Handbook for Sound Engineers

914 Chapter 25

down the system noise floor will go and how suscep-
tible the mix stage is to extraneous fields and earth
currents. In this age of digits, ground paths are espe-
cially crucial. Remember from Fig. 25-65 how the
ground noise on the noninverting input of an op-amp
mix stage gets amplified up by the noise gain of the
stage? This implies that a ground noise of 100 dBu
will end up at about 70 dBu for a 32-source mixer,
which is hardly adequate.
A simple, but so often ignored, rule with
virtual-earth stages is to make sure that the ground
reference has got the same dirt on it as the signal and
vice versa. Yes, ground follows signal. If both ground
and signal have the same noise in the same phase, there
is a chance that the noise will get ignored as common
mode and not amplified in the mix-amp. So, for each
mix bus, there should be a parallel ground bus being fed
by the last relevant ground reference from each channel.
Avoiding a major bus-length ground loop (otherwise
known as a single-turn transformer!) means that all the
heavyweight signal current in the channel proper (e.g.,
fader/mute/mode switchers) has a direct wire to central
ground while the mix-amp has a respectable output
referenced ground to work against, clean of channel
signal currents but representative of the reference of the
buffer amplifiers. The mix-amp does not take a direct
system central ground of its own.

25.13.5 Passive Mixing

There are, of course, alternatives to single-bus
virtual-earth mixing. Passive resistor mixing, Fig.
25-88, is quite viable for fixed-assignation systems that
are not going to be chopped, changed, or switched in
and out. A major advantage is that bus capacitance is
merely something to be taken into account in terms of
frequency response and phase rather than directly
imperiling the stability of the mix-amp. For passive
mixing, the mix-amp is just a buffer amplifier to make
up the loss in the resistor tree; RF filtering becomes
simple with known filter source and load impedances
together with the ability to refer against ground. A
primary weakness is that the bus is unbalanced and is of
some impedance at audio (albeit fairly low due to paral-
leled sources). As such it lays itself wide open to
induced noise and capacitatively coupled crosstalk.
Despite this, it is a method used with considerable
success for many years in quite a few production
mixers.
In all cases but especially in small mixers, say with
fewer than eight sources, there is a theoretical noise

advantage to passive mixing over virtual earth. As an

extreme example, simple summing of two sources of

passive mixing calls for 6 dB of gain to make up for the

loss in the summing network. A virtual-earth mixer

needs around 10 dB. Beyond eight sources this advan-

tage tends to the insignificant.

25.13.6 Devolved Mixing

Distributed or devolved mixing, Fig. 25-89, uses local

mix-amps to sum relatively small blocks of channels;

the outputs of these local amplifiers is then taken to a

common summing point. This quite neatly obviates

having to deal with a long bus but does create a prac-

tical problem of locating the distributed summers.

Both passive and devolved systems have the advan-

Figure 25-88. Passive mixing arrangement.

Figure 25-89. Distributed or devolved mixing.

Mixer resistor tree

Bus input filter Gain makeupamplifier

R

R

R

32

1

Inter module sections

may be screened

(For any source, mixing

attenuation looks like this)

R

R/ 32

R

Short conventional bus

1

8

9

16

Etc.

Etc.

Master

summing

amplifier

Local May be screened cable

summers