PSI A23-M Review “A Resounding Success”

PSI A23-M Review

Swiss speaker gurus PSI have never failed to impress us with their technical innovations, and their latest model is no exception!

Swiss–based PSI Audio are one of a group of smaller high-end monitor manufacturers that perhaps fly somewhat under the radar of larger, higher–profile brands, but that doesn’t mean that they’re not highly capable engineering-led organisations. As evidenced by the new A23‑M, little could be further from the truth. I last reviewed a PSI monitor back in the November 2013 issue when I looked at the A25‑M and, as it happens, was extremely impressed. Now, all of five years later, I’m looking at the A25‑M’s new sibling, the A23‑M. Although smaller than the A25‑M, the A23‑M still probably falls more towards the midfield monitor category than the nearfield, and in terms of both weight and size it’s just about as big as I can comfortable fit in my studio room either side of my DAW. It’s not really a monitor for bijou studio spaces, so be warned. It’s also a monitor that demands deep pockets.

Let’s Twist Again

The three driver components of the A23‑M are arranged in a similar manner to the A25‑M, a manner that seems to be becoming increasingly popular. A bass driver is accompanied by a mid–range/tweeter module, which can be rotated by 90 degrees. Rotation of the mid/tweeter module is useful because it facilitates installing the A23‑M in either portrait or landscape orientation, while retaining vertical alignment of the mid–range driver and tweeter. This is important because the wavelength at the crossover frequency between the two drivers is typically comparable to the physical distance between them, so at positions vertically off axis, their individual path lengths to the listening position will vary by a significant proportion of wavelength, and that will result in phase changes and response anomalies. Now, while such response anomalies can be relatively benign when vertically off axis, they could constitute a bit of a problem in the horizontal plane, so rotating the mid/tweeter provides a simple fix.

Perhaps you’re wondering why the orientation of the mid–range and bass drivers is not so important? It’s simply because the wavelength at their crossover frequency is much longer, so off-axis changes in path length to the listening position constitute a significantly smaller proportion of wavelength.

Dome On The (Mid–)Range

I’ve spent a paragraph on the arrangement of the A23‑M drivers without describing the drivers themselves, so I’m going to put that right now. Somewhat unconventionally, however, rather than start at the top or the bottom, I’m going to start in the middle. The middle is particularly interesting because it’s where PSI’s major technical development since the A25‑M is to be found: their in-house mid–range dome driver. Most mid–range drivers in moving-coil speakers are conceptually smaller versions of bass drivers, and that’s certainly the case for the A25‑M. Being smaller versions of bass drivers, they tend to display the same set of problems — it’s just that the problems kick in at a higher frequency. Dome mid–range drivers are more akin to larger versions of dome tweeters, and while they display a different set of problems, they also have a couple of inherent advantages. Firstly, their total diaphragm and voice–coil mass tends to be significantly lower than equivalent cone drivers (significantly less than half is typical), and secondly the fact that a dome diaphragm is driven around its edge rather than from its centre means there’s, in theory, far closer control of diaphragm movement.

Despite their inherent advantages, however, dome mid–range drivers have a reputation among speaker engineers for being difficult to get right. It’s not a trivial job to realise the full benefit of the advantages such that they outweigh the potential issues. And those issues? Well, the two significant ones are firstly that in having a much lower moving mass, dome drivers have a much higher fundamental resonance, and this will limit their low–frequency bandwidth and power handling. A cone mid–range driver will have its fundamental resonance most likely below 100Hz, which means its operational bandwidth can extend easily down to take over from a bass driver at, say, 300Hz. A typical dome mid–range driver’s fundamental resonance is likely to be north of 300Hz, which means the bass driver bandwidth will be required to extend somewhat into mid–range territory. The second potential issue is that getting a mid–range dome to perform well at the upper end of the required bandwidth is tricky. Typically, as frequency rises, the centre section of the dome will begin to become mechanically disconnected from the outer edge and cause a variety of response and dispersion anomalies.

The solution is brute force engineering of the iterative trial and error kind, and that’s both time and resource intensive. Having hinted that speaker engineers tend to shy away from dome mid–range drivers, those who have accepted the challenge and succeeded in making the concept work have come up with some extremely fine drivers, and it’s no coincidence that dome mid–range drivers feature in some extremely well-regarded monitors from, for example, ATC, PMC and Neumann, to name just three.

At 56mm in diameter, PSI’s dome mid–range driver is a relatively small device compared to, for example, ATC’s renowned 75mm dome, and not unexpectedly that means the bass–to–mid crossover frequency is set at a relatively high 620Hz. The PSI dome also takes a very different approach to diaphragm material compared to the ATC driver, by employing a thermoplastic material where the ATC dome uses a coated fabric. One advantage of the relatively small mid–range dome is that it enables the mid driver and tweeter to be mounted relatively close together. This helps with vertical dispersion consistency around the crossover frequency.

The natural details of female voices and higher–pitched acoustic instruments, if they have survived the tracking and mixing processes, are reproduced spectacularly. And if they haven’t survived the tracking and mixing processes, the problems are obviously revealed.

All About That Bass

A bass–to–mid–range crossover at 620Hz means that the A23‑M bass driver is required to make a significant contribution beyond the low–frequency end of the audio bandwidth. Middle C, remember, is at 261Hz, so the A23‑M bass driver is more often than not going to be responsible for a major proportion of the actual ‘tune’ in pretty much any piece of music you care to imagine. So, if pushed, I think I’d be tempted to call the A23‑M’s 210mm bass driver a ‘bass/low-mid driver’. Said driver is a relatively conventional–looking unit sporting a polypropylene diaphragm terminated by a rubber surround mounted in a traditionally profiled die-cast chassis.

Having said that mid–range domes can be viewed as larger versions of dome tweeters, it’s good to be able to report that this seems to be the case with the 25mm A23‑M tweeter. The tweeter diaphragm material appears to be the same on both drivers, and the relationship of surround and dome dimensions also looks similar. It seems to me that PSI have taken their tweeter expertise and scaled it up for their mid–range driver, and this is no bad thing because my previous experience of the A25‑M tweeter was very positive. Both the tweeter and mid–range drivers are loaded by a short conical waveguide. The waveguides will offer a degree of horn loading and consequent sensitivity increase in combination with some narrowing of dispersion.

Let It Flow

The A23‑M’s low–frequency system is port–loaded, with the port itself located alongside (or, in portrait mode, beneath) the bass driver. The port is created by a shelf and slot incorporated within the enclosure in the same manner as the port in the A25‑M (and, it has to be said, quite a few other monitors), but the A23‑M incorporates an innovation that PSI call ‘Flow Guide’. The concept of Flow Guide is to add a profiled wedge within the port that, effectively, results in the port’s cross-sectional area varying with length. The aims are to maximise laminar airflow while at the same time suppressing organ pipe resonances. Diagram 1 shows the output captured by a measuring microphone inserted a couple of centimetres into the port, and it’s clear that the latter aim has been achieved. The port output is very clean, with no evidence of organ-pipe effects. The port tuning frequency just above 40Hz is clearly apparent too.

Diagram 1: The A23–M's output, measured a few centimetres inside the port.

Diagram 1: The A23–M’s output, measured a few centimetres inside the port.

On the more fundamental question of the pros and cons of ported monitors, PSI’s stance is that if ports are designed appropriately and thoughtfully, both in terms of their dynamic behaviour (the low–frequency delay and overhang that they add) and of their potential distortion and compression effects, there’s no reason not use them. I guess the proof is in the listening, but one thing that is apparent from the published specification, and my analysis with FuzzMeasure, is that, for a ported monitor, the A23‑M has relatively low levels of LF group delay (around 12ms at 50Hz). This, combined with the obvious care taken in the Flow Guide technique, augurs well for dynamic, accurate and tuneful bass.

Living In A Box

So I’m halfway through and, having been distracted by the A23‑M’s drivers and its low–frequency characteristics, I’ve not really finished describing the monitor. While the A23‑M standard finish is PSI’s corporate maroon (as pictured), my review pair were in a silver–flecked black (which I actually much prefer). The enclosure is of conventionally manufactured rectilinear MDF construction with slightly rounded front edges. The enclosure panels feel rigid and well damped.

The A23‑M follows the usual format in having an amplifier heatsink and connection panel around the back. Along with the connection sockets found there, input sensitivity and low–frequency level adjustment knobs are also present. The A23‑M amplification is rated at 140 Watts, 70 Watts and 50 Watts for the bass driver, mid–range driver and tweeter respectively. Interestingly, while the bass and mid–range amplifiers employ Class–G and Class–H variable–supply–rail technology, the tweeter amplifier is a traditional Class–A/B design. I came across a similar mix of amplifier technologies on the KEF LS50 Wireless, where the bass/mid driver used a Class–D amp and the tweeter a Class–A/B. A further interesting amplifier feature, one that’s widely employed by PSI, is their Adaptive Output Impedance (AOI) error–correction feedback system. AOI is actually an enhanced version of the ‘Current Drive’ idea that I believe was first floated by Malcolm Hawksford and Paul Mills in an Audio Engineering Society paper of 1989. It’s also employed on the Kii Three monitor. The idea is to sense the current flowing in the driver voice coils by measuring the voltage across low–value series resistors. Now, as driver output is actually proportional to the current flowing in the voice coil (rather than the voltage across its terminals), the sensed current signal can be fed back to the amplifier and employed for error correction. It’s a neat idea and its use by PSI speaks of an appreciation that active speakers should be considered as integrated systems rather than, as all too often seems to be the case, drivers and amplifiers that just happen to share a common enclosure.

Finally, there’s one more subtle feature that I really like: the electronics module is housed in its own enclosure, insulated from the acoustic battering it would otherwise receive from the rear of the drivers, and it’s compliantly mounted too and so is also isolated from mechanical vibrations.

Electric Avenue

The electronics within the A23‑M are analogue all the way through and incorporate only a balanced input on an XLR socket. Monitors that remain resolutely in the analogue domain, in defiance of a trend towards internal D–A, DSP and A–D stages, are becoming less common. On this subject, Roger Roschnik, PSI’s CEO, explained to me that the PSI philosophy is very much based on optimising the analogue electronics and electro-acoustics before considering any opportunities that DSP might offer. It’s a philosophy I have some sympathy with: a better mid–range driver, for example, more fundamentally advances the art than DSP–based correction of the existing one. That’s not to say that all applications of DSP within monitors are misguided, the Kii Three (again) being a prime example of where DSP brings significant benefits.

Diagram 2: The on– and off–axis (20 degrees vertically) response of the A23–M.

Diagram 2: The on– and off–axis (20 degrees vertically) response of the A23–M.

Despite being entirely analogue, however, the A23‑M does incorporate some signal processing beyond the expected crossover filtering: it’s phase compensated from 200Hz upwards so that the various time–domain errors introduced by the drivers and filters are equalised. Although, as with any multi-driver speaker, such time–domain compensation begins to become less effective off axis as relative driver path lengths to ears change, it is still, I feel, a very worthwhile element of signal processing.

Along with measuring the A23‑M’s port characteristic (Diagram 1), I used FuzzMeasure to analyse a couple of further characteristics. Firstly, Diagram 2 shows the A23‑M’s frequency response from 200Hz upwards on axis and, simply for interest’s sake, 20 degrees vertically below axis (in landscape mode). There are no surprises: the axial response is impressively flat to within just over ±1dB, and the off-axis curves shows the expected dip around the mid–range/tweeter crossover region followed by reduced tweeter level further up the band. The off-axis curve is very tidy in the grand scheme of things and suggests good driver integration.

Diagram 3: The A23–M's frequency response, in its default setting (purple) and with the LF response set to -10dB (blue).

Diagram 3: The A23–M’s frequency response, in its default setting (purple) and with the LF response set to -10dB (blue).

Diagram 3 shows the A23‑M’s low–frequency character, measured using a microphone very close to the monitor and equidistant from the bass driver and the port. The A23‑M is unusual for port–loaded speakers in that it is possible, thanks to the port and driver being adjacent, to use a close-mic technique to get a reasonably accurate picture of low–frequency performance. I measured the nominal low–frequency cutoff (-6dB) at 36Hz, which isn’t too far from the published 34Hz. The low–frequency roll-off slope appears to be around 36dB/octave, which suggests that the A23‑M includes a 12dB/octave high-pass filter (the natural roll-off of a ported monitor is around 24dB/octave). The blue curve in Diagram 3 shows the effect of the A23‑M’s rear–panel low–frequency level control set at minimum (-10dB). For what it’s worth, because all rooms and installations are different, I preferred the LF level set to -4dB.

Listening In

Which brings me finally to what I found on listening to some old favourite CDs and Pro Tools sessions. Some monitors have quirks that require a little learning, but not the A23‑M. I felt completely comfortable with it almost immediately, but more than that, the A23‑M displays an air of quiet competence. It’s not at all showy in its tonal balance; there’s no emphasised mid–range or overcooked tweeter, it seems completely trustworthy. And behind the reliability, there’s a serious ability to resolve detail and offer clarity over a very wide bandwidth. Elements of a mix, from low frequencies right through to the limits of audibility, are clearly revealed, with explicit clarity, at any volume level from sensible to really very loud. But it’s in the mid–range that the A23‑M really excels. The natural details of female voices and higher–pitched acoustic instruments, if they have survived the tracking and mixing processes, are reproduced spectacularly. And if they haven’t survived the tracking and mix process, the problems are obviously revealed. The new dome mid–range driver is clearly a success.

And the A-23M’s subjective bass performance? Well, as a dyed-in-the-wool fan of closed–box monitors, I’d love to hear a non-ported version of the A23‑M, but with my hopes for its performance raised by the low group‑delay specification and careful port design, I was not disappointed. The A23‑M’s bass seemed subjectively beyond reproach — tuneful, dynamic and very extended.

It’s not just in terms of tonal characteristics that the A23‑M excels, however: stereo images are tremendously well focused with great portrayal of depth and space, and not just centrally but right across the sound-stage. To my mind, strong stereo image performance such as this requires a monitor to display both a tight time–domain performance and low ‘noise’ — so no spurious resonant output from cabinet panels, for example. The A23‑M is extremely good in both those respects and I think it shows. It’s actually extremely good in pretty much all respects.


If you’re in the fortunate position of considering the A23‑M, there’s a few other very fine monitors you also probably ought to be considering from manufacturers such as ATC, Dutch & Dutch, Genelec and Neumann, to name just four.


  • Wide bandwidth.
  • Faultless electro-acoustic performance.
  • Great imaging and clarity.


  • None.


There’s clearly a huge level of electro-acoustic know-how and experience built into the A23‑M, and it’s undoubtedly apparent from the first note. It’s a resounding success.