You are here

Ex Machina Soundworks Pulsar

Three-way Active Monitor By Phil Ward
Published March 2021

Ex Machina Soundworks Pulsar

The physics of loudspeakers have hardly changed in a century — but DSP and material science have come a long way, and the Pulsar takes full advantage.

The last pair of monitors I reviewed were priced around what I consider to be the entry level for serious active monitoring. The next pair, and the subject of this review, retail at about six times that price — entry level they are not. They are the Pulsars, from New York‑based startup Ex Machina Soundworks. Although pretty large for a nearfield monitor, and heavy too, at 25kg each, at first glance they look relatively plain: near black, rounded‑edge boxes with a couple of drivers on the front and a metal panel on the back. But delve a little deeper and there’s some serious DSP and materials technology within. A graphene tweeter, for instance...

Before I get on to the specifics, a correction: I said that the Pulsar front panel carries two drivers, however one of those drivers is a 176mm‑diameter, compound dual‑concentric unit, in which the tweeter is mounted at the apex of the midrange diaphragm. So the Pulsar is in fact a three‑way system. Beneath the compound driver (or alongside it, if you install the Pulsar in landscape orientation) is a conventional, closed‑box‑loaded, 210mm aluminium diaphragm bass driver. There is no reflex port or ABR.

The rear panel houses switches to select between high‑latency and low‑latency operation, and LF correction options for full‑, quarter‑ or half‑space operation.The rear panel houses switches to select between high‑latency and low‑latency operation, and LF correction options for full‑, quarter‑ or half‑space operation.The aluminium plate around the back serves the dual purposes of amplifier heatsink and connection panel. Despite its 192kHz internal digital signal flow, the Pulsar offers only a balanced analogue XLR input. Also unusual is that there is no input sensitivity adjustment, and I found the sensitivity subjectively a little lower than normal (see ‘Sense & Sensitivity’ box). However there are a couple of option switches, one offering sub‑300Hz EQ profiles for corner, wall‑mount and free‑space installation environments, and one offering low or high latency options. If you’re wondering why on earth you might want to use monitors on high latency mode, it’s because the Pulsar’s DSP power can be used to equalise both amplitude and phase response. This time‑ and frequency‑domain correction results in 40ms input‑to‑output delay, which is potentially an issue if the Pulsar is to be used for tracking or in an audio/visual context. In the low‑latency mode, only the amplitude response is equalised, leaving the time domain response au naturel and yielding an input‑to‑output latency of just 2ms. I’ll investigate the latency options with a little FuzzMeasure analysis a few paragraphs down the page.

Graphene Tweeter

So, to the Pulsar’s graphene tweeter. If you’ve not had your finger on the pulse of high‑end materials science in the last 15 years you might not know that graphene was first isolated by Andre Geim and Konstantin Novoselov of the UK’s University of Manchester. In pure form, graphene is a sheet material formed by a hexagonal lattice of carbon atoms just one atom thick, so despite the extraordinary mechanical properties that make it, in terms of tensile strength and stiffness, the strongest known material, it’s not a lot of use. However, it can be incorporated in a number of different ways into composite sheet materials that, while not as mind‑bendingly strong as the pure form, still offer mechanical properties that few other materials can approach. The tweeter diaphragm of the Pulsar is made from one of those materials: GrapheneQ, manufactured by Ora Sound in Canada.

Ora Sound describes GrapheneQ thus: “GrapheneQ is over 95 percent graphene oxide by weight and is formed by depositing flakes of graphene into thousands of layers that are bonded together with proprietary cross‑linking agents.” The result, say Ora, is a sheet material that’s relatively easy to form but has bending strength of up to twice that of aluminium, combined with around 60 percent of the density. Let’s call it twice as strong and half as heavy. GrapheneQ also displays an order of magnitude higher self‑damping and offers very high thermal conductivity.

The implications of GrapheneQ for driver diaphragms, in particular tweeters, are significant. The high bending strength means that a typical 25mm tweeter dome will remain pistonic (ie. moving as a whole rather than with some parts of its surface area lagging behind others) to well above the limits of audibility, and the high self‑damping means that when the tweeter dome finally reaches a resonant mode, its Q will be lower than would otherwise be the case. The other two GrapheneQ characteristics also result in significant benefits. Reducing the moving mass of a diaphragm has a fundamental pay‑off in terms of efficiency and extended frequency response, and improving its thermal conductivity will help with power handling.

So GrapheneQ sounds as if it’s the perfect material from which to make a tweeter diaphragm, however there’s more to designing a high‑performance tweeter than specifying a high‑tech diaphragm material. For example, quite a few of what are generally regarded as the best moving‑coil tweeters around (those from Dynaudio or ATC, for instance) employ soft textile dome diaphragms with mechanical properties that could hardly be further from GrapheneQ. They can be made to work so effectively because numerous aspects of tweeter design (the voice‑coil and magnet architecture, dome suspension and management of rearward radiation, for example) can all be just as significant as the diaphragm material. A GrapheneQ tweeter dome is undoubtedly noteworthy, but unless the rest of the design and engineering is up to the job, it’s not a magic bullet.

Tex Support

It’s not only the tweeter that employs a high‑tech diaphragm material; the midrange driver does also. The material goes by the trade name Textreme and it’s manufactured by a Swedish company called Oxeon. Textreme comprises a thin, woven, carbon‑fibre‑ply composite that is said to offer significant advantages over traditional carbon‑fibre‑based diaphragms in that it has a significantly higher ratio of fibre to epoxy binder. This makes the diaphragm both usefully more rigid and lighter, while still offering good self‑damping, so that when it does enter resonant modes towards the top end of its frequency band, it does so relatively benignly. Having said that, Pulsar claim that the midrange driver diaphragm remains pistonic to two octaves above its operating band.

Somewhat ironically, a midrange driver with pistonic motion extending up to higher frequencies than is usual will likely require a crossover to the tweeter at a lower frequency than might otherwise be the case. The reason is that, in remaining pistonic, the directivity of the diaphragm will be defined primarily by its 140mm diameter, which will result in off‑axis loss beginning to kicki n not far beyond 1kHz. Less rigid midrange diaphragms that lose their pistonic motion at lower frequencies often delay the onset of directionality because, as the outside region of the diaphragm becomes decoupled from the inside region nearer the voice coil, the effective radiating diameter becomes smaller. The Pulsar’s midrange/tweeter crossover frequency is specified as 2kHz (with extremely steep, 48dB/octave filter slopes), so it will be interesting to see, when I come to taking some FuzzMeasure data, if there is any sign of an off‑axis dispersion dip just below that frequency.

Another unusual element of the Pulsar midrange driver is its small ‘underhung’ surround component. The surround on a bass/midrange driver is one of the most complex and important components, not least because it has to do two very different but equally vital jobs. First, it has to allow the diaphragm unimpeded axial freedom of movement at low frequencies. Simultaneously, it also has to provide damping at midrange frequencies to dissipate the vibrational energy that travels through the diaphragm. As soon as you go for a three‑way speaker architecture, though, and remove the low‑frequency duty from the midrange surround, its job becomes far easier and it can be made both much smaller and exclusively suited to a midrange role.

This is one reason why the Pulsar midrange surround is small, but a second reason, one that also accounts for its underhung profile, is the dual‑concentric nature of the Pulsar’s midrange and tweeter drivers. In this kind of dual‑concentric arrangement (pioneered in the 1980s by Laurie Fincham and his team at KEF), the midrange diaphragm effectively behaves as a waveguide for the apex‑located tweeter, so its profile is vitally important (another benefit of the Textreme material is that its rigidity enables the diaphragm profile to be optimised for the tweeter waveguide role). However, a large surround component located around the circumference of the diaphragm can make a complete mess of the waveguide function by creating a big circumferential diffraction feature. This results in major response aberrations, whereas making the surround very small, and recessed beneath the plane of the diaphragm, effectively takes it out of the equation.

In the context of the Pulsar’s use of a dual‑concentric mid driver and tweeter there is a further benefit that arises as a result of its three‑way format: the midrange diaphragm movement is massively reduced by it not being required to play bass. Almost the last thing you need of a midrange diaphragm working as a tweeter waveguide is for it to pump backwards and forwards at low frequencies. Significant level‑dependent amplitude modulation and intermodulation distortion would then result, as the relative position of the waveguide and the tweeter change. With only mids to reproduce, the diaphragm movement is likely to remain less than ±1mm and the resulting distortion effects made innocuous.

Sailing The High SEAS

All of which brings me to the Pulsar bass driver. The driver is a custom‑manufactured 220mm unit from SEAS in Norway that’s very obviously designed for low‑frequency duties. In the same way that the three‑way format enables midrange drivers to be optimised for midrange, so it is for bass drivers. A glance at the huge roll surround on the Pulsar bass driver shows just what can be done if a bass driver doesn’t have to worry about midrange. The driver has an aluminium diaphragm and is capable of ±14mm linear excursion – that’s an unusually big number. The crossover from bass to midrange is at 200Hz and, as with the mid‑to‑high frequency crossover, the filter slopes are notably steep at 48dB/octave.

The bass driver is, as I mentioned earlier, loaded by a closed box so its low frequency roll‑off will be relatively shallow and its response to transients will be unhindered by the delay, overhang and miscellaneous non‑linearities that can afflict ported or ABR systems. In its phase‑equalised mode, Ex Machina specify Pulsar phase linearity of ±15 degrees down to 35Hz. This, by my calculations, equates to a maximum low frequency group delay of under 2ms, which is not far short of an order of magnitude less than a typical reflex‑loaded monitor (and about half that of a Yamaha NS10 at 70Hz). And while I’m on the subject of the Pulsar specification, Ex Machina claim harmonic distortion at an impressively low <0.5 percent from 100Hz upward.

SHARC Attack

There’s still two remaining elements of the Pulsar engineering and technology that I’ve not covered so I’ll get to them now. Firstly, as I’ve mentioned in passing, the Pulsar relies heavily on digital signal processing to flatten both its amplitude and phase (time) responses. It employs SHARC processors from Analog Devices to handle the amplitude and phase EQ, the crossover filters and protection limiter functions, and AKM Velvet Sound A‑D and D‑A processors to convert between analogue and digital domains. Downstream of the DSP, the Pulsar employs three channels of Hypex Ncore Class‑D amplification rated at 250W, 100W and 75W for the bass driver, midrange driver and tweeter respectively.

Finally, on the subject of the Pulsar’s design and engineering, its enclosure is neither an MDF or birch ply construction. Instead it’s fabricated in a proprietary wood composite material that goes by the name of Valchromat. Valchromat is described by its manufacturers as an “evolution of MDF”. It is more dense than MDF and offers higher mechanical rigidity, but perhaps just as significantly, it is available in a range of colours that extend through its thickness, so Valchromat can be profile machined without needing to be re‑finished. Valchromat, as used in the Pulsar, quite definitely gives the monitor a ‘hewn from solid rock’ feel, and adds not a little heft.

Call The Fuzz

Diagram 1: The Pulsar’s frequency response from 200Hz‑20kHz, measured on axis (blue) and 20 degrees horizontally off axis (purple).Diagram 1: The Pulsar’s frequency response from 200Hz‑20kHz, measured on axis (blue) and 20 degrees horizontally off axis (purple).

As is now traditional, I explored a couple of aspects of the Pulsar’s performance with some FuzzMeasure analysis, and the results are presented here. Diagram 1 illustrates, in blue, the Pulsar’s axial amplitude frequency response in high‑latency, phase‑corrected mode, from 200Hz to 20kHz. It’s one of the flattest responses I’ve measured and other than the odd wiggle fits between ±1dB limits. That’s really outstanding. Diagram 1 also shows, in purple, a 20‑degree off‑axis curve, and that’s just as impressive. There’s a slight hint of the beginning of midrange driver directionality showing just above 1kHz, and a slight lump that appears between 7kHz and 8kHz in the tweeter band, and of course, the very top end of the tweeter decays a little, but the Pulsar’s off‑axis response is just as well sorted as the axial response. Being a dual‑concentric mid/HF design, the Pulsar’s off‑axis response won’t in practice vary much between the vertical and horizontal axes.

Diagram 2a: The Pulsar’s impulse response, in its low‑ (orange) and high‑latency (blue) modes.Diagram 2a: The Pulsar’s impulse response, in its low‑ (orange) and high‑latency (blue) modes.

Diagrams 2a, 2b and 2c illustrate some of the consequences of the Pulsar’s DSP‑based phase equalisation. Correcting phase effectively means equalising the consequences of the various frequency‑dependent time‑delay phenomena that occur naturally within all speakers. So, Diagram 2a illustrates the Pulsar’s raw impulse response in unequalised (orange) and phase‑equalised (blue) modes. An extra 32ms (less than the specified 38ms) of overall latency in phase‑equalised mode is clearly apparent. Also just about apparent is that the phase‑equalised impulse response (blue) appears to be more compact.

Diagram 2b: An impulse response, measured in low‑latency mode and zoomed in. The arrival times of the three drivers are clearly visible.Diagram 2b: An impulse response, measured in low‑latency mode and zoomed in. The arrival times of the three drivers are clearly visible.

To investigate that further, Diagrams 2b and 2c show the two impulse responses separately on an expanded timescale. It’s now obvious that the phase‑corrected impulse response is significantly less smeared. In the unequalised impulse it’s actually possible to discern the signal arrivals from the three drivers: the first peak is the tweeter, the second peak is the midrange driver (with a bit of tweeter resonance/diffraction imprinted) and the third peak is the slower‑to‑arrive bass driver. In the phase‑equalised impulse (Diagram 2c), the separate signal signatures are combined into one. This is exactly what I’d expect to see from phase equalisation. The big question is: will the equalised and unequalised modes sound different?

Diagram 2c: In high‑latency (phase‑ and EQ‑corrected) mode, all three drivers reproduce the impulse simultaneously.Diagram 2c: In high‑latency (phase‑ and EQ‑corrected) mode, all three drivers reproduce the impulse simultaneously.

Before I answer that question there’s one more FuzzMeasure curve to describe. Diagram 3 shows the results of positioning a measuring mic right up against the bass driver diaphragm. This technique provides a reasonably accurate measure of the Pulsar’s low‑frequency roll‑off (the falling response above 70Hz is a measurement artefact and doesn’t really exist; the faster roll‑off at 150Hz however is real and reveals the bass/mid low‑pass crossover filter). The technique works quite well with a closed‑box monitor but isn’t really practical with a ported or ABR‑loaded monitor because it often isn’t possible to get the measuring mic equally in the very close field of driver and port/ABR simultaneously. The close‑mic measurement here reveals a ‑3dB low‑frequency bandwidth to around 44Hz (‑6dB at 32Hz), which isn’t too far from Ex Machina’s spec claims. The curve also shows the expected 12dB/octave high‑pass roll‑off of a closed‑box speaker.

Diagram 3: A close‑mic measurement of the Pulsar’s LF driver.Diagram 3: A close‑mic measurement of the Pulsar’s LF driver.

Listening In

So there’s a lot of technology involved, but of course that doesn’t necessarily mean the Pulsar is going to sound any good. We can rest easy though, because the Pulsar sounds remarkable. It’s difficult to know where to start. Perhaps the most striking aspects are the midrange detail, image focus and depth, and a quality of vocal reproduction that makes well recorded voices sound spookily real. My partner, who works in theatre sound design, spent some time using the Pulsar during the review period working on a recorded two‑part conversational play, and on numerous occasions while overhearing her working, my brain would tell me that there were two other people in our house. An ability to reproduce voices with such a degree of accuracy, apparently completely free of speaker artefacts, is an extraordinarily difficult trick for any monitor, and to my mind is perhaps the ultimate test. Not only is the human voice an incredibly complex mix of acoustic elements, in both time and frequency domains, but evolution has ensured we’re finely tuned to be sensitive to every smallest detail.

Perhaps the most striking aspects are the midrange detail, image focus and depth, and a quality of vocal reproduction that makes well recorded voices sound spookily real.

The Pulsar’s ability to pull off the voice trick so well is I think down to three aspects of its design and technology. Firstly, the combination of a resonance‑free midrange diaphragm material and surround in a dual‑concentric arrangement, with a genuinely pistonic tweeter diaphragm. Secondly, the use of a properly inert cabinet material that doesn’t play along. Thirdly, phase compensation. The last of those factors is perhaps controversial because the audibility of phase ‘errors’ has long been a matter of intense debate in speaker design and psycho‑acoustic circles, but for me, there was a significant subjective difference between phase equalised and unequalised modes. In the latter mode the Pulsar lost a little of its magic on voices: they became simply less convincing — a little dulled and with something slightly awry in the balance of consonant and vowel sounds. The stereo image became slightly more diffused too.

I’ve concentrated on the midrange and vocal performance of the Pulsar so far, but that is not the only striking aspect of its performance. The Pulsar’s bass, for example, is not only immensely powerful and extended but has a faultless grip on dynamics and pitch. Bass sometimes felt a little detached from the lower midrange, but I suspect that is because the Pulsar doesn’t suffer the cabinet panel resonance and driver distortion ‘fug’ of less capable monitors. When material comes along that lives in the upper bass/lower mid band, the Pulsar plays it appropriately and any thoughts of detachment dissipate completely.

Back on Pulsar bass, in my workstation environment it felt occasionally a little too ambitious and I quickly settled on using the wall‑mounting LF EQ option, and sometimes even the corner option, to drop the level a little. In bass bandwidth terms the Pulsar definitely lives in midfield rather than nearfield monitor territory, and I think it ideally needs a little more room to breathe than my 4 x 5 m studio room offers. I didn’t quite have review time available to experiment with Sonarworks or Dirac Live to see if they would help the Pulsar integrate with the room more effectively, but I suspect both would do so. If the Pulsar were to be considered for monitoring in a small(ish) space I think experiments with Sonarworks, Dirac, ARC or even Trinnov products might be worthwhile.

I’ve not mentioned the Pulsar’s high‑frequency performance so far, but, in truth, there’s little to say. The exceptional performance of the Pulsar on vocals couldn’t happen without all the high‑frequency details being present and correct, and the tweeter simply does what’s necessary without drawing particular attention to itself. If you do give the tweeter some attention, however, it is extravagantly detailed yet naturally balanced and smooth in its general tonality.

I’ve experienced very few other monitors that offer so much easy‑to‑hear mix detail over such a wide bandwidth as the Pulsar. It doesn’t seem to care what genre of material you throw at it, or how loud you wish to play it, the Pulsar just spreads everything, warts and all, out in front of you without apparently adding anything of its own character. As a nearfield/midfield mix tool I think the Pulsar probably has only a few equals, and even fewer betters.

Sense & Sensitivity

Having been slightly surprised that the Pulsar offers no input sensitivity adjustment, and then finding it subjectively a little quiet, I enquired of Ex Machina what their reasoning is. This was their reply: “There is a bit of a lower input sensitivity compared to some other speaker models on the market. This is more noticeable side by side with another monitor. It’s something that can be adjusted by all but a couple of monitor controllers, using the gain settings for their different outputs. One of the reasons we don’t offer trim adjustment is because the individual signal‑to‑noise calibration is designed to make sure the power average is all but identical from unit to unit. We felt this obviated the need for trim adjustment. The input sensitivity overall is designed to allow for +14dBu of input at full scale, which we felt better leveraged the +18dBu outputs that are common on modern converters. Especially with the advent and spread of digital attenuators, we wanted to keep the standard operating range of the monitors in the highest‑resolution portions of attenuators throws.”


The Pulsar is a remarkable monitor but it is expensive. If you’re lucky enough to be considering such an investment there are some other similarly remarkable options that you’ll want to hear: the ATC SCM25A, the Kii Three, the Dutch & Dutch 8C, the Neumann KH420, the Genelec 8351A, and the PSI A23‑M, for example.


  • Incredible detail and stereo imaging.
  • Deep, accurate bass.
  • Natural tonal balance.
  • Very low levels of midrange coloration.


  • None.


In the past I’ve felt that cutting‑edge technology and electro‑acoustics don’t always get along. That’s not the case with the Pulsar. The combination of its DSP correction and driver diaphragm technology, that’s pretty much as advanced as things get, results in one of the most capable nearfield/midfield monitors that I’ve had the privilege of hearing.


£7999 per pair including VAT.

SX Pro +44 (0)800 6522 320.

$7999 per pair.

Ex Machina Soundworks +1 646 662 7041.