Handbook for Sound Engineers

Consoles 865

responses. Common-mode unrejected signals still

appear at the amplifier input as if nothing had happened.

25.10.6.6 Input Impedance

As determined earlier, we would end up with better

noise performance and cleaner sounds if the micro-

phone looked into a high, preferably infinite, imped-

ance. Preferences apart, we have already had to define

the reflected load (input), impedance by the resistor

needed to keep the front-end stable under unplugged

conditions (R 0 ), but at least it is an order of magnitude

or so above working impedances, so its effect is small.

It does, though, act as part of an attenuator of input

signals along with the source impedance and winding

losses, Fig. 25-44. This is the major factor responsible

for worsening front-end noise performance using trans-

formers. Any attenuation before the optimized amp

directly degrades the noise figure, typically between

1 and 6 dB, depending on the transformer.

If the transformer were perfect, it could be assumed

that the reflected impedance, as seen by the micro-

phone, would be constant over the audio band. At the

low-frequency end, Fig. 25-45, the diminishing induc-

tive reactance of the transformer windings (tending to

zero with frequency) becomes a term of greater impor-

tance, affecting parallel impedances, attenuation, and

accordingly, efficiency. Winding self-capacitances and

the passive compensation networks are largely to blame

for the high-frequency droop, although the list of

contributing mechanisms is nearly endless.

A good rule of thumb is that the midband input

impedance should exceed ten times the source imped-

ance, or about 2 k: for a dynamic microphone. Any

wild variation in this impedance is obviously going to

result in frequency and phase response aberrations,

which are probably the greatest single drawback to

transformer front ends. Things aren’t quite as bad as

they seem; examples of performance shown here have

been deliberately of a marginal transformer to highlight

the illeffects, notably in response and impedance flat-

ness; good transformers from good reputable manufac-

turers such as Jensen, Lundahl, and Sowter generally

show much nicer results, but the design criteria to eke

the best from them remain nonetheless.

25.10.6.7 Attenuator Pads

Attenuator pads, regrettably necessary in many

instances to preserve head-room and prevent core satu-

ration with elephantine sources, should maintain

expected operating impedances when introduced. The

transformer primary should still be terminated with a

nominal 200: , while the microphone should still look

at 2 k: or above. Departure from these will cause the

microphone/amplifier combination to sound quite

different when the pad is thrown in and out, as would be

expected from altering source and load impedances in

and around complex filter characteristics. A significant

downside to pads is that although the differential

(desired) signal is being attenuated to the expected

degree, any common-mode signal isn’t.

Figure 25-43. Frequency response of taming network.

Figure 25-44. Input losses which worsen noise figure.

Taming alone

With Ro

6 dB/octave Peak frequency

of resonance

Output levelWith microphone

plugged in

Frequency

Microphone

L

R RP RS Amplifier

Attenuation at this

point due to all series

elements into

Transformer

primary/secondary

winding resistances

Source

VS

Ro

Ro

Figure 25-45. Typical input impedance curve.

2 k

1 k

Zin

–ohms

Frequency–Hz

20 100 1 k 10 k