[sdiy] LED current source needed for 'digital' outputs?

[rburnett at richieburnett.co.uk]

> My experience is: They don't manage it. As soon as the interfaces are
> USB-powered you will suffer a lot of digital noise via groundloops
> much likely. In our live impro-project we have forbidden USB-powered
> audio interfaces.

That's been my experience also.  USB is very noisy and once it becomes 
part of an analogue signal path you get all sorts of hums, whistles and 
whines that originated as current fluctuations deep inside the PC!  
Colin Fraser told me about a neat optically isolated USB chip that is a 
good solution at the design stage, so i'd guess that the best USB audio 
interfaces have these inside, and keep the analogue and digital grounds 
seperate.  Running the laptop from batteries (with the SMPSU 
disconnected) also improves the situation immensely!

FWIW, i've also found that a great big ferrite clamp over the USB cable 
really helps attenuate the high frequency hash you get with USB audio.  
When i've been working on dsPIC projects I got a bit of USB noise on the 
audio whilst the Microchip ICD2 debugger was connected.  Breaking the 
USB connection obviously killed the noise, but on some occasions I 
wanted to run in debug mode to analyse some software nasties, and I 
found that winding the USB cable through a large ferrite (Mn/Zn) power 
toroid helped kill off most of the annoying hissing and whistling.  It's 
called a common-mode choke, and works well for high frequency USB noise, 
but sadly will do little for traditional 50/60Hz "hum loops". Computers 
vary a lot too, so your mileage may vary ;-)

-Richie,