Handbook for Sound Engineers

Consoles 849

resulting in a fake ground potential 60 dB below the
output of the amp. An earlier stage in the chain A1 (in
this example, a microphone amplifier, with a consider-
able amount of gain) has its feedback leg (amplifier
reference) tied to the same fake ground. Its input ground
reference (here lies the problem) is taken from a sepa-
rate bus supposedly to provide a nice, clean ground.
This it does admirably, the bus being tied straight to
reference ground and having no sources of great
substance going to it.

Any signal present on the fake ground is duly ampli-
fied by the microphone amplifier (in its inverting mode)
and is attenuated at the line amplifier output back into
the fake ground. Naturally, as soon as the microphone
amplifier gain exceeds the output attenuation, the entire
chain bursts into oscillation.

A very similar mechanism was responsible for an
owner’s criticism of his well-known console that when-
ever he attempted to use the track routing on any
channel modules, the sound of that channel discernibly
altered. It was found that ordinarily nothing in the
channel drew much current; all ground impedance
requirements were quite light. Light, until the track
routing line amp with its load of routing resistors and a
terminated output transformer was accessed, demanding
a relatively large ground current. This output stage
current shared the only ground access point of the
module (two paralleled connector pins) with all the rest
of the module electronics, with the notable exception of
the microphone and line input transformer ground
returns. The resultant feedback, although nowhere near
enough to promote oscillation, did by virtue of the
phase shifting of the output transformer at both high and
low frequencies result in distinct coloration.

A purist answer to these fake and loop problems is to
choose one grounding point for the entire console and to
take every reference and ground return directly to it
through separate ground wires.

A few less than minor problems would ensue. The

enormous number of ground lines would soon outstrip

the capacity of the module connectors, and the mass of

wiring would cause apoplexy from the wiremen and

aggravate an already critical world shortage of copper.

Fortunately, a working compromise suggests itself

based on separating the different classes of ground

requirements by impedance.

Bucket grounding refers to tying fairly high-imped-

ance sources to a common ground point, bus, or line

(since the ratio of their impedances is so great that

resultant fake ground potentials can hopefully be made

low enough to ignore). Anything that is likely to draw

current (any kind of output or line amplifier stage)

should go directly to ground, will not pass through any

bus, and will not collect shared ground paths on the way

to the bucket.

Any ground bus will have a measure of resistance

and must, therefore, be fake to a certain degree. If we do

our sums right, ground bus signal levels can be kept

acceptably low, below 100 dBu.

Smugly, we can expect to ignore figures like that

until we (almost inevitably) amplify them up.

25.8.4.4 Ground Noise in Virtual-Earth Mixers

A virtual-earth mix-amp unavoidably amplifies ground

noise. Fig. 25-28A tells the story. For instance, a multi-

track mix-amp can typically have 32 sources applied to

it; the through gain from any source is unity (assuming

the source resistors equal the feedback resistor), but the

real electronic gain of the circuit is 33 or a touch over

30 dB. Redrawing the circuit slightly in Fig. 25-28B

shows exactly what this 30 dB is amplifying. Consider

as a clue that which is directly applied to the nonin-

verting input of the op-amp—the ground! True, it is

amplifying the noise due to the resistors and the internal

noise mechanisms of the device, but for our argument

here, it is amplifying ground. In any reasonably sized

console, providing no sources are grossly out of propor-

tion to the majority, ground noise is pretty random and

noisy in character. The result is that, on being amplified

up, it serves to make the mix-amp apparently much

noisier than would be expected from calculation. In

suspect systems it has been found to be the predominant

noise source. It is no accident that the real electronic

gain of a mix-amp is also known as its noise gain.

It is truly astonishing what attention to virtual-earth

mixer grounding can have on bus noise figures. For

mix-amps, practical noise performance has little to do

with the device employed and nearly everything to do

with grounding.

Figure 25-27. Feedback and oscillation via poor grounding.

Mic

input

Separate clean ground bus

Feedback leg

Variable

gain

Oscillation

loop

600 7amplifier

termination

Fake ground

0.6 7 ground loss

Reference ground