OS Acoustics have followed up their impressive DB7 with an even more capable twin‑woofer design. We put the new DBS8 to the test.
Back in April 2018 I reviewed OS Acoustics’ debut monitor, the DB7, and very much admired it. The newly developed and launched DBS8 not only extends the OS Acoustics monitoring range upwards in terms of price and performance aspirations, but also offers a somewhat different electro‑acoustic design philosophy. Where the DB7 is a conventional reflex‑loaded two‑way system, the DBS8 is a closed‑box design, sporting twin bass/mid drivers in a D’Appolito configuration. These changes ought to endow the DBS8 with a significantly different performance character, both objectively and subjectively, compared to the DB7.
Like that of its sibling, the DBS8’s cabinet is unusually deep, but I was relieved to find when the delivery driver dropped off the boxes that it’s not quite as big as I feared. It’s only a little taller and wider than the DB7 and, at 15.8kg, it’s also not the heaviest monitor I’ve had to deal with recently. It’s still not a compact or lightweight nearfield monitor though, and to my mind it definitely strays into midfield monitoring territory. The enclosure itself is constructed from a mix of panel thicknesses and is braced internally, with finite element analysis apparently being used to establish the optimum brace locations. The look of the cabinet is relatively understated, other than the perhaps slightly ‘blingy’ chrome‑plated driver fixing screws, but the gently curved bevel on the front edges of the enclosure gives it a visually distinctive and attractive aesthetic. As with the DB7, the DBS8 electronics module is housed in its own isolated space within the enclosure. This is a good thing, as it significantly reduces the possibility of microphonic effects within the electronics.
The DBS8 shares both amplifier and driver technology (and their respective manufacturers) with its predecessor. Being of D’Appolito format, the DBS8 incorporates two bass/mid drivers, located either side of a tweeter. The bass/mid drivers are 18cm aluminium‑diaphragm units manufactured by SEAS in Norway, and are closely related to the unit used in the DB7. The DSB8 tweeter, manufactured by Eton in Germany, is also related to the unit used in the DB7, but its 28mm‑diameter magnesium/ceramic dome is slightly larger than the DBS7’s 26mm item. While a 2mm diameter increase might appear minimal, it actually represents a reasonable increase in radiating area, and that ought to bring benefits in terms of power handling and distortion performance. The negative side‑effect of a slightly increased diaphragm diameter — narrower dispersion above, say, 10kHz — is insignificant.
It may not have escaped your attention that both the bass/mid and tweeter diaphragms are metallic in nature. The choice of metallic diaphragm materials is made in order to delay the onset of diaphragm break‑up to a frequency above the operating band (in the case of the bass/mid driver) or above the audible range (in the case of the tweeter). A quick look at the data sheets published by the respective manufacturers reveals that the first break‑up resonance of the bass/mid diaphragm occurs at around 7kHz, while tweeter diaphragm break‑up appears to occur above audibility.
The 7kHz diaphragm break‑up of the bass/mid driver is significantly higher than a thermoplastic, paper or composite diaphragm will typically reach before it stops moving as a whole, and in fundamental terms this falls into the column marked ‘advantages’. But there are a couple of potential disadvantages. The first is that break‑up resonances in metallic diaphragms tend to be severe and, again looking at the SEAS data sheet, that’s certainly the case with the DBS8 bass/mid driver. The 7kHz resonance results in a high‑Q response peak that rises about 15dB. The only way to mitigate such a resonance is to ensure that the low‑pass crossover filter starts to do it’s thing at least an octave below, and that its filter slope is steep. That’s the case with the DBS8: its crossover frequency is at 2.1kHz with fourth‑order (24dB/octave) filter slopes. The second potential disadvantage of bass/mid drivers that remain pistonic to higher frequencies is that they tend to become noticeably directional further down the frequency band than bass/mid drivers with less rigid diaphragm materials. This, again, tends to require a lower crossover frequency to mitigate.
Talk of a lower crossover frequency brings me neatly on to the tweeter. The disadvantage of lowering the crossover frequency in a two‑way speaker is that it asks more of the tweeter, and that perhaps is one reason why OS Acoustics chose to go a step up in terms of tweeter diaphragm size compared to the DB7. Doing so will help the tweeter at the lower end of its working band where the diaphragm movement required is greatest.
Back to the bass/mid drivers, these share an approximately 23‑litre closed box that, in parallel with some gentle LF EQ, results in a claimed extension that is down just 2dB at 35Hz. As I’ve explained quite a few times in the past, a closed box ought to result in superior time‑domain performance and an absence of the distortion and compression effects often introduced by reflex ports. The downside of a closed box, ie. greater mechanical load on the bass/mid drivers, is of course countered on the DBS8 by the simple expedient of there being two bass/mid drivers. Their combined diaphragm area adds up to somewhere around that of a single 20‑25 cm driver, but it’s also significant that twin drivers double the thermal power handling of the system.
Moving on from the drivers to the DBS8’s electronics, as with the DB7, the amplifier and DSP module comes from Danish Class‑D amplification specialists Pascal. The amplification provides 500 Watts for the two parallel bass/mid drivers and 150 Watts for the tweeter, but also of significance is the DSP power incorporated within the electronics module. This provides a variety of optional frequency‑domain EQ presets and time‑domain compensation. The EQ presets offer a ‘flat’ option, a slightly warmer ‘voiced’ option and a ‘room boundary’ option that provides some compensation for the low‑frequency gain resulting from locating the monitors close to room boundaries.
DBS8 connection facilities are, as is traditional, located on the rear panel adjacent to a substantial heatsink. Audio connections comprise just a balanced analogue input on XLR with a link output. Despite its lone analogue input, the DBS8 internal signal flow is all digital and runs at 24‑bit/96kHz using Cirrus Logic A‑D and D‑A converters. A couple of network sockets are also fitted, but these are reserved for service only. Alongside the XLR sockets are the input sensitivity and EQ preset selection knobs. It’s a shame that the input sensitivity knob isn’t stepped, and also that it adjusts over a relatively wide range (30dB), because unless both monitors of a pair are set to 0dB (I can’t really imagine why you’d need to use ‑30dB), it’s not easy to be sure they’re both working at exactly the same level.
The time‑domain compensation offered by the DBS8 aims to equalise the frequency‑dependent latency effects from about 200Hz upwards that are fundamental to the electro‑acoustics of speakers and result in portions of the audible band being radiated at different times. OS Acoustics actually use the term ‘linear phase’ to describe time‑domain compensation; as I suspect the concepts of phase and time, and their relationship to frequency, cause a certain amount of confusion I’ve had a go at offering an explanation in the Time & Phase side‑bar.
Time‑domain compensation seems a relatively hot topic in contemporary monitor design, with companies such as HEDD Audio, Kii Audio and Ex Machina Soundworks all employing similar techniques. In my experience of products from all three of those companies (along with the OS Acoustics DB7), time‑domain compensation has benefits. It’s a relatively subtle subjective effect but definitely seems to help with stereo image focus and clarity of mix details.
I’ve glossed over the D’Appolito format of the DBS8 so far, and although I’ve explained it in previous reviews they were all from a little while ago, so it’s probably worth revisiting the subject. The D’Appolito configuration is named after the American speaker engineer who first drew attention to the fact that the arrangement of twin bass/mid drivers mounted either side of a single tweeter offers interesting and potentially useful characteristics in terms of dispersion, along with increasing low/mid frequency power handling. These characteristics arise in much the same way that the vertical dispersion of line‑array speakers is controlled. As a listener moves vertically away from the perpendicular axis (assuming the speaker is in portrait orientation), the path length from each bass/mid driver to the ears changes and destructive interference between the drivers begins to occur. The results of this off‑axis destructive interference is that vertical dispersion through the midrange is somewhat restricted, which helps suppress ceiling and floor (or desk) reflections. I’ll illustrate this ‘D’Appolito effect’ with some FuzzMeasure analysis further down the page.
Over years of designing and reviewing speakers I’ve come to the conclusion that the D’Appolito format does indeed offer benefits, but it’s not just that narrow vertical dispersion offers potential advantages in portrait orientation; it can do so in landscape orientation, too, by suppressing side wall reflections. So the D’Appolito format effectively provides an opportunity to orientate monitor dispersion to suit room layout and reflective characteristics.
So, to the all‑important FuzzMeasure analysis. Diagram 1 illustrates the DBS8’s on‑axis amplitude frequency response from 200Hz upwards, in both ‘flat’ and ‘voiced’ EQ modes. The first thing to note is that the response is impressively flat and well controlled. The voiced EQ simply adds a little extra low‑mid energy. All of the Diagram 1 data was taken with the DBS8 time‑domain compensation engaged. I checked and it makes little or no difference in the amplitude response. In the time domain it’s a different matter, and that brings me on to Diagram 2.
Diagram 2 illustrates the DBS8’s impulse response, both with and without time‑domain compensation. The first thing to note is that compensation inevitably introduces a little extra latency — around 5ms. Partly this is due to DSP not being instantaneous, and partly it’s due to the fact that if you’re going to equalise the arrival time of different signals, the ‘cause precedes effect’ rule of our universe requires that the faster signals are delayed rather than the slower ones advanced. The second thing to note on the two impulse responses is that the delayed one is less smeared than the earlier one; the time‑domain compensation draws together the various non‑synchronous elements of the signal.
Perhaps an important thing to appreciate with time‑domain compensation implemented in a monitor that uses multiple drivers is that it’s only ‘valid’ at centrally perpendicular measuring positions. Moving to a different measuring (or listening) position will change the relative distance between the drivers and consequently change the relative time it takes for their acoustic energy to reach the mic or listener. All time‑domain compensation can really achieve is to ensure that the audio energy across the entire frequency band at least leaves the speakers in sync.
Diagram 3 illustrates this. It shows the DBS8’s time‑domain‑corrected step response, measured on both central perpendicular axis and also 20 degrees vertically off‑axis (I’ve illustrated this with step responses rather than impulse responses because the smearing effect is easier to see). The off‑axis impulse response is less clean because the distance from the two bass/mid drivers to the microphone is now unequal and the drivers have begun to go ‘out of phase’. Ironically, in the context of time‑domain compensation, the restricted vertical dispersion offered by the D’Appolito format is also a consequence of the two drivers going out of phase away from the central perpendicular axis.
And speaking of dispersion, Diagram 4 illustrates how the D’Appolito format influences things vertically. It shows the axial frequency response overlaid with vertical and horizontal 20‑degree off‑axis curves. The 20 degrees horizontal response is hardly less flat than the axial response, but the vertical response shows a significant (10dB deep) dip beginning at a few hundred Hz and ending as the tweeter takes over above 2kHz. Further off‑axis vertically the dip will increase in depth. It’s this dip that can suppress ceiling/desk reflections (in portrait monitor orientation) or wall reflections (in landscape orientation) in a region of the music band that’s both typically very busy and corresponds to maximum human hearing sensitivity.
Diagram 5, the last from my FuzzMeasure analysis, shows a very close‑mic (less than 1cm from the bass driver diaphragm) measurement of the DBS8’s low‑frequency characteristics. A close‑mic measurement can provide a pretty accurate measure of a monitor’s low‑frequency bandwidth and reveals the DBS8 to have a low‑frequency ‑3dB cutoff at around 44Hz (a little higher than claimed) with a roll‑off thereafter of 18dB/octave. 18dB/octave is slightly steeper than the inherent 12dB/octave slope of a simple closed‑box speaker, and suggests that the DBS8 incorporates a 6dB/octave high‑pass filter at around 50Hz within its electronics. The filter will almost certainly be present to offer some protection against driver overload at very low frequencies.
One last check I made on the DBS8’s low‑frequency characteristic was to look at its low‑frequency group delay. I measured this at around 12ms at 50Hz, which is slightly higher than I’d expect for a closed‑box speaker, and actually approaches reflex‑loaded speaker numbers. There’s two things to bear in mind, however. Firstly, a contributing factor of the increased group delay will be the additional high‑pass filter I mentioned in the last paragraph, and secondly, while group delay might not be down at ‘normal’ closed‑box levels, the DBS8 won’t suffer the low‑frequency overhang and port distortion and compression that tends to characterise reflex loading. It would theoretically be possible to compensate for the DB8’s low‑frequency group delay, but overall system latency would then probably become impractically high.
As usual, a combination of my own mix work and favourite CDs provided the raw material, and I listened to the DBS8s mostly in time‑domain compensated mode (which, I can confirm, added an extra degree of clarity and image focus to proceedings). It was obvious right from the start that the DBS8 is an extremely competent monitor. It has a great feeling of authority over the music. There’s no hint of sluggishness about leading edges, or overhang when notes stop.
Starting with the bottom end, the DBS8 does exactly what’s needed of a monitor. It plays its bass without artifice or emphasis. It’s perhaps not the most generous monitor of its size and price in terms of low‑frequency bandwidth or maximum volume level, but for music mix work, I’d much rather have timing and pitch accuracy over another half octave of extension or another few dB of volume. DBS8 bass makes for a really secure foundation on which you can safely build a whole raft of mix decisions further up the frequency spectrum. If your expectations of a monitor are for generous chunks of fruity bass, you’ll perhaps be a bit underwhelmed by the DBS8; if you want to make informed mix decisions about low‑frequency mix elements, which I guess is more likely as a Sound On Sound reader, I think you’ll really appreciate what the DBS8 offers.
The level of detail rendered audible by the DBS8 is genuinely exceptional.
Moving on up the frequency band to the midrange, the DBS8 continues to offer seriously impressive clarity and detail. It’s one of those monitors that in some circumstances might actually feel a little unhelpful because it is so adept at revealing flaws (it certainly had me completely rethinking a particular mix). The tonal balance through the midrange seemed perhaps a little emphasised, which tends to exaggerate detail, but that’s no bad thing in a mixing context (it didn’t do the NS‑10 any harm), and the level of detail rendered audible by the DBS8 is genuinely exceptional. The audibility of reverb tails, for example, is particularly striking. Midrange imaging is well focused and explicit also, with mix elements properly located in their own space within the stereo landscape.
Moving again to the high end of the frequency spectrum, I remember when reviewing the DB7 being impressed with the subjective qualities of the Eton tweeter, and I feel exactly the same about the slightly larger version used in the DBS8. It delivers an impressive level of detail, but does so without drawing undue attention. There’s a satisfying sense of finesse and accuracy about it.
So the DBS8 pressed a lot of the right buttons for me. There’s great precision, detail and explicit accuracy about its performance throughout the entire audio band. Despite its reasonably large dimensions it doesn’t try too hard with low‑frequency bandwidth, concentrating instead on accuracy, and its D’Appolito format offers usefully a narrow vertical dispersion which I believe helps with room integration. Factor in very high‑quality drivers and electronics, a thoughtfully engineered enclosure and top‑notch manufacturing, and there’s very little — nothing, in fact — not to admire.
OS Acoustics describe the DBS8 as a ‘linear phase’ monitor whereas I prefer to talk about ‘time‑domain compensation’ because I think it’s an easier concept to grasp. But what’s the difference? The concept of phase is a way of linking time and frequency. Imagine a clock face with a hand spinning around it once a second. If you were to project the vertical position of the tip of the hand onto a graph, on which time is the X axis, you’d see a sine wave with a frequency of 1Hz. One cycle of the sine wave takes 1 second and corresponds to a 360° trip of the hand around the clock face. Half a cycle corresponds to 180° and 0.5s, and quarter of a cycle is 90° and 0.25s. So we have time expressed as degrees of cyclic angle. But the cycle time at different frequencies takes a different length of time. At say, 1kHz, a full cycle takes 1ms, which means that 180° takes 0.5ms. So in an electronic, or electro‑acoustic device such as a monitor, described as ‘linear phase’, the in/out latency (known technically as the group delay) remains constant with frequency, which in turn implies that the change of phase with frequency is linear.
The DBS8 is priced at a point where the competition is particularly strong. Monitors such as the PSI A‑21M, PMC TwoTwo 6, Genelec 8341A, Dynaudio Core 59, Focal Trio6 Be, Unity Audio Super Rock and ATC SCM20A are all high‑performance alternatives.
- Great bass performance.
- Impressive detail and clarity throughout the audio band.
- Useful D’Appolito dispersion characteristics.
The DBS8 perhaps marks a change of design philosophy at OS Acoustics and it’s one that to my mind has definitely delivered the goods. The DBS8 easily deserves a place, I think, among the best‑performing monitors in its price bracket.
£4999 per pair including VAT.
OS Acoustics +44 (0)845 299 0213
£4999 per pair (approx. $6900).