Much better how? It really feels like this ESS interface is just the newest flavor of the month, after savitech/c-media/xmos/now ess usb chip originally made for sound bar application...
It's just DDC, can't imagine it makes a big difference on analog output.
A lot of myths around the USB interface. Thought this can be the good opportunity to clarify some technical issues around the USB interface.
First of all, at the physical layer, i.e. the cable transmission, USB is not error free. It has an error rate around 10 to the negative 12. This is not a problem for keyboard or mouse input yet for audio data streaming, it is a concern. You need to add CRC check sum, etc, along with the digital data to verify the correctness at the recipient end.
Roughly USB transmits data in either synchronous or asynchronous mode. For audio streaming, synchronous mode isnt good because it has problems such as jitter, etc. It also eats up a lot of CPU time. Asynchronous mode is what mostly used today. Audio data are sent as packet bursts in asynchronous mode. Depends on whether you want to do error checking or not, if you want to ensure an error free transmission, the PC side will have to calculate the CRC code based on the audio data then send out the data along with the CRC. On the receiver side, the data will have to be verified against the CRC code. If there's any mismatch, the receiver will ask the PC to resend the data packet. All of those activities will have to be done within a fixed time window due to the nature of audio playback. This requirement call for certain engineering solutions.
The next step is to convert those PCM or DSD audio data packets into I2S signals to drive the DAC. The multi-bit audio datum will have be converted to bit stream through a device called "serdes".
Since the audio data packets is coming in as bursts, and the I2S stream to DAC is continuously clocking, a FIFO is needed to regulate the data conversion flow. This is like what you see in a car wash station. Each car is washed at fixed pace yet new cars may come in at any time, at irregular intervals. You need to provide enough waiting area for incoming cars to park so the car washing process can progress at a steady rate.
XMOS, Amanero, etc, they are actually just generic processors. Take XMOS as an example, the new one has 8 or more small cores with a dedicated USB interface. The make may boast the processing power as 4000MIPS yet this is the total processing power of multiple cores. The computation capability of each core is much less than that.
XMOS does have a dedicated USB yet this interface does not support any CRC checking, etc. The CRC error checking will have to be done by a core. The core also has to handle the data resend, etc. Yep, the core may not have enough processing power to handle the data transmission job on time. Cheap XMOS chip only has 128K memory on board. This memory has to used for data processing as well as the FIFO for incoming audio data. This means the size of the FIFO is very limited.
Now, to convert multi-bit audio data packet to I2S serial data, yep, it has to be done by a core again. For DSD512, the I2S is clocking at 22.8MHz or higher, depends on the DAC design, another challenge to the core. Even worse, this process may introduce new jitters to I2S, and I have found, this is the main reason why XMOS will sound differently from Amanero and others.
If you pay attention to Amir's tests, you will see in devices with bad USB interface, the HF regions have a lot of spikes and noises. This is why.
For the new ESS chip, it is a true USB interface design, not a generic processor. It has a dedicated USB interface which does CRC error checking, for instance, in hardware directly. It uses outboard memory as FIFO so you can have a FIFO as large as you want it to. The audio data to I2S conversion is done through a real serdes which has less extra jitter.
In chip industry, this is a very simple design. It's just the market is so small, nobody was interested in creating such a dedicated audio USB interface device then finally ESS decided to do it.
With the new ESS interface, I can hear the background is darker, the roughness on top is gone, the stage is wider, etc, typical improvement with a lower jitter level.
Sure, the PC side, the driver side has entirely another set of stories yet I have to stop here before making the post too long.