Author: Mark Wheeler - TNT UK
Written: March-June, 2006
In part 1 we learned of the woeful inadequacy of the passive crossover, that burdens so many domestic loudspeakers, and touched on the vast tranche of problems such a clumsy archaic device adds to the task of music reproduction in our homes. In February '04 Scott Faller wrote eloquently for TNT-audio about the full-on suck of passive crossovers, and things haven't got any better since then. Home cinema, and some stereo (the 60s combination of Jordan Modules and Basset Bass springs to mind as an excellent example), systems reduce the demands of big bass reproduction from amplifier channels and speakers by provision of active subwoofers, but that is only half a solution if the front speakers in your 5.1 set still have passive crossovers.
Ted Jordan used nifty line-arrays of 50mm drivers that managed 150Hz to 20kHz sans x-over back in the 60s. Is it ironic the the inability of small passive speakers to reproduce cinematic climaxes (surely secondary to the visual action) should prompt this domestic acceptance of active operation, rather than high quality music reproduction for its own sake? The background rumble of a screen explosion is apparently enough to motivate manufacturers to persuade the public to buy decent-sized active bass reproducers in a way that never occured to them to hear music properly reproduced in the home. That famous movie mogul Sam Goldwyn once said that no-one ever went out of business by underestimating public taste, and moments like this in the domestic audio-visual scene seem, depressingly, to confirm this.
Domestic audio has seen the active crossover appear within the upgrade hierarchy only from manufacturers whose emphasis has been on system building rather than universal compatibility. From the early days, in the first age of flared-trousers and platform-soles, the Linn-Naim upgrade ladder included the Naim NAXO active crossover with both the Linn Isobarik and Linn Sara speakers, long before Naim made their own speakers. In the late 70s Nytech included active crossovers at much more modest price levels, and even Clive Sinclair (yes, he of '70s affordable pocket calculator, ZX Spectrum computer, and C5 urban transport fame) included active filters in his low-cost modular audio range. Linn and Naim are still two big players who keep the active option as an important step in their upgrade ladders even if they're not making them for each other anymore.
In the 1980s, while the British audio dogma dictated a hierarchy of Linn-Naim (six-pack 'briks) or Pink-Exposure that both culminated in big multi-amped active systems, and the Meridien alternative jumped straight into active classics like the Meridian M3, without pasing go, but certainly parting with much more than £200. The domestic audio world has moved a long way since then and now punters seem happy to drop 40k€ on a pair of 3-way loudspeakers driven by a 12k€ power-amplifier via a large cluster of giant chokes (inductors) & capacitors & wire-wound resistors that would look more comfortable in a certain 193## Fritz Lang movie. When did the paradigm shift back to this arrangement? Is the ONE-AMP+MULTI-DRIVERS hegemony really a superior alternative?
Big multi-way speakers, with complicated passive crossovers gradually evolved, in the late 60s to mid 70s, to address perceived problems of driver bandwidth, intermodulation distortion, horizontal dispersion, cone colouration and power handling. Sadly the increasing complexity of manipulating audio signals in the whole-volts domain through large componant passive filters had the unexpected, and I suspect largely unnoticed, effect of sapping all the life from the music itself. I well remember an enthusiastic demonstrator of a well-regarded complicated 3-way speaker with high-slope filters telling me that the reason I found them to be unexciting compared with my own simple 2nd order 2-way speakers, was because the complicted 3-way system would be less distorted and less coloured, and that I was missing the distortion. I was actually missing dynamics and phase coherence; the fancy 3-ways were simply less musical.
The late 70s backlash against this design trend brought to the UK market fierce little belters like the Linn Saras and big simple 2-ways like the Mission 770. Some of these were so idiosyncratic that they demanded only specific source and amplification components to sound tolerable. The Saras were part of the Linn-Naim hierarchy and their ultimate incarnation used the Naim NAXO active crossover in place of the Linn passive x-over, and set up like this they could really sing.
Liberated from the bottleneck of the passive crossover, the NAXO fed the Sara's parallel-driver isobarik low-impedance loading directly with a Naim NAP250. Now they could work in a way that was unbelievable to anyone who had suffered their passive incarnation. Distortion, transient response and even phase response were now much improved, even if the active little Linn Saras (named after Ivor's then baby daughter who was also described as small and loud) still made no sacrifices on the alter of accurate frequency response.
Yes, in theory...
Just measure the voltage/frequency curve at each driver's terminals, and emulate that using an interstage filter. Simple, really?
If the passive crossovers are removed from a conventional system and replaced by active
x-overs and multiple power amplifiers the system will change radically. This is not a small effect in the order
of a new mains wire or interconnect. This is more radical than a new amplifier. Sometimes even the basic
character and voicing of the speaker will dramatically alter, often unpleasantly. This is because a well
designed passive system takes into account the series resistance of the main bass driver's series inductor
(first or second order x-over) or two series inductors (third or fourth order x-over). These big inductors
commonly have resistances of over 1 ohm.
The Q of the bass alignment is altered by the series resistance between the driver and the driving amplifier. Obviously the output impedance of the amplifier also has an influence on the magnitude of the effect. The designer of a commercial loudspeaker system juggles the cost of expensive low-resistance inductors with the costs of other performance influencing components. It is possible to reduce the size and the dc resistance of a given inductor value, but at the expense of other trade-offs. The designer chooses what they believe are the right set of compromises to fit the chosen performance/price priorities for that system. By removing the passive crossover you just upset the delicate balance of decisions.
Once an active filter is installed in the system it is possible to rebalance a new set of decisions that will restore, or even create for the first time, accurate voicing and frequency balance. Removing some cabinet volume by placing objects (kid's wooden blocks are about the right size) inside, one- at-a-time until the bass alignment is similar to the original, works for some sealed systems as the series resistance of the bass inductor merely affects the system Q and power handling in this respect for sealed boxes.
The quality of bass is not just defined by its Q (even if Q does stand for quality). The amplifier (regardless of its output stage topology) is now in a better position to start and stop the driver movement with the signal. Furthermore, the lack of reactive components between amplifier and driver results in the phase shifts merely being those of the driver itself (easily read from the impedance curve). The bass "group delay" problems often described by reviewers are simply these phase shifts caused by the combinations of crossover inductors and capacitors with the shifts between inductive (slope up with f increase), capacitive (slope down) at bass resonance (twice over in reflex or ported speakers) of the drive unit impedance, which then becomes gently inductive with increasing frequency.
Already, from the above, active drive should have a big effect on transparency and fidelity to the original signal and we're only an octave above the fundamental bass resonance of the system.
Connecting an amplifier directly to each driver goes several steps in the right direction. Furthermore, each amplifier is driving only a part of the bandwidth. Thus a design that excels at delicacy and low distortion can be specified for the treble while a powerhouse can be specified for the bass. Providing their basic voicing and total circuit input-to-output phase shift are similar they will integrate well. Their individual gain can be adjusted for the sensitivity of each driver.
If you rip the passive crossovers out of your loudspeaker cabinets, the speakers will sound better anyway as the x-overs are in a less hostile environment. Replace them with filters between pre-amp and power-amps, and what will you expect to hear?
The first big difference listeners tend to notice is the bass. Comments range from bass-shy to lean or lightweight. The system Q has just reduced, perhaps (in a good system) from a maximally-flat Q=0.707 to a lean Q=0.5. This could reduce bass output by up to 3dB over a whole octave. A sealed system now needs a smaller box to return to the original balance. A ported system could use an increase in port length to achieve the same result.
Alternatively, the loudspeaker system probably had a Q more like 1. This gives a boost to the response around the system resonant frequency and makes the bass sound fuller and more impressive in the shop. It is less accurate than Q=0.707, and also sounds slower. In this case the active conversion improves the bass without any disadvantages. There are arguments that with good bass extension the best transient response and PRaT is achieved by a Bessel function Q=0.577.
Dynamics are different with active speakers. Macro-dynamics tend to be much better, especially if cheap passive x-over components had been used. The whole gamut of problems caused by heating all those coils and resistors, causing compression selectively in different parts of the frequency spectrum vanishes. When the bass coil heats up and its resistance increases, causing bass peaks to compress, the treble is unaffected so the passive crossover modifies the frequency response according to signal level. When the treble resonance- tuning LCR network heats up it shifts the filter frequency too.
Micro-dynamics are much better too. They are more consistent because they are less affected by the variable heating of the crossover, and the voice coils, at any given frequency or time affecting the tiny parts of the signal while the simultaneous big swings heat the crossover. Thus with a passive filtered speaker, the subtle nuances of a soloist's great performance are being modulated by the accompanying orchestral climax. Hence the active crossover speaker has more accurate instrumental timbre. intermodulation distortions tend to be better controlled, which improves pitch stability. Timing benefits from better phase performance of filters terminated by pure resistance in the active filter compared with the drivers' reactive load presented to the passive filter.
That dramatic quality improvement that accompanies the giant leap (not a small step) from passive to active loudspeaker operation really has to be heard. Naim always recommend that it is undertaken at the end of their passive upgrade pathway, but my experience, even with Naim amplifiers, is that two lesser stereo amplifiers actively driving a pair of 2-way loudspeakers is superior to a better stereo amplifier feeding a passive crossover with the same loudspeaker. Driving my late 80s Decca London ribbon treble (a development of the original Decca Kelly DK30) with Focal 10N501 bass-mid, crossed over at 1.8kHz, I compared various active and passive set-ups. The passive crossovers were designed by me, with advice from Malcom Jones, and modified from Decca's own design 4th order 2.5kHz treble section (with a mutual-inductor increasing the rate of roll-off), and a 3rd order low-pass bass combining with the driver roll-off to emulate 4th order. This was constructed with the best air-cored inductors, Solen polypropylene capacitors and no resistors. It was designed to be single-wired or bi-wired or bi-amped with the simplest of connection changes. The crossover was mounted in a sturdy external box and all wire types were matched.
The active crossover boards were constructed, to the same voltage curves as the passive, by Falcon Electronics. Their 24v power supply needs were fed by Naim SNAPS power supply as a family match with the, similarly SNAPS fed, Naim 42.5 pre-amp, and the boards of the crossovers were populated by similar branded and specified components as the Naims, so I could be more certain that differences were due to the crossover changes.
Despite Naim's admonitions to the contrary, two modest NAP110 power-amps (horizontally split), were compared against a NAP250 (chrome bumper). Frequency response in-room was the same. While I could hear the greater authority of the Naim NAP250, the whole active set-up was so much clearer and more musical, that there was no contest. At that time the pair of used NAP110 cost much less than one used NAP250.
The two NAP110 were tried bi-amped through the passive crossovers. This was a significant improvement over the single NAP110. The NAP250 into the passive crossover was the best of the passive crossover arrangements, but not as good as the active crossover pair of NAP110. This was so much better than any of these that I wished to try a pair of NAP250 with the active filters, but I could not borrow a second NAP250.
Horizontal split (one stereo amp for L&R bass, the other for L&R treble) was reccommended by Naim dealers at that time for their NAXO set-up. I compared horizontal split with vertical split (one stereo amp for left channel treble & bass, the other stereo amp for right channel treble & bass). Horizontal split sounded much more musical and revealing of subtle nuances and details and micro-dynamics. Vertical split sounded better at big-scale demands and macro-dynamics and produced a much wider and much deeper soundstage. In the context of Naim amplification the horizontal split makes more sense, playing to its strengths.
Doug Dunlop, of Concordant Audio, set up 4 Exultant power-amps and an Exquisite pre-amp in his own listening room to try the options with the Decca/Focal speakers. The comparison was passive vs active using four monoblock power amps in each configuration. The Exquisite had 2x2 pre-amp outputs making this an easy task to switch between each configuration. To my ears (and another listener present) there was absolutely no contest. The active crossover blew the passive crossover into the weeds in every respect. Doug kept complaining that he thought he could "hear transistor" from the solid-state active crossovers, which was an anathema to him. I too could hear the crossover transistor electronics, and understood what he meant, but it did not detract from the otherwise big improvement for me. Doug chose to use the passive crossovers with the Decca/Focal speakers at the Chesterfield show that year. If he'd made a valve active crossover I expect he would have changed his mind!
I cannot think of any reason why the upgrade path ever stops short of active loudspeaker operation, except for the vested interest of manufacturers in keeping all their components universal so that separates buyers remain on the upgrading treadmill at one-componant-at-a-time.
It may simply be a failure of imagination by manufacturers, or retailers, but it is just as likely to be the innate coinservatism of audio buyers. We don't accept new ideas easily. despite the superficial gloss of monthly me-too improvements in the audio media, this is just a veneer of piecemeal improvements to existing technology. The most couture of the really-high-end uses technology from the 1930s!
The cheapness & cool-running of digital amplifiers ofers the best opportunity to adopt active loudspeaker operation as the new hegenomy. Equalisation at line level, or in the digital domain, has to be more cost effective than lumps of glop on drive unit cones. The long-throw bass unit may at last be liberated from intermodulating midrange signals as compact 3-way speakers with x-over frequencies below 200Hz become more viable and less hampered by passive filters.
It is very hard to accept any passive system as being any more than "good for a passive", once listenners have experienced an active system at a similar quality point.
Some Music I have enjoyed more with active loudspeakersFirst four extensively played during those Decca/Focal experiments...
Test Recordings used that emphasised the difference
Copyright © 2006 Mark Wheeler - www.tnt-audio.com