Reference

Two wires, four wires, and three ways to wire three pins

Both names are wrong as counts. A "2-wire" party line runs on three conductors (a pair and a screen) and carries both directions of speech and the DC that powers every pack on it. A "4-wire" circuit is two pairs, one each way, and carries no power at all.

What the names describe is how many directions share a conductor. Two-wire is one path with everybody talking and listening on it at once, which is why it needs hybrids, nulling and termination. Four-wire gives each direction its own pair, which makes it electrically trivial and unable to be a party line without a matrix behind it.

And then there's the part that costs a day on site: Clear-Com, Audiocom and RTS TW all use a 3-pin XLR and all wire it differently. Plug any one into another's line and it powers up and passes some audio, often on the wrong channel, which is considerably worse than not working at all. The matrix below computes that from the pin tables rather than listing it.

What a party line actually is

It's a distributed amplifier. There's no mixer. Every station has its own microphone amplifier and its own line driver, and each drives the shared line through a current source. Every station also listens to that line at a very high bridging impedance (more than 10 kΩ), so joining or leaving doesn't change the level anybody else hears.

That's the property the design is built around. Adding a fifteenth pack to a party line doesn't make the other fourteen quieter; it just draws more current from the supply. So the limit is amperes rather than audio.

The DC and the audio share the pair. A main station or power supply puts 30 V nominal onto the line, and the same conductors carry the duplex speech on top of it. One termination per channel is needed across the whole network, normally sitting in the main station.

Sidetone is the price. Because your own microphone is driving the same line you're listening to, you hear yourself, and a speaker station would howl. That's what the nulling adjustment on a speaker station is for: it cancels the station's own contribution out of what it hears.

Three systems, one connector

System	Pin 1	Pin 2	Pin 3	Channels	Call signal
Clear-Comand compatible party line	Common	DC 30 V	Audio ch 1	1	DC on the audioSteady lamp
AudiocomTelex Audiocom	Common	Audio ch 1 + DC 30 V	Audio ch 1 + DC 30 V	1	20 kHz toneFlashing lamp
RTS TWTelex/RTS two-wire	Common	Audio ch 1 + DC 30 V	Audio ch 2	2	20 kHz toneFlashing lamp

The call signal is a fourth axis of incompatibility, and it doesn't follow the audio. Audiocom and RTS raise a call with an inaudible 20 kHz tone on the line; Clear-Com adds a DC voltage to the audio instead. So even in the cross-pairings below where speech gets through, the call light generally won't, and Clear-Com's lamp means something different anyway. Theirs is steady and doubles as a cue: held on is standby, released is execute. A flashing RTS or Audiocom lamp just means "put your headset back on".

The DC method is corroborated from the other direction by the open-source ComClone beltpack, which raises a call by switching its supply rail onto the bus through a transistor, with an RC network limiting the slew rate, because a DC step onto a line everybody's listening to arrives as a thump if you let it.

Clear-Com — Power and audio on separate pins, one duplex channel per line. The audio on pin 3 is unbalanced, referenced to the pin 1 common.
Audiocom — One channel carried balanced across pins 2 and 3, with the DC riding on both. The odd one out, and the reason "it is a three-pin XLR" tells you nothing.
RTS TW — Two duplex channels down one XLR, by putting channel 1 on the same pin as the DC. This is the trick that makes RTS TW look like Clear-Com and behave differently.

Transcribed from Table 2.2 of RTS's own Handbook of Intercom Systems Engineering, which sets the three side by side — a useful thing for a manufacturer to have published about its competitors.

What happens when you cross them

Row is the station you're holding, column is the line you plug it into. It's computed pin by pin from the table above: whether the pack finds DC where it expects it, and whether its audio lands on the channel it asked for.

station ↓ line →	Clear-Com	Audiocom	RTS TW
Clear-Com	Workspowered · 1/1 ch audible	Workspowered · 1/1 ch audible	Works, wrong channelpowered · 1/1 ch audible, 1 misrouted pin 3: expects channel 1, finds channel 2
Audiocom	Partly workspowered · 1/1 ch audible pin 2: expects audio, finds DC only	Workspowered · 1/1 ch audible	Works, wrong channelpowered · 1/1 ch audible, 1 misrouted pin 3: expects channel 1, finds channel 2
RTS TW	Works, wrong channelpowered · 1/2 ch audible, 1 misrouted pin 2: expects audio, finds DC only pin 3: expects channel 2, finds channel 1	Works, wrong channelpowered · 2/2 ch audible, 1 misrouted pin 3: expects channel 2, finds channel 1	Workspowered · 2/2 ch audible

The dangerous cell is the one that half works

A Clear-Com pack on an RTS TW line powers up and talks. It finds DC on pin 2 where it expects it, and audio on pin 3 where it expects it, except that pin 3 on an RTS line is channel 2. So the pack works, on the wrong channel, and the person holding it has no way to know: there's no fault, no noise, and no indication. They're talking to a different crew from the one they think they're on.

The reverse is quieter still. An RTS pack on a Clear-Com line gets its channel 2 from pin 3 and finds channel 1 silent, because Clear-Com's pin 2 carries DC and no audio at all. Half the pack works and half of it is dead, which reads as a broken beltpack rather than a wiring incompatibility.

Audiocom doesn't rescue the situation either. It expects its one channel balanced across pins 2 and 3, so on a Clear-Com line it finds audio on one leg and bare DC on the other. It hears something, single-ended and at the wrong level, rather than failing cleanly. On an RTS line it's powered and audible on channel 2 while believing it's on channel 1.

Not one of the 6 cross-pairings fails in a way you'd notice from the pack. All 6 of them power up, 6 pass some audio, and 4 put a working station on a channel it didn't ask for. No combination here behaves like a broken cable, which is why the answer is labelling and discipline rather than diagnosis.

Four-wire, and the hybrid

Four-wire is two balanced pairs: one carries audio out, one carries audio in. No DC, no shared path, no sidetone, no nulling and no termination argument: each direction is an ordinary balanced line-level audio circuit going one way. It's what every matrix intercom port is, what connects an intercom to a mixing console or a codec, and what a wireless base station presents to the wired world.

It also can't be a party line by itself. Four-wire is point to point; the "everyone hears everyone" part happens inside the matrix, which is a mixer with a port count. That's the underlying architectural difference, and it's why matrix systems scale to hundreds of users and party lines don't.

The RJ45 pinout for 4-wire is manufacturer-specific. There's no standard: as with the audio multipins, the connector is standardised and the assignment isn't. What is universal is the rule for joining two systems: TX+ to RX+, TX− to RX−, in both directions, because one box's output is the other box's input.

Joining 4-wire to a 2-wire party line needs a hybrid. RTS's handbook describes it as a traffic cop: it puts the matrix's talk signal onto the bidirectional party line while blocking that same signal from coming back into the matrix. That blocking is the nulling, and when it's wrong the returning signal goes round the loop and the system howls or echoes.

Older hybrids are nulled by hand with a pot, against the line with its packs connected, so a null done in the shop is wrong on site once the cable length and pack count change. Newer ones null automatically and continuously, which is the biggest practical improvement in this part of the industry.

What is actually inside a beltpack

The part that makes a party line work is the line driver, which is a voltage-controlled current source rather than a voltage amplifier. RTS call theirs bilateral, meaning it can both source and sink, which is what lets one pair carry everybody in both directions at once.

Because each station injects a current rather than asserting a voltage, the signals just add. They're converted to a voltage once, at the single termination resistor, which is why there's exactly one termination in a system and why it matters so much. A current source can't short out its neighbours the way a voltage source would, so stations can be added anywhere along the line without any of them fighting.

The numbers behind it, from RTS's handbook: the current source has an output impedance of 10 kΩ or greater and a transconductance of about 3.3 millisiemens (5 mA out per 1.5 V in). The microphone preamplifier ahead of it reaches 54 dB of gain, and an electronic switch sits between the two so that a closed talk switch disconnects the preamp from the line entirely rather than just muting it. That's a noise decision: sixteen idle preamps all connected to one pair would audibly hiss.

On the listening side, a headphone amplifier of 30–40 dB drives a 50 Ω headset to around 105 dB peak, which is what a concert or a sports floor needs, with the headset's own isolation doing the rest.

The sidetone null is a differential amplifier, and that's why it drifts. The open-source ComClone beltpack does it the textbook way: the local microphone signal goes to one input of an op-amp and the audio coming back off the bus goes to the other, with a trimmer setting how much of your own voice is subtracted from what you hear. The null is therefore a balance between your own signal and the line's response, so it holds only for the line it was set against. Change the cable length, the pack count or the termination and the balance moves, which is the same behaviour people report from hybrids further up the system.

Worth noting that not every party-line station is built the way RTS describe. The handbook says these systems use current sources "or similar electronics", and ComClone shows what the hedge covers: it couples its audio onto the bus through a capacitor rather than a current source, and interoperates on a Clear-Com-style line regardless. The current source is the reason the topology scales: a simpler driver works fine until enough of them are sharing a line.

Stations	1	5	10	20	40	75
Line impedance	196 Ω	182 Ω	167 Ω	143 Ω	111 Ω	80 Ω
Level change	-0.17 dB	-0.83 dB	-1.58 dB	-2.92 dB	-5.11 dB	-7.96 dB

Checked at build time against the handbook's own arithmetic: one station on a 200 Ω line gives 196 Ω and about 0.175 dB, and it takes 20 stations to cost 3 dB. RTS put the practical ceiling at 75 stations given enough DC, with a switch on the power supply that doubles the system impedance to help. Their claim of a 6 dB maximum spread across 2–75 stations is a little tighter than this model gives, which puts it nearer 7.6 dB; the two firm figures above match exactly.

How many packs, and how far

Both limits are DC. How many is supply current divided by per-station current; how far is voltage drop along the pair against the minimum a station will run on.

Stations per supply

Station	Draw	1 A supply	2 A supply
Beltpack	33 mA	30	60
Speaker station	100 mA	10	20

Checked against Clear-Com's published capacity: a 2 A main station runs 60 beltpacks or 20 speaker stations, and the build fails if this model stops reproducing both. Mixed loads fall on the line between them.

Reach at 50 mA

Gauge	100 m	300 m	1000 m	3000 m
24 AWG	29.2 V0.8 V lost	27.5 V2.5 V lost	21.6 V8.4 V lost	4.8 V25.2 V lost
22 AWG	29.5 V0.5 V lost	28.4 V1.6 V lost	24.7 V5.3 V lost	14.1 V15.9 V lost
20 AWG	29.7 V0.3 V lost	29.0 V1.0 V lost	26.7 V3.3 V lost	20.0 V10.0 V lost

Volts left at the far station, from a 30 V supply (Clear-Com's nominal; an RTS TW supply is nominally 32 V and buys two more volts). Red is below the 18 V minimum the RTS handbook gives for a user station in high-impedance mode.

The handbook's own worked example doesn't add up

RTS give this under IR Drop: 10,000 feet of 22 AWG pair is 320 Ω, and "the DC voltage drop at the end of the wire due to a user station using 50 milliamperes of current is 0.040 amperes times 320 ohms equals 12.8 volts."

Fifty milliamperes is 0.050 A, not 0.040. The arithmetic is right for 40 mA and the sentence says 50. At the stated 50 mA the drop is 16 V, leaving 16 V from the 32 V supply the example assumes (RTS TW's nominal, where the table above uses Clear-Com's 30). That's below the 18 V minimum the same handbook gives two entries later. The conclusion that a beltpack works at the end of 10,000 feet of 22 AWG holds at 40 mA and fails at 50.

The 320 Ω itself is sound: copper resistivity gives 322 Ω for that run, and this page's model is checked against it at build time. It's only the current that slipped. Worth knowing if you have ever designed a long run against that example.

Will this rig actually work?

All three models above at once, on your numbers, because they constrain each other. A supply that can power the packs may not reach them, and a line long enough to reach may not have the volts left when it gets there.

This calculator needs JavaScript. The tables above give the same three models separately.

The volt-drop figure loads the whole run with the total current, which is the pessimistic case: stations distributed along the cable draw through less of it than that. It's the right way to be wrong here, and it's what the RTS worked example does too. Per-station current varies by model and with whether a pack is talking, so leave headroom on top.

The other things on the line

A conference channel is only part of what these systems do. Three other functions ride on the same infrastructure, and each answers a problem the plain party line creates.

IFB

Interrupted FeedBack — also Foldback, or Interrupted Return Feed

One way, and the clue is in "interrupted". Talent normally hears programme in their earpiece; when the director keys the IFB, the programme is interrupted and replaced by the director's voice. The talent has no way back on this path, which is what you want when they're live on air.

The chain is a hot microphone feed, a control panel, a central electronics unit doing the switching and mixing, and a talent receiver at the far end. Where talent must converse with other talent elsewhere, the programme leg has to be a mix-minus (everything except themselves), or they hear their own voice back and can't speak over it.

Stage announce

SA, or the second channel doing a different job

A two-channel system is very often one channel of crew conference and one channel feeding loudspeakers: the director cueing actors, calling the house, or talking to the floor during rehearsal. It's the same electrical party line; only the destination differs.

Worth thinking about which channel that is, because it's the one where whatever is said leaves the headsets and enters the room.

Mic kill

And the very specific reason it exists

A command that switches off every microphone on a channel at once. It exists because of one recurring failure: somebody goes on a break, leaves their talk switch latched, and puts the headset down on top of something.

A live dynamic microphone is an excellent magnetic antenna. Laid on a power transformer, a dimmer, or an older monitor's deflection yoke, it injects hum into the channel for everybody, and with the pack unattended, nobody nearby knows it's theirs. Mic kill is the remote fix for a fault the person who caused it has walked away from.

There is a distance note attached to IFB worth knowing: RTS reckon on at least one talent receiver working at the end of two miles of 22 AWG pair, or two stations at one mile. Where only a single pair is available, a talent receiver can be run in "pseudo-stereo" (pins 2 and 3 commoned as the high side against pin 1), giving a mono feed with each ear separately adjustable.

Beyond the analogue party line

Digital party line — Clear-Com HelixNet and RTS's digital partyline run on the same two-conductor-plus-screen cable already in the building, and carry several channels plus data down it. The selling point is that the cable infrastructure doesn't change; the packs and the main station do.
Matrix — RTS ADAM and ODIN, Clear-Com Eclipse. Every user gets a port, every port is 4-wire, and who hears whom is a routing decision rather than a wiring one. This is where large facilities live, and where party lines appear as just another kind of port via a hybrid.
Wireless — DECT at 1.9 GHz is the workhorse (Clear-Com FreeSpeak), with 5 GHz and 2.4 GHz options where the spectrum allows. The base station presents 4-wire or party-line ports to the wired system, so everything above still applies to the wired half.
IP — Dante, AES67 and OMNEO carry intercom between rooms and buildings, and phone-app panels put a key panel on a crew member's handset. The failure modes move from voltage drop and nulling to latency, PTP and QoS, which is a different discipline rather than a simpler one.
One earth, in one place. An analogue party line is unbalanced and referenced to its own common, so every additional earth connection is a loop. RTS are unusually specific about it: clear all earth grounds from the circuit return, and let the only ground be the 22 kΩ resistor in the power supply. A second earth arriving anywhere (through a rack, a camera, an interface) is the usual cause of a hum that nobody can localise.
Do not tape beltpacks to metalwork. Fastening a portable station to a structure grounds its case to whatever that structure is bonded to, which either injects local ground currents or completes a loop large enough to act as an antenna. It's a tempting thing to do on a truss or a camera dolly, and it's a reliable way to make a channel noisy.
Unbalanced goes a very long way, until it runs beside mains. RTS report acceptable noise on unbalanced runs of up to two miles. What changes that is proximity: cable sharing ductwork with power feeds over about 300 m wants converting to balanced operation, or running "dry line, balanced" with each station powered locally instead of from the line.
Reversed DC isn't fatal, and reversed audio is common. Stations carry a protective diode, so a swapped supply just means the pack doesn't work (negative is ground in these systems). Phase reversals, by contrast, are routine on portable cable that has been made up or repaired without being put on a tester afterwards, and they're far harder to spot.
The packs outlive their support, and then their plastic. Clear-Com no longer service the RS-500 series, and the reported failures are almost all mechanical rather than electronic. A headset jack wire pulled off its solder joint is the common one, and the ABS bezels go brittle and disintegrate long before the boards do. With a batch of dead packs, harvesting is the realistic route: the electronics mostly still work, and what you're short of is enclosures. Community-sourced, from practitioners rather than documentation. See the sources below.
The 4-pin XLR headset is the one thing that mostly is standard. Pins 1 and 2 are the microphone, 3 and 4 the earpieces. The compatibility trap is the microphone element: a pack expecting a dynamic headset and a headset with an electret element will be quiet, distorted, or both.

Sources

The three-way pinout, station currents, the 18 V minimum and the hybrid description — RTS, Handbook of Intercom Systems Engineering, read directly. Table 2.2 is the connector wiring by manufacturer; the 30–65 mA headset and ~120 mA speaker-station figures, the ≥10 kΩ bridging impedance and the IR Drop glossary entry are all from it.
Supply capacity, the 30 V line and the termination rule — Clear-Com, Party-Line Intercom System Installation Manual, read directly. The 60-beltpack / 20-speaker-station capacity of a 2 A supply is their published chart, and this page's power model is pinned to it.
That Clear-Com and RTS two-wire are not natively compatible — stated by Clear-Com themselves in their own Solution Finder note on TW partyline, which also describes the TWC-701 interface for combining two Clear-Com channels onto one TW circuit.
Beltpack internals, line loading and the call-signal methods — the same RTS handbook. The bilateral current source, its ≥10 kΩ output impedance and 3.3 mS transconductance, the 54 dB microphone preamplifier and the electronic switch ahead of the line driver are from its glossary and Chapter 2; the loading arithmetic this page computes is worked longhand there; and the 20 kHz call tone versus Clear-Com's DC call voltage is from its Call Lights section.
IFB, stage announce, mic kill and the grounding practice — the same handbook again, which is the most substantial single document on this subject that is freely available. Chapter 3 carries the IFB chain and its two-mile / one-mile talent-receiver figures, the pseudo-stereo wiring for a single pair, and the magnetic-antenna explanation for why mic kill exists; the single-earth rule (only the 22 kΩ resistor in the power supply), the warning against fastening portable stations to metalwork, and the 300 m threshold for converting to balanced operation are from its noise and grounding sections.
An independent implementation — Richard Crowley's ComClone 2, an open-source party-line beltpack with a published circuit description. It corroborates the Clear-Com pin assignment and the 28–30 V line from the other direction, and its self-nulling duplex hybrid and DC call circuit are the concrete versions of what the manufacturer documents describe in the abstract. An open design rather than a manufacturer specification, and it's used here to illustrate mechanism rather than to define behaviour.
Another open project in the same space — BlueClone, on the Blue Room forum, a single-channel open-source beltpack aimed at compatibility with the wired systems in common UK use. Its microphone front end offers switchable bias and gain specifically so that either dynamic or electret headsets can be used, which is the headset trap above seen from the designer's side.
Further reference — Best Audio publishes intercom pinouts, adapter schematics and theory notes. Most of it sits behind a free member login and has not been drawn on here. Practical Show Tech's two-part "Keep Comm with Pete" (episodes 79 and 80) covers the same ground as a recorded session. It's video rather than text, so it's a pointer rather than a source for this page.
RS-500 series service status and failure modes — ControlBooth, "Clear-Com rs-501/502 repair". Community-sourced and anecdotal: practitioners reporting that Clear-Com declined to service a 501, that pulled headset-jack wiring is the usual fault, and that the ABS bezels fail before the electronics. Useful as a picture of what actually breaks, not as a specification.
Four-wire RJ45 assignments are manufacturer-specific. No standard exists, so none is given here; the TX-to-RX rule is the part that generalises.

Assembled 7 September 2026 with AI assistance. The compatibility matrix, the power budget and the reach table are computed from the pin tables and the manufacturers' own published figures, and each is validated at build time against a number those manufacturers print. But nothing here has been measured on hardware, and no station has been deliberately plugged into a foreign system to confirm what the matrix predicts. Per-station current varies with model and with whether a pack is talking; treat the counts as a design starting point and leave headroom.

Companion page: XLR, AES3 and DMX — intercom is one of the four unrelated circuits that share the three-pin shell, and the cross-system hazards there are the same shape as the ones here.