I try to answer the question of what makes a streamer, a high quality one.
There are many technical characteristics that differentiate the quality of a streamer:
a very important element is the power supply since the DNA of these machines is based and developed in any case on an operating system which is nothing more than a skein of applications and therefore of programs in continuous necessary execution, these compete for time with the CPU which will never be able to devote itself exclusively to the application that manages the audio player. Paradoxically, everything goes as if the only running program, which is what interests us, was running on a less powerful machine than its theoretical potential.
That's why the linear power supply in high performance streamers needs to be taken care of: I recently heard one that has three custom transformers that provide four main power rails from which all the separate voltages are regulated for each part of the streamer. The reasons that lead to the use of three (smaller) transformers instead of perhaps a larger one are justified precisely in the "unstable nature" of the system, in fact the smaller transformers are more stable in the presence of high current oscillations (typical of PCs) and have less hum during peaks. In machines built with "intelligence" it is ensured that each transformer is used for its dedicated operating area: One for the processor, one for the motherboard and one for the peripherals. Since a streamer power supply has a high peak current load and all the currents alas add up in the core of the transformer interacting with each other, separating the areas of action decreases the interaction and improves the quality of the final sound.
Precisely from this point of view, in these products, the use of high inductance power transformers characterized by a low field saturation in the core is aligned and the application of filters to keep the peaks to a minimum and therefore the noise... . how to say the less trash enters and the less I have to filter at the end!! Obviously the same design philosophy is pursued for the loads that join the power source and by interacting produce harmonics and intermodulation noise.
This is the reason that drives "some manufacturers" to implement separate tracks for dedicated areas and, in the best cases, a separate regulator for each load. Of course, regulators also have noise which is a factor in the final power supply output signal. But actually what is more "important" are the noises of the main power supply and the noise generated by the rectifier and the inevitable current peaks of the capacitors.
All these sources of noise must be mitigated in some way (since they cannot be completely eliminated..) and then also in this case adequate and well-made solutions are needed such as the use of special low inductance transformers, Schottky diodes, localized current capacitors, shielded power planes on PCBs, wiring, etc,etc.
Obviously digital without watches does not exist! Its role is FUNDAMENTAL and the same attention goes in particular to phase noise. To maximize performance, the streamer motherboard can be equipped with two clocks, the system clock (chipset) and the processor clock. The system clock (chipset) is required to synchronize all components/clocks on the motherboard. This clock also synchronizes the clock for the CPU, but the CPU clock is usually only used on the chip (PROCESSOR) itself. Since the CPU has to do more things at a time than the motherboard, the clock frequency will be multiplied in the CPU.
Again the general criterion in well-designed machines is to separate using a clock on well-chosen key areas whose implementation has the purpose of lowering the mentioned noise consequently making the jitter lower in the whole system considerably increasing the quality of the listening experience .
Last but not least is the software: particularly noticeable are the heavily modified versions of Arch Linux (low latency / realtime kernel / headless). The advantage of using Linux is that you start from scratch and configure only what you really need for audio playback.
The reasons that determined the choice of some operating systems, mainly reside in the "guarantee" of greater stability.
Linux is a very stable and reliable operating system that can be modified substantially. Furthermore, and most importantly, with Linux it is possible to achieve incredibly low latency (audio latency and processor latency) and latency is one of the key elements to obtain a sound really fluid and natural; the lower the latency, the better the overall playback quality. That's why the buffer size is always turned off or set to the lowest possible setting in the playback software (always dealing with a bunch of applications running all the time!!)
So what differentiates a high-end streamer from other products is the quality of the project capable of intelligently minimizing the weak points of these machines which derive from their very nature or that of being in fact PCs dedicated to music. It is therefore inevitable that the problems of these implementations involve issues such as the player used or the CPU not "able" to follow the rhythm imposed by the DAC, or not perfect in respecting the deliveries of the samples to the DAC in the timing imposed by the frequency of sampling, not to mention the injections of jitter on the USB data, therefore of signal variations that have roots in problems that specifically concern the hardware of these configurations.
Not all streamers on the market achieve these objectives or rather, obtain results such as to obtain an excellent level of reproduction quality. This is inevitably paid for in the final price.