Monitoring - Active Crossovers (Part 4)
For control rooms, active systems come in two formats. The first is simply an extension of the normal arrangement - you change over the speakers and add some extra amplifiers plus the active crossover to the monitoring rack. Packaging (i.e. hardware, mechanics) accounts for a large proportion of the price of sound equipment, and by building a speaker enclosure which contains all the necessary amplifiers, plus the active crossover, a lot of expense is saved.
The discrete, rack-mounting active crossover was originally developed for Rock PA in the early 70's. The first serious active crossover for professional use arrived in 1978, from Brooke-Siren Systems, a company that had been formed by ex-Midas console designer, Chas Brooke. The BSS (Brooke-Siren Systems) crossovers were flexible, modular units, based in a card-frame system. Other than optional limiters and LED PPM (Peak Programme Metering), they featured plug-in frequency cards, allowing crossover points (frequencies) to be changed with ease, to accommodate a variety of speakers.
This was a great improvement over the messy 'variable' crossovers, where the crossover points were set with knobs; the variable frequency-set involved severe compromise to the filter accuracy. This was essentially on account of the poor calibration on the front panel, and wide pot tolerances, which led to gaps at the crossover points, or massive overlap, hence severe peaks or dips in the sound, which tended to be rendered painful! There was also the ease with which the knobs could be tweaked, accidentally or otherwise; an event which resulted in the expiry of many a tweeter as it attempted to reproduce 200 watts at 200Hz!
The problem with the card-frame approach, much as it made expansion (from a 3 to 4 way system say) easy, was one of expense. Moreover, the flexibility offered by modular crossovers is rarely required in the studio; monitoring here doesn't suffer the upheavals of a touring PA.
Nevertheless, many control room crossovers do come as a discrete 'box of filters', in a rack-mounting format, à la PA variety. In a three way system, bass, mid and treble (stereo) outputs feed into 3 separate stereo amps, or six mono amplifiers, or even into a single package containing six amplifiers, known as a stereo tri-amp. The amplifier outputs then feed directly to the speakers. These may be discrete components again - using carefully chosen bass, mid and treble drivers and/or enclosures from different manufacturers, or even monitors purpose built for tri-amping ie. a single enclosure with separate input sockets for each driver.
With the former approach, components and enclosures from a variety of speaker manufacturers may be used. For example, one of my own enclosure designs uses ElectroVoice in the bass and top, with a Fane driver in the mid. This 'roots' approach is apt to be tiresome though, unless you have a firm, practical knowledge of drivers and enclosures which are likely to work well together.
On the other hand, if you need to build an active system cheaply, and have some experience with PA (where mixing and matching of drivers is routine) it can offer great rewards. Designing speakers for 'active duty' is also easier than designing a passive system in that aside from not having to design a (passive) crossover, you don't need to worry too much about matching the sensitivity of the drivers. Instead, the crossover output levels can be adjusted to balance up the drivers without waste of amplifier power. Give a musician enough rope and... yes, too much flexibility can be bad news, and reduced hassle comes in the shape of monitors ready-wired to plug your 4 or 6 amplifiers into.
Sadly, the number of manufacturers making conventional, established designs in 'active compatible' format is small: ATC, Eastmill, ElectroVoice, Malcolm Hill, Turbosound and Tannoy. (Other manufacturers please write to us at HSR if your monitor speakers are built to plug into a Bi- or Tri-amp system). Rack mounting crossovers of the non-modular variety are more common: The BSS FDS-300 model illustrated on the September cover (Fourth up in the rack) of HSR is but one of several accurate, low-cost designs.
With these, the crossover and amplifiers are built into the speaker enclosure. Often known as active speakers, or tri-amplified (bi-amplified for a 2 way system, etc.) cabinets, the idea here is not merely that of elegant minimalism; it's also concerned with making active-speakers accessible to people who don't want to become involved in assembling an active system from component parts. As with conventional speakers, the manufacturer has done the work for you, and all you need to do is plug in the line level monitor signal, and switch on.
Physical integration also cuts costs: basic active speakers begin at around £500, but as you don't need to buy an amplifier, these speakers potentially compete with conventional units costing over £350. Integrated systems also cost a fraction of the price of a modular, card-frame crossover alone. In fairness, however, when put into perspective alongside a £20,000 PA system, a £1500 crossover is good value.
Some speaker manufacturers (eg. ElectroVoice) supply crossovers to suit their drivers or enclosures, in which instance choice of crossover frequency should present no problems. More often, the crossover is a universal design, from a specialist manufacturer, and the crossover frequencies will need to be set up, to suit the drivers or enclosures in use.
Classic two (and three)-way speaker systems tend to have crossover points in the 2 to 4kHz and 200Hz to 400Hz region(s) respectively, but if you're ever likely to divide up the signal for 4 or more drivers (eg. sub bass, or high treble), crossover points will be spread across the audio spectrum, so flexibility in setting the crossover points will be necessary. Good, modern crossover designs make a wide range of frequencies available by exchanging plug-in cards, modules or DIL plugs. These contain resistors, capacitors or both, to accurately set or program the crossover point(s), and even the alignments.
Less sophisticated crossovers feature only a limited number of switchable crossover points, and whilst these are fine for more flippant PA applications, they don't allow the fine tuning of the crossover point(s) that's usually necessary for the best results. Crossover frequency changes of +/- 10% are very significant in some systems, especially between the tweeter and midrange driver.
Drivers and speakers must also be matched for sensitivity. For instance, a horn-loaded tweeter might be 6dB more sensitive than a 15" Thiele-loaded bass driver. To confuse matters, the top-end power amplifier might have a lower sensitivity, requiring say 1.2 volts to drive it to it's rated power, compared with, say, 776mV (0dBU) for the bass amplifier. In a passive system, power amplifier sensitivity obviously doesn't enter into the equation, whilst excess sensitivity in one or more drivers has to be padded down with a series attenuator. This involves throwing away power, headroom and hard won efficiency. The active approach has a different answer. For example, with a 'hot' (excessively sensitive) tweeter, the input signal to the top-end amplifier is attenuated with the gain control on the crossover. Whilst this still involves attenuation, at least it doesn't come hand in hand with a loss of headroom in the power amplifier, where it's most needed.
On the assumption that drivers are roughly matched for sensitivity, with say +/- 4dB overall difference between bass, mid and treble units, some up-market crossovers feature balance controls with, say, +/- 6dB range rather than the (unnecessary) 0dB to minus infinity range of the usual 'volume control' attenuator.
Setting crossover attenuators requires some careful listening to establish accurate tonal balance, and this fine tuning is best done over a period, and with a wide range of music. In a system with a separate set of controls for the left and right channels, balance is more difficult to achieve, but the rewards are greater. Usually, it's best to roughly balance each side individually, with the other channel muted. At the same time, stereo pots are notorious for left-right mistracking at settings below half-way, and separate mono-attenuators can offer greater accuracy, as well as compensating for small sensitivity differences between, say, the right and left channel midrange drivers. Control settings can also be tweaked to provide 'Tilt' balance, ie. subtle broadband EQ to make the sound a touch 'brighter' say. Figure 1 illustrates the sort of responses you can introduce with an active system; curves which aren't readily available from any conventional EQ.
In last month's HSR, the advantage of -24dB/octave slopes was briefly discussed. Although a -24dB/octave (called 4th order from now on) crossover provides electrical phase coherency at the driver terminals, the acoustic phase relationship between adjacent drivers isn't necessarily zero degrees.
For instance, if the midrange driver's response is already falling (acoustically) at -6dB/octave (ie. with a first order slope) at the crossover point, a 3rd order (electrical) slope in the crossover will be required for phase coherency. The acoustic and electrical roll-offs add, to give an overall 4th order rate of attenuation, and so phase coherency at the crossover point. By comparison, if we'd used a 4th order crossover here, the sum of the electrical and acoustic responses would be 5th order (-30dB/octave). This spells a 90 degree phase error, and some very audible colouration.
High power active systems of the professional variety tend to use drivers well within their pass band (the area where their response is substantially flat), so there's no acoustic roll-off about the crossover point to complicate matters. Nevertheless, there are often smaller phase differences between drivers, related in particular to their spatial location. Thus the 0 degree phase difference is only an approximate requirement.
In a similar vein, there are several ways of achieving fourth order crossover slopes, some of these being slightly out-of-phase, by say 25 degrees at the crossover point, depending on the damping factor of the filters. This topic is too large to be discussed in depth, but essentially, some fourth order crossovers suit certain speaker systems better than others, depending on how well their respective phase characteristics marry up. The various filter functions responsible for these differences are known as alignments, and as their names are often long winded, they're commonly abbreviated in a code, which also gives information about the slope. Alignment codes are also used to specify the characteristics of speakers, particularly at the bottom end, where of course, the roll-off referred to is purely acoustic.
Table 1 gives some of the more common alignments. Meanwhile, Table 2 delineates the differences between two common fourth order crossover alignments. Often there will be no choice; all you can do is be aware of the difference, and assess crossovers on the basis of how they perform with your speakers. With adjustable crossovers though, it's possible to ask the manufacturer to supply cards for alternative alignments, and work on a trial and error basis.
It's well known that steep filter slopes - anything in excess of -6dB/octave - can imbue sound with unnatural qualities. The -36dB/octave anti-aliasing filter used in A-D convertors is a prime example. This argument often arises in defence of second and third order crossovers (-12 and -18dB/oct), which have shallower slopes, and hence should sound better, or have better 'transient response' than fourth order crossovers. In fact, the argument is spurious; their phase will be lagging or leading by 180 or 90 degrees respectively, and these slopes therefore exhibit the worst transient response. With a competently set up fourth order crossover, the slopes aren't audible, despite their steep sides, because the integrated response is flat.
Feature by Ben Duncan
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!