Product test

High-end DAC: eargasm or money down the drain?

David Lee
8/9/2023
Translation: Julia Graham

We need digital-to-analogue converters (DACs) in order to play music. But notebooks and other devices already have an in-built DAC. I wanted to know what sets expensive DACs apart from their cheaper counterparts.

Sitting in front of me is a small black box that costs over 1,000 francs – in Germany, the price point is actually 1,200 euros. This is a DAC or digital-to-analogue converter that, like many of its kind, is also a headphone amplifier.

Before I start the test itself, here’s some background information, in case you’re not aware of what devices like this are used for.

What’s a DAC?

DAC is the abbreviation for digital-to-analogue converter. It’s essential for playing music, as sound waves are analogue. Digital data needs to be converted into an analogue signal so that your headphones or speakers can produce sound waves from it.

Digital-to-analogue conversion can be placed at different points. If you plug a 3.5 mm audio cable directly into your notebook, the conversion will take place in the computer. When you connect headphones via USB or Bluetooth, digital data is sent to the headphones and converted there. This means USB and Bluetooth headphones have a DAC built in.

However, a DAC can also be a standalone device that’s connected between the source and headphones. It receives digital data via USB, converts it into analogue signals and outputs them at the headphone jack.

Why use an external device?

Many audio purists think built-in DACs in notebooks and smartphones aren’t up to the job, and that you can hear the difference between them and external DACs. In other words, the music should sound better.

Since an external DAC is usually also a headphone amplifier, it has the advantage of providing more juice. Smartphones and, to some extent, other mobile devices don’t have enough signal voltage for high impedance headphones. They might not be loud enough, and the sound may not be optimal either – regardless of the DAC’s quality.

Another reason is that a lot of external DACs have additional connectors, so you can use different plugs without an adapter. There’s also extra features, such as a simple bass boost or a tonne of additional features, as is the case with the RME unit sitting on my desk.

Incidentally, USB hubs with a headphone output also contain a DAC. However, this is hardly likely to be any better than the DAC built into your computer. In this instance, the only advantage is that the headphone jack might be in a more accessible place.

The USB hub under the Mac Mini boasts a headphone output and therefore a built-in DAC.
The USB hub under the Mac Mini boasts a headphone output and therefore a built-in DAC.
Source: David Lee

Why buy such an expensive DAC?

That’s the million dollar question. This test should give us the answer. High-end devices are always disproportionately expensive, especially for all things audio. Because music that sounds good is so amazing that, for a lot of people, price isn’t an issue. They just want the best quality.

But as I wrote in the article below, companies don’t make DACs that sound bad these days. Certainly not external ones. This raises the question of what exact added value these expensive DACs deliver. As a side note, the RME Audio device sitting in front of me is actually on the cheaper end of the scale when it comes to expensive DACs.

I wanted to hear an expensive device like this with my own ears. Will the sound knock my socks off? Or will it be a huge disappointment? Perhaps its added value has nothing to do with the sound at all.

Getting to grips with the device

Even unboxing the RME ADI-2 DAC leaves a positive impression thanks to the remote control as well as the detailed user manual. The English version alone contains 72 pages and explains the features along with background knowledge. This is necessary as the subject matter is complex.

The device has its own power supply. In standby mode, the power unit it comes with makes funny noises, almost like a faint siren. When switched on, it doesn’t make any audible noise.

I’m a big fan of the screen on the front of the ADI-2. I like it when things also happen visually as the music plays, as is the case with records or tape machines. This lets me see the amplitude at different frequencies in real time.

There are two headphone outputs on the front. «Phones» is a 6.3 mm jack with an exorbitant 10 V output. This is significantly more than even a high-impedance headphone requires. The manufacturer was obviously aiming to supply every non-sensitive set of headphones with enough power. The second port with a 3.5 mm jack is called IEM, which means in-ear monitor. It provides less output voltage but is supposed to reduce ambient noise to an absolute minimum for particularly loud headphones.

On the back, you’ll find even more ports, including the line-out for RCA (unbalanced) and XLR (balanced), plus S/PDIF coaxial and optical. These are inputs, which also allow you to use the DAC for recording sound. But I’ll leave that for now. It’s complicated enough as it is.

Array of features

The ADI-2 boasts a lot more features than most DACs. The majority don’t have an equaliser at all. Here, you have a separate one for each output. You can even create two different settings for the left and right channels. I think that makes a lot of sense, as we don’t usually hear the same in both ears. The EQ (equaliser) is top of the range. Its five frequency bands can be adjusted in terms of frequency and slope Q. There are also low- and high-pass filters. The whole thing is nicely displayed in curves. And the settings can be saved as presets, which in turn can be used to standardise the sound profile of different headphones based on preset data.

The stereo sound range can be narrowed down to mono. The crossfeed effect simulates the sound of loudspeakers in your headphones. This means the sound from the left channel also reaches the right ear, albeit slightly delayed.

The ADI-2 remembers all settings, for example what volume a port was last set at. If you plug in a cable, the DAC switches to this source – at the last set volume.

I don’t want to list all the features here. But one thing I have to mention is that you have six different reconstruction filters to choose from. These determine how the analogue signal is formed from digital data.

Reconstruction filters in detail

The user manual goes into the various DA filters, as it calls them, in more detail. They’re known as SD Sharp, SD Slow, Sharp, Slow, NOS and Brickwall.

The manual shows the filters and their impulse responses.
The manual shows the filters and their impulse responses.
Source: David Lee

Graphics illustrate the differences in impulse response. But what does that entail? The test sound file consists of a very short, intense impulse – it’s bookended by absolute silence. It looks something like this. Theoretically, the sound pressure level should be zero until right before the impulse, then briefly go to maximum and then immediately back to zero. But in practice, this is impossible. The graphics show that there are slight swings before and after – known as the attack and release times. These imperfections are slightly different depending on each filter.

The impulse response of the Brickwall filter, according to the manual.
The impulse response of the Brickwall filter, according to the manual.
Source: David Lee

The graphics give you the impression there are huge differences. However, they’re actually very slight, as the whole thing takes place in under a millisecond. I’m not sure if you can even hear it. Latencies (site in German) of less than 10 milliseconds aren’t audible to humans. We perceive it as simultaneous.

The filters also influence the frequency response. While the NOS filter boasts the best impulse response, it loses the most volume at high frequencies, starting as early as 10 to 12 kHz. In other words, still in the clearly audible range. Conversely, the SD Sharp filter has the worst impulse response, but it doesn’t lose any highs up to 20 kHz.

Comparing the frequency response of different filters. Source: user manual
Comparing the frequency response of different filters. Source: user manual
Source: David Lee

Filter: can I hear it?

I sampled the various filters before looking at all these graphics and researching the data. Prior knowledge like this influences what you think you hear.

In terms of headphones, I mainly use the Beyerdynamic DT 990 Pro and Sennheiser HD 660S2. However, these headphones aren’t in the DAC price bracket. A Stax or Audeze set would be more appropriate. Unfortunately, I don’t have anything like that.

At first, I don’t hear any difference between the filters. With time, it becomes clear that the NOS filter sounds a bit duller than Sharp, SD-Sharp and Brickwall. The Slow filters seem to fall somewhere in between, but the differences are very subtle.

In a test I carried out with a synthesiser, I could hear frequencies up to about 16 kHz. That being said, even at 15 kHz, they need to be quite loud so I can still hear them. So it’s not surprising that I can still just about detect the NOS’ treble losses, but not those of all other filters. I’m not aware of the fractions of milliseconds in question during oscillation, even when listening to a click test.

Now that I’m writing up this report, I once again can’t hear any distinctions at all. Somehow, it seems to depend on my mood that day. What’s clear is the differences are minimal.

DAC comparison: can I hear it?

At first, I can’t unequivocally say whether I can detect a difference between the RME Audio ADI-2 DAC and my inexpensive Fiio E10K-TC. I feel like there’s a difference, but I may be imagining it. Once again, it could be partly due to the psychological effect, but also because the volume isn’t exactly the same.

I want to know for sure, so I try to do a blind test.

In the audio MIDI setup within macOS, I can create a multi-output device. As a result, the Mac emits the sound on both DACs simultaneously. This is important in order to switch quickly between them both, which in turn is essential for distinguishing differences clearly.

The same model of headphones also needs to be connected to both DACs. For one thing, every set of headphones sounds different. And also because it wouldn’t be a blind test when I can tell I’m wearing a different set of headphones.

I don’t have any duplicates. After all, it’d make no sense under normal circumstances. However, I do have three sets of the Beyerdynamic DT 770 Pro with different impedances. The 32-ohm version and the 250-ohm version sound almost the same, apart from the volume. I add the same fabric cushioning (from the 80-ohm) to the 32-ohm version so I don’t feel the difference. Then I adjust the volume using a gauge, and now I can finally get started.

My experimental set-up for the blind test.
My experimental set-up for the blind test.
Source: David Lee

I listen to five tracks. I correctly identify the expensive DAC on three of them, while I get it wrong on the other two. The three I get right also happen to be the ones I heard samples of before. The other two I had to guess without pre-listening.

I repeat the test another day, this time previewing all the tracks first. Once again, I got three right and two wrong. At 60%, the hit rate is only slightly higher than the statistical chance probability of 50%. Moreover, one of the guesses doesn’t even count at all. I didn’t hear any difference but noticed that the cable on the 250-ohm headphones was heavier, so I knew what I was hearing. For my next attempts, I added additional weights to the lighter cable.

The distinction is so slight that it could just as easily stem from the headphones. Because the 250-ohm and 32-ohm models sound very, very similar, but not exactly the same.

Long story short: I can’t reliably distinguish the expensive DAC from the cheap one in a blind test. Even when pre-listening, I could hardly notice any difference in parts.

A glance at the chips

The RME Audio ADI-2 DAC has been around since 2019. Initially, it boasted the AK4493 chip. This has been unavailable since a fire at AKM’s factory in late 2020, and RME Audio now uses ESS’s ES9028Q2M instead. This chip is also used in current FiiO devices, such as the K5 Pro ESS
and the KA3. In contrast, the FiiO I used in the blind test features the PCM5102 from Texas Instruments.

The manufacturer has done an excellent job of being transparent about the change of chip in the ADI-2 user manual. RME Audio also adds that «ADI-2 DACs with AK4493 and ES9028Q2M are neither sonically nor visually easily distinguishable.»

The ES9028Q2M supports custom filters. This means two things. Firstly, that RME Audio was able to transfer its filters to the new chip. At the same time, the company swapped out a less popular filter. Secondly, it means that not every DAC with this chip sounds the same. The FiiO DAC boasting this chip might sound slightly different, but I can’t even tell the difference from a DAC with a completely different one.

High sampling rates and DSD

The ADI-2 DAC can play audio at a sampling rate of up to 768 kHz. But I don’t have any music files with rates as high as that. It wouldn’t help anyway, because even with tracks at 192 kHz, I can’t hear a distinct difference to 44.1 kHz.

The device also processes music in DSD format. You can even record in direct stream digital (DSD) via the SPDIF inputs. But its practical use is confined within narrow limits – mainly due to the format itself. It’s practically impossible to process DSD digitally. This means any post-processed recording first needs to be converted to ordinary PCM format and then back to DSD. With any advantages of DSD lost in the process. As a result, tracks are limited to unprocessed live recordings of acoustic instruments. And even then, DSD is often converted back to PCM, for example, if you want to use an equaliser.

DSD requires a special audio player, such as JRiver Media Center, where I need to enable the bitstreaming option otherwise no DSD will go to the DAC.
DSD requires a special audio player, such as JRiver Media Center, where I need to enable the bitstreaming option otherwise no DSD will go to the DAC.
Source: David Lee

NativeDSD offers a starter pack that also includes six DSD files as FLAC to give you a comparison. However, the DSD files sound a lot louder to me, meaning a blind test isn’t an option. But given the limitations of DSD, it doesn’t matter if the files would’ve had slight advantages in terms of sound.

Verdict: stands out from the crowd but not from a sound point of view

The RME Audio ADI-2 DAC sounds excellent. However, the same can be said of much cheaper DACs. Either way, I can’t reliably distinguish this device from my much cheaper DAC in a blind test. Similarly, I can only detect at most minimal differences between the individual filters. If it were just a question of sound, the ADI-2 would definitely be too expensive compared with its competition, at least for me.

However, the device has much more to offer than a cheap 08/15 DAC. If we’re purely talking about performance, even the quietest headphones in the world will sound loud enough with this device. There’s a remote control, recording feature and an outstanding equaliser. Along with countless other functions, not all of which I’ve used, even after a number of weeks. Anyone who buys this DAC clearly gets more than they would from other models. But whether these additional features are worth the extra price is something only you can decide.

I can picture the ideal place for the RME Audio ADI-2 DAC being somewhere a wide variety of headphones are used. For example, in a music studio. The EQ lets you create headphone profiles, and both headphone outputs offer rich performance and zero noise disruption for all types of wired models.

Header image: David Lee

160 people like this article


User Avatar
User Avatar

My interest in IT and writing landed me in tech journalism early on (2000). I want to know how we can use technology without being used. Outside of the office, I’m a keen musician who makes up for lacking talent with excessive enthusiasm.

These articles might also interest you

  • Product test

    The Sennheiser HD 660S2 sounds delightful

    by David Lee

  • Product test

    Ontrac headphone review: stupendous audio quality, but the software needs improvement

    by Stefanie Lechthaler

  • Product test

    Sennheiser Momentum True Wireless 4: amazing undistorted sound

    by Florian Bodoky

Comments

Avatar