ICEinLAVA's Domain

When you say the phrase “bi-amped,” most newbie sound guys are in awe that you have such a system; they think bi-amping can cure the common cold, or at least all vestiges of bad sound and feedback. What I want to do in this column is introduce the bi-amping method, and show through example how you can do it correctly without a lot of information like speaker processor presets and driver response curves.

Looking at the flip side, using an existing speaker cabinet’s passive crossover network solves all your problems, as the system becomes one-amp, one speaker, plug-and-play kind of operation. For a majority of users, the compromises of a passive crossover network inside the cabinet are acceptable. These compromises are: loss of damping factor and efficiency due to passive component impedances, a blurring of fidelity at the crossover frequency point due to non-phase aligned high and low frequency (HF and LF) drivers, bumps in the frequency response due to a non-optimal crossover frequency point and to keeping the passive crossover component count low. And lastly, the imprecision of the HF driver efficiency matching and the necessary protection circuits, as low frequency amplifier clipping can produce harmonics that could damage the HF driver.

Bi-amping may solve most of these problems, but if not done correctly, it may do more harm than good. In the next section, I am going to use a basic example and extend it to bi-amping with an analog crossover. Then I’ll take it the next step by using a speaker processor.

A Working Example

Let’s say that we have inherited someone’s homebrew (a.k.a. “proprietary”) monitor wedge loaded with an Eminence Delta Pro-12 LF driver and a Peavey 22XT HF driver for bi-amping. The first thing to do would be to ensure that a suitable four-wire connector is in place (such as a Neutrik NL4 Speakon), with completely separate runs from the driver connections to the connector. Making sure the +1/-1 contacts go to the LF driver and the +2/-2 contacts go to the HF driver reduces the risk of blowing the HF if connected to a two-wire passive amplification system. Another protection would be to add a non-polarized 10-microfarad capacitor (50-volt minimum rating) in series on the +2 wire, to prevent accidental low-frequency energy from getting to the HF driver.

Bi-amping is more than two amps to two drivers, and even more than adding an active crossover unit before the amplifiers. First of all, the HF amplifier can be one-half to one-fourth the RMS power rating of the LF amplifier, so bi-amping is conducive to running pairs of bi-amped speakers with two stereo power amplifiers, each sized for their types of drivers. In the example, the 22XT HF driver can handle 75 watts continuous (RMS) from 1.5kHz and up, with the Delta Pro-12 handling 400 watts continuous above 100Hz. With both drivers having an 8-ohm impedance, amplifier sizing can be from the RMS rating to twice the RMS rating, or about the “program” rating of the drivers. This means the LF amplifier should be about 400 to 800 watts at eight ohms, and the HF amplifier should be 75 to 150 watts at eight ohms.

Crossover Point

Without a preconceived notion of the proper crossover frequency point, there are some rules of thumb to use. HF drivers with 3- or 4-inch diaphragms should work fine at a 1.5kHz point, and HF drivers with 1- or 2-inch diaphragms should work well at the 2.5kHz point (such as the 22XT). But before you dial ‘er up and kick on the power amps, there is one more item to consider. The efficiencies of LF and HF drivers vary quite a bit, with the HF usually many dB higher than the LF. So first time trials should back off the analog crossover gain by around 6dB to the HF section relative to the LF section.

For you seat-of-your-pants operators, this is all you need to know. By using your most precise measuring instrument (your ears), you should be able to tweak in the HF and LF gains to taste, and trim the crossover frequency points within a few hundred Hertz of the “thumb” values. For those of you who would like a more scholarly view of bi-amp setup, more information needs to be acquired or measured.

Information Overload

A handy set of specifications would be the efficiency ratings of the each of the drivers. In the working example, the Pro-12 LF driver is rated at 100dB SPL at one watt, one meter, with the 22XT rated at 111dB SPL at one watt, one meter. Thus this 11dB efficiency difference could be factored into analog crossover gain settings or in a digital speaker processor. If this data is not available, you can measure similar results using a “Real-Time Analyzer” (RTA) and a pink noise source. At moderate volume levels, an RTA fed by an omni-directional measurement mic can show the approximate frequency response curve when the crossover is fed a pink noise signal. Pink noise is the hissing signal source that has equal energy in each octave of the audio bands. Most RTAs use a display screen or a matrix of LEDs with a gain control to depict the frequency response.

Measuring and tweaking the bi-amp system can be done by placing the LF driver gain at unity (zero dB) and adjusting the HF driver gain so that a reasonably flat response is in the area around the crossover frequency. Moving the crossover frequency point a little bit may further squelch some bumps and unevenness in response curve. When using analog crossovers, beware of the “CD correction” button. This feature is for placing a high-frequency shelving filter boost to correct for losses when using Constant Directivity (CD) horns on the HF driver. Use this feature if it fits well, but avoid it if the one-size-fits-all setting becomes too shrill in the presence (2kHz to 8kHz) and high (8kHz to 20kHz) frequency bands. By using a graphic equalizer before the analog crossover, you may be able to tune in a more optimal response.

With the crossover frequency and driver gains, the previous setup information still applies. Using the limiter setup menus, you may use your amplifier’s full power sensitivity input signal levels as guides to limit the driver excursion just as or slightly before the amplifier hits its clipping/limiting threshold. More advanced speaker processors are offering compression or overshoot limiting features to keep your driver on the safer side when the speakers are being pushed to the maximum. Low frequency drivers can have softer limits or a more gradual entry into compression by a couple dB, with HF drivers needing none to maybe one dB of hard compression before brickwall limiting.

With most passive and analog crossover bi-amp systems, frequencies near the crossover point come out of both drivers. But most speaker cabinets have the HF driver behind the horn assembly, which does not line up with the voice coil of the LF driver. These few inches of HF to LF voice coil misalignment can create some phase cancellation and blurring of the music in that frequency band. By using the digital signal processor, a fraction of a millisecond of LF delay can be added to bring into alignment the drivers’ phasing. For example, if the 22XT’s voice coil (centered in the magnet) is about six inches behind the Pro-12 voice coil, then a 0.46-millisecond delay (0.5/1090 feet per second) can be added to align the drivers. Note that the crossover frequency had no impact on the calculation, just the misalignment distance divided by the speed of sound.

Tweak-Qualization

The nicest thing about digital signal processors is that they can make a graphic equalizer on main mixes virtually obsolete. Of course, if you have guest mixers, then a “sacrificial” graphic equalizer is desired to keep the hands off the speaker processor that you have tweaked to perfection. By using your ears, an RTA/pink noise setup, or Smaart, you can flatten up the response to your needs or add that “special sauce” curve that brings out the act’s best show. Besides precision CD horn equalization, common parametric filter corrections are applied to mid-frequency hills or valleys and HF horn resonances. Also, adding Bessel or Butterworth response high and low pass filters at reasonable upper and lower driver frequency limits help save your drivers from un-needed signal reproduction that is annoying at the least or destructive at the worst.

With digital signal processors, the sheer quantity of parameter choices can be daunting. When in doubt, choose Linkewitz-Riley crossover point filters with 24dB/octave slopes just like most analog crossovers implement. If you are transferring filter settings from one brand of processor to another, beware that the filter bandwidths are not always stated in the same terms. Some will use “Q” or Quality Factor in defining filter bandwidths, while others will use dB/Octave. These representations all get to be brain mush when implementing low Q (wideband) filtering, so use your eyes and ears first until you are up to the task of understanding the mathematics of filtering, which is a possible Theory and Practice topic for the future.