A telephone transmitter is the speech-to-current transducer at the mouthpiece end of a telephone — usually a carbon-granule microphone that varies its electrical resistance in step with sound pressure on a diaphragm. Thomas Edison patented the practical carbon-button form in 1877 (US 474,230 and related), and Emile Berliner's loose-contact design followed the same year. A DC bias current flowing through the granules turns those resistance changes into a modulated voice-frequency current that travels down the line. The result drove every Bell System handset for nearly a century.
Telephone Transmitter Interactive Calculator
Vary carbon resistance and bias current to see how a carbon-granule telephone transmitter converts diaphragm compression into current modulation.
Equation Used
The calculator treats the rest bias current as the current through the carbon pile at the rest resistance. When sound pressure compresses the granules, resistance falls, so the same bias voltage produces a higher current. The difference between rest and compressed current is the voice-frequency current swing.
- Bias voltage is set by the rest condition and remains constant during the loud compressed peak.
- Compressed resistance is the loud-voice peak value of the carbon granule pile.
- Line-feed resistance and induction-coil losses are ignored for this simple transmitter model.
The Telephone Transmitter in Action
The carbon-granule transmitter works on a deceptively simple principle. A thin metal diaphragm sits in front of a small chamber packed with roasted, sized carbon granules — typically anthracite coal carbonized at around 1,000 °C and screened to 0.5 to 1.0 mm. A DC bias current, traditionally 30 to 60 mA from the local battery or the central-office line feed, flows through that granule pile from a front electrode bonded to the diaphragm to a rear electrode in the cup. When you speak, sound pressure pushes the diaphragm in and out by a few microns. That tiny motion compresses and relaxes the granule pile, the contact area between granules grows and shrinks, and the bulk resistance swings from roughly 50 Ω at rest down to 20 Ω on a loud peak. The DC current modulates around its bias point at voice frequency, and that AC component is what couples through the induction coil onto the line.
The design exists because pure dynamic microphones — Berliner's first attempts, for example — produced output too weak to drive miles of open-wire copper without amplification, and amplification was not commercially available until the audion arrived in 1906. The carbon transmitter is effectively an acoustic-to-electric conversion stage with built-in gain: the diaphragm only has to throttle a current that's already there, so a 10 µPa whisper can produce a measurable line current swing. That's also where the failure modes hide. If the granules pack down — a common problem with handsets stored mouthpiece-up for years — sensitivity drops and the user sounds distant. If moisture gets in, granules clump and the transmitter goes raspy or cuts out entirely. If the diaphragm corrodes or warps beyond about 0.05 mm of flatness, the front electrode loses good contact and you get the classic carbon-mic frying noise. Western Electric's fix on the F1 capsule was to seal the chamber and specify a phosphor-bronze diaphragm 0.05 mm thick — anything thinner rang, anything thicker killed the high frequencies above 3 kHz.
Key Components
- Diaphragm: Thin metal disc, typically phosphor-bronze or aluminium 0.05 to 0.10 mm thick and 35 to 45 mm in diameter, that converts sound pressure into mechanical displacement. Flatness must hold within 0.05 mm or the front electrode contact patch shifts and sensitivity drops 3 to 6 dB.
- Carbon Granules: Roasted anthracite or hard coke screened to 0.5 to 1.0 mm grain size, filling a sealed chamber roughly 6 to 10 mm deep. The granule pile's bulk resistance varies with diaphragm motion — typically 20 to 80 Ω swing around a 50 Ω rest value.
- Front Electrode (Carbon Button): A polished carbon disc bonded to the rear of the diaphragm, transferring diaphragm motion into the granule pile. Edison's 1877 patent specified lampblack pressed under 2,000 psi; later designs used machined retort carbon.
- Rear Electrode and Cup: Brass or nickel-plated cup that holds the granules and forms the second current path. Cup depth controls granule pile compression — too shallow and the granules pack hard, too deep and they shift under handset orientation changes.
- DC Bias Source: Local battery (early designs, 3 to 6 V) or central-office line feed (modern, 48 V through a 400 Ω feed resistor) that drives 30 to 60 mA through the granules. Without bias, the transmitter produces no output — it is not a self-generating device.
- Induction Coil: Step-up transformer, typically 1:5 to 1:20 turns ratio, that couples the AC voice-frequency component onto the line while blocking the DC bias. This is what gives the carbon transmitter its line-driving voltage swing.
Industries That Rely on the Telephone Transmitter
The carbon transmitter dominated voice telephony from the 1880s until electret condenser microphones took over in the 1980s. Even today you'll find carbon transmitters in places that need ruggedness, simplicity, and immunity to electromagnetic pulse — and in any working heritage telephone exhibit where authentic voice quality matters. Field sensitivity, environmental sealing, and the bias-current requirement decide where the technology still earns its keep.
- Telecommunications Heritage: Connections Museum Seattle operates a Panel switch and a No. 1 Crossbar with original Western Electric 323-type carbon transmitters in the F1 handsets used by demonstration callers.
- Military Communications: The TA-312/PT field telephone, still in NATO service through the 1990s, used a carbon transmitter capsule because it survived shock, water immersion, and worked with a hand-cranked magneto.
- Industrial Plant Paging: GAI-Tronics Page/Party handsets in refineries and steel mills used carbon transmitters into the 2000s for explosion-rating compliance and survivability against hydrogen sulphide environments.
- Mining and Tunnelling: Federal Signal mine phones on coal-face haulage loops used sealed carbon capsules because the bias-current design tolerated 3 km of wet, low-quality cable without the line driver an electret would need.
- Museum and Restoration: Vintage payphone restorers rebuilding Western Electric 233G three-slot sets specify NOS T1 transmitter capsules to keep the period-correct voice timbre.
- Aviation Ground Comms: David Clark headset carbon-element transmitters served airline ground crews from the 1960s through the 1990s before electret retrofits became standard.
The Formula Behind the Telephone Transmitter
The useful equation for sizing a carbon transmitter is the AC current swing it puts onto the line for a given sound pressure level. This tells you whether your diaphragm-and-granule combination will be heard clearly, sound thin, or saturate into harmonic distortion. At the low end of the typical 30 to 60 mA bias range, the transmitter runs quiet and clean but loses sensitivity below conversational levels. At the high end of bias current, sensitivity climbs but granule heating accelerates packing and shortens capsule life from years to months. The sweet spot sits around 45 mA for a Western Electric F1-class capsule.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| iac | AC voice-frequency current swing on the line | A (rms) | A (rms) |
| Ibias | DC bias current through the granule pile | A | A |
| ΔR | Resistance swing of the granule pile under sound pressure | Ω | Ω |
| R0 | Quiescent resistance of the granule pile at rest | Ω | Ω |
| Sd | Diaphragm sensitivity (resistance change per unit sound pressure) | Ω/Pa | Ω/(lbf/in²) |
| p | Sound pressure on the diaphragm | Pa | lbf/in² |
Worked Example: Telephone Transmitter in a Western Electric F1 handset rebuild
A working telephony exhibit at the Museum of Communications in Seattle is rebuilding a Western Electric F1 handset with a fresh T1 carbon transmitter capsule. The conservator needs to predict the AC line current at normal conversational sound pressure (1 Pa, about 94 dB SPL at the mouthpiece) so the museum's restored 1930s magneto exchange drives the receiver at a realistic loudness. The capsule sees 45 mA bias from the line battery, has a quiescent granule resistance of 50 Ω, and a measured diaphragm sensitivity of 8 Ω/Pa.
Given
- Ibias = 45 mA
- R0 = 50 Ω
- Sd = 8 Ω/Pa
- p (nominal) = 1.0 Pa
Solution
Step 1 — at nominal conversational level (1 Pa, 94 dB SPL), compute the resistance swing the diaphragm produces:
Step 2 — compute the AC current swing on the line at nominal speaking level:
That is a healthy, museum-correct line current — enough to drive a U1 receiver at clear conversational loudness through several hundred metres of 24 AWG cordage. The handset will sound exactly like a 1935 Bell call.
Step 3 — at the low end of the operating range, a quiet talker at 0.3 Pa (about 84 dB SPL):
This is the threshold of intelligibility on a long line. The remote receiver will sound thin and the listener will lean in — exactly the experience museum visitors get when they whisper into the mouthpiece, which is part of the demonstration.
Step 4 — at the high end, a shouted 5 Pa (108 dB SPL):
In theory. In practice the granule pile saturates above about ΔR ≈ 30 Ω because the carbon contact area can only grow so much before adjacent granules are already touching. Above 3 Pa the response goes nonlinear and you hear the classic carbon-microphone harmonic distortion — the unmistakable bark of a 1940s long-distance connection under heavy speech.
Result
The nominal AC line current at conversational level works out to 7. 2 mA rms — exactly in the band a 1930s magneto exchange and U1 receiver were designed around. At a quiet 0.3 Pa input the current drops to 2.2 mA and intelligibility hangs on, while a shouted 5 Pa would mathematically demand 36 mA but the granule pile saturates near 27 mA and the output bites into harmonic distortion. If you measure significantly less than 7.2 mA on the bench at 1 Pa, the three usual suspects are: (1) granule packing from long mouthpiece-up storage, which raises R0 above 80 Ω and chokes the swing — fix by tapping the capsule firmly on a hardwood block; (2) oxidation on the front carbon-button-to-diaphragm contact, which adds a series resistance that does not modulate; (3) a bias current below 30 mA from a tired line battery or a wrong feed resistor, which shrinks every term in the equation linearly.
Telephone Transmitter vs Alternatives
Carbon transmitters held the field for a century because nothing else gave you that combination of built-in gain, ruggedness, and dirt-cheap construction. They lost ground to electret and dynamic capsules on fidelity, noise floor, and lifespan. Here's how the three compare on the dimensions that actually matter when you're choosing a transmitter for a heritage rebuild or an industrial handset.
| Property | Carbon Granule Transmitter | Electret Condenser Microphone | Dynamic (Moving-Coil) Microphone |
|---|---|---|---|
| Output level (typical, into 600 Ω line) | −10 to 0 dBV (very high, self-amplifying) | −40 to −30 dBV (needs preamp) | −55 to −45 dBV (needs preamp) |
| Frequency response | 300 Hz to 3 kHz, peaky | 20 Hz to 16 kHz, flat ±3 dB | 50 Hz to 12 kHz, smooth |
| Distortion at conversational level | 3 to 8 % THD | < 1 % THD | < 0.5 % THD |
| Power requirement | 30 to 60 mA DC bias mandatory | 1.5 to 10 V at < 0.5 mA | None (self-generating) |
| Lifespan in continuous service | 3 to 10 years before granule packing | 20+ years | 30+ years |
| Cost (capsule, 2024) | $15 to $40 NOS | $0.20 to $5 | $8 to $50 |
| Best application fit | Heritage telephony, mil-spec field phones, EMP-hardened comms | Modern handsets, headsets, consumer electronics | Broadcast, recording, hostile-environment paging |
| Sensitivity to handset orientation | High — granules shift, output varies 3 to 6 dB | None | Very low |
Frequently Asked Questions About Telephone Transmitter
Granule heating. The 45 mA bias current dissipates around 100 mW in the granule pile, and as the carbon warms it dries out any residual moisture and the granules sinter — they fuse at their contact points. That raises R0 and shrinks the resistance swing, so output drops 4 to 8 dB over the first 5 to 10 minutes of use.
Tapping breaks the sintered bridges loose. Long-term fix is to drop bias to 35 mA if your exchange will tolerate it, or replace the capsule with one that has properly graded granule sizing — a single grain size sinters faster than a graded mix.
The bias current is what matters, not the voltage. Pick your feed resistor so 30 to 60 mA flows through the capsule's 50 Ω quiescent resistance. At 12 V that means a feed resistor of around (12 − 2.5) / 0.045 ≈ 210 Ω, dropping about 9.5 V across itself and the rest across the capsule and induction coil primary.
The catch is headroom. With only 12 V on the loop, large speech peaks can pull the capsule voltage low enough that the granule swing clips on the negative half-cycle. You'll get audible asymmetric distortion. Either limit speech levels with a foam windscreen on the mouthpiece, or move to 24 V if you want clean output above 1 Pa.
Almost always the diaphragm sensitivity Sd is higher than the assumed 8 Ω/Pa. Two real-world causes: a thinner-than-spec diaphragm (a 0.04 mm replacement instead of the 0.05 mm original moves more under the same pressure), or a partially loaded granule chamber where the pile is shallower than design and individual grain motions translate more directly to bulk resistance change.
Diagnostic check: measure R0 by passing a known 10 mA through the capsule at rest. If R0 reads 35 to 40 Ω instead of 50 Ω, the chamber is under-filled and you're seeing inflated sensitivity that will collapse when the granules redistribute under the next handset orientation change.
If the exhibit is purely visual and the goal is reliability for thousands of visitor uses, fit an electret retrofit (Mod-comm and a few aftermarket suppliers make drop-in T1 form-factor electret cartridges with built-in line-matching). They last 20+ years and never pack.
If the exhibit's whole point is authenticity — visitors comparing a 1955 call to a modern call — keep the carbon. The peaky 300 Hz to 3 kHz response and the slight harmonic edge are the sound of mid-century telephony, and an electret capsule simply does not reproduce that timbre. Budget for a fresh capsule every 3 to 5 years under museum duty.
The granule pile has settled away from the front electrode. With the handset face-up, gravity pulls the granules toward the rear cup and the carbon-button contact patch loses pressure — bulk resistance shoots up to several hundred ohms and the bias current drops below the threshold where speech modulation is audible.
This is intrinsic to the technology and was a known nuisance even in Bell System days. The fix is mechanical: the chamber should be filled to about 95% capacity with the diaphragm slightly preloading the pile. If your capsule fails this orientation test, the granule charge is low. Some restorers add a tiny amount of fresh graded carbon through the rear vent, but a NOS replacement capsule is the cleaner path.
Diaphragm leak. The carbon transmitter chamber needs to be acoustically sealed at low frequencies — there's a small vent for atmospheric pressure equalisation but it's tuned to a few Hz, not a few hundred. If the diaphragm gasket has perished, or the front cap is not seated, low-frequency sound pressure equalises through the leak instead of moving the diaphragm.
Quick check: cover the mouthpiece holes with your finger and lightly tap the back of the capsule. You should hear a clear thump in the receiver. If you hear a thin tick instead, the seal is gone. A 0.5 mm cork or rubber gasket cut to the bezel diameter usually restores low-end response.
For a Western Electric F1-class capsule in residential use (a few minutes of speech per day, handset stored face-down on the cradle), R0 typically rises from 50 Ω to 65–75 Ω over 12 months. Plug that back into the formula: at 65 Ω the nominal output drops from 7.2 mA to about 5.5 mA, a 2.3 dB loss. Past users described the result as the line getting 'a bit further away' over time.
In heavy industrial paging service the same capsule can hit 100 Ω in under a year because granule heating accelerates sintering. That's why GAI-Tronics published a 12 to 18 month replacement interval for the carbon capsule in their Page/Party handsets, regardless of whether it had failed outright.
References & Further Reading
- Wikipedia contributors. Carbon microphone. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
