Re: Lossless SD digital Player is computer audio now reborn?
Posted: Wed Apr 06, 2016 11:42 pm
Here's my experience with isolation
When I was developing the Ciunas a couple of years ago, I did a number of prototypes
I did a prototype which was fully decked out in isolation. USB 2.0 high speed isolators weren't available then so I isolated the I2S signals coming from the Amanero board to the downstream chips (Dac or SPDIF transmitter chip). I reclocked the I2S signals just before the input to these chips. I used the GMR isolators as my research also determined they were the lowest jitter isolators available. I powered each side of the isolators by an individual battery, no other power. The idea of using the isolators was about checking the audible effect of isolating noise on the ground lines on the USB cables, not about noise on the power lines.
So the audio clocks were on the "downstream" side of the isolators & fed back through the isolators to the Amanero chips. This is synchronous clocking. So this is the best configuration that can be used - clocks on the "clean" side & sent back to the "dirty" side, reclocking of I2S signals after the isolator (because even the GMR ones add some jitter). This is the configuration used by the best designs (they term it "galvanic isolation") except they don't use isolated battery power on each side of the isolator
Now when I compared this with the exact same configuration but without the isolators, there was no audible difference. I believe the use of batteries probably is the underlying reason why there was no audible effect
Now, here's the interesting part -now that USB 2.0 high speed isolators are available, using one in front of the Ciunas makes an audible difference as it does with all DACs tried.
So the isolation of the USB digital signals themselves is the important factor - it appears that is what the noise is riding on, not the USB ground line. And the USB signals coming out of the isolator are improved by regenerating using a USB hub chip.
Now another factor - if the "dirty" side of the USB isolator is powered by VBUS it is audibly noticeable. If the "dirty" side of the isolator is powered by a linear supply it sounds better but better again if it is powered by battery
When I was developing the Ciunas a couple of years ago, I did a number of prototypes
I did a prototype which was fully decked out in isolation. USB 2.0 high speed isolators weren't available then so I isolated the I2S signals coming from the Amanero board to the downstream chips (Dac or SPDIF transmitter chip). I reclocked the I2S signals just before the input to these chips. I used the GMR isolators as my research also determined they were the lowest jitter isolators available. I powered each side of the isolators by an individual battery, no other power. The idea of using the isolators was about checking the audible effect of isolating noise on the ground lines on the USB cables, not about noise on the power lines.
So the audio clocks were on the "downstream" side of the isolators & fed back through the isolators to the Amanero chips. This is synchronous clocking. So this is the best configuration that can be used - clocks on the "clean" side & sent back to the "dirty" side, reclocking of I2S signals after the isolator (because even the GMR ones add some jitter). This is the configuration used by the best designs (they term it "galvanic isolation") except they don't use isolated battery power on each side of the isolator
Now when I compared this with the exact same configuration but without the isolators, there was no audible difference. I believe the use of batteries probably is the underlying reason why there was no audible effect
Now, here's the interesting part -now that USB 2.0 high speed isolators are available, using one in front of the Ciunas makes an audible difference as it does with all DACs tried.
So the isolation of the USB digital signals themselves is the important factor - it appears that is what the noise is riding on, not the USB ground line. And the USB signals coming out of the isolator are improved by regenerating using a USB hub chip.
Now another factor - if the "dirty" side of the USB isolator is powered by VBUS it is audibly noticeable. If the "dirty" side of the isolator is powered by a linear supply it sounds better but better again if it is powered by battery