Just do it. It’s how teenagers dare one another to do something stupid. It’s also how deep believers in a given audiophile solution tell others to just do it. Today’s ‘it’ is the opposite of adolescent stupidity. It’s super smart. Until now it simply wasn’t that easy to actually do. What am I on about? Achieving better bass plus superior dynamic expression and lower distortion across the midband. That path leads to adaptive active bass. The vast majority of speakers behave omnidirectional at low frequencies. Their bass reflects off hard surfaces – so all walls, floor, ceiling and furnishings – in our room. While obviously delayed in time because reflections travel longer than direct sound—how to minimize those reflections acoustically I briefly touch upon in the postscript below–bass reflections add loudness. That’s called room gain. How much room gain we see and exactly where depends on the size and dimensions of our room and speaker setup. To keep it simple, here we just say that this loudness interaction between room and bass is a far bigger variable than how the mid and high frequencies bed in which ‘see’ less of the room because their dispersion pattern keeps narrowing toward the treble.
How the bass of a passive speaker beds into any given room is unpredictable. It’s why bass which can’t be adjusted independently from the mid/treble–to correct for how our room alters its response versus that of another listener’s with the same speaker–isn’t that brilliant. At the very least we should want a separate level control to tweak our hifi’s overall response where the room interferes by taking over the most. Passive speakers don’t offer that. What controls them is our master volume on a variable source, preamp or integrated. It works the treble, midrange and bass as a singular entity. To conquer that limitation, we must divide.
Passive bass suffers another weakness. It inserts inductors, resistors and capacitors between its woofer/s and amp. Those parts make up the low-pass crossover which defines the bandwidth across which the bass drivers work. Such passive crossovers soak up amplifier power, undermine their control over the woofer voice coils, inject phase shift and cause other nonlinearities. They’re middlemen that exact a toll where things get lost in translation. Active bass does away with the middlemen. Active bass inserts its filter function ahead of the amplifier. Over just a bit of wiring that now gains a direct path to the woofer voice coils for far superior control over their stop-and-go. Inherent in this concept is the woofer amp being no longer tasked to handle anything but bass. It needn’t be the world’s most refined ultra-wideband sophisticate. All it does now is yeoman duty in the bassment. We can optimize it across far narrower bandwidth with crazy power and high current likely involving copious amounts of negative feedback and class D. For just bass, beast mode is ideal.
If we add up the above, an active subwoofer becomes the obvious answer. But how do we integrate it perfectly with our speakers for a sophisticated music not movie-mayhem system? In the latter sector, preamp/processors and HT receivers already incorporate smart bass management functions to pre-empt today’s feature. In the supposedly sophisticated 2-channel and high-end realm meanwhile, things are still far more primitive. At best our preamps offer a second pre-out. Our speakers see the same full-range signal they always have. So does our sub when fed from the second pre-out. Now its plate amp’s low-pass filter is in charge of blending our sub at a hopefully mirror-imaged slope with how our speakers happen to passively attenuate in our room.
That’s not very accurate. It also relies on a plate amp’s likely DSP filter. What’s more, we’ve done nothing to ease our main speaker’s bass burdens. Where recorded, they still see signal down to 20Hz. Never mind that they can’t reproduce it. Their voice coils still warm up in response. That exacts an inverse payment of impedance. Rising voice-coil impedance begins to choke dynamic expression. Then there are larger bass excursions that attempt to track the incoming full-range signal. If our speaker is a two-way monitor or compact floorstander, its woofer is in fact a mid/woofer that hands over to a typically ~2kHz tweeter. Whatever distortion we can reduce from lowered excursions; whatever heat we can keep away from voice coils to run with less dynamic compression; it all impacts our majority bandwidth all the way up to the tweeter. That’s not a marginal win.
A superior way to integrate a subwoofer is a separate active low/high-pass filter. That splits the full-range signal into bass to the sub; and everything but bass to the speakers. No longer do our speakers dictate where we must tack on the sub to avoid undue overlap and resultant bloat. We get to decide exactly where to optimally implement the handover. Even though our speaker might be good to ~50Hz, our ears could show that it sounds rather more uncompressed, dynamic and resolved when cut at 90Hz. By using the same filter frequency and slope on speakers and sub, we achieve a properly dovetailed frequency response. We also avoid the subwoofer plate amp’s inferior filter by using its LFE 0.1 input or mode. Even better, our speaker amp no longer sees low or low/mid bass. That has it perform better. In fact, if we plan this out well in advance, we can pursue a smaller more sophisticated amp of lesser power and smaller more sophisticated or affordable speakers which will never do any real bass. Didn’t I promise that today’s just do it is very smart indeed?
Let’s get at the ‘how’. How do we generate the desired purist analog active high-pass filter for our speakers? Until now, options were slim. Beyond the €500 Sublime Acoustics, we had the €3’500 JL, spl or €4’500 Wilson Audio candidates. DACs from Bel Canto, Linn and TotalDAC implement equivalent filter options in the digital domain with dedicated hi/lo analog outputs. FutureFi integrateds from NAD and Lyndorf high-pass their speaker outputs and low-pass their sub out via smart menus. What if we want to add perfectly integrated active bass to what we already own, just for less dosh than the big boys? Now our tale catches us up with my own dilemma from a few years ago. Then I’d asked Pál Nagy, maker of the icOn range of autoformer passive preamps from the UK’s Manchester, whether he could install a fixed 40Hz 4th-order Linkwitz-Riley hi/lo-pass filter in my icOn pre. I asked for that frequency because my speakers at the time were good to 35Hz; and a 4th-order LR (defined by its -6dB point) sums perfectly in phase. Pál could do that — and my results were sterling. His were less so once some of my readers wanted their own; except at different filter frequencies. Building one-up custom boards wasn’t the best use of Pál’s limited time so he decided to expand the concept into a tidy range of so-called Gradient Boxes. Those are active analog precision filters of varying functionality. They accommodate simpler needs as well as power users like reviewers and retailers who change components like underwear to work on optimizing new combinations on a possibly daily basis. Now our meandering intro has arrived. Let’s describe the hardware that accomplishes the job we’ve set for ourselves today: to properly integrate a subwoofer (or two) into a music system.
The maxed-out Gradient Box offers one input on RCA and XLR and sub-out and sat-out on RCA and XLR. Due to space constraints, the balanced sub-out is on mini XLR but cabled mini-to-standard XLR adapters are included. A rear toggle selects between the inputs after which we can use any combination of outputs. Also on the rear is a power toggle, an auxiliary 6V power inlet for an optional linear power supply, the standard power inlet for the stock switching power supply, an update port and a sub-level pot which fully clockwise adds 10dB of voltage gain via op amp to the sub out. On the belly of the beast sit A and B bays of filter frequency sliders. Available turnovers come at 40, 60, 70, 80, 90, 100 and 120Hz. Each bay offers separate high and low pass adjustments. In belly-up mode, the frontal display auto-rotates 180° to show right-side up again.
Unlike commercial competitors, this deck has remote control over most of its functions which the display then confirms. The only features which don’t run through the microprocessor to thus be within the remote’s grasp are the manual frequency adjusters. Originally those were switched via the microprocessor but, during the pandemic, a part crucial for that became unobtainium. A fully functional prototype was therefore scrapped and a comprehensive redesign of the PCB was conducted. While switching between the A and B settings remains on the remote, which two filters settings one compares relies on manually setting them first. From-the-seat setup for the best turnover frequency thus follows a process of elimination. For example, start setting the A side LP/HP at 120Hz, the B side LP/HP at 40Hz. Back in the seat, switch between either setting until you’re clear on which is better. Leave that alone, change the other one. Rinse and repeat. For the duration of A/B comparisons, the auditions will be in mono. Once the ideal value is determined, A and B become right and left and in most scenarios will sit at the same frequency. However, it’s possible to offset L/R and HP/LP. A channel offset like 60Hz for the right LP/HP and 70Hz for the left LP/HP might account for an asymmetrical room or unequal boundary distances. An LP/HP offset could set a deliberate ‘hole’ to compensate for a severe room mode. For example, the left/right LP might sit at 70Hz, the left/right HP at 80Hz to minimize a room peak right in-between.
But there’s more. Let’s inspect the remote and clarify each button. EQ activates an 80Hz shelving filter to be either +3dB at 20Hz (EQ1) or +6dB (EQ2). That subsonic lift can compensate the roll-off of particularly dipole/Ripol subs. Mute is self-explanatory. LP- reduces the sub volume, LP+ increases it. ±VOL is the master volume to eliminate a preamp where desired. A and B we already covered. BYP removes the sub from the equation and sends the incoming full-range signal uncut to the speakers. This is ideal for comparisons between with/without sub without rewiring a thing. PH is short for subwoofer phase which can be either 0° or 180°. LP can isolate the left and right sub channel and mute the sub altogether. HP does the same for the speakers. This selective muting is particularly useful when two subs are used and we want to track best placement.
Because the Gradient Box is firmware driven, my personal unit runs the same hardware slightly differently with an alternate firmware. That’s because my analog master volume sits inside the DAC. I’d never use the Gradient Box’s master volume. Hence my version uses the high-current master attenuator chips differently. Now LP± is a pure sub attenuator up to -68dB while VOL± has converted to the equivalent HP± attenuation. Depending on which speaker/sub combo I review or listen to, either the sub or speaker amp will play too loud. That imbalance simply reflects different voltage gain in the sub/speaker amps and different speaker/sub sensitivities. If the sub plays too loud, I cut LP gain. If the speaker plays too loud, I cut the HP. The related ‘+’ buttons simply reverse direction up to zero attenuation. They never cross into active gain. So if you sit at – 11dB on the LP and decide that the sub should really play just a bit louder, hit LP+ for 1dB-stepped increases all the way back to 0 for no attenuation. If in a different scenario that’s not enough, leave the LP at 0 and instead cut the high pass to make the speakers less loud. Because Pál was happy to do it, I also asked him to change the default display of the original firmware to show the word joy rather than a number. Night mode can change that to a single small decimal point; and there’s two different display intensities.
How much you ask? With three variations of the Gradient Box on the cards, pricing at time of writing remains TBC please see Srajan’s 6moons-landed update here. For clients with simpler needs, Pál is working on a purely single-ended version with his typical autoformer master volume, one input and fixed sub/sat outs. He’s also working on a 7-filter version without A/B switching. To wrap my how-to coverage, the analog thus zero latency Gradient Box’s filter choice of 4th-order Linkwitz-Riley creates phase-consistent summing so the phase control of our active sub should sit at zero. Because our low pass now executes inside our external smart crossover, we must bypass the one built into the sub. On my Dynaudio’s menu for the low-pass settings, there’s an option for ‘none’. Other subs offer a dedicated LFE or 0.1 input which anticipates being preceded by a home-theater pre/pro with its own bass management. Our Gradient Box steps in for such a pre-pro so we too use the LFE input.
For our purple snake to swallow its tail and segue us back at the beginning, Pál Nagy’s Gradient Box concept under his Passive Preamp icOn umbrella practices purist analog apartheid. It puts us in charge to set what for our room and speakers is the ideal frequency divider. Our main speakers destress to perform better. Our bass exploits dedicated LF weaponry of heavier high-excursion cone/s mated to high power of low impedance and superior damping. We get remote control over our relative bass balance from the seat. This even accommodates late-night whisper sessions where we can compensate for our hearing’s loss of bass sensitivity by bumping up LP gain. If we plan out this 2.1 or even 2.2 music system well in advance, there should be no real enthusiasm for big floorstanding speakers which our Gradient Box would knee-cap. For the same money we can get a superior far smaller speaker that need only be good to ~70Hz if our filter setting puts it already 6dB down at 80Hz; or even higher. In my experience of doing this for two systems, an 80Hz transition does not suffer mono bassiosis as any presumed toll on imaging. That frequency is still too low to betray directional detection. With this excellent new solution on hand, it’s more opportune than ever for proponents of music subs to dare fellow ‘philes to just do it. This is not about adolescent infatuation with boom trucks though you could indulge in that, too. This is about superior ‘audiophile’ performance, period. Not only does proper full-range bass increase the perceived black value of our tonal palette. It enhances the sense of recorded ambiance and scale. And properly dovetailed, it makes our speakers more dynamic and easeful. That’s a win/win all around and how, after 20+ years on the audiophile beat, I roll these days.
Postscript. Minimizing slow bass created by DSP requires knowing our sub’s digital latency and then compensating physically. For example, my upstairs Dynaudio 18S suffers 2.5ms of latency. Sound travels roughly 1m in 3ms. It’s why my Dynaudio sits ~86cm closer to the seat than the speakers. Even if a sub were to suffer zero latency perhaps because its circuity was purely analog, placing it in a corner so farther away from the seat than the speakers recreates time delay now in the acoustic domain. Anything arriving late is slow. If you don’t want slow bass, make sure to practice proper time alignment. Minimizing slow bass created by late reflections from omni dispersion wants a sub with dipole or better yet, cardioid dispersion. Dipole’s figure-8 radiation pattern eliminates side-wall reflections to cancel our room’s lateral mode. Cardioid bass like the Kii Three BXT system or Ripol subs from ModalAkustik, sound|kaos and Voxativ even attenuate front-wall reflections to soften our room’s longitudinal mode which reduces time blur and overhang still further. I use 80Hz filters both upstairs and down. Upstairs is because its mini monitors are only good to ~60Hz. Put differently, nothing bigger was needed or desired and this smaller room benefits cosmetically from barely-there monitors. Downstairs is because of the big 2×15” Ripol sound|kaos sub. Exploiting its directionality across two octaves cancels my room’s 40/70Hz modes without any unsightly acoustic bass traps. That’s literally a huge win.
Further information: IcOn