constant resistive impedance to the amp. Poorly designed ones are all over the place!

Here is a graphic to identify the parts of the Klipsch "AA" network and a listing of its problems by the numbers. The total effect of all these design problems is the reason for the strange impedance curves and several other bad characteristics of the AA network.

Some details about the design problems, by the numbers: The solution to the problems with the AA:

The initial goal was to develop a network to replace the AA in my own Belle Klipsch speakers that would be similar in scope to the "AA" but would bypass all these problems. Using the filter design techniques learned through many years of designing similar but far more aggressive networks at microwave frequencies, where these sort of errors could never be tolerated, a correct design was done. The result was the Universal network. It is basically a "corrected" AA. The schematic is below showing the actual changes to the AA to transform it into the ALK Universal. All the part values have changed so it makes no sense to just modify it. You might as well build an entirely new one!

Then later.....

As time went on and more investigation was done it became obvious that more aggressive slopes to control interference between drivers would be a good thing. The literature always implied that filter slopes beyond about 18 dB / octave didn't sound good. It was also obvious to me that loudspeaker people simply didn't know how to design simple networks let alone more aggressive ones! This is most likely why attempts at sharper slopes were failures. Experience at microwave frequencies with L-C filters FAR more involved than anything that could fit in a loudspeaker would make an extreme-slope network having about 20 - 25 dB ultimate rejection at audio frequency a cake-walk. The extreme-slope network series was the result.

Al K.