[LAU] ground-loops -- was Re: USB audio interfaces >= 8 channels

Fons Adriaensen fons at linuxaudio.org
Wed May 16 09:22:32 UTC 2012

On Tue, May 15, 2012 at 11:43:17PM +0200, Robin Gareus wrote:

> On 05/15/2012 10:59 PM, Fons Adriaensen wrote:
> > Very few multimeters are capable of measureing AC current with any
> > level of accuracy. If you have 150 mV between two points, and zero
> > current when you short-circuit them (as a current meter is supposed
> > to do) then at least one of the two measurements is bogus.
> They were not measure simultaneously. I first measured the voltage and
> _then_ the current. In both cases the device was not directly connected:
> I measured between USB-plug and socket shield.

You can't meaure them simultaneously anyway... the voltage measurement
requires open circuit, the current measurement requires closed circuit
(via the current meter).

But if there is a voltage of 150 mv between A and B then there must
be a current if A is connected to B. Unless that voltage is purely
'static' (i.e. very high impedance) in which case it's meaningless
in this context.

>   Screen <-VGA-> Laptop (TP X60s) <-USB-> UA-25 <-TRS/jack-> Speaker(s)

According to the specs the speaker inputs are balanced. Assuming this
is true, then using a balanced TRS -> XLR cable between the UA-25 and
the speakers should not give any bzzzzzz. Any current between the two
grounds will flow over the cable screen and not via the signal wires.
Even a balanced cable, RCA or TS -> XLR with the 'cold' wire (XLR pin 3)
connected to ground at the RCA or TS side should work. An ubalanced
connection (not using XLR pin 3, or connecting it to ground at the input)
will produce bzzzzzz.

> All devices are connected to one power-strip (except the UA25 which
> takes power from USB). To take one more thing out of the equation, I run
> the computer on battery. So it's Screen vs. Speaker.

The resistance of a power cable (including contact resistance etc.) will
be well below 1 ohm. So if two of those are in the same power strip, and
there is 150 mV between the grounds at the other ends, then there must
quite a heavy current flowing in at least one of them. Which would mean
that the device producing that ground leak is probably lethal if not
grounded. Or the points where you measure that voltage are not really
connected to ground at all.

If your audio cables are OK (balanced), check the resistance between
the power input ground pin and any points on the same equipment that
should be grounded (USB connector shields, the S of a TRS, pin 1 of
an XLR.



A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)

More information about the Linux-audio-user mailing list