SPL Phonitor xe headphone amplifier

Glowing VU meters, a deep red face, a machined volume control fitted to an actual potentiometer, what’s not to love with the SPL Phonitor xe headphone amplifier. SPL is a German pro audio outfit which entered the consumer space around 2014. Its consumer products are grouped under its “Professional Fidelity” banner. The xe is a purely analogue unit, although the model I reviewed had the DAC upgrade module fitted.

tl;dr

• Large unit, solid construction – 278mm wide, 100mm tall and 330mm deep, 5kg
• Made in Germany
• Available with red, silver or black front panel
• Inputs on standard unit: stereo balanced XLR line inputs, stereo RCA unbalanced inputs
• Additional inputs with optional DAC768 module: USB Type-B for computers, Optical S/PDIF via TOSLINK, Coaxial S/PDIF via RCA, AES/EBU via XLR
• Outputs: 2 x 4-pin XLR balanced headphones; 2 x 6.35mm TRS unbalanced headphones
• Outputs front or rear panel, switchable
• Rated power output (balanced): 2 x 8 watts, 600 ohms, 1kHz, 1% THD; 2 x 3.5 watts, 250 ohms, 1kHz, 1% THD, 2 x 700mW, 32 ohms, 1kHz, 1% THD
• Rated power output (unbalanced): 2 x 2.7 watts, 600 ohms, 1kHz, 1% THD; 2 x 5 watts, 250 ohms, 1kHz, 1% THD, 2 x 1 watt, 32 ohms, 1kHz, 1% THD
• Measured power output (unbalanced): 2 x 920mW, 300 ohms, 1kHz, 1% THD; 2 x 263mW, 16 ohms, 1kHz, 1% THD
• Optional DAC768 module based on AKM AK4490 DAC chip
• Tested digital audio standards with optional DAC768 module installed: up to 24-bits and 768kHz PCM; up to DSD256 (via DoP only)
• No MQA support
• Interesting and effective “Matrix” analogue processing to reduce weaknesses in headphone playback
• Centre-focused balance control allows fine tuning of left vs right
• Stylish analogue VU meters with 3-position sensitivity switch
• All switches are mechanical
• Machined volume control with powered potentiometer
• No remote included, but supports third-party remote controls for volume control
• Summary: lovely sound, powerful sound, regardless of headphones. Exceptionally low output impedance and high output means you can choose any headphones you like without worrying about whether they will work with this unit. And it’s bullet-proof. If needed, it can pump out 16 volts into a pair of recalcitrant high-impedance headphones.
• $3,999 base price, DAC768 module extra
• Available here (no DAC), or here (with DAC768 module)

A little more on the SPL Phonitor xe

The above covers most of the details. But pay attention to the photos. This is a handsome device. Look, the VU meters are useless. If, that is, you consider aesthetic considerations to be useless. I don’t. A big part of high-end audiophilia is visual style. I don’t normally comment too much on looks, but if you’re going to buy one of these, get one with the red face. Really, it’s so cool. (But I’ll forgive you if, unaccountably, you wish to go silver or black).

The SPL Phonitor Xe is only for headphones. It does not have pre-amp outputs. If you want dual function, consider the SPL Phonitor X.

While I had this unit on my desk, I had the opportunity to participate in a Zoom presentation on the SPL Phonitor series from Hermann Gier, one of the founders of SPL. Amongst the things I learned was that this unit – like a lot of SPL devices – uses a 120-volt rail rather than the usual 30 volts. Herr Geir says that this allows a significantly greater signal to noise ratio than the industry norm. That 120-volt rail decision is now termed "VOLTAir Technology".

Perhaps the one feature that sets this unit apart from a dozen – hundred! – other headphone amplifiers is the “Matrix” system. Herr Gier described how some years ago he’d been preparing some music using headphones. But the result was weak bass and a generally unrealistic sound. That was because when you’re listening with headphones, your left ear is getting the unadulterated left channel and nothing of the right, and vice versa for the right ear. The Matrix system bleeds a tonally-adjusted signal from each ear to the other. Since this is a totally analogue system, I figure it does not add any time delays which could enhance the effect or perhaps move the stereo “image” out to in front of the listener.

The volume knob is “Milled from solid” what, I’m not quite sure. I’d guess aluminium More importantly, behind the knob is an analogue Alps RK27 potentiometer for controlling the output level. Some swear by digital volume control. I’m agnostic on this. A quality pot for volume control has been the norm for decades, up until very recent times. It sounded fine to me, and allowed me to fine-tune the output level quite precisely. Not being digital, I wasn’t restricted to decibel or half-decibel steps. The volume control can be adjusted by remote control. The unit learns the volume up and down codes for whichever IR remote you want to use.

Balanced headphones?

Hermann Gier also discussed the merits of balanced headphone outputs. Or, you might say, the demerits. He is far from enthusiastic.

First, he points out, “balanced” headphones aren’t balanced in the same sense as balanced connections between components. Proper balanced connections essentially use differential amplifiers to neutralise noise picked up on long lines. But that’s not what happens with balanced headphones. What these do is merely ensure that the left and right signals are entirely separate. Of course, they also roughly double the amount of power available to the headphones.

Mr Gier noted that one result of the elimination of a common return between left and right is that the baseline signals are untethered from each other, resulting in subtle phasing effects. These may be pleasing but are not accurate. I’m not certain that this is a real problem. After all, all our loudspeaker systems are “balanced” in the same sense that some headphones are “balanced”.

Finally, Mr Gier pointed out that in all the professional audio gear produced by his and many other companies, and all of it marketed to people who are extremely picky about sound quality, virtually none includes balanced headphone outputs.

Given that he was arguing against interest, that is, against a feature in some of his own products, this is certainly a very interesting perspective.

I for one have been trying to use balanced headphone connections wherever possible in recent months but have yet to become convinced of any sonic improvement, or indeed any sonic demerit. They just sound the same to me.

Phonitor Matrix

What wasn’t the same was the “Matrix” sound processing system. Perhaps the word “processing” is a little too much given that it was entirely analog. The idea behind it is that the shape of the human head modifies the signal which reaches one ear, compared to the other, depending on the angle. Treble is somewhat muted for the ear furthest from the sound source. In addition, bass tends to be emphasised with headphones, compared with loudspeakers.

The Matrix system is intended to deal with this. How did it do?

I should first note that without Matrix switched on, the sound was first class with the half-dozen different headphone models I used. It easily drove all of them with the volume control rarely straying beyond 9pm on the dial.

For a lot of music the Matrix effect was fairly modest, but on occasion it was … wow! For example, the tracks “Jesse James” and “Uncle Charlie Interview #1” on Nitty Gritty Dirt Band’s 1970 album, Uncle Charlie & His Dog Teddy, start hard in the right channel, then the voice switches over to the left channel while a banjo starts up in the right. This whole thing is quite wearing on the ear, thanks to the wide, unrealistic separation. Matrix brought everything closer to the centre and consequently made it much more listenable. And much more realistic sounding.

This is something of a reductionist test, but it was solid proof of the concept. Indeed, I’d recommend to users to get ahold of a track like this and use it to tune the Matrix settings to your preference.

One of the things which really dates the early stereo Beatles recordings is the frequent practice, then, of sticking some elements of the mix hard in one of the channels. Even with modern turntable equipment, stereo separation is rarely much above 25dB at 1kHz, and worse at other frequencies. Back in the 1960s, separation on playback equipment was presumably even worse, so it made sense to push for a really wide separation, if only to demonstrate this new-fangled stereo thingo. But on CDs, channel separation is routinely 90dB plus across pretty much the full audio bandwidth.

All of which is a prelude to saying listening to Revolver with this headphone amplifier and the Matrix engaged was a revelation. It sounded like a much more recent recording, with a sensible stereo sound stage – albeit one that was still located in my head rather than in front of me. There was even a little air around some of the percussive effects in the right channel on “Taxman” and Ringo Starr’s drums sound realistic once they’re no longer crushed over to the left.

With more modern recordings the changes are less dramatic, and for me offered less. For example, on the track “Too Many Puppies” from the Primus album Frizzle Fry I preferred to have Matrix switched off since the stereo mix, while still offering a good sound stage, is somewhat centre focussed. The cymbal accompaniment is one of the wider elements and adds interest by remaining wide. Bringing it all in towards the centre – and softening the bass slightly – worked against my enjoyment.

Which is why, of course, there’s an on/off switch for this process on the front panel.

Signal handling

I mostly used the unit with the optional DAC768 module, but also checked out both the RCA and XLR analogue inputs. All worked as they should. Since there’s an input selection knob on the front panel, you can leave all your devices plugged in and just switch between them.

I installed the SPL drivers on a Windows computer for playing high resolution audio.

I ran through all my various resolution test tracks. 44.1kHz, 16 bits? Yes, of course. 88.2kHz and 96kHz at 24 bits? Yes. 176.4kHz and 192kHz; them too. How about 352.8kHz and 384kHz. Yes. And for the first time in a review, I was able to use my newly-created 705.6kHz and 768kHz 24-bit test tracks to confirm that a DAC under review fully supports them.

How about DSD? The first track I tried failed. I went into the settings of JRiver Media Center and found that the “DSD bitstream in DoP format” box under “Device Settings” was unchecked. I ticked that box and then DSD played. That included standard DSD64 – the same stuff found on SACDs – double-speed DSD128 and quad-speed DSD256. That ought to cover everyone’s needs.

One thing I didn’t like is the absence of any indicator on the unit as to what digital audio signal it is receiving. Setting up a computer to deliver pristine digital audio to a DAC is not always straightforward, and all too easy to get wrong. That’s why I like the DAC to confirm which signal it is receiving. Still, this unit’s drivers do place an icon on your Windows computer’s status bar. Click on that and the SPL USB Audio Device Control Panel will open up. The status tab shows the current sample rate. If your computer is set up correctly, the sample rate will match the resolution of the audio playing. With PCM signals you can see at a glance what rate the driver is receiving and passing onto the DAC. Because DSD only works with this unit in DoP format – that is, disguised as PCM – the panel will show DSD64 as 176,400Hz, DSD128 as 352,800Hz and DSD256 as 705,600Hz. This is what that looks like:

See here for our guide on how to make sure your DAC is getting the correct digital signal from a Windows 10 computer.

Measurements

The first time I ran a formal test, some of the results were dreadful. The one that first caught my eye was stereo crosstalk. It was shown as just -16.2dB. Then I remembered that I had the Matrix processing switched on. Whoops! I reran the test, and with 44.1kHz, 16 bit signals – CD standard – the results were excellent. Noise was at -97.1dBA, crosstalk was in the vicinity of where it ought to be, at -89.6dB. THD was just 0.0007% and IMD was at 0.0041%. Here’s the frequency response with the Matrix processing both off and on:

The impulse response showed a standard level of diminishing rippling after the impulse, thus:

That suggests to me that the Short Delay Sharp Roll-off filter has been employed. Except that Herr Gier assured me that it uses a FPGA – Field-Programmable Gate Array – to manage this side to things. That said, the SPL website seems oddly simplistic. It says that the “analog output of the DAC chip must be filtered by a low pass filter. Phonitor xe has two of them: One for PCM audio and one for DSD audio, since different roll-off frequencies are required.” As we’ll see, it clearly has more than two, since different filters are also required for different PCM formats.

With 24-bit signals, the SPL Phonitor xe allowed through a just a little of the noise my Surface Pro 2017 picks up from my Ethernet wiring when its plugged into the Surface Dock. I use this as a test of the ability of a DAC to isolate its analogue outputs from such noise. In both cases, the noise levels were well below audibility, but it’s interesting to check them out anyway. I plugged the unbalanced headphone output into my RME ADI-2 PRO FS R BE ADC and set the xe’s output level to within a decibel under 4dBu – about 1.2 volts RMS – and the RME’s input sensitivity to match. With the Surface Pro running on battery, the noise levels were at -107.9dBA. With it plugged in, it was -105.3dBA. Here’s a graph showing the difference across (and beyond) the audio spectrum:

Several features of this graph seemed familiar, so I had a look through some recent product reviews and found that there were odd square notches and almost identical high frequency spikes in a similar measurement of the Focal Arche DAC/Headphone amplifier. It also employs the AK4490 DAC chip. A wild guess by me: could both devices be employing the same DAC board? I doubt that it’s the chip alone doing this. Whatever. None of that stuff is at all audible.

And here’s the frequency response with 192kHz sampled signals:

I suspect that the Short Delay Slow Roll-off filter may have been engaged for this one.

Measurements of the Matrix

There are two controls for adjusting the operation of the Matrix: Crossfeed and Angle. To see the effect, I ran my usual measurement for each of the six Crossfeed levels with Angle set to 22 degrees, and then for each of the four angles with Crossfeed set to the highest level.

Here’s the first of those:

Of course, what this is measuring is stereo crosstalk. The higher a point on the graph, the greater the crossfeed. The white line is the crosstalk with Matrix switched off. The rest show the differing levels. Each higher level increases the crossfeed by a decibel or two. I should note that at its worst, the system increased total harmonic distortion from an inaudible 0.0017% to an equally inaudible 0.0018%, and IMD from an inaudible 0.0026% to an equally inaudible 0.0042%. Noise was within 0.1dB across the board.

And here’s the effect of the Angle setting. First, this also seems to fiddle around a little with crossfeed:

And it tweaks the frequency balance a little:

Again, the distortion measurements are only a little changed, and not in an audible way, from what they were without Matrix switched on.

Levels and impedance

The output levels the SPL Phonitor xe were consistently impressive. Into a 300-ohm test load, the unit pumped out close enough to 75 milliwatts. That would give any such headphones a level of more than 18dB above their sensitivity rating. That output was gain-limited. That is, my digital test signals were sine waves peaking at 0dB. Which means they were as loud as a sine wave could get. And I had the volume control on the xe maxed-out, with no hint of distortion. I was wondering what it might have achieved had the volume knob been able to rotate even more.

Into a 16-ohm test load – representing about the minimum impedance you’re likely to see in earphones – the unit’s output was limited by clipping rather than gain. In the past my technique has been to turn things up as loud as they’d go, until just short of visible distortion of the waveform viewed on an oscilloscope. With this unit I’ve added another check: a spectrum analyser (on a computer, of course) with a THD display. So with this one, I turned up the level as high it would go until either the THD hit 1% or the waveform was visibly deformed. For the most part 1% THD was the limiting factor, but with the 10kHz test signal, the waveform was looking wonky – although not clipping – well before 1% THD was reached.

I mention all this so you can judge the worth of the measures for yourself.

Anyway, the 100 hertz test signal managed a 339mW output, or +25.3dB. At 1,002 hertz the output was 263mW/24.2dB, and at 10,000 hertz a relatively modest 101mW/20dB. As I keep pointing out, there is no real-world signal which would call on a headphone amplifier to deliver any but the tiniest fraction of that output at 10kHz. Real world audio falls away at around 6dB per octave, so the 10kHz level will be at least 36dB less than the 100Hz level. If you were destroying your ears with a signal making full use of the 339mW into 100 hertz – that’d be an output of 125dB for modestly sensitive headphones – the 10kHz component of the signal would require less than 0.1mW.

But impressive as those output levels are, there was one problem. They didn’t come close to what the SPL Phonitor xe is specified to deliver: 5,000mW into 250 ohms and 1,000mW into 32 ohms are the closest specified impedances.

I consulted the manual. It turns out that there are two DIP switches on the underside of the unit. One changed the default input sensitivity on the RCA sockets from -10dBV (0.32 volts RMS) to 0dBu (0.77 volts RMS). Ho hum. But the other boosts the headphone output level by an incredible 22dB.

So a re-measure was in order. Into the 300-ohm load, the output was no longer gain limited. It was instead limited by a weird distortion appearing in a part of the waveform. I couldn’t get anywhere near 5 watts, but it did deliver a very strong 965mW/29.8dB at 100 hertz, 922mW/29.6dB at 1,002 hertz and 706mW/28.5dB at 10kHz. There was no change into the 16-ohm load, since that had been limited by clipping or other distortion, not by gain.

Last measure: I calculated the output impedance to be 0.32 ohms. If that isn’t the lowest value I’ve ever determined, it’s very close to it. And for this, lower is most definitely better.

Conclusion

The SPL Phonitor xe produces lovely sound, powerful sound, regardless of headphones. It’s exceptionally low impedance output means all headphones will receive a perfectly balanced signal. And its high output means you can choose any headphones you like without worrying about whether they will work with this unit. If needed, it can pump out 16 volts into a pair of recalcitrant high-impedance headphones.

Plus, it looks cool in a nicely retro sort of way.