AUDIO-TECHNICA AT-HA5050H high-res dock/preamp/DAC
With the continued rise of head-fi for those who love music as a personal experience, this thrilling headphone amplifer provides a benchmark for performance.
This thrilling headphone amplifier provides a benchmark for head-fi performance.
I’ve recently written in our sister magazine Australian Hi-Fi about the output impedance of headphone amplifiers — about how this can affect the frequency performance of headphones, and how difficult it is to even find out what the figure might be.
Well, the Audio-Technica AT-HA5050H Hybrid Headphone Amplifier and DAC hides nothing on that front. Indeed, I’d have to say that you’ll probably be unable to find a set of headphones for which this unit isn’t an ideal fit.
Equipment
The first hint about that are the eight 6.5mm headphone sockets in a slightly recessed section of the front panel. These are grouped into four pairs. Each pair is labelled with its output impedance: 120 ohms, 82 ohms, 33 ohms and 0.1 ohms. Audio-Technica says of this: “All headphones are not the same. Each model has its own character, its own sonic signature. That’s why we’ve created a headphone amp that lets you choose from four different output impedances... to experience a sound that best suits you and your headphones.”
I am of the view that the ‘correct’ output impedance for a headphone amplifier is 0.1 ohms, or otherwise as low as possible. Very few headphone makers publish their models’ impedance curves, so how significant output impedance affects their performance will be a guess. That said, this unit lets listeners experiment. In particular, quite a few headphones have a bass peak in impedance. Using one of the higher impedance outputs will boost the bass in the area of the impedance peak.
But of course there’s more to the AudioTechnica AT-HA5050H than lots of outputs. As you may have guessed from the $8999 price tag, this is a premium piece of equipment. It weighs 11 kilograms, so construction is solid. It’s also one lovely looking bit of kit. When you switch it on, the slightly warm-coloured brushed aluminium front panel is enhanced by the orange glow of two analogue VU meters. There’s a 16mm-tall wood-finish bar across the bottom front. (I assume that it’s wood, anyway. I would have had to scratch into it to be certain.) It has a near luminous, golden brown finish, and a super-gloss look like the trim on a high-end luxury car.
Between and under the VU meters is a large analogue volume control. Yes, real analogue. It has physical stops at 7 o’clock (for no sound at all) and 5 o’clock for maximum, just like an old-fashioned analogue stereo amplifier. More revealingly, there were no 1dB or 0.5dB jumps
in volume as it was adjusted. As far as I could measure, it’s continuously variable.
To the right of the volume control is a four-position rotary switch with which you can adjust the level of the VU meters. It does nothing to the sound, but it lets you keep the needles in the middle of the scale so that they flicker more attractively. As if often the case, I see the meters as an aesthetic flourish rather than useful. But a lovely flourish they are!
You might think of all that stuff as the analogue side of the panel. The left-hand side seems to be more the digital side. It has a power switch — hardwired, it seems — and an input selector with four positions. Above that are a dozen small LEDs, arranged in a six-rung ladder. One of these is labelled ‘Bit Depth’. Nine others indicate the sample rate of the PCM signal being handled: 32kHz, 44.1kHz, 48kHz, 88.2kHz, 96kHz, 176.4kHz, 192kHz, 352.8kHz and 384kHz. The top two indicate DSD64 and DSD128 signals.
The input selector matches the four rearpanel inputs: S/PDIF in the form of coaxial digital audio, USB Audio from a computer, analogue line input (via stereo RCAs) and balanced analogue line input (via XLR sockets).
The unit is missing a few connections that you might expect: no optical digital input, no support for balanced headphones and, most surprisingly, no line outputs.
Inside the Audio-Technica AT-HA5050H is what the company calls ‘hybrid’ technology. It covers the gamut from old to new, from analogue to digital. The pre-amplifier stage is tube-based. The power amplifier stage is a Class-A bipolar transistor design. It has a maximum stated output of 2000mW into 16-ohm loads, and 62mW into 600-ohm loads. It has a ‘Rated Output’ of 125mW into 16 ohms, 3.3mW into 600 ohms.
The digital stage supports up to 192kHz sampling and 24 bits over S/PDIF, and up to 384kHz and 32 bits, plus DSD64 and DSD128 over the USB input.
Installation
There was a bit of guesswork on my part in setting up this unit. There was no manual provided with the review unit, no manual was available on the product website, nor was it otherwise available.
Nor was there a CD with any drivers included, nor any drivers provided on the website. I imagine that there are supposed to be drivers for Windows at least. While Windows 10 now supports many of this unit’s capabilities, it didn’t originally, and I find Windows 10 remains a little flaky with high-end digital audio, so I prefer to use product-specific drivers if possible.
So it was a matter of plug it in and see what it would do.
I mostly used a Windows computer, although I tried a Mac later. Windows recognised the unit straight away. It automatically became the default speaker. Clicking through to the settings, Windows confirmed that it supported audio from 16 bits and 32kHz through to 32 bits and 384kHz, capturing all the important sampling rates in between.
With high-end gear like this I avoid Windows audio handling. Windows has to resample audio, mix in notifications and so on, so there’s no way it can be bit-perfect. Likewise, Windows can’t handle DSD.
I use Foobar 2000 to bypass all that. To provide good sound at anything other than standard 44.1kHz material I selected the ‘WASAPI (push):HA5050H Asynchronous PCM384/DSD128’ setting. (Note, that’s ‘(push)’, not ‘(event)’. The latter makes the sound very crunchy indeed with this unit.) You will have to have installed the WASAPI component for Foobar.
But to play DSD, one needs to install the Super Audio CD Decoder and choose the ‘DSD:WASAPI (push):HA5050H Asynchronous PCM384/DSD128’ setting.
Macs have long-supported USB Audio Class 2.0 devices, so it was comfortable with this DAC. On my Mac Mini I used Audirvana player software. It was able to deliver all the various outputs with the minimum of configuration.
I also did some listening with optical media. First I put my CD player next to it... and then realised that it only had an optical output, not coaxial. So I switched to a Blu-ray player to spin the discs. It worked perfectly with audio up to the 192kHz sampling maximum available on disc.
Performance
Listening to music using the Audio-Technica AT-HA5050H was a delight. Part of the delight was due to the indicators on the front panel. I could confirm at a glance that the audio was coming through from the computer or Blu-ray source in its proper format and correct sampling frequency. There’s no wondering with this unit.
There was one more piece of information beyond that offered by other equipment. The ‘Bit Depth’ LED glowed blue when the signal was 16 bits, and purple whenever it was anything else — 1 bit for DSD or 24 or 32 bits for PCM.
But mostly it was just the music. In addition to four different sets of headphones I had to hand — two open-backed, two closed — the distributor included the Audio-Technica ATH-ADX5000
open-backed headphones ($2699). I played a little with the higher impedance outputs, but I did all my critical listening and most general listening with the 0.1-ohm output. I really don’t see the point of burning up power in an in-line resistor without a compelling reason. And it meant no unpredictable frequency-response anomalies.
First, volume level. Even my low efficiency Sennheiser HD-535 headphones were at listenably-loud levels with the volume control around 10 o’clock. Later I would determine that clipping starts to occur with it up past 2 o’clock. That’s many many decibels higher than 10 o’clock! All the other headphones were significantly louder, even the AT model, which has an impedance of 420 ohms. I predict that anyone who uses this headphone amplifier will be pleased with the levels.
And with the detail that’s revealed, the clean, clean delivery of all manner of music. I went from prog rock — Genesis’ ‘Nursery Cryme’ opens with a very subtle equipment buzz that I’d not previously noticed — through high resolution jazz, such as Kent Poon’s audiophile recordings in DSD128 and PCM384. Thence to some Primus in standard CD format, and Alison Krauss in DSD64. Bizet/Shchedrin, back in CD format. The entire live album ‘Ricochet’ by Tangerine Dream (DSD64) and some recent acoustic work by Laura Marling (CD).
Balance was perfect. Headphones with a tendency to brightness preserved their character, but seemed slightly tamed by a seemingly limitless power and extension in the bass. Especially at high levels.
The unit is rated at 16 ohms minimum headphone loads. Which raises the question: can you use two such low-impedance headphones at a time? Or more?
I don’t know for certain. No manual, remember. My guess would be that one could fully populate the outputs. Indeed, as I’m writing this paragraph I have arrayed before me four sets of high quality headphones, some of 32 ohms impedance, each of which is plugged into one of the four lowest-impedance outputs. The unit seems perfectly happy with the situation. Of course, the relative output levels of all the headphones would be fixed. Generally, those connected to the higher impedance outputs will be lower in volume than the others (unless they have a high internal impedance). So I can’t see most people actually doing this.
Measurements
I ran some objective tests on a few aspects of performance. The first was perhaps the least important: the headphone output impedance. The outputs are, after all, labelled. But it was an opportunity to test whether my methodology is worthy of my confidence! To cut a long story short, using three frequencies (100Hz, 1002Hz and 10,000Hz) and two different load resistances (295 and 15.9 ohms), my measurements for the three high-impedance outputs averaged, rounded to whole numbers, 33, 82 and 120 ohms. The lowish RMS voltage resolution of my DSO became an issue with the 0.1-ohm output. With calculated values of 0.00 through to 0.57 ohms, and an average of 0.31 ohms with a standard deviation of 0.26 ohms, I certainly would not dispute the 0.1-ohm claim.
With 44.1kHz signals, Audio-Technica has clearly gone for a quite gentle filter (above right). The roll-off starts at around 12kHz. It’s down by 1dB at 17kHz and 3dB at 20kHz.
For 96kHz material the -1dB point is at 35kHz, with -3dB at 43kHz. With 192kHz content -1dB is pushed out to 30kHz and -3dB a touch above 70kHz.
What was surprising was that the noise floor was relatively high. I pushed the output level up quite high to ensure there was little corruption of the signal from outside sources, but the best I could manage (with 24-bit signals) was around -97dBA. As usual, I start to worry when my measurements are fairly mediocre so, while leaving the test rig unchanged, I tested a different headphone/ DAC in its place, one is less than a tenth the cost. It tested better than -107dBA.
The graph comparing the two results in instructive (see below). The cheaper DAC basically had its noise floor just under -138dB across the spectrum, apart from a couple of spikes (most likely from the ADC in the test rig). The AT-HA5050H almost reached that low level across the midrange and treble, but started to ascend below 3kHz. By 1kHz it was 18dB higher. Across the bass it was around 24dB higher.
Fortunately, even a -97dBA SNR means inaudible noise. I certainly heard not the slightest hint.
As to output levels, well, wow. Just before clipping into a 15.9-ohm load using the 0.1-ohm output, the unit was delivering well over 1500mW per channel. That’s 32dB above a headphone’s sensitivity rating. With a 295-ohm load that fell away to a ‘mere’ 105 milliwatts for +20dB. They are the highest clean output levels I’ve measured from any headphone amplifier. And purely for reference, that’s 5V RMS or more. Few modern devices manage much more than 2V.
Conclusion
The AT-HA5050H may have a small market, given the price, and being for headphone-only use, lacking a line output. But it delivers glorious sound, should suit any non-balanced headphones on the market — and yes, is surely among the most gorgeous headphone amps on the market. Stephen Dawson