Audio Reconstruction

This is the second post in a three part series, after Rabbit Holes.

I had not wanted to invest too much into a music player before even knowing if there’s a well audible difference with music sources using higher sampling rates – as theorised last year – so I had chosen a low cost one.

Now it was time to gear up. Gear is cool. :) Not that expensive equipment and instruments are automatically good. But when only using cheap ones you never know if you’re missing out as you cannot compare. So an upgrade was inevitable. Can you feel how happily I wrote this?!

My focus is on portable devices. I live outside, the whole year, and a music system in my living room would not serve me well. The verandah is my de facto living room. Also, a portable playback system is less demanding. I use in-ear monitors (IEM), which are easier to drive than loudspeakers. Of course, I also lose out on certain experiences, such as the bodily perception of the music, the reverbs in the room, and picking up sounds in the left channel with the right ear, and vice versa. But then again I also avoid many problems that can result from exactly these conditions and effects.

Electrically controlling a loudspeaker membrane well is so much more demanding than the tiny drivers inside an in-ear monitor. In general, making good electronics becomes way more difficult if you have to push out a several Watts, not just about one or three tenth of a Watt (100 to 300 mW).

In-ear Monitors

My Unique Melody in-ear monitors are custom-made to fit my ears perfectly. You visit an audiologist to get impressions made of your ear canals and the outer ears using some gooey material that hardens into shape within a few minutes. The audiologist then creates a 3D laser scan of these impressions,¹ and sends the data to the manufacturer. I think today they at least partly use 3D printing to create a shell, into which the audio drivers and crossovers get installed. Some ten years ago, I needed to mail in the actual physical impressions, which I think they used directly in the manufacturing process.

There are ten so called balanced drivers in my current model, four each for bass and mid-range, and two for the highest frequencies. There exist also in-ear monitors with dynamic drivers, which use bigger diaphragms to produce the sound. These drivers are bigger and look more like conventional loudspeakers. They usually require a vent opening in the shell to allow sufficient air flow. I have tried them, but clearly prefer the balanced driver type. It’s a personal choice, some people prefer the dynamic ones.

Each driver group blasts into a separate physical tube inside the shell, which exit close to my eardrums. It’s astounding how these little drivers can produce good sound, even down to low frequencies, which seems to defy physics considering the little moving surfaces available to create the required air pressure waves.

The perfect fit of the shells in my ears is crucial for good sound, and bass in particular.² The fit is also responsible for dampening the surrounding noises by about 20 dB, which allows to listen at lower levels, and still get the details. Noise-cancelling earphones are amazing (I own and use AirPods Pro, just not for music³), but simply mechanically shutting out noise is better, as no additional audio processing (read: potential distortion) is required for that purpose.

Music Player

The portable high resolution music player market is dominated by relatively small Chinese and Korean companies. For example Shanling, Astell & Kern, Cayin, or HiBy. Not exactly household names.

There are two main concepts and technologies for the digital-to-analogue conversion, delta-sigma modulation and resistor ladders. Intuitively I was drawn to the resistor network (R-2R) DACs, as they appear as direct and analogue as you can make that process. Reading reviews confirmed by bias. And a bias it is. But I had to make a decision. The prices of the products do not exactly invite A/B testing with both technologies at my home.

HiBy offers three portable players with R-2R converters. These are discrete networks of resistors with extremely low tolerance values, complemented by additional circuitry in an FPGA to control the linearity of the ladder and thus the conversion.

R-2R is pretty unusual these days. Most DACs use delta-sigma modulation. Your smartphone uses it. Your CD player as well. If you even remember what a CD player is. R-2R was an early technology, among other reasons abandoned due to the drive to ever cheaper consumer HiFi equipment, which is easier to achieve simply using a pair of sophisticated delta-sigma chips.⁴ It’s not easy to make a good R-2R DAC, especially a discretely built one. R-2R is now making a comeback, also in “big” DACs⁵ for the living room. Again, it’s a personal choice. They sound different than delta-sigma DACs.

The three HiBy players – RS8, RS6, and RS2, from left to right⁶ – come at, well, different price points, from just under 500 to over 3,000 Euros, ie. from reasonable to crazy. Reading the specs, the sizes are from small, with the RS2 being comparable to a pack of cigarettes, to pretty big, with the RS8 being about the footprint of my iPhone 13 Pro (not the Max), but twice the thickness. Barely pocketable. The RS8 is also heavy as a brick, mostly due to its huge 12,000 mAh battery. Lots of analogue electronics is power-hungry.

The basic music reconstruction architecture is roughly the same on a coarse “block diagram level” viewed from some distance, and if I squint my ageing eyes. The latest generation RS8 has a more evolved R-2R conversion circuit. Just like on the recording side, also with the playback, ie. the reconstruction of the analogue signal from digital data, there are tradeoffs to be made. So the players offer different filters,⁷ depending on your taste, and the downstream equipment used, such as different types of headphones.

Apart from the RS8’s improved R-2R converter, a big difference between the players comes on the analogue side. That is, the anti-imaging analogue low-pass filter, the amplifier stages, the volume control – all the places where good design and high quality components will form the final listening experience. Good parts and good engineering are crucial. For example, in the spec sheet for the integrated amplifier used in the RS2 you find the manufacturer’s proposed layout for the circuit board that optimises the amp’s performance. The RS8 offers a discrete class A amplifier stage.

Apart from the music signal path, there are other important components. For example the clocks. The quality of the clock, with low jitter, translates directly into the linearity of the D-to-A conversion. The RS8 uses separate power supplies for the different digital and analogue stages, in order to minimise any mutual interferences. For example, a multi-100 kHz oversampling clock on the digital side can easily propagate through power supply lines. It’s a nasty problem known to any engineer.

Of course, to state the obvious, the player per se does not “sound” anything at all. It’s only the combination with my in-ear monitors that produces the air pressure waves in my ear. Other earphones or headphones will produce a different sound, even with the same reconstruction in the player.

Next Up

Now, with a decent playback system in place,⁸ we’ll check out the differences between music recordings as experienced in more detail in the next – and last, for now – post in this series of three. It will not surprise you that the quality is in general way better than what you can experience with, say, an iPhone with its digital to analogue dongle to drive wired listening devices, such as in-ear monitors. That little white connector that plugs into the phone contains the DAC and the amplifier. Were the DAC and amp still built in, as in the olden days, they would not get more space inside the case.

Music listening using smartphones is not made to produce high quality sound, since most clients simply don’t want this, as exemplified by the fact that music these days is all about streaming. Streaming at low quality, that is. It appears that most customers don’t mind. You can stream losslessly compressed music from Apple Music at sample rates up to 192 kHz, for example, but the iPhone cannot go higher than 48 kHz sample rate, so you can get at least CD quality, not just lossy AAC. The iPhone can dish out higher sample rate digital music data, but you’ll need an external DAC. Apple also recommends to use wired playback equipment to avoid the quality loss of wireless transmissions.

Is the sound quality improvement proportional to the price of my new player? No.