Focal Arche with Focal Celestee headphones

Focal is the French firm known for first class loudspeakers and headphones. It has made occasional forays into electronics, especially with devices that are tuned for its own products. That brings us to the Focal Arche DAC and headphone amplifier. Within its settings it has particular modes for several of Focal’s higher headphone models.

Focal Arche with Rubik's Cube for scale

I have previously written an extensive review of the Focal Arche DAC in September 2019 for Australian HI-FI, where it was published some months later. I considered simply posting the review here but there are complications. Sometime around the start of 2020, before the review was published (and before I was paid for it) Australian HI-FI was acquired by the British parent company of What Hi-Fi? While I had never assigned copyright of my material to AHF prior to the acquisition, I was required to do so in a contract with AHF’s new owners in order to continue publishing there. Read the whole 3,000+ word original review at What Hi-Fi? (Which, I note, kind of rudely does not show me as author.)

What follows is an update with some new measurements, some things I didn’t cover in the original review and a surprising finding that a former problem seems to have disappeared. Incidentally, with the original review I used an RME Babyface Pro analogue to digital converter for the measurements. I’ve upgraded to an RME ADI-2 PRO FS R Black Edition. That means lower noise floor, and support for 384kHz and even 768kHz sampling. As we’ll see, that lower noise floor makes a marked difference in the measurements.


  • Inputs: 1 x USB-B (for computer audio), 1 x optical digital audio, 1 x coaxial digital audio, 1 x analogue stereo (RCA)
  • Outputs: 1 x 6.35mm stereo headphone, 1 x balanced headphone (4-pin XLR), 2 x line output (unbalanced RCA and balanced XLR)
  • Claimed Digital Formats: PCM to 32 bits and 384kHz sampling; DSD to 11.2MHz
  • Confirmed Digital Formats: PCM to 24 bits and 192kHz sampling; DSD to 5.6MHz
  • Excellent performance with plenty of power
  • Excellent isolation from noise, especially on XLR outputs
  • One significant usability problem in that volume control is for both line and headphone outputs. Great care should be taken if switching between modes
  • Dimensions: 200mm wide by 60mm tall by 297mm deep; 321mm tall with headphone stand fitted
  • Weight: 4.65kg
  • Price: $3,999
  • Available: here

A brief description of the Focal Arche

See the dimensions above. This is a solid slab of electronics. What’s that, you say, about “321mm tall with headphone stand fitted”? Why yes, included in the box is a shapely strap of aluminium which slots into the heatsink on the top and holds your headphones, when not in use, in a most attractive and space-efficient way. (The same shapely piece of aluminium, supplied with a base, sells for $449 as a standalone product.) At your end is a blue-on-black display, a digital volume control, a 6.35mm standard headphone output and a 4-pin XLR output for balanced headphones. The volume knob also acts to set the unit to standby, wake it up, invoke the menu and select items therein.

Focal Arche front

At the far end is a USB Type-B input for connecting to your computer (or other USB Class 2.0 Audio source, such as an iPhone … but see below). There are also:

  • Two RCA sockets for analogue audio input
  • One RCA socket for electrical S/PDIF from, for example a CD player
  • One TOSLINK socket for optical S/PDIF

Between the two ends are two AKM AK4490 high-end DAC chips, and a pure Class A, dual mono amplifier capable of a claimed one-watt output into 32 ohms.

The XLR and phono line outputs are permanently ruled by the volume control. There is unfortunately no setting to bypass that. If you are using the Focal Arche in a dual capacity, you’ll probably want to put the output up to or near maximum for your main system. Take great care to bring the volume down a long, long way before you plug in your headphones! Plugging in headphones does not mute the line level outputs.

Focal Arche back

There’s a “Low Gain/High Gain” switch in the setttings. This applies equally to the headphone and line outputs. There is about 12dB difference in level between the two settings.

Testing a few capabilities

Since I now have the ability to capture 384kHz sample signals, I now have the ability to measure them. So I duly tried sending a 24-bit, 384kHz PCM signal to the DAC. JRiver reported that it would not work. So I tried with DSD128 and DSD256. DSD128 worked, DSD256 didn’t.

JRiver error message for 384kHz PCM

This is quite the mystery. When I reviewed the unit earlier, I wrote: “Windows reported supported PCM resolutions of 16 and 24 bits from 44,100 up to 384,000 hertz.” Now it’s reporting support of PCM resolutions of 16 and 24 bits from 44,100 hertz up to 192,000 hertz. The unit is running on the same firmware – 01.01.08 – which remains the most up to date. The dual AKM AK4490 DAC chips definitely support those signals, so I suspect a driver problem.

JRiver error message for DSD256

Does this matter? Well it does if you have an extensive collection of PCM352.8 and PCM384 and DSD256 content. Do you? I don’t. I only have a few test tracks in these formats which I found by dint of a lot of searching. I think the great majority of even high-def fans will be fine with PCM192 and DSD128.

For a while I just assumed that the DAC would work with a portable device. To check, I plugged the Lightning Camera Adaptor into my iPhone and the USB cable from the DAC into that. “Cannot Use Accessory”, said the iPhone, continuing that “FOCAL ARCHE AUDIO 2.0: This accessory requires too much power.” Although why it would require any is beyond me. The adaptor I used also had a Lightning socket, so I supplied power to the iPhone that way, upon which the iPhone happily used the Arche as a DAC.

iPhone message about needing power

Measured performance

I didn’t redo the measurements of the output impedance and power levels. The earlier measures stand. In “Voltage” mode – the output impedance is around 2.5 ohms. That’s a little higher than I would prefer, but still low enough to have virtually no effect on the sound of most headphones. In the Hybrid or Elear modes, the output impedance was 10 ohms. In the Utopia, Clear, Elegia or Stellia modes, it was 17 ohms. For most headphones, just leave it in Voltage mode.

There are two gain levels. In low gain Voltage mode the Focal Arche delivered around 12mW, or 10.8dB above headphone sensitivity rating, into 300-ohm loads, and 170mW or +22.3dB into 16-ohm loads.

Focal Arche front with balanced headphone plug

In high gain voltage mode – I was only able to grab a few measurements because the output protection system would kick in – expect around 195mW into 300-ohm loads, which is +22.9dB. And for 16-ohm loads, the Focal Arche can manage 1.7 watts, which is +32.2dB.

Were things to remain linear, which they surely wouldn’t, that would mean that the Campfire Audio Solaris 2020 in-ear monitors, driven this way, would deliver up to 151 decibels SPL. Heh.

I should note at this point, the unit runs pretty hot, even when idling. That’s the nature of Class A amplification: it biases the output above the maximum voltage it can deliver so it’s running a near full power the whole time.

Noise and frequency response

The Focal Arche adopts fairly standard treatments for filtering the output. With 16-bit, 44.1kHz (ie. CD standard, which is how the overwhelming majority of music you will listen to is presented) signals, the output was flat to 10kHz, then drooped a little to -0.25dB at 20kHz, with a brick wall filter kicking in pretty much immediately above that. Rather than showing the identical performance delivered via the headphone, RCA and XLR outputs, I just chose the XLR outputs so that you could see the left and right channels. They match to within, what 0.01dB? Effectively perfect:

Focal Arche frequency response, 44.1kHz sampling

How about high resolution 192kHz signals? The output was similar up to 20kHz, down by 0.25dB at that point. It was down by 1dB at a hair under 40kHz and -3dB at 66kHz. If you’re concerned about whether or not you’ll be able to hear 90kHz tones, why yes you will … if your ears are up to it! The output is at -6dB at 90kHz. Note, this graph shows the high-resolution frequency response to be effectively identical regardless of whether the signal is emerging from the XLR, RCA or (unbalanced – I still haven’t developed the gear required to measure the balanced headphone output) headphone outputs:

Focal Arche frequency response with 192kHz sampling

With CD-standard signals, the noise levels were down around the theoretical minimum possible: -97.7dBA, ±0.1dB depending on whether they were coming via XLR or RCA, or whether the Microsoft Surface Pro computer supplying the digital signal was connected to power or Ethernet. If you look at the graph, you’ll see that the XLR output was very slightly better, but I defy you to hear the difference:

Focal Arche noise performance, 16 bits

It was when I checked noise performance with a 24-bit signal that I was surprised. You see, in the original review the product performed quite poorly from its RCA outputs back then. With the XLR outputs, it scored what I thought to be an impressive noise level of -108dBA. And that was regardless of whether my Surface Pro 2017 was plugged into power and a noisy Ethernet, or running off battery and connected via a nicely isolating Wi-Fi.

But when I measured the RCA outputs, things weren’t so good. Here’s what I wrote back then:

“But when I was using the RCA outputs, things were very different. First, they were inconsistent. From measurement to measurement [the noise level] ranged from -79dBA to -97.6dBA. Again, it didn’t matter whether the computer was plugged in or not. To double check that, I pulled out a network streamer and connected it to the Focal Arche by optical digital audio, just to ensure that there was no way any electrical interference could be carried. It resulted in a middling -89.8dBA noise level.”

This is the kind of stuff that gives me heartburn. Could it be that my measurement methodology was messing things up somehow? I switched in another DAC with RCA outputs. Its noise level measured at -114.5dBA!

I fully expected similar weak results this time around. But I was wrong. First, with the XLR line outputs it still made no difference at all whether my Surface Pro 2017 was plugged into power and Ethernet or not. It kept all that noise completely away from the analogue outputs. In both cases it measured at -118dBA, a 10dB improvement over my previous measurements. I put that down to a lower noise contribution from my new ADC.

But how about the RCA outputs? This time around, well, yes, there’s a little more noise with these outputs, but the -112.4dBA result is significantly better than the XLR result with old measurement regime.

And even the 6.35mm headphone output managed utterly inaudible noise results of around -100dBA (measured at one volt output, which fits nicely into the 4dBu minimum level setting of the RME ADC). Here’s the graph showing the noise levels at various output states:

Focal Arche noise performance, 24 bits

I guess the one remaining point of interest is that strange noise at the extreme right of the graph. But if you can hear extremely quiet – -90dB at their worst – sounds between 60kHz and 70kHz, I’ll just have to doff my hat to you.


I liked the Focal Arche back in 2019, but was concerned about that apparent noise form the RCA. Now I kind of love it.

DacsDesktop headphone amps and dacsFocal