Achieving great sound quality with a music server is mainly about two things.
The first factor is the minimisation of noise interference. Sound quality is diminished if high frequency noise sits on top of (is at a higher frequency than) the bit rate of the digital signal. Why? Because it creates uncertainty for the digital receiver in your DAC, about the transitions between bits. It is important to appreciate that a digital signal is ideally a perfect square wave, and that this means there is a need for low noise interference up to very high frequencies, well beyond the bit rate, in order to “square out” the wave.
If we were talking about just transferring data from one device to another, then the noise only needs to be below an unchallenging threshold that enables the bits to be read accurately. But we are not talking about just moving data from one point to another. We are talking about playing a digital audio signal, and the integrity of the signal, not just the bits, impacts the performance achievable by the DAC. If DACs were perfect, then perhaps music servers wouldn’t have to be. But with real world DACs the difference between servers is easily heard.
The Antipodes EX and CX are both designed for extremely low noise, but importantly, this is achieved by minimising the noise that is created, rather than filtering noise after it has been created. Filtering noise above the bit rate is used by most of our competitors, but it means reducing the squareness of the wave, and therefore the precision of the signal. A square wave is generated by first generating the sine wave version of the signal and then adding several harmonics above the signal to square it out. If you attenuate the higher harmonics then you reduce the squareness of the wave.
In our view, the first factor – minimisation of noise interference – is critical. Without adequately addressing it then the timbre of voices and instruments will be unnatural, and typically harsh. Addressing it by minimising the noise generated, as opposed to filtering it, leaves the immediacy and life in the music intact. When noise filtering is applied we typically hear pleasant tonality but which fails to be musically engaging. For this reason both the EX and CX employ state of the art techniques to minimise creation of noise interference with the digital signal, without any noise filtering.
The second factor is minimisation of contention between processes. This is why so many high-end music servers use Linux, since you can build your operating system up from the ground to run only the services you need and to give suitable priority to the processes.
But a key part of this is also to reduce the contention between the server app and the renderer app. Each music server solution requires both of these apps. You can think of the server app as managing your music and sending the file to be played to the renderer app, and the renderer app as converting the file to a digital audio signal. Sometimes the renderer app is in the DAC, which has the advantage of being separated from the server app, but it is now potentially adding and suffering contention with DAC processes. The CX sounds better than the EX because the greater computing power enables the server app and the renderer app more resources and therefore suffer less interference from one another. This is not to say that more power means better sound, because more power can generate more noise, but because the CX is uniquely very low-noise the additional power is made into a benefit by the CX.
When using the CX+EX solution, the server app runs on the CX and the renderer app runs on the EX. Separating the apps this way provides a major lift in performance. But a critical issue is the potential for noise to be introduced by the need to send the signal from the CX to the EX. This is dealt with by enabling a direct Ethernet link between the CX and EX. Using a short and high quality cable in this position will also make a significant difference to sound quality. Both the CX and EX have two Ethernet ports. This enables one of the CX ports to be connected to your network for control purposes, and the other to provide a dedicated low noise link to the EX, or to your Ethernet DAC.
We are sometimes asked why the EX is part of our top solution when the CX sounds better. The CX sounds better mainly due to its performance running the server app. But removing the server app from the EX unleashes its true potential running the renderer app.
Contention between the server app and the renderer app results in sound that has an element of confusion. The impact on tonality and speed is small, but the images are not solid or stable, there is less clear space in the soundstage, and the ear/brain has to work a little harder to make sense of what it is hearing. Some people describe this as coherence, and it also affects transparency. As you go up from the EX to the CX to the CX+EX, it is this increase in coherence and transparency that is most noticeable.