Telephones are older than vacuum tubes and of course transistors. How was signal amplification done?

I mean the technology, not the details.


Some additional informations that I should have given at the beginning :

  • The question is restricted to telephony
  • I'm not interested in experimental equipments Say for example that the equipment should have been done in quantity more than 30 pieces, or better that it has been a commercial product.
  • I'm not only interested in purely electric solutions : it can be mechanical, hydraulic ... (but without human repeaters !)
  • impedance adaptation (acoustic or electric ...) are not considered here as amplification


  • There wasn't any amplification between the microphone and the headphone, even for a 1200km transmission, but yelling was necessary at one side, and absolute silence at the other side (see WhatRoughBeast answer)

  • the carbon microphone itself is a amplifier. The general definition of a amplifier is not always very easy (see Ali Chen answer, and BillF second answer, if you can follow), but it suffice to say that a electrodynamic speaker coupled to a carbon microphone is a electrical amplifier (see next-hack answer and BillF first anwer). I add that the other kinds of microphones are attenuators (hence the question)

  • the best telephonic lines have a loss of only 0.04dB/km at audio frequency. (compare with 10dB/km at 300kHz for our telephonic lines)

  • The loutest sound that a human can stand is more than 80dB above the lowest he can ear. (RussellBorogove comment). It is possible that the sound inside the horn (see peufeu answer) is even more loud than what a human can stand.

Thanks for the contributions.

  • 4
    \$\begingroup\$ Carbon granule mike to electromagnetic earpiece did not need amplification. You'd be surprised what amplifiers were around before tubes and transistors. Look up the rotating drum mechanical-acoustic amplifier - PA rather than telephone, capstan, not free energy! \$\endgroup\$ – Neil_UK Oct 3 '17 at 5:21
  • 2
    \$\begingroup\$ Search at magnetic amplifier or saturable reactor. en.wikipedia.org/wiki/Magnetic_amplifier \$\endgroup\$ – Szymon Oct 3 '17 at 5:56
  • 4
    \$\begingroup\$ A relay would work quite well for some applications. No it's not linear, and yes it's slow, but it does amplifies a signal. \$\endgroup\$ – user3528438 Oct 3 '17 at 18:45
  • 1
    \$\begingroup\$ @user35284338 Your comment is a joke \$\endgroup\$ – andre314 Oct 3 '17 at 18:55
  • \$\begingroup\$ Thanks to @next-hack answer, I'm discovering that carbon microphones are capable of large power amplification (X100 according to a not verified internet source). \$\endgroup\$ – andre314 Oct 3 '17 at 19:26

10 Answers 10


It wasn't.

In the early days (ca 1890) long-distance telephony was done from sound-proofed booths and often using non-standard (4-wire) telephones. New York to Chicago was the limit on range. By 1911, the use of loading coils allowed communication from New York to Denver. Notice, though, that this was pretty frustrating and involved a lot of yelling.

In 1915 the first transcontinental call was made, with amplification using the audion vacuum tube.

  • 106
    \$\begingroup\$ And then 90 years later, the "smartphone" was invented and, by cunningly placing a poor quality microphone in an animated tablemat a considerable distance from the mouth, the technique of "pretty frustrating and a lot of yelling" was reinvented. This has now been augmented by the "headset", which using advanced electronics is able to simulate the effect of the person speaking being trapped at the other end of a long drainpipe with a sock crammed in their mouth. \$\endgroup\$ – Ian Bland Oct 3 '17 at 12:56
  • 2
    \$\begingroup\$ Could you expand on the "non-standard (4-wire) telephones"? Was it some kind of Kelvin sense arrangement, or just one pair for each direction, or something else? \$\endgroup\$ – pericynthion Oct 3 '17 at 15:36
  • 8
    \$\begingroup\$ @pericynthion - 1 pair per direction. \$\endgroup\$ – WhatRoughBeast Oct 3 '17 at 15:38
  • 5
    \$\begingroup\$ You mean 1200km without any amplification ! ? \$\endgroup\$ – andre314 Oct 3 '17 at 18:59
  • 2
    \$\begingroup\$ @andre - Yup. But note the need for a soundproofed room. Received signal was very weak. Plus, the 4-wire system meant no losses in the hybrid coil. \$\endgroup\$ – WhatRoughBeast Oct 4 '17 at 1:24

Amplification, before tube amplifiers were available, could be achieved in several ways. Some of them are:

  • Using a electrodynamic speaker coupled to a carbon microphone, to form a repeater.
  • Using the negative resistance of a mercury lamp (even if it the physical nature of the negative resistance was not understood).
  • Using a magnetic amplifier (still used well after the tube- and transistor-era, and even today in some applications). Magnetic amplifiers were initially used as a switch in late 1800, and they were used as amplifiers when tubes were already available, but high-reliability was very important.
  • 6
    \$\begingroup\$ Magnetic amplifiers provided the energy to move 1,000 ton battleship turrents, for precision servo-controlled aiming. \$\endgroup\$ – analogsystemsrf Oct 3 '17 at 16:46

The funnel you're supposed to yell into is actually a horn:

enter image description here

Horns are acoustic transformers, they are usually used the other way around: a high-pressure transducer ("compression driver") is mounted on the throat, and the horn outputs a low-pressure acoustic wave on the other end with a much larger surface area, which results in high acoustic power. Basically a megaphone:

enter image description here

But they work both ways. When mounted on a microphone, the whole surface area of the "mouth" of the horn is used to input acoustic waves, and at the other end it outputs much higher pressure, which increases the signal available to the primitive microphones of the day...

This is not amplification as it is purely passive, but I'm sure it helped.

Note: horns also distort due to internal reflections and resonance modes, which is why they usually sound a bit "honky".

  • 5
    \$\begingroup\$ But they work both ways - Plugin your earphones into your microphone jack, fire up a sound recorder and see for yourself. Disclaimer: May involve yelling to get satisfactory results. \$\endgroup\$ – Tejas Kale Oct 3 '17 at 20:13
  • \$\begingroup\$ Audio devices are transducers - so like antennas can indeed work both ways. One thing I would add is how hearing aids concentrate sound in the same way the microphone horns on the old phones. Of course it is all done with electronics now. But I always crack up when I see old photos of the elderly men holding up big horns to their ears, saying, "Eh?" \$\endgroup\$ – SDsolar Oct 10 '17 at 3:50

The pre-vacuum-tube telephone system used exactly the carbon-granule microphone as the amplifier. What Bell invented was the use of essentially the same device as speaker and microphone. Now we call this the dynamic microphone. Bell's system had no gain mechanism; it had no intrinsic advantage with respect to range of communication over two tin cans connected by a string. The practical range was a small number of city blocks.

The carbon microphone has the required property of an amplifier: it functions basically as a valve. Thus, the electrical signal can have a far larger power than the acoustical signal. This was the invention that took the telephone from being a toy to being a telecommunication system. (By the way, the famous quote from the executives of Western Union, that they could not see any practical application of the Bell's telephone 'at the present time' was specifically in reference to Bell's patents, which is what they had been offered. Their objection was precisely the limited range.)

The carbon microphone was invented by Thomas Edison, and he and everyone else involved knew exactly what was going on (amplification) and why it was necessary for the growth of the telephone system. The story is detailed, at least from Edison's point of view in a book written by his patent attorney, Frank Lewis Dyer, Edison His Life and Inventions, Chapter 9. (Available from http://www.gutenberg.org/files/820/820-h/820-h.htm ). The carbon microphone was also used in an early electrical analog amplifier called the "telephone relay" which was essentially a telephone receiver mechanically coupled to a carbon button. There is a lot of information about early amplifier technologies in a book by H. Peter Friedrichs called Instruments of Amplification: Fun with Homemade Tubes, Transistors, and More (2003).

If you are wondering why it is that we only know of Bell's contribution to the telephone, when it was really Edison who made the system scalable, blame J. Pierpont Morgan. He imposed the structure of early 20th century high-tech quasi-monopolies in a reorgainzation of companies that enforced the necessary patent licensing. In essence Bell was given what became AT&T and Edison was given GE. It was Morgan who enforced a broad scope interpretation of Bell's patents, not anyone connected with the U.S. Patent system.

My own interest in this history goes back a few years when I was looking into whether any technological system had ever succeeded in scaling (by more than say a factor of 10) without the use of a gain element. The reason that this was and is of interest are the claims for reversible and quantum computing, for example, which cannot have gain elements. I include in "amplification elements" such things as the stirring stick in the Paleolithic campfire, the paddle valves in canal locks, and the steam engine throttle. It appeared that the early telephone system was the exception. In fact it turned out to be the absolutely classic example of a system concept that was going nowhere until the gain element was invented.

  • \$\begingroup\$ My english vocabulary is limited, but I would say that a valve (I mean the vacuum tube) is rather a diode. Don't you mean rather a triode ? \$\endgroup\$ – andre314 Oct 4 '17 at 19:45
  • 3
    \$\begingroup\$ By "valve" I mean a mechanism that can externally control a rate of flow. Think of the valves with handles that control water flow. The British used the word "valve" initially for Fleming's vacuum diode, and here the piping analogy is the "check valve", which permits flow in only one direction, and has no externally movable part. When de Forest invented the triode the British use of "valve" was applied to these new devices as well. Note that both control valves and check valves were important parts of the high technology of the time, the steam engine, and therefore useful analogies. \$\endgroup\$ – BillF Oct 4 '17 at 20:02

Telephones are older than vacuum tubes and of course transistors. How was signal amplification done?

I remember when I was a kid you could buy a plastic handset that connected to another plastic handset over a pair of wires that were a hundred feet (or so) in length. You could speak into the handset (no batteries required) and at the other handset the person could hear you. It worked in both directions i.e. there was a small loudspeaker (that doubled as a microphone) connected with two wires to another small loudspeaker in the distant handset.

It relied on the loudspeaker being about 10% efficient i.e. it coupled about 10% of the acoustic power it received and this was converted to about 1% of the original sound power at the other end. It was enough to hold a conversation and it didn't need batteries.

There is a company who (last time I looked) produced "sound-powered" telephones for flammable gas environments because they are intrinsically safe and won't cause an explosion i.e. this tech is still in use today.

  • 2
    \$\begingroup\$ ... and you wanted to make the sound louter and that's why you are analogue electronics engineer today. \$\endgroup\$ – andre314 Oct 3 '17 at 20:01
  • 2
    \$\begingroup\$ And the military uses them because they work even when the power is out. \$\endgroup\$ – Loren Pechtel Oct 4 '17 at 3:00
  • \$\begingroup\$ 100ft ... I know it's not the fun for ~9yo boys, but wouldn't yelling have worked better? :-) \$\endgroup\$ – yo' Oct 7 '17 at 16:43

If you could call it amplification, it was done in the carbon microphone into which you speak. A voltage was applied across the carbon element of the microphone. Sound waves altered the resistance of the element producing a varying current. This can produce an electrical signal more powerful than the original sound. From there, pretty much everything the signal went through would attenuate it to some degree - wires, transformers, etc.. As noted in the Wikipedia article, carbon microphones can form the basis of an amplifier, but not a very good one.

  • \$\begingroup\$ Totally agree. It is not clear for me how to calculate a power gain between a acoustic domain and a electric domain, but because "electrodynamic speaker coupled to a carbon microphone" has gain (purely electric), it is clear that something has gain, and this something is necessaly the microphone \$\endgroup\$ – andre314 Oct 4 '17 at 6:01
  • \$\begingroup\$ @andre not clear to me either how to calculate gain for a carbon microphone, but there is a relationship between sound pressure as it is applied to the carbon element and electrical resistance of the element. Acoustic energy is driving the time-varying pressure on the element; a DC input voltage is driving the line current through a time-varying resistance in the element. The acoustic energy itself doesn't drive the electrical signal, it only shapes the current driven by the power supply voltage. Like when you turn on a light - your finger on the switch doesn't power the light. \$\endgroup\$ – Anthony X Oct 7 '17 at 0:30
  • \$\begingroup\$ @andre the common unit is power. 1 Pa * 1 m^3/s = 1 N * 1 m/s = 1 V * 1 A = 1 W :) \$\endgroup\$ – hobbs Oct 7 '17 at 9:04
  • \$\begingroup\$ Estimating the flow rate (the m^3/s term) seems like the hard part, but maybe it's fair to use the speed of sound times the effective aperture area of the mouthpiece, times a term to convert peak to RMS :) \$\endgroup\$ – hobbs Oct 7 '17 at 9:09

Before vacuum tubes and transistor amplification, most amplification of sound was done at acoustical level, by means of exponential horns. The amplification occurs by means of creating directional sound, and better match between acoustic impedance of open air and transducer membrane. One simplest example of this technique is to hold both palms around your mouth when shouting into public, or use of a paper cone.

A better design was the Edison Phonograph, wikipedia.

enter image description here

  • \$\begingroup\$ Yes, but the horn is a passive device (since it is not powered). It helps to solve the problem of lack of power in the signal, but the point of my question is where were are the elements (if any) that afford power amplification (these elements absolutly needs a power source, otherwise we have invented the perpetual movement) \$\endgroup\$ – andre314 Oct 6 '17 at 19:11
  • \$\begingroup\$ In other words : there should amplification of a quantity expressed as Watts (Not Volt, Ampere, Pascal, Bar,meters (for Displacement) ... any quantity that can be added when at a same frequency). Our ear is sensible to displacement (of a internal membrane in the ear), so we feel an amplification but it's not really one (it's a impedance adaptation) \$\endgroup\$ – andre314 Oct 6 '17 at 19:27
  • \$\begingroup\$ ... a impedance adaptation that minimize the power loss between the source and the load, but don't create power. \$\endgroup\$ – andre314 Oct 6 '17 at 19:38
  • \$\begingroup\$ @andre, then you probably need to clarify your question. \$\endgroup\$ – Ale..chenski Oct 6 '17 at 19:49
  • \$\begingroup\$ Done. Thanks ! (please don't remove your answer, it's a opportunity to clarify this point) \$\endgroup\$ – andre314 Oct 6 '17 at 19:59

Here's a supplement to answers already offered, to clarify the role of a third factor sometimes linked (confusingly) with attenuation/amplification.

As 'andre' already wrote in one of his comments, 'attenuation is not the only limiting factor' (in the usability of the telephone signal).

For the early telephone lines in the era before amplification, probably the next most important factor in usability was the extent of signal distortion, especially phase distortion.

Loading-coils, designed and applied in accordance with Heaviside's theory of transmission-line behavior, greatly reduced distortion. This description of the 'Heaviside condition' (+ references) shows what was done and makes clear that the design-aim was 'no distortion', not 'no loss'.

The inductive loading-coils used to approximate the 'Heaviside condition' were essentially passive elements. They were naturally also somewhat resistive, hence even marginally increasing power/amplitude losses. But the advantage in reduced distortion outweighed that.

In some sources, this good and even dramatic result has been rather confusingly described, if not as 'amplication', then certainly as that the coils 'reduced attenuation' (e.g. here). Of course the bad thing reduced was the garbling and unusability of the signal due to distortion, not the power- or amplitude-loss as such.

(If the meaning of 'attenuation' can be taken broadly enough to cover loss of intelligibility as well as loss of power / amplitude, then maybe ok, but the sources don't make that clear.)


Let me outline a general theory of active systems and gain. An active system is one which contains a "generalized engine," which is a mechanism that draws upon some source of Helmholtz free energy to push another system (the "load") away from thermal equilibrium. In electrical technologies voltage is specifically Helmholtz free energy (which some are now calling "exergy") per charge. And thermal equilibrium is every node at ground potential.

Useful engines will have some mechanism to control their action, and a key question is how much energy or power does the control mechanism absorb? Obviously, if it takes more power to turn on a motor, say, than the power it delivers, you have a losing proposition. The ratio of energy delivered to the load to the energy absorbed by the control mechanism is the gain. (Of course the systems that we call amplifiers meet this definition of "engine.")

Now, there is a question about whether we need to use energy or power. If the action of the engine is continuous, the output is measured as a power. If the control mechanism also continuously absorbs energy, the input is also a power, and the gain is just the ratio of these quantities. But, we often have cases where a discrete input of energy can control a continuous power. Large-geometry MOSFETs work this way. What we can quote is output power over input energy, and we get a number which has units of frequency. Of course we know how to interpret this as a gain-bandwith product: the gain depends upon how fast you want the system to work. But, this is still a relevant measure of gain.

Take the example of a steam locomotive: The output power is rather obvious, but the control mechanism is the throttle, which requires a one-time input of energy to change its state. Thus a locomotive can be characterized by its gain-bandwidth product.

The remaining case is where a continuous input of power is required to maintain some net energy in the load. Muscles work this way, but hardly any technologically useful systems work this way.

You can use this conceptual framework to describe virtually anything that does something. A particularly amusing exercise is to go back and look at the cartoon mechanisms of Rube Goldberg, and identify the sources of energy, the control mechanisms, and some estimate of the gain involved. If the overall gain is less the one, the action will fizzle out and the machine would stall before the final action is triggered.

  • \$\begingroup\$ That's remind automatics, where the impulse response unit is Hz (!), when input and output are homogen (for example Volts) \$\endgroup\$ – andre314 Oct 5 '17 at 19:28
  • \$\begingroup\$ Not the same thing ? \$\endgroup\$ – andre314 Oct 5 '17 at 19:30
  • 1
    \$\begingroup\$ Your interesting "answer" goes beyond the question, though I think your motivation is not to harvest reputation points. \$\endgroup\$ – andre314 Oct 5 '17 at 19:45
  • \$\begingroup\$ There was some hesitancy to call the acoustic-to-electronic conversion process "gain," so I simply tried to show how gain is in fact a universal property of technologies. \$\endgroup\$ – BillF Oct 5 '17 at 20:04
  • \$\begingroup\$ Thanks. My hesitancy was more basic : how to define the real power received by the microphone membrane (without speaking of acoustic impedance adaptation, another difficulty). But never mind. \$\endgroup\$ – andre314 Oct 5 '17 at 20:16

There is a wonderful book you should see: "Instruments of Amplification" by H. P. Friedrichs (AC7ZL). It gives a lot of information on carbon mike audio amplifiers, flame triodes, carbon arcs, and other weird ideas. Electromechanical relays began their life as amplifiers of discrete telegraph signals, but there were attempts to use them as audio power amplifiers.

For the many of these negative feedback was required to get something like a linear response.

Google Voice of the Crystal, or go to the ARRL website. Have fun, but don't go blind or burn the house down with your carbon arc!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.