Quote:
Isn't USB 1.0's bandwidth supposed to be overkill for 24/192? Then 2.0 even more - the problem is supposed to be in the OS drivers, no?
Stuff like this pop up months after I read the thread on the subject and then I can't find the dang thread again.
USB has bandwidth for 12Mbps, 192K/24 consumes 9,216,00bps, which is pretty close to full usage, and assuming latency, interrupts, etc. probably is not an easy load for USB 1, and definitely couldn't handle simultaneous I/O. USB 2 does 480Mbps which obviously is an easy fit...
So USB 1 is definitely NOT overkill, it's really close to the limit for 96KHz bi-directional, which is probably what a lot of chipmakers target, so for cost and reliability they shoot lower.
Also (my assumption, USB guru may correct) I believe the 12 Mbps is for asynchronous and as I mentioned, conventional wisdom was that to minimize jitter you needed synchronous transmission, which will usually reduce bandwidth. For most users these days, they have 256K audio, not even CD, so the volume market, which is what chip people care about, has little need for 24/96, and from a marketing perspective probably a good call though being interested in high-end I think it's deplorable.
There is a USB 2.0 audio device class, and I looked and there are a couple of chips that purport to support this, not sure why they don't have more uptake. Probably these are in the more expensive desktop units that have high-speed 96 or 192K USB interfaces.
Bottom line, this explains why the D12 USB interface is limited to 48K, and why it still sounds great over USB is the jitter-reduction of the CS chip. For the pricepoint of this device, using a pricier USB chip probably cost more than offering the optical interface, wouldn't yield sonic benefts, and would have pushed the price over $300.