Renting a two-channel Clearcom power supply and some headsets for the last Voyd gig provided the opportunity to do a much-needed analysis of its electrical characteristics. The supply was an older one from ALPS, with deVeer's yellow logo still painted on top of the smallish case. Inside was a simple two-sided circuit board with plenty of open space, allowing easier circuit tracing and component examination. This unit is simply a power supply, not an entire intercom station itself -- one must still plug in a beltpack + headset to join the conversation. The unit allows for two channels and includes two line outputs on the front that can be assigned to one of channels A or B. Professional show-control intercom of this type has the XLR3 pinout 1 ground 2 24 - 30 VDC regulated 3 audio, about 1V PP centered about ground level Ground and +30 are invariant and common to all units in the entire system. Obviously the power supply must be very well-regulated and quiet. The audio bus has a characteristic impedance of 200 - 400 ohms AC-coupled for the audio itself, and about 5K in the DC domain. The connections to every station are the same and in/out direction does not electrically matter -- it's a simple parallel bus, with all line connector pins wired straight through to each other. Output from the power supply generally appears on a male connector, which is ill-advised due to the ease with which one could short the supply to ground or the audio, but the intent was probably to run wiring the same way as sound wiring, with the "downstream" direction being the same way the pins point. Again, it doesn't matter -- one female/female turnaround at the power supply would allow running all the cables 100% backwards but most people don't bother to worry about it. System ground [pin 1] and chassis/AC ground are not the same, but connected with a .01 uF capacitor for RFI filtering in parallel with a 1M resistor to provide a bleed path for stray DC. Touching metal objects while wearing a headset often produces soft clicks or static on the audio line as voltages couple to the line cable shield. And we know all too well that the system is not immune to strong radio signals. The headset has a 4-pin female XLR connector to mate with the male on the box, wired thus: 1 mic - 2 mic + 3 earpiece - 4 earpiece + Earphone impedance is around 400 ohms, and the dynamic mic impedance appears to be 100 - 200 ohms. Boxes and headsets built to this general spec can freely interoperate across brands/models, allowing headsets and beltpacks to become mixed up and still work for the most part. Stations vary in their power demand. I had two models to test -- the Clear-com model 501, a small unit with the "soft" talk button that a double-press latches on, and the larger, boxier Production Intercom BP-1 like most of the units we're used to. The 501 pulls 24 mA in listen mode and 28 mA or more in talk; the BP-1 only wants 9 or 10 mA and doesn't seem to change much to talk. Any unit will draw slightly more when producing loud audio. With units turned up all the way and feeding back to produce maximal loud noise, a two-unit system was pulling slightly over 50 mA. With a .8A supply breaker, this would theoretically limit this particular system to 32 stations at maximum level, but in real life nobody would be able to tolerate the racket. Signaling, to light the "call" or "signal" light, is done by raising the audio line's DC level above about 2.5 VDC -- in practice, to about 15V. This can be done with a DC bias circuit with about 5K source resistance, which works against an equivalent ground-bias "termination" circuit in the power supply to divide the applied voltage in half at the audio bus. If several stations signal at once, the audio line DC voltage will rise even higher since their source impedances add in parallel. Presumably anything up to near the supply rail is valid, and signaling is generally brief. Audio rides AC-coupled on top of the DC offset so that signaling doesn't break the audio path. The change in DC offset is done very gently and slowly, buffered through R/C networks, to avoid loud clicks on the audio line. This is why the signal light lags the button press by an easily perceptible time, on the order of a quarter second or so. Some units pull up a little faster and produce soft "thumps" in the audio. Signal voltage can be applied un-buffered and works to instantly light the signal lights just fine, but is definitely unpleasant in the ear. Audio is simple single-ended analog voltage, whose level would be considered "loud" at about .5VRMS, or a little less than normal +0dBu line level in audio gear. It is therefore possible to listen to comm traffic by simply tapping the audio line with a high-impedance line input to a preamp and turning the gain up a little more. This could provide a way for additional people to hear show-control traffic but not be able to talk into it, which may be useful for training, without needing additional full headsets. Shorting the audio line to ground causes the system to go silent, but not in a way that an individual user can easily detect. Sidetone audio in a station's own earpiece becomes slightly louder, but there's no good way to tell that it isn't going out the line without listening from another station. Signaling, however, will not work in this state; thus the signal button also tests DC integrity of the audio path. If the audio line is completely floating, headsets tend to screech. This can happen as an annoying transient if pin 3 does not have contact while 1 and 2 do, which is easy to do while plugging in. A "channel" is simply a separate audio bus with its own terminator. The "party line" button that joins the two channels on the supply in question actually just switches all the A channel units to the B bus. This supply doesn't have the ability to keep channels separate while listening and talking independently into both like a master station; that requires many more active components. Channel crosstalk is eliminated simply by the fact that the different audio buses do not connect to each other, even though all units share the same power and ground. The Clear-com 501 beltpacks have a *much* noisier mic preamp than the BP-1s. Evidently a whole lot of additional circuitry is turned on when "talk" is pressed on the 501, whereas the BP-1 just leaves its nice quiet mic preamp enabled and just connects in the mic itself. The additional hiss from the 501 is really obvious. The double-click method to lock the mic on is also klunky and unreliable. In general, the BP-1s seem to suck a lot less than the Clear-com units for a variety of reasons, which is possibly why the Production Intercom gear seems to be in higher favor these days. One advantage the Clear-com stuff still has, since the talk button is a "soft" latch circuit, is a remote kill-all-mics feature. Power supply DC ground reference for audio line, which still allows stations to pull DC voltage up softly [through a similar network but switched to +30, most likely] and ride audio on top: pin 3 audio bus O-----/\/\/\/---*---/\/\/\/--- 220 ohm +| 4.7 K | --- | --- | 10 uF | | `------------* _|_ \ / gnd v More info [and schematics!] can be found at www.clearcom.com and www.beltpack.com, with many other pointers to related things at www.epanorama.net/links/intercom.html. An interesting rundown on one person's efforts to build "com clones" is at www.rcrowley.com/comclone. _H* 040202