Red Rover, Red Rover, Tell Me ‘Bout the Crossover
For speaker designers, crossovers are some of the most problematic issues they face.
By definition, the crossover splits and directs the electrical energy from your amplifier to the multiple drivers in your speaker. I use the term “driver” to mean the transducer that converts electricity to sound and “speaker (or speaker system)” to include the driver(s’) box and all the other parts of a complete speaker system. But are there optimal crossover points, or are they dependent on the capabilities of the drivers involved?
Unless you have a full-range single driver, you will have to deal with crossovers. The reason to have two or more drivers is the fact that you must move a large volume of air (for each cycle of bass) to achieve normal sound levels in most rooms, while you need a very small amount of air for treble since you have so many more cycles. Since the small tweeter cannot take the high power needed for the big woofer, you need to split up the power and send it to the best suited driver.
The Right Driver
To have the same sound level, the same amount of air must move back and forth in the same amount of time. If that time is say one second, the frequency is the number of times a smaller amount of air is moved both in and out in that second. For each cycle, in and out, at lower frequencies more air must be moved than at higher frequencies since there are fewer cycles in each second. In fact, the amount of air moved is inversely related to the frequency: if you double the frequency, you only need move half the amount of air in each cycle. In the highest audible treble, 20 kHz, you need only 1/1000 as much air movement per cycle as you need at the lowest bass, 20 Hz. You also need a small driver to have dispersion in the treble match the dispersion in the midrange and bass. Plus, you need a driver with a light voice coil for good treble efficiency and a heavy voice coil for good bass efficiency.
Thus, it is easier to make different drivers for different frequency ranges: from subwoofers for very deep bass to super-tweeters for very high treble. Then you need to split the full-range signal from your amplifier and sent the correct part to the driver that is best in that frequency. You particularly need to protect the small treble drivers from the low frequency drive from your amplifier. The massive voice coils and cones used on many low frequency woofers reduce their high-frequency output even without crossovers.
I believe the audible benefits of coherence make it worth the design effort to have the both drivers be the same distance away from the listeners through mechanical design of the cabinets (usually putting the tweeter behind the face of the woofer). With amplified systems, it is possible to introduce a time delay to the closer driver as part of the electronic crossover; but that is a topic for a future blog. If the designer is not concerned with coherence, some major trade-offs like correct phase and frequency response must be considered in the crossover design. Only when the signals from both the tweeter and woofer arrive at the same time can you reproduce the electrical signal in the acoustical signal. I will let another speaker designer defend the position that reproducing the input signal is not a goal and I will assume an in-time design in the rest of this discussion.
In an ideal crossover, the crossover frequency is the frequency where the woofer and tweeter each contribute half of the energy to the total signal reproduced. Below that frequency, the woofer is producing more of the energy and above the crossover frequency, the tweeter is producing more of the energy. In a less than ideal crossover, phase becomes a problem. This is similar to the phase problems with complete speakers we discussed earlier in Phased Out: Part 1 and Phased Out: Part 2 – Absolute Phase. Now the problem is between the individual drivers, not the complete speakers systems.
When the two drivers are in phase with each other at the crossover frequency there is no cancellation of output of one with the output of the other and the frequency response is flat throughout the region of the crossover. If the phase between the drivers is shifted, they both need to put out extra output to add up to a flat response. In the extreme case and the output of one is 180 degrees out-of-phase, you get a deep hole in the response through the range as one totally cancels the other.
For a tweeter, a simple crossover of a single capacitor (and maybe a few resistors for level matching) can match the natural high frequency roll-off (or enhanced with an inductor) of a woofer. This simple design has been used in many popular speakers over the years. Models such as the Acoustic Research AR-4x, KLH 17 and Ohm E all utilize the design. Both Acoustic Research and KLH have gone through a number of corporate and ownership changes since those models were current, but these links are a place to start. Please note these three are all smaller models designed for smaller amplifiers. As the need for more output, the tweeters with this simple a crossover tend to be overdriven. If the input below the crossover point can be reduced without affecting the phase in the crossover region, the tweeter can handle greater power. In our Ohm Walsh speakers we generally use the simple capacitor design until we get down at least 12 dB; below that frequency we add extra protection to the tweeter as it is contributing very little to the output at these lower frequencies.
For a woofer, no additional parts may be needed if the response has a natural roll-off due to the inductance and weight of the voice coil. If there are some high-frequency resonances present, additional low-pass or notch-filtering may be needed. An example of such a design is the current Walsh 2000 that uses an aluminum cone and rubber surround. The cone has a bell-like resonance that the rubber cannot damp; so we add a filter to eliminate the excess ringing. On almost all the Ohm Walsh systems, we add Sub-Bass Activatorstm (SBAs) to extend the deepest bass, reduce the distortion and increase the dynamic head room; but this too is a topic for another blog.
In all cases, Ohm uses pulsed square waves (similar to step functions) to test our designs. With them we can see the combination of time, phase and amplitude. Any error shows up immediately.
Yes and Yes are the quick answers the teaser questions “…are there optimal crossover points or are they dependent on the capabilities of the drivers involved?” but, I have covered just the fundamentals and ran longer on this blog than I usual. More to come in the future…
Enjoy & Good Listening!