I am NOT trying to derail the thread! Seriously considering an upgrade for my DDC - the 2 main contenders are Denafrips (likely Iris) and the DI-20HE.
Even a basic DI-20 (non-HE) stands up against Denafrips implementation. I will try to explain it below. For now let me put a few words in general why we use DDC.
TL;DR,
A main problem with USB connection is spreading ground loops which PC is a great source. USB purifiers do not work in general, as USB standard require DC level handshaking. It makes galvanic isolation very difficult to implement, A solution is to place galvanic isolation after converting it to I2S, leaving USB receiver chip on the dirty side. Well, but what about clock? This is a big question!
There are three modes of USB isochronous transfers (all audio transfers): synchronous, adaptive and asynchronous. All modern USB implementation support the last one for the best quality. There are two methods of a clock synchronisation in this mode: push and event (asynchronous with explicit end point - in USB terminology).
- Push mode means that a host decide a moment when to send a next frame, it means a speed of the stream is determined by a sender. It has an inherent jitter and an average speed is not matching an internal clock speed. A difference in a speed will cause under or over flowing FIFO buffers. There are number of methods of synchronising clock speed, PLL, reclocking or even adding/dropping zero samples in quiet passages. Each one has drawbacks.
- Event mode require receiving feedback frames from the sink (receiving side) for synchronising speed with USB controller clock which becomes a reference clock. Most of modern USB implementation support this mode. It allows to use a fixed frequency DAC clock and no PLL or reclocking is required, it is ideal for a low jitter.
Operating systems Windows and MAC OS prefer push mode, but event driven mode can be selected in a good player. However so far there was no clear advantage of using event mode. Why? It is because almost all of DAC implementations always apply reclocking methods, this is a problem. I don't remember the brand, but it was a device in the price range $15k that was using full advantage of event driven mode.
/TL;DR,
Now coming back to the Denafrips. All Denafrips DACs (except Terminator Plus) do not have galvanic isolators on the USB connection. In this situation Denafrips DDC's bring solution to the problem, improving quality of the sound. It is still doing internal reclocking, so it is not the best solution and I see SMPS on the PC board Gaia.
DI-20 was introduced one year ago and immediately gained a highest praise from the users. Check DI-20 thread for details. It wasn't communicated with us clearly, but now we have more details. It is on the occasion of giving explanation why users should upgrade their DACs to the 2021 standard. All 2021 models receive the same mods as DI-20 had from the beginning.
It is what is happening there. As I mentioned before, USB controller is on a dirty side, so it is not a good idea to use controller clock as a reference clock. All others deploy reclocking. Audio GD decided to develop bi-directional galvanic isolator for I2S lines, feeding the internal high-precision fixed frequency clock back to the USB controller board (as opposed a feed-forward). In this way a full advantage of event driven mode can be achieved, no PLL or reclocking is used.
Denafrips DDC's do a job in their segment, their all product range, but what we get from Audio GD is standing up against most of competitors in the 10 times more expensive segment. It will change soon, as others will follow, but at the moment it is exceptional. Starting from entry level DI-20 and all DACs 2021 level, to the DI-20HE that can be used with other brands high-end solution due to the regenerative power supply.