Jan 1, 2020

On Crossing Over…

By Bryan Geyer

Loudspeaker systems are designed with the expectation that each internal driver will be operated within its optimum frequency range. The need to comply with this restriction is traditionally accomplished by sectioning the input signal into several (commonly 2 or 3, sometimes 4) frequency-defined passbands that feed the individual drivers. These bands present sufficient overlap (“cross over”) to create a flat and seamless integrated output. The analog crossover networks that formulate these passbands and link them to the allocated drivers are generally derived by either of two basic means…

(1) As a low impedance passive filter network, composed entirely of discrete passive parts, that’s inserted into the high current signal path between the power amplifier output and the loudspeaker drivers. (The parts are normally secured inside the sealed loudspeaker enclosure.) A significant advantage implicit with this passive network is that it facilitates the use of a single power amplifier to feed all of the derived passbands and related drivers serving that channel.

(2) As a separate active filter network (with high Zin), composed of active solid-state circuitry in a dedicated external control box (with its own power supply), that is inserted into the line-level signal path after the preamp stage (or after the master volume attenuator when there’s no preamp) and before the ensuing (hence multiple) power amplifiers. Each of the derived frequency-defined passbands will then connect (via low Zout) directly to its own independent power amplifier, and each such amplifier will output directly to a designated driver within the speaker system. An obvious disadvantage with this active approach is the escalating cost. There’s the initial expense incurred in creating the active filter circuits + enclosure, as well as the cost of providing a captive power amplifier for each of the output passbands.

Passive low impedance crossovers can be effective when they’re properly implemented with care, but doing so becomes a challenge. Because the passive elements will be inside the path between the power amplifier and the loudspeaker, they must be able to tolerate high current signals and still stay stable. That dictates beefy parts, i.e. the use of wire-wound resistors, low gauge (fat) wires for the inductor chokes, and large value non-polarized capacitors. These comprise the sort of parts that aren’t normally stocked in tight tolerances, although precision is essential to assure that the filter network will be accurate. This mix of requirements isn’t mainstream; obtaining and/or creating such components often becomes difficult.

Passive crossover design can be deceptive. Erroneous assumptions and unexpected complexities are frequently experienced. Below are four such examples, as previously cited by prolific DIY contributor Tom Perazella, in a letter that appeared in the September, 2017 issue of the Boston Audio Society’s “Speaker” journal…

(a) Using the driver manufacturer’s designated impedance as a guideline for the selection of the required passive component values might lead to significant error. The real impedance of the woofer at the intended point of crossover is likely to be appreciably different in value than the manufacturer’s stated impedance. Using the labeled value could cause significant initial error.

(b) Matching the different drivers’ relative passband sensitivity is inherently difficult when using a passive low impedance network. In order to keep the crossover points stable you must use L-pads to achieve the corrected levels. Again, the manufacturer’s stated sensitivity might not be entirely accurate, so the components required for the L-pad should be selected only after measuring individual driver sensitivity at the intended crossover frequency.

(c) After all of the requisite driver measurements are made, and all of the passive components fully specified, it is very likely that the desired values won’t be available. Complex series/parallel groupings will probably be required to approximate the target values. This component selection challenge becomes particularly vexing in the case of air core inductor chokes, where you’ll need to specify higher-than-needed values, and then remove some of the wire turns while making constant in-process measurements.

(d) Last, when the construction phase is done and you conduct final measurement and listening tests you will generally want to make some adjustments. This will often require exchanging expensive passive components. The consequent rewiring will likely force layout changes, and the related delay makes new comparison measurements more difficult, less reliable. For these reasons, completing those critical final adjustments gets quite arduous.

Clearly, it’s apparent that some of the pitfalls described above might go unrealized—or get ignored—or simply remain unsolved—when a low impedance passive crossover network is created. In such case, the related speaker system’s output will forever reflect the inaccuracy of that oversight. Personally speaking, I believe that numerous commercially marketed “high end” speaker systems do exhibit some of these same errors.

Active crossovers are relatively more complex than passive networks, but easier to enable. They’re generally more precise, also inherently cleaner. (Separately amplified passbands foster less IM than with a passive network. The latter uses only one amplifier.) But the most obvious and vital advantage that an external active crossover bestows is that it’s both accessible and controllable. It’s not buried inside a sealed speaker enclosure, and the output levels can be varied, they’re not fixed. Given this combination of convenience and flexibility, it’s easy to alter the tonal subtlety of the sound to favor a particular program or genre, and then rapidly reset to “standard profile” when desired. It’s this versatile control advantage that makes an external active crossover network so compelling in use.

For more insight on the many advantages implicit with active analog crossover networks, check some of these informative sites…

Application…
Both types of analog crossovers, active and passive, are often applied together. One very effective case is to use an active crossover to split the incoming line-level signal at 70 to 100 Hz, and send the low-pass output to a pair of self-powered subwoofers with the high-pass output to the main stereo power amplifier and main speakers. The latter’s internal passive crossover will then generate new frequency-defined passbands and link to a designated driver. (In such case, any crossover that was supplied as an integral feature of the self-powered subwoofer is simply bypassed* and unused. Bypass mode option is normally enabled via the subwoofer’s control panel.)

Of course, it’s also possible to apply active analog crossovers at the main speakers too. In such case, the high passband from the initial subwoofer/main speaker split would feed a cascaded active network, with those outputs linked to multiple captive power amplifiers. Each such power amplifier would then connect directly to a driver inside the main enclosure. (The internal passive crossovers would have to be deactivated or removed.) However, let’s review some basic fundamentals first…

In terms of the potential benefit, it’s particularly rewarding to apply active crossovers at the bass end of the spectrum. This is because…
            …the low bass is where passive networks become more difficult; there are vexing component problems.
            …it’s important to confine high energy low bass to the subs, so effective high-pass filtering is essential.
            …the bass range is where an active Linkwitz-Riley full 4th order filter slope is most helpful.

Further, at mid-to-treble range frequencies the use of a traditional passive crossover becomes more practical, or more fitting, because…
            …the passive parts get smaller in size than those needed for the low bass, so they’re easier to implement.
            …the passband “handoff” frequencies becomes less critical, more forgiving with respect to accuracy.
            …a passive crossover avoids the need to provide a captive power amplifier for every passband.

That last aspect is decisive. Dedicating a captive stereo power amplifier (the “self-power” amp) to the subwoofers while also providing another stereo power amplifier to drive the ensuing stages seems prudent and practical. But adding still more power amps in order to serve the main speaker’s upper bass, mid, and treble drivers borders on obsessive. This seems especially so when considering that the primary audible benefit derived with captive bandpass power amplifiers fades at frequencies beyond the middle bass (400 to 600 Hz) region. In truth, the upper bass, mid, and treble drivers can be effectively served by properly designed passive low impedance crossovers. Such networks have proved acceptable in that role for decades. Their continued use seems logical and warranted.

Main speakers only…
All of the applications noted above assume the use of paired subwoofers in concert with the main speakers. That combination is especially helpful when “mini-monitors” comprise the primary speakers, but active networks are just as helpful when full range speakers are used without any supplementary subs. The existing passive crossover networks within the full range speaker enclosures would then be (wholly or partially) deactivated, and the related drivers linked directly to the active network’s independent power amplifiers. A single active crossover controller with just two output passbands could serve to drive the woofer and all of the higher frequency drivers (requires two stereo power amplifiers). Or a dual active crossover controller—such as this product: https://sublimeacoustic.com/products/k231-stereo-3-way-active-crossover—could serve to provide three distinctly separate output passbands to drive the woofer, the mid-range, and the tweeter sections independently (requires three stereo power amplifiers). In either case, convenient access to fully independent ± gain controls would be provided for each separate passband. This latter feature represents a critical advantage that can’t be achieved by means of a passive network.

BG (December 2019)

*The purported “crossover” that’s inside many commercial self-powered subwoofers might not be a true crossover. It’s often just an active low-pass filter that rejects frequencies that are higher than the internal “crossover” option. It might not provide any complementary high-pass output to keep the low frequencies out of the input to the main power amplifier. Or, when provided, that high-pass output might be routed only via a simple series capacitor, with grossly inadequate (-6dB/octave) bass rejection slope. Measure carefully if you actually intend to use one of these built-in subwoofer “crossovers”. Some are worthy, many are not.

No comments:

Post a Comment

Thank you for your comment. It will be published after review.