MICS, PREAMPS and splitters
Apr 1, 1999 12:00 PM, Bill Whitlock
Let me debunk a common myth. Impedance matching is a concept that does notapply to mics and preamps. Because the electrical noise floor, heard ashiss, of a preamp's input stage is constant, S/N ratio increases as theinput signal gets larger. Therefore, preamps are normally designed torecover as much of the mic's output voltage as possible. As shown in Figure1, the mic's output impedance and the preamp's input impedance form avoltage divider.
Loading loss, usually expressed in decibels, compares the output voltagewith a specified load to that under open-circuit or no-load conditions. Forexample, a 150 V mic loaded by a preamp having a 1.5 kV input impedancewill deliver 91% of its open-circuit voltage to the preamp (loading loss =0.8 dB). As a rule of thumb, loading losses become negligible (under 1 dB)when the load impedance is ten or more times that of the source. Connectinga 150 V mic to an impedance-matched preamp with an actual input impedanceof 150 V would waste half the mic's output voltage, unnecessarily degradingS/N ratio by some 5 dB.
System frequency response will be affected by the low-pass filter in theFigure 1 equivalent circuit. Generally, the most dominant filter is formedby the mic's internal resistance (RM) and inductance (LM) and the mic'stotal capacitive load (CC + CL). In most systems, cable capacitance makesup most of this capacitance. Common mic cable has a capacitance of about 25pF/ft (82 pF/m) between the two shielded conductors. Star quad cable (moreon this later) has about twice this capacitance.
Like loudspeakers, real mics do not behave like resistors. Figure 2 showsthe frequency response effects of cable length on a simulated Shure SM57dynamic mic. The response peaks are due to the dynamic mic's inductance,which forms an under-damped low-pass filter when it interacts with cablecapacitance and preamp input impedance. Responses for an ideal 150 V sourceare shown for comparison.
Figure 3 uses the same simulated SM57 mic to show the effect preamp inputimpedance has on damping the response peaks. The upper curves, 10 kV and 3kV, are typical of transformerless mic preamps, while the lower curve, 1.5kV, is typical of a mic preamp using an input transformer. Active orphantom-powered mics, such as condenser types, generally appear lessinductive than dynamic types and are less prone to these peaking effects.Their responses will generally be closer to the ideal curves in Figure 2.
Mic splittersMics, unlike most other audio system devices, have no coupling whatsoeverto the power line. This neatly avoids ground loop hum and buzz problemscaused by power line currents that might otherwise flow in interconnectcables. Problems are rare because the mic housing and cable shield aregrounded only at the preamp input. If a single mic is parallel connected(sometimes called a "hard-wire split") to two or more preamps, however, amyriad of problems crop up.
So-called mic "splitters" are used to provide additional, isolated outputsfrom a single mic. As shown in the schematic of Figure 4, the mic isdirectly connected to preamp A via the splitter's direct output. Only thisdirect output can pass phantom power from the preamp back to the mic. Themic also drives the primary winding of the splitter's transformer, whichmagnetically couples the signal to the isolated secondary windings, whichthen drive preamps B and C while avoiding any problematic directconnections between preamps. From its point of view, each preamp isconnected to a normal floating mic source. From the mic's point of view,however, the transformer effectively parallels all cable capacitances andpreamp input impedances. This produces additional loading loss, which mayreduce the direct output level by 1 dB or 2 dB. Loading losses, inpractice, limit transformer-type splitters to four-way or less. Transformerwinding resistances may cause an additional 1 dB loss at each isolatedoutput. These effects are insignificant in most systems. Because cablecapacitances are effectively paralleled, it is generally a good idea tolimit total cable length, which includes all the direct and isolated outputcables, to about 500 ft (152 m) of standard cable or 250 ft (76 m) of starquad cable when using dynamic mics.
RFI and magnetic fieldsA mic cable can inadvertently become an RF receiving antenna. Largeconductive objects, including steel or concrete slab floors and equipmentracks, tend to behave as localized ground planes. Routing cables in or nearthem can reduce RF pickup. Where fields are especially strong, such as nearbroadcast transmitter sites, high RF voltages can appear at the ungroundedends of cables. Although the induced voltage is theoretically equal in allconductors (common-mode), normal impedance imbalances will convert part ofthe energy into signal. Preamps vary widely in their tolerance of RFsignals. Many transformerless designs become radio receivers at rathermodest RF levels, but all designs will eventually misbehave if the RFfields are strong enough. See my February 1999 column for more on RFI.
One purpose of a mic splitter is to isolate the shields of the isolatedoutput cables from the input/direct output shields and each other. Althoughavoiding ground loop problems, this lifting can allow each cable to becomea whip antenna. If you are particularly unlucky, the cable length will betuned to the frequency of a powerful nearby transmitter, producing high RFvoltages at the ungrounded or lifted end of the cable. This voltage can bedrastically reduced by using an RF terminating network consisting of aseries connected 0.01 mF (10 nF) ceramic capacitor and 51 V resistor, atthe lifted end. This terminates the line for RF frequencies but looks openat audio frequencies. See Jensen application note AN005 for details on thisand other details about mic splitter design and construction.
Hum and buzz can enter the signal path via magnetic induction. Any varying(AC) magnetic field will induce an AC voltage in any conductor exposed toit. Balanced systems will reject such pickup as long as the voltages in thetwo signal carrying wires is identical. Any difference, by definition,becomes a signal. Induced voltage is proportional to magnetic fieldstrength, which falls rapidly with distance from the field source. Twistingvirtually eliminates magnetic pickup because it makes the average distanceof each conductor to any external magnetic field source the same. Star quadcable takes this a step further by averaging with four twisted conductors.Note that a mated pair of XLR connectors leaves almost 2 inches (51 mm) ofcable untwisted, therefore vulnerable to magnetic pickup. Strong ACmagnetic fields are produced by any power cables operating at high current,power transformers, motors, computer CRTs or TV receivers. All mic cabling,and especially connectors, should be routed to avoid such areas. Forexample, it is a bad idea to place a pair of mated XLR mic cable connectorson top of a power amp.