A question primarily to Sonny but everyone please feel free to chip in.
Do you often see what you would consider poor design in audio equipment?
What would you consider poor design, would omitting galvanic isolation from a USB DAC be poor practise?
Keith.
Depends on the implementation. So far there's been many generations of USB interface implementations. The most basic is like the Amanero with no galvanic isolation, and powered by the USB bus. This is the worst as it's powered by dirty power and has nothing to isolate the noise from the USB bus from getting into the DAC. let alone the mediocre clocks. Then along came galvanic isolation. But it's no free lunch either. The galvanic isolators, although they isolate the grounds, they also add jitter themselves. So if you power the USB interface from a clean power source, and connect a low noise streamer up to it, it can actually sound better without the galvanic isolation. But then it was discovered that you can re-clock with flip flops after the galvanic isolation, power the USB interface with ultra clean power after the isolators, and the jitter is reduced to levels lower than anything before. So not only do you get the low jitter, you also get the isolation. The only sacrifice is you must use clocks of double the frequency as the flip flop halves the frequency.
But that's not the only issue with USB interfaces. The clocking scheme itself needs to be considered. USB interfaces use 2 clocks for Async operation. The 1st clock clocks the incoming signal, and then there's an FPGA, or Async USB chip such as XMOS or Cmedia that buffers the data into RAM and re-clocks the data with clock# 2. But then the DAC chip has it's own master clock, so the data has to be clocked again by it. This is a poor way to do things. The better way is to just use very good clocks in the USB interface, and syncronously slave the DAC chip to clock # 2 of the USB interface. This is the lowest jitter way to do things. However you need to make sure the DAC chip is located in close proximity to the USB clocks, because the further the clock has to travel, the more jitter is introduced.
The DAC in the PD2 uses the very best of all the strategies I mentioned above for the USB. It had undergone several prototypes to get to this point and I had tested some as well. This is by far the best yet, and it seems to really be immune to variances in the connected computer finally.
You can see how close the DAC chip is to the USB clocks in the photo. Building the USB interface onto another board you wouldn't be able to achieve jitter this low. And using a fancy high end external clock, good luck getting jitter this low by the time it makes it to the DAC chip. Those are made for the pro sound world when several devices need to be clocked in sync. However some manufactures purposely make the internal clocks in their DAC's poor, so they can sell you an external clock and you actually hear a difference.