Rob Watts
Member of the Trade: Chord Electronics
- Joined
- Apr 1, 2014
- Posts
- 3,144
- Likes
- 12,421
So Chord DAC's are unlike some DAC's that only require 5V VBUS for initial handshake but will keep on working if 5V is turned off later?
Yes the USB VBUS powers the USB decoder and the source side of the galvanic isolator. The benefit of this approach is the USB device whilst decoding creates RF noise and correlated and distorted noise, so keeping that device away from the DACs ground plane is key here. As far as the FPGA is concerned it merely sees I2S type data, where the data clock comes from the FPGA, so the FPGA has the minimum amount of processing to do and being (almost completely) immune from source noise.
Hi Rob, in this example though, while you've removed one loop by removing the USB reclocker, let's say you are left with 3 x SMPS's and the USB source is earthed and integrated amplifier is earthed...
The USB galvanic isolation of Qutest can still be defeated here right, as the ground currents / leakage currents can see go through Qutest and amp?
Sometimes when I think about the different paths these currents can go I wonder how important galvanic isolation is in the big scheme? Sure you may block these currents on USB input but they still have a pathway via PSU's (including Qutest's?)?
Especially if the system has bad grounding overall, i.e. multiple earthed PSU's?
Galvanic isolation is absolutely crucial in the scheme of things; what we are talking here are second order or more accurately third order effects. My job is to make the quality of the source (with bit perfect data of course) become irrelevant to ensure maximum overall performance; and I am not quite there yet as sources still do make a difference. So given that source jitter certainly is irrelevant, then the only thing that can make a difference is RF and correlated noise being injected into the DACs ground plane. The only way this can happen with the galvanic isolation and the RF filtering built is is via external ground loops and currents being induced into the ground and mains. What I am trying to get over is adding extra complexity to the source will actually make matters worse, as you are creating ground loops that didn't exist; the resultant performance may be easily mistaken for an improvement - increased perceived width because the depth has collapsed, a brighter more analytical SQ due to more noise floor modulation - but of course degraded timbre variation, more listening fatigue and things being perceived to be louder.