Yet another PCM2707C USB DAC

[Blueridge]

Hello everybody!

Having a little spare time and because I also bought a new pair of headphones (I'm not going to name them, they're not audiophile rated ) I decided to try my hand at building a small USB DAC based on the PCM2707C.

Why? It's plug and play, driver-less, and 16bits are enough for me since they're enough to cover the dynamic range required by all music genres.
24bit vs 16bit, the myth exploded!

Ah... and it also includes a headphone amplifier that can drive 32ohms headphones (the sort I just bought), but it's better if you take a look at the datasheet:
https://www.ti.com/product/pcm2707c.
I chose the PCM270xC flavor because it has improved volume control under windows vista and above (a greater number of steps compared to the PCM270x).

I've attached the whole eagle project, I'll post some PDFs later if needed. I followed closely TI's datasheet, or at least I tried to.
Meanwhile I'd like your opinions on the layout. What I tried to do:
- separate the digital and analog ground, while also keeping a continuous ground plane underneath the USB traces;
- have a separate shield connection, isolated from the ground, with the possibility of connecting the USB shield to the ground in multiple ways (I've found a number of different proposals and recommendation on the topic, all having their good points, see here).

My aim was to create a SMD only PCB, with the routing on top only while keeping the bottom for ground plane. The PCB is hand made, using the photosensitive method, so I wanted to have as little trouble as possible when aligning the mask for top and bottom.

I did not have a SMD 3.5mm jack at hand so I had to use a THD one. Also, originally I planned on using ceramic caps only (I had some 100uF ceramics at hand - but misplaced them ) so the output caps became an aluminum electrolytic cap, 150uF from Epcos in parallel with 4.7uF and 10uF ceramic caps in 1206 package.
Al ceramic caps are X7R except the 22nF and 18pF which are C0G.
Al resistors are thick film, except the ones used in the audio path (3.3k, 47R) which are metal film resistors.

The USB connector is from samtec (mini USB type B), I modified the library and I removed the centering pins, also from the connector itself.
The 3.5mm is from Multicomp I think, nothing fancy, the crystal is the most stable I could find in a friendly SMD package.

In the end I used 22nF ballast caps for the crystal as I also could not find the 18pf at the time I soldered all components, and I was eager to get the DAC running.

The LED near the USB connector is fro signaling if there's power on the USB port and the LED connected to the SSPND signal I use it as a confirmation that the device has been enumerated and is active.

Originally I made a fault, by pulling the 1.5k resistor connected to Data+ to GND instead of VDD (3.3V from the internal regulator) but the layout in the archive already contains a correction.

I'll make some pictures of the build later. I'll also try to make some oscilloscope screenshots with how the signal looks like at different frequencies.

So, what do you think?