from the coherer to dsp - ebu technology & innovation the coherer to dsp m. lemme and r....

63EBU Technical Review Spring 1995Lemme & Menicucci

From the coherer to DSPM. Lemme and R. Menicucci (Vatican Radio)

This article reviews the developmentof electronic devices used over thelast century in wirelesscommunication. It looks at earlyreceiving devices such as thecoherer, the magnetic detector andthe cat’s whisker, progressing to thethermionic valve, the semiconductor,the microchip and digital signalprocessing.

On the transmission side, the earlydevices discussed include thespark-gap generator, the voltaic-arcgenerator and static frequencymultipliers. This is followed by abrief description of more modernpower devices, including thermionicvalves and electron-velocity controltubes.

1. Introduction

The electromagnetic-wave generator that wasfirst conceived by Hertz, in 1887, consisted of acopper rod 3 m long which had a 30 cm zincsphere attached to either end. The rod wasdivided in the middle to include a spark gap of0.75 cm between two small brass spheres. Thisassembly – which we now call a Hertz dipole –was connected across the secondary winding ofan induction coil. When the primary coil was con-

nected to a battery, a spark would appear acrossthe gap, generating an oscillating current. As aconsequence, damped electrical waves wouldpropagate across the room.

The wavelength produced by the Hertz resonatordepended on the inductance and the capacitanceof the circuit (the large zinc spheres could bemoved along the rod to vary the wavelength ofthe oscillations). Hertz was aiming to reproducephenomena similar to light and the shortestwavelength he managed to generate was about30 cm, equivalent to an oscillation frequency of1000 megacycles/second (1000 MHz in modernparlance).

The energy radiated by the dipole was called“Force Diffusion” by Hertz; it was Lord Kelvinwho later gave it the name electromagnetic wave.Sir William Crookes was very surprised when henoticed that electromagnetic waves could passthrough “the walls and the fog of London”. In1892, he gave an explanation of this phenomenonand was even able to foretell wireless telegraphy.

2. Early receiving devices

2.1. Hertz resonator

The first and simplest detector of electromagneticwaves was the Hertz resonator. It consisted of anopen metal ring with a small metal sphere attachedat either end. A spark jumped across the small gapbetween the two spheres, whenever a spark wasgenerated in the transmitting apparatus.

From the coherer to DSPM. Lemme and R. Menicucci (Vatican Radio)

This article reviews the developmentof electronic devices used over thelast century in wirelesscommunication. It looks at earlyreceiving devices such as thecoherer, the magnetic detector andthe cat’s whisker, progressing to thethermionic valve, the semiconductor,the microchip and digital signalprocessing.

On the transmission side, the earlydevices discussed include thespark-gap generator, the voltaic-arcgenerator and static frequencymultipliers. This is followed by abrief description of more modernpower devices, including thermionicvalves and electron-velocity controltubes.

1. Introduction

The electromagnetic-wave generator that wasfirst conceived by Hertz, in 1887, consisted of acopper rod 3 m long which had a 30 cm zincsphere attached to either end. The rod wasdivided in the middle to include a spark gap of0.75 cm between two small brass spheres. Thisassembly – which we now call a Hertz dipole –was connected across the secondary winding ofan induction coil. When the primary coil was con-

nected to a battery, a spark would appear acrossthe gap, generating an oscillating current. As aconsequence, damped electrical waves wouldpropagate across the room.

The wavelength produced by the Hertz resonatordepended on the inductance and the capacitanceof the circuit (the large zinc spheres could bemoved along the rod to vary the wavelength ofthe oscillations). Hertz was aiming to reproducephenomena similar to light and the shortestwavelength he managed to generate was about30 cm, equivalent to an oscillation frequency of1000 megacycles/second (1000 MHz in modernparlance).

The energy radiated by the dipole was called“Force Diffusion” by Hertz; it was Lord Kelvinwho later gave it the name electromagnetic wave.Sir William Crookes was very surprised when henoticed that electromagnetic waves could passthrough “the walls and the fog of London”. In1892, he gave an explanation of this phenomenonand was even able to foretell wireless telegraphy.

2. Early receiving devices

2.1. Hertz resonator

The first and simplest detector of electromagneticwaves was the Hertz resonator. It consisted of anopen metal ring with a small metal sphere attachedat either end. A spark jumped across the small gapbetween the two spheres, whenever a spark wasgenerated in the transmitting apparatus.

64 EBU Technical Review Spring 1995Lemme & Menicucci

2.2. Coherer

An improved detector, the coherer, was inventedby the Frenchman, Edouard Branly. It was basedon the discoveries of the Anglo-American, D.E.Hughes, and the Italian, Calzecchi Onesti – whichhad been further updated by the Englishman, SirOliver Lodge, as well as the Russian, A.S. Popov.

The coherer consisted of a tube filled with finemetal filings (Marconi used 95% nickel and 5%silver). When RF energy was passed through thefilings, the particles cohered and the resistancedropped. The coherer could be made responsiveto a change of state, i.e. to the presence orabsence of a signal. When a signal was present,a battery-powered circuit attracted the movingelement of a relay and, consequently, the wholeassembly could be used to receive Morse codesignals from a remote transmitter. In order toseparate the particles inside the bulb, so that theapparatus could be operated again, some form of

“de-coherer” such as a solenoid tapper or amechanical shaker was necessary.

A coherer with an antenna and an earth connectionformed the receiving part of the equipment used byMarconi in 1896/7 to carry out his experiments atPontecchio near Bologna, Italy, and at LavernockPoint near Cardiff, Wales. A circuit diagram of thelatter installation is shown in Fig. 1.

The coherer was also used in the system Marconideveloped to communicate with ships at the turnof the century. In this system, according to hispatent No. 7777 (Fig. 2), Marconi connected thetransmitting antenna to the secondary winding(jigger) of a “Tesla transformer” (formerly theantenna had been connected to ground via themetal spheres of the exciter). The primary wind-ing of this transformer was connected in serieswith the exciter and a capacitor, formed byLeyden jars. At the receiving station, the anten-na was connected to ground through the primarywinding of a Tesla transformer, whose jigger wasconnected to the coherer. The first apparatus ofthis type was installed by Marconi in the Ameri-can liner St. Paul at the end of 1899.

Although the coherer was used successfully forseveral years, it soon showed its poor efficiencyand sensitivity. In December 1901, at St. John’s inNewfoundland, Canada, Marconi replaced it witha mercury drop detector.

2.3. Magnetic detector

The early detectors were rather unstable andirregular in operation. These problems, alongTransmitter Receiver

Jigger

Coherer

Figure 1Circuit diagram of thetransmit-receivesystem used duringMarconi’s LavernockPoint trials in May1897(as published by theGeneral Post Officein the Engineer-in-Chief’s report of1903).

Figure 2Marconi’s system forlong-range radioreception, as fitted tothe American linerSt. Paul in 1899(Patent No. 7777).

2.2. Coherer

An improved detector, the coherer, was inventedby the Frenchman, Edouard Branly. It was basedon the discoveries of the Anglo-American, D.E.Hughes, and the Italian, Calzecchi Onesti – whichhad been further updated by the Englishman, SirOliver Lodge, as well as the Russian, A.S. Popov.

The coherer consisted of a tube filled with finemetal filings (Marconi used 95% nickel and 5%silver). When RF energy was passed through thefilings, the particles cohered and the resistancedropped. The coherer could be made responsiveto a change of state, i.e. to the presence orabsence of a signal. When a signal was present,a battery-powered circuit attracted the movingelement of a relay and, consequently, the wholeassembly could be used to receive Morse codesignals from a remote transmitter. In order toseparate the particles inside the bulb, so that theapparatus could be operated again, some form of

“de-coherer” such as a solenoid tapper or amechanical shaker was necessary.

A coherer with an antenna and an earth connectionformed the receiving part of the equipment used byMarconi in 1896/7 to carry out his experiments atPontecchio near Bologna, Italy, and at LavernockPoint near Cardiff, Wales. A circuit diagram of thelatter installation is shown in Fig. 1.

The coherer was also used in the system Marconideveloped to communicate with ships at the turnof the century. In this system, according to hispatent No. 7777 (Fig. 2), Marconi connected thetransmitting antenna to the secondary winding(jigger) of a “Tesla transformer” (formerly theantenna had been connected to ground via themetal spheres of the exciter). The primary wind-ing of this transformer was connected in serieswith the exciter and a capacitor, formed byLeyden jars. At the receiving station, the anten-na was connected to ground through the primarywinding of a Tesla transformer, whose jigger wasconnected to the coherer. The first apparatus ofthis type was installed by Marconi in the Ameri-can liner St. Paul at the end of 1899.

Although the coherer was used successfully forseveral years, it soon showed its poor efficiencyand sensitivity. In December 1901, at St. John’s inNewfoundland, Canada, Marconi replaced it witha mercury drop detector.

2.3. Magnetic detector

The early detectors were rather unstable andirregular in operation. These problems, alongTransmitter Receiver

Jigger

Coherer

Figure 1Circuit diagram of thetransmit-receivesystem used duringMarconi’s LavernockPoint trials in May1897(as published by theGeneral Post Officein the Engineer-in-Chief’s report of1903).

Figure 2Marconi’s system forlong-range radioreception, as fitted tothe American linerSt. Paul in 1899(Patent No. 7777).

with some polemics, induced Marconi to aim hisresearch at a new type of detector. Inspired by apaper published in 1896 by Prof. Rutherford,Marconi developed the magnetic detector whichwas a significant improvement over the coherer.The magnetic detector employed clockworkmechanical power to magnify the weak incomingpulses from a spark transmitter so that they werestrong enough to drive a pair of headphones.

Basically, the magnetic detector operated as fol-lows (Fig. 3). A loop of hard-drawn iron wires(a) , wound together in the form of a “rope”, wasmade to move endlessly through two coils, bymeans of a clockwork drive. The inner coil (b)– a single-layer winding – was connected be-tween the aerial and the earth via a tuning filter.The outer coil (c) was made from a large numberof turns of fine wire and was connected directlyto the headphones (T). Two permanent magnets(d), with their “like-poles” together, produced aquadropole field that caused the moving ironrope to divide into two magnetic domains sepa-rated by a “wall” which was stationary relative tothe two coils.

Whenever a high-frequency “pulse” was receivedfrom the transmitter, the resultant magnetic fieldacted – via the primary coil – on the wire rope.Incoming half-waves of one polarity would causethe magnetic domain wall to “flick” backwardsand produce –via the secondary coil – a “click” onthe headphones, while incoming half-waves of theother polarity had little effect on the wall. Thus,the device could reproduce the dots and dashessent out from a spark transmitter, just like a crystal

detector – except that the magnetic detector alsooffered considerable gain.

Marconi’s magnetic detector was patented in 1902and was widely used for about 20 years.

Looking at the pictures and circuit diagrams ofthe time, an observer with today’s technicalknowledge could be excused for thinking thatsuch devices were rather simple and rudi-mentary. However, reading the journals of thoseearlier experimenters, one can get a feeling of theproblems they had to overcome. Every circuitcomponent caused problems, the solution ofwhich needed months of work and repeated

Figure 3Diagram from

Marconi’s PatentNo. 10245 of 1902

showing themagnetic detector.

Figure 4A very early receiver

using a galena crystaland cat’s whisker as

the detector.

with some polemics, induced Marconi to aim hisresearch at a new type of detector. Inspired by apaper published in 1896 by Prof. Rutherford,Marconi developed the magnetic detector whichwas a significant improvement over the coherer.The magnetic detector employed clockworkmechanical power to magnify the weak incomingpulses from a spark transmitter so that they werestrong enough to drive a pair of headphones.

Basically, the magnetic detector operated as fol-lows (Fig. 3). A loop of hard-drawn iron wires(a) , wound together in the form of a “rope”, wasmade to move endlessly through two coils, bymeans of a clockwork drive. The inner coil (b)– a single-layer winding – was connected be-tween the aerial and the earth via a tuning filter.The outer coil (c) was made from a large numberof turns of fine wire and was connected directlyto the headphones (T). Two permanent magnets(d), with their “like-poles” together, produced aquadropole field that caused the moving ironrope to divide into two magnetic domains sepa-rated by a “wall” which was stationary relative tothe two coils.

Whenever a high-frequency “pulse” was receivedfrom the transmitter, the resultant magnetic fieldacted – via the primary coil – on the wire rope.Incoming half-waves of one polarity would causethe magnetic domain wall to “flick” backwardsand produce –via the secondary coil – a “click” onthe headphones, while incoming half-waves of theother polarity had little effect on the wall. Thus,the device could reproduce the dots and dashessent out from a spark transmitter, just like a crystal

detector – except that the magnetic detector alsooffered considerable gain.

Marconi’s magnetic detector was patented in 1902and was widely used for about 20 years.

Looking at the pictures and circuit diagrams ofthe time, an observer with today’s technicalknowledge could be excused for thinking thatsuch devices were rather simple and rudi-mentary. However, reading the journals of thoseearlier experimenters, one can get a feeling of theproblems they had to overcome. Every circuitcomponent caused problems, the solution ofwhich needed months of work and repeated

Figure 3Diagram from

Marconi’s PatentNo. 10245 of 1902

showing themagnetic detector.

Figure 4A very early receiver

using a galena crystaland cat’s whisker as

the detector.

experiments. Only rough materials were used:for example, the prototype of the magneticdetector was enclosed in a Havana cigar box.There were no suitable instruments to measureRF currents, nor was there any well-groundedtheory to aid the design of these early wirelessinstallations.

2.4. Cat’s whisker

Another simple detector around this time was thefamous cat’s whisker (Fig. 4). It consisted of a finewire electrode whose pointed tip was pressedagainst the crystal of a detector (very often acrystal of galena). When connected to a suitabletuned circuit, an antenna, an earth point and aheadphone, this simple circuit formed a veryuseful receiver for use with local stations.

3. Thermionic valves

3.1. Diodes

All the above detectors eventually gave way tothe thermionic valve (Fig. 5) which was con-ceived by Prof. Fleming in 1904. Marconi soonrealized the practical importance of this inven-tion and told one of his collaborators at Clifden:“the future of the radio will be based upon thissmall lamp. Its use will require the help of specialdevices. Many experiments will be needed toimprove it”.

In 1905, Fleming’s diode was installed in thereceivers of the Marconi company. The diodeconsisted of a glass bulb containing a carbon fila-ment (the cathode) surrounded by a cylinder of

thin metal (the anode or plate). When the filamentwas heated until white hot, the emitted electronsreached the anode plate. If the latter was at apositive potential with respect to the filament, itattracted the electrons so that a current flowed inthe filament-plate circuit. If the plate becamenegative, no current flowed through the valve.Thus, by using a thermionic diode, very small HFcurrents could be rectified into unidirectionalcurrents.

3.2. Triodes

In 1906, Lee de Forest experimented withFleming’s valve and ingeniously conceived a thirdelectrode, the control grid, in order to control theflow of electrons. When the grid was negative withrespect to the filament, the electrons were repelledso that the anode current was reduced. When thepotential of the grid changed to positive withrespect to the filament, the anode current increasedso as to produce an amplifying effect. At first, thetriode valve proved to be a sensitive detector; itwas not until 1912 that Lee de Forest appreciatedand exploited its possibilities as an amplifier.

In the first triode receivers, the radiofrequencysignal was taken direct from the antenna to thevalve input via a Tesla transformer with a tuningcapacitor, thus providing a tuned-resonancecircuit. Rectification was achieved by means ofeither a single crystal or diode, if only half-waverectification was required, or a pair of crystals ordiodes if full-wave rectification was required.

Around this time, it had also been “discovered”that the triode valve could be used to generatecontinuous radiofrequency oscillations and,moreover, these oscillations could be modulatedat audio frequencies. The stage was now set forrapid developments in the field of radiotelephonyand Marconi was just one of many to exploit thisnew communications medium. In 1914, he usedvalve oscillators to carry out regular radio-telephony transmissions for the Italian Navy.Marconi was able to achieve distances of around110 km by using devices based upon theoscillating circuit of Captain Round (Fig. 6).

A triode oscillator works on the principle of feed-back from the anode circuit to the grid circuit.When an oscillating voltage of suitable amplitudeand phase is fed back from the anode plate to thegrid so as to increase the grid signal, the circuitstarts to oscillate with a frequency determined bythe circuit constants (inductance and capacitance)of the anode and the grid.

Figure 5An early thermionicdiode, circa 1904.

experiments. Only rough materials were used:for example, the prototype of the magneticdetector was enclosed in a Havana cigar box.There were no suitable instruments to measureRF currents, nor was there any well-groundedtheory to aid the design of these early wirelessinstallations.

2.4. Cat’s whisker

Another simple detector around this time was thefamous cat’s whisker (Fig. 4). It consisted of a finewire electrode whose pointed tip was pressedagainst the crystal of a detector (very often acrystal of galena). When connected to a suitabletuned circuit, an antenna, an earth point and aheadphone, this simple circuit formed a veryuseful receiver for use with local stations.

3. Thermionic valves

3.1. Diodes

All the above detectors eventually gave way tothe thermionic valve (Fig. 5) which was con-ceived by Prof. Fleming in 1904. Marconi soonrealized the practical importance of this inven-tion and told one of his collaborators at Clifden:“the future of the radio will be based upon thissmall lamp. Its use will require the help of specialdevices. Many experiments will be needed toimprove it”.

In 1905, Fleming’s diode was installed in thereceivers of the Marconi company. The diodeconsisted of a glass bulb containing a carbon fila-ment (the cathode) surrounded by a cylinder of

thin metal (the anode or plate). When the filamentwas heated until white hot, the emitted electronsreached the anode plate. If the latter was at apositive potential with respect to the filament, itattracted the electrons so that a current flowed inthe filament-plate circuit. If the plate becamenegative, no current flowed through the valve.Thus, by using a thermionic diode, very small HFcurrents could be rectified into unidirectionalcurrents.

3.2. Triodes

In 1906, Lee de Forest experimented withFleming’s valve and ingeniously conceived a thirdelectrode, the control grid, in order to control theflow of electrons. When the grid was negative withrespect to the filament, the electrons were repelledso that the anode current was reduced. When thepotential of the grid changed to positive withrespect to the filament, the anode current increasedso as to produce an amplifying effect. At first, thetriode valve proved to be a sensitive detector; itwas not until 1912 that Lee de Forest appreciatedand exploited its possibilities as an amplifier.

In the first triode receivers, the radiofrequencysignal was taken direct from the antenna to thevalve input via a Tesla transformer with a tuningcapacitor, thus providing a tuned-resonancecircuit. Rectification was achieved by means ofeither a single crystal or diode, if only half-waverectification was required, or a pair of crystals ordiodes if full-wave rectification was required.

Around this time, it had also been “discovered”that the triode valve could be used to generatecontinuous radiofrequency oscillations and,moreover, these oscillations could be modulatedat audio frequencies. The stage was now set forrapid developments in the field of radiotelephonyand Marconi was just one of many to exploit thisnew communications medium. In 1914, he usedvalve oscillators to carry out regular radio-telephony transmissions for the Italian Navy.Marconi was able to achieve distances of around110 km by using devices based upon theoscillating circuit of Captain Round (Fig. 6).

A triode oscillator works on the principle of feed-back from the anode circuit to the grid circuit.When an oscillating voltage of suitable amplitudeand phase is fed back from the anode plate to thegrid so as to increase the grid signal, the circuitstarts to oscillate with a frequency determined bythe circuit constants (inductance and capacitance)of the anode and the grid.

Figure 5An early thermionicdiode, circa 1904.

Marconi was right when he foretold that Fleming’svalve would open up new horizons for radio-telephony. However, there were still many prob-lems to be solved, first of all concerning thevacuum in the valve bulbs. A small amount of gasin the glass bulb of de Forest’s Audion seemed atfirst to be advantageous for it to function. How-ever, the operation of the de Forest and Flemingvalves was irregular due to the gas within them.On the one hand, the amount of gas increased withtime; on the other hand, a variation of the biasbetween the anode plate and the filament undersuch conditions was sufficient to cause the break-down of the valve, due to the increased bombard-ment of positive ions onto the filament.

Langmuir, an American, discovered that vacuumreduction in the bulb was caused by gas atomscontained in the molecules of the filament and inthe metal and glass parts of the valve. He con-structed new pumps to create a hard vacuum in thebulb. He also studied ways of expelling the gasatoms from the metal and glass parts of the valveby the so-called “clean-up effect”, and he deviseda method to seal the filaments inside the glass bulbat a very high temperature. Lee de Forest was thefirst to introduce metal filaments in place of thecarbon filaments used previously. Further experi-ments led to the metal filaments being coated withalkaline oxides which improved the efficiency ofthe thermionic triode even more.

3.3. Tetrodes and pentodes

The plate-grid capacity – the cause of many prob-lems for technicians – was reduced to almost onehundredth that of a triode by providing a secondgrid (the so-called screen grid) to act as a screenand prevent the control grid from being influencedby the anode voltage changes. The screen gridreduced the anode-grid feedback and, hence, itprevented instability.

Later, the pentode (containing a third grid calledthe suppression grid) and other types of valveswere developed, leading to further progress in thedesign of receivers and transmitters. In addition tothe amplification and detection abilities of thesedevices, they could also be used to generate con-trolled oscillations within the receiver. Thus, forexample, these more sophisticated valves allowedProf. Fessenden to construct the first heterodynereceiver, which achieved signal detection by beat-ing the incoming RF signal against one producedby a local oscillator in the receiver. In this way, anunmodulated RF carrier was made audible, as thebeat note was at an audible frequency.

4. Superheterodyne receivers

Thereafter, the superheterodyne (superhet)receiver was developed. It had several importantadvantages over earlier types of receivers. In thesuperhet circuit, the incoming RF signal waschanged to a lower frequency known as the inter-mediate frequency (IF). The major part of theamplification then took place at this lower fre-quency before detection occurred in the normalmanner. In this way, it was possible to avoid thedrawbacks of tuned radiofrequency (TRF) cir-cuits, which were simple to design and constructbut which presented the risk of instability andhad a problem of the RF gain being dependentupon frequency.

5. Early transmitting devices

5.1. The spark gap

Let us turn now to the first transmitters. Followingon from the early experiments of Marconi andothers, a lot of improvements were achieved inboth the circuits and the power supplies of trans-mitters, due to the cooperation of Fleming. Onesuch circuit is shown in Fig. 7. In order to avoidtoo high a voltage at the secondary of the trans-former, a dual transformation technique was used.Here, a capacitor was first charged with a voltageof 20 kV, then its oscillatory discharge was

Figure 6The oscillating circuit

developed by CaptainH.J. Round.

Marconi was right when he foretold that Fleming’svalve would open up new horizons for radio-telephony. However, there were still many prob-lems to be solved, first of all concerning thevacuum in the valve bulbs. A small amount of gasin the glass bulb of de Forest’s Audion seemed atfirst to be advantageous for it to function. How-ever, the operation of the de Forest and Flemingvalves was irregular due to the gas within them.On the one hand, the amount of gas increased withtime; on the other hand, a variation of the biasbetween the anode plate and the filament undersuch conditions was sufficient to cause the break-down of the valve, due to the increased bombard-ment of positive ions onto the filament.

Langmuir, an American, discovered that vacuumreduction in the bulb was caused by gas atomscontained in the molecules of the filament and inthe metal and glass parts of the valve. He con-structed new pumps to create a hard vacuum in thebulb. He also studied ways of expelling the gasatoms from the metal and glass parts of the valveby the so-called “clean-up effect”, and he deviseda method to seal the filaments inside the glass bulbat a very high temperature. Lee de Forest was thefirst to introduce metal filaments in place of thecarbon filaments used previously. Further experi-ments led to the metal filaments being coated withalkaline oxides which improved the efficiency ofthe thermionic triode even more.

3.3. Tetrodes and pentodes

The plate-grid capacity – the cause of many prob-lems for technicians – was reduced to almost onehundredth that of a triode by providing a secondgrid (the so-called screen grid) to act as a screenand prevent the control grid from being influencedby the anode voltage changes. The screen gridreduced the anode-grid feedback and, hence, itprevented instability.

Later, the pentode (containing a third grid calledthe suppression grid) and other types of valveswere developed, leading to further progress in thedesign of receivers and transmitters. In addition tothe amplification and detection abilities of thesedevices, they could also be used to generate con-trolled oscillations within the receiver. Thus, forexample, these more sophisticated valves allowedProf. Fessenden to construct the first heterodynereceiver, which achieved signal detection by beat-ing the incoming RF signal against one producedby a local oscillator in the receiver. In this way, anunmodulated RF carrier was made audible, as thebeat note was at an audible frequency.

4. Superheterodyne receivers

Thereafter, the superheterodyne (superhet)receiver was developed. It had several importantadvantages over earlier types of receivers. In thesuperhet circuit, the incoming RF signal waschanged to a lower frequency known as the inter-mediate frequency (IF). The major part of theamplification then took place at this lower fre-quency before detection occurred in the normalmanner. In this way, it was possible to avoid thedrawbacks of tuned radiofrequency (TRF) cir-cuits, which were simple to design and constructbut which presented the risk of instability andhad a problem of the RF gain being dependentupon frequency.

5. Early transmitting devices

5.1. The spark gap

Let us turn now to the first transmitters. Followingon from the early experiments of Marconi andothers, a lot of improvements were achieved inboth the circuits and the power supplies of trans-mitters, due to the cooperation of Fleming. Onesuch circuit is shown in Fig. 7. In order to avoidtoo high a voltage at the secondary of the trans-former, a dual transformation technique was used.Here, a capacitor was first charged with a voltageof 20 kV, then its oscillatory discharge was

Figure 6The oscillating circuit

developed by CaptainH.J. Round.

supplied through a Tesla transformer to a secondoscillating circuit provided with a spark gap. Thiscircuit was connected to the antenna via thejigger (secondary winding) of a second Teslatransformer.

The spark gap consisted of two aligned steeldisks, almost tangential to each other with anarrow spark gap between their edges. Thelower portion of each disk was dipped in its ownseparate vessel filled with mercury, to avoid theoverheating of the disks during electrical dis-charges. It was an installation such as thiswhich succeeded in repeatedly transmitting thefamous Morse code signal for the letter “S”(dot-dot-dot) from Poldhu in Cornwall, Eng-land, to St. John’s in Newfoundland, Canada, inDecember 1901.

This momentous event gave new impetus to theresearchers. The simple spark gap needed to beimproved. After arcing, it did not lose its con-ductivity immediately, because of ionizationeffects, and this prevented the capacitor fromrecharging quickly, resulting in a greater energyconsumption and the generation of fewer wave-trains. The simple spark gap was duly replacedby the quenched spark gap and, afterwards, bythe rotary discharger (Marconi, 1905) in orderto cause the electrode to quench and the sparksto be interrupted more quickly.

However, there was still the problem of dampedwaves and low efficiency, which allowed only thetransmission of telegraphy signals; intermittentoperation was not adaptable to telephony modu-lation.

5.2. Voltaic-arc generators

There was a need to go back to basic theory. In1850, Sir W. Thompson had defined the relation-ship between the resistance, inductance and capac-itance that was needed to provide an oscillatorydischarge from a circuit. Based on those ideas,generators of undamped waves were eventuallydeveloped, along with voltaic-arc generatorswhich used carbon electrodes.

The operating principle of the voltaic-arcgenerator (Fig. 8) is based upon the negativeresistance of the arc. A series capacitor and induc-tor, connected in parallel with an arc, are chargedby a DC power supply. This causes the voltageacross the arc and the LC circuit to increase. How-ever, the capacitor is charged to a greater voltagethan that of the arc, due to the effect of the inductor.When the capacitor starts to discharge, the currentacross the arc increases, thus causing the voltage todecrease. When the voltage of the capacitorequals that of the arc, the inductor maintains thedischarge current, causing the voltage across thecapacitor to become lower than that of the arc. Atthe end of the discharge cycle, the capacitor ischarged again and the process is repeated.

ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ

Teslatransformer

Powermeter

Figure 7Circuit diagram of aMarconi transmitteraround the time of thePoldhu - St. John’strials (1901).

supplied through a Tesla transformer to a secondoscillating circuit provided with a spark gap. Thiscircuit was connected to the antenna via thejigger (secondary winding) of a second Teslatransformer.

The spark gap consisted of two aligned steeldisks, almost tangential to each other with anarrow spark gap between their edges. Thelower portion of each disk was dipped in its ownseparate vessel filled with mercury, to avoid theoverheating of the disks during electrical dis-charges. It was an installation such as thiswhich succeeded in repeatedly transmitting thefamous Morse code signal for the letter “S”(dot-dot-dot) from Poldhu in Cornwall, Eng-land, to St. John’s in Newfoundland, Canada, inDecember 1901.

This momentous event gave new impetus to theresearchers. The simple spark gap needed to beimproved. After arcing, it did not lose its con-ductivity immediately, because of ionizationeffects, and this prevented the capacitor fromrecharging quickly, resulting in a greater energyconsumption and the generation of fewer wave-trains. The simple spark gap was duly replacedby the quenched spark gap and, afterwards, bythe rotary discharger (Marconi, 1905) in orderto cause the electrode to quench and the sparksto be interrupted more quickly.

However, there was still the problem of dampedwaves and low efficiency, which allowed only thetransmission of telegraphy signals; intermittentoperation was not adaptable to telephony modu-lation.

5.2. Voltaic-arc generators

There was a need to go back to basic theory. In1850, Sir W. Thompson had defined the relation-ship between the resistance, inductance and capac-itance that was needed to provide an oscillatorydischarge from a circuit. Based on those ideas,generators of undamped waves were eventuallydeveloped, along with voltaic-arc generatorswhich used carbon electrodes.

The operating principle of the voltaic-arcgenerator (Fig. 8) is based upon the negativeresistance of the arc. A series capacitor and induc-tor, connected in parallel with an arc, are chargedby a DC power supply. This causes the voltageacross the arc and the LC circuit to increase. How-ever, the capacitor is charged to a greater voltagethan that of the arc, due to the effect of the inductor.When the capacitor starts to discharge, the currentacross the arc increases, thus causing the voltage todecrease. When the voltage of the capacitorequals that of the arc, the inductor maintains thedischarge current, causing the voltage across thecapacitor to become lower than that of the arc. Atthe end of the discharge cycle, the capacitor ischarged again and the process is repeated.

ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ

Teslatransformer

Powermeter

Figure 7Circuit diagram of aMarconi transmitteraround the time of thePoldhu - St. John’strials (1901).

Poulsen then experimented with arcs dipped into alarge tank of hydrogen. This type of generatorproduced quasi-continuous waves which could beamplitude modulated for telephony applications.The system offered the great advantage of simpli-city and economy, although it generated har-monics and consequently, RF interference. Inspite of its drawbacks, many researchers deemedthat the Poulsen arc was the successor of the spark-gap system and devoted themselves to improve iteven further. Prof. Majorana – who experimentedwith multiple-spark generators for radiotelephonyapplications in 1903-1904 – used the Poulsen arcto transmit telephony signals over a range of about400 km (the maximum distance that could bereached by radiotelephony at the time).

5.3. Static frequency multipliers

Special alternators were developed to generatecontinuous RF waves, but they had manymechanical problems due to the very high fre-quencies that had to be reached. One method ofovercoming such problems was to use a “staticfrequency multiplier” – a magnetic-core device,similar to a peaking transformer, which providedharmonics by distorting a sinewave. The multi-plier was supplied directly by the alternator andoperated as an oscillation exciter in a circuitwhich was tuned to the desired frequency andcoupled to the antenna.

This type of system was limited to longwaveapplications. The cost of the installation wasprohibitively high because, in addition to thealternators, several 200 m high towers wererequired to support the horizontal longwaveradiators. This prompted the researchers to de-velop a simpler, more economic, generator of RFoscillations.

6. Thermionic valvegenerators

It was around this time that the first thermionic-valve generators appeared. A lot of time wasdevoted to their development, before the advan-tages of their application could be enjoyed. At thatpoint, there were many clashes between theproponents of valve generators and the supportersof systems which used RF alternators.

In 1915, considerable progress was obtained by thestations at Arlington and Honolulu. An oscillatoryvalve was coupled to the grid circuit of a valve ofgreater power. The plate current of the latter drovein turn the grid circuits of a further 10 valves in

CIsolatingchoke

Isolatingchoke

Carbonrods

Oscillatorycircuit

DC powersupply

parallel. An array of a further 300 valves with theiranodes in parallel was driven by the above circuitsand supplied its output energy to the high antennaof the Arlington station.

Marconi had already experimented with valvetransmitters for the navy in 1914. With his ownexperience and with that of Arlington and Hono-lulu, he became convinced that a transmitter whichused alternators, a voltaic-arc generator or rotarydischarger could not provide a commercial servicemore convenient than the wired service. There-fore, he decided to develop systems that wouldrequire new valve designs which were capable ofdelivering power levels of greater than 1500 W.

In order to achieve the required power levels,Marconi connected 48 valves in parallel andincreased the anode voltage to 10 kV. The resul-tant current in the aerial was then 330 A. A newproblem to be overcome was that of keeping thefrequency constant.

In 1921, Franklin devised the “Master Oscilla-tor” consisting of an oscillating valve circuitwhich generated signals of constant frequency,independent of temperature changes. It was con-nected to a number of amplifier stages in order toreach the desired power for supplying the aerial.Thereafter, as it was easier to stabilize an oscilla-tor that operated at a low frequency, the “driveoscillator” was arranged to generate a frequencywhich was a submultiple of the frequency of thetransmitter. The drive oscillator was followedby several frequency doublers and triplers whichalso operated as amplifiers. Very stable oscilla-tors with “quartz crystals” (based upon piezo-electrical phenomena) were later developed: asuitably-cut quartz plate is a resonator of excel-lent quality, especially if it is kept at a constanttemperature.

In addition to the problem of producing high-power radiofrequency energy, solved by the use ofvalves, a new problem arose: which frequency wasthe most suitable for a given transmission?

Figure 8Circuit diagram of a

voltaic-arc generator.

Poulsen then experimented with arcs dipped into alarge tank of hydrogen. This type of generatorproduced quasi-continuous waves which could beamplitude modulated for telephony applications.The system offered the great advantage of simpli-city and economy, although it generated har-monics and consequently, RF interference. Inspite of its drawbacks, many researchers deemedthat the Poulsen arc was the successor of the spark-gap system and devoted themselves to improve iteven further. Prof. Majorana – who experimentedwith multiple-spark generators for radiotelephonyapplications in 1903-1904 – used the Poulsen arcto transmit telephony signals over a range of about400 km (the maximum distance that could bereached by radiotelephony at the time).

5.3. Static frequency multipliers

Special alternators were developed to generatecontinuous RF waves, but they had manymechanical problems due to the very high fre-quencies that had to be reached. One method ofovercoming such problems was to use a “staticfrequency multiplier” – a magnetic-core device,similar to a peaking transformer, which providedharmonics by distorting a sinewave. The multi-plier was supplied directly by the alternator andoperated as an oscillation exciter in a circuitwhich was tuned to the desired frequency andcoupled to the antenna.

This type of system was limited to longwaveapplications. The cost of the installation wasprohibitively high because, in addition to thealternators, several 200 m high towers wererequired to support the horizontal longwaveradiators. This prompted the researchers to de-velop a simpler, more economic, generator of RFoscillations.

6. Thermionic valvegenerators

It was around this time that the first thermionic-valve generators appeared. A lot of time wasdevoted to their development, before the advan-tages of their application could be enjoyed. At thatpoint, there were many clashes between theproponents of valve generators and the supportersof systems which used RF alternators.

In 1915, considerable progress was obtained by thestations at Arlington and Honolulu. An oscillatoryvalve was coupled to the grid circuit of a valve ofgreater power. The plate current of the latter drovein turn the grid circuits of a further 10 valves in

CIsolatingchoke

Isolatingchoke

Carbonrods

Oscillatorycircuit

DC powersupply

parallel. An array of a further 300 valves with theiranodes in parallel was driven by the above circuitsand supplied its output energy to the high antennaof the Arlington station.

Marconi had already experimented with valvetransmitters for the navy in 1914. With his ownexperience and with that of Arlington and Hono-lulu, he became convinced that a transmitter whichused alternators, a voltaic-arc generator or rotarydischarger could not provide a commercial servicemore convenient than the wired service. There-fore, he decided to develop systems that wouldrequire new valve designs which were capable ofdelivering power levels of greater than 1500 W.

In order to achieve the required power levels,Marconi connected 48 valves in parallel andincreased the anode voltage to 10 kV. The resul-tant current in the aerial was then 330 A. A newproblem to be overcome was that of keeping thefrequency constant.

In 1921, Franklin devised the “Master Oscilla-tor” consisting of an oscillating valve circuitwhich generated signals of constant frequency,independent of temperature changes. It was con-nected to a number of amplifier stages in order toreach the desired power for supplying the aerial.Thereafter, as it was easier to stabilize an oscilla-tor that operated at a low frequency, the “driveoscillator” was arranged to generate a frequencywhich was a submultiple of the frequency of thetransmitter. The drive oscillator was followedby several frequency doublers and triplers whichalso operated as amplifiers. Very stable oscilla-tors with “quartz crystals” (based upon piezo-electrical phenomena) were later developed: asuitably-cut quartz plate is a resonator of excel-lent quality, especially if it is kept at a constanttemperature.

In addition to the problem of producing high-power radiofrequency energy, solved by the use ofvalves, a new problem arose: which frequency wasthe most suitable for a given transmission?

Figure 8Circuit diagram of a

voltaic-arc generator.

The first works of Marconi were based upon theuse of medium and long waves. Apparatus of highpower and large antennas supported by hightowers were needed, which required considerableinvestment and resulted in a very expensive radioservice. The theory of Loewenstein (concerningthe composition of the atmosphere and theinfluence of its layers upon the propagation of theelectrical waves), along with experiences made inthe short-wave field, prompted Marconi to confideto one collaborator: “Only the short waves will beable to save the radio from the blind alley of themedium and long waves”.

In 1923, Marconi carried out important experi-ments with Franklin between the ship “Elettra” atanchor by St. Vincent in the Cape Verde Islands,(off the coast of West Africa) and the station atPoldhu in Cornwall, England, where a short-wavetransmitter (25 – 90 metres) had also beeninstalled. At a distance of 4130 km the received

signals were strong and clear, even if the trans-mission power was reduced to only 1 kW. Undersuch conditions, the signals were stronger thanthose transmitted from Caernarvon (Wales) andNauen (Germany) on a wavelength severalhundred times greater, and with a power of200 kW.

Thereafter, studies were made to improve themodulation percentage (i.e. modulation index)and to reduce the distortion. One such methodwas grid modulation which was easier to producebut which had low fidelity. Heising introducedanode modulation where the signal, after suitableamplification, was superimposed on the anodevoltage of the final radiofrequency valve. In con-trast to the other methods, a modulated wave witha high modulation percentage and a low distortionwas obtained.

Finally they had to solve the problem of how tocool the valves which operated at power levels ofabove 2500 W. Around 1930, the anode wascooled by oil circulation. Later, after the solutionof anode-insulation problems, distilled watercirculation was used to keep the anode cool. Thus,the high power required in the final stages of thetransmitter, especially for long waves, could nowbe obtained by a lesser number of valves operatingin parallel. After a lot of improvements, thetetrodes of today are capable of delivering severalhundreds of kilowatts.

7. Modern transmittingdevices

Increases in the power needed for AM trans-mitters, accompanied by increases in the cost ofenergy supply, has prompted the development ofhigh-efficiency RF power amplifiers and modu-lation systems (Pulse Duration Modulation,Pulse Step Modulation, etc). More recently,completely-digital AM transmitters with veryhigh efficiency have been developed.

Over a few decades, the increased application ofradio-telecommunication systems, and the grow-ing need for more radiofrequency spectrum toaccommodate these systems, has led to the use ofever higher frequencies: hundreds of MHz, tens ofGHz ... up to the frequencies of light. New devices,based on the control of the electron beam current,have taken the place of the old electronic tubes.

The maximum operating frequency of triodes,tetrodes and pentodes is limited by the reactancesand resistances associated with the electrodes.These effects, for instance, reduce the input and

Figure 9A high-voltagemercury-arc rectifierfrom 1933.

The first works of Marconi were based upon theuse of medium and long waves. Apparatus of highpower and large antennas supported by hightowers were needed, which required considerableinvestment and resulted in a very expensive radioservice. The theory of Loewenstein (concerningthe composition of the atmosphere and theinfluence of its layers upon the propagation of theelectrical waves), along with experiences made inthe short-wave field, prompted Marconi to confideto one collaborator: “Only the short waves will beable to save the radio from the blind alley of themedium and long waves”.

In 1923, Marconi carried out important experi-ments with Franklin between the ship “Elettra” atanchor by St. Vincent in the Cape Verde Islands,(off the coast of West Africa) and the station atPoldhu in Cornwall, England, where a short-wavetransmitter (25 – 90 metres) had also beeninstalled. At a distance of 4130 km the received

signals were strong and clear, even if the trans-mission power was reduced to only 1 kW. Undersuch conditions, the signals were stronger thanthose transmitted from Caernarvon (Wales) andNauen (Germany) on a wavelength severalhundred times greater, and with a power of200 kW.

Thereafter, studies were made to improve themodulation percentage (i.e. modulation index)and to reduce the distortion. One such methodwas grid modulation which was easier to producebut which had low fidelity. Heising introducedanode modulation where the signal, after suitableamplification, was superimposed on the anodevoltage of the final radiofrequency valve. In con-trast to the other methods, a modulated wave witha high modulation percentage and a low distortionwas obtained.

Finally they had to solve the problem of how tocool the valves which operated at power levels ofabove 2500 W. Around 1930, the anode wascooled by oil circulation. Later, after the solutionof anode-insulation problems, distilled watercirculation was used to keep the anode cool. Thus,the high power required in the final stages of thetransmitter, especially for long waves, could nowbe obtained by a lesser number of valves operatingin parallel. After a lot of improvements, thetetrodes of today are capable of delivering severalhundreds of kilowatts.

7. Modern transmittingdevices

Increases in the power needed for AM trans-mitters, accompanied by increases in the cost ofenergy supply, has prompted the development ofhigh-efficiency RF power amplifiers and modu-lation systems (Pulse Duration Modulation,Pulse Step Modulation, etc). More recently,completely-digital AM transmitters with veryhigh efficiency have been developed.

Over a few decades, the increased application ofradio-telecommunication systems, and the grow-ing need for more radiofrequency spectrum toaccommodate these systems, has led to the use ofever higher frequencies: hundreds of MHz, tens ofGHz ... up to the frequencies of light. New devices,based on the control of the electron beam current,have taken the place of the old electronic tubes.

The maximum operating frequency of triodes,tetrodes and pentodes is limited by the reactancesand resistances associated with the electrodes.These effects, for instance, reduce the input and

Figure 9A high-voltagemercury-arc rectifierfrom 1933.

output impedances, and the efficiency as well.Additionally, when the electron transit time iscomparable with a quarter of the period of themicrowave, the total efficiency of the valve, whenoperating either as an oscillator or an amplifier, isreduced to less than half. As a result, it is difficultto produce triode or tetrode amplifiers for useabove 1 GHz which have an output power ofseveral kilowatts and an acceptable efficiency forcontinuous operation.

7.1. Electron-velocity controltubes

Tubes which operate by modulating the velocity ofan electron beam have been developed for micro-wave use. Such tubes are mainly of two types:

– klystrons and travelling wave tubes, where theelectron beam flows linearly;

– magnetrons, where the electron beam follows acurved path under the action of orthogonal elec-tric and magnetic fields.

Klystrons and travelling wave tubes find a wideapplication in broadcasting and microwave tele-communication systems. Both types of device usean axial magnetic field to “collimate” the electronbeam, i.e. to prevent the beam from spreading outas a result of small initial divergent angles, and therepulsion effect of the negative charges whichmake up the beam. The speed of each movingelectron is controlled by an electric field and iscalculated from the following equation for kineticenergy:

Ve� 12

where: V = the accelerating voltage acting onthe electron

e = the charge on the electron(–1.6 x 10–19 C)

v = the velocity of the electronm = the mass of the electron

(9.1 x 10–31 kg).

From this:

v � 2Vem 107� (cm�sec)

And hence:

v � 5.93� 107 V� (cm�sec).

7.1.1. Klystron

The Klystron was invented in 1937 and basicallyoperates as follows (Fig. 10). Electrons are firedfrom an electron gun which contains a cathodeheated by a filament, and a voltage source whichproduces acceleration of the beam. The micro-wave input signal at the first cavity creates volt-ages across the gap traversed by the beam, causingthe electrons in the beam to accelerate or deceler-ate. As it travels towards the collector, the electronbeam is thus “velocity modulated” by the inputsignal; groups of electrons bunch together to form“electron packets”. When these electron packetsarrive at the output cavity, they generate a micro-wave resonance at an increased power level rela-tive to the input signal, due to the energy suppliedby the electron beam.

Additional “intermediate” cavities are providedon some klystrons to improve the bunchingprocess, resulting in a higher gain and improvedefficiency. By varying the size of a cavity, it ispossible to adjust its resonant frequency. Thus ina multi-cavity device, if all the cavities are tunedto the same frequency, the gain is maximized butthe bandwidth is reduced. On the other hand, ifthe cavities are tuned to slightly differentfrequencies, the gain is reduced but the bandwidthis increased.

Klystrons with two or more cavities are used asnarrow-band linear power amplifiers. They cangenerate about 10 kW under continuous operatingconditions, at a frequency of several GHz. Themaximum gain achieved by a klystron is some tensof decibels and its efficiency ranges from about30 % to 50 %.

Electron gun

Input cavity Output cavity

Collector

Waveguidecoupling

Coaxial linecoupling

Inputsignal

Figure 10Schematic diagram of

a two-cavity klystronamplifier.

output impedances, and the efficiency as well.Additionally, when the electron transit time iscomparable with a quarter of the period of themicrowave, the total efficiency of the valve, whenoperating either as an oscillator or an amplifier, isreduced to less than half. As a result, it is difficultto produce triode or tetrode amplifiers for useabove 1 GHz which have an output power ofseveral kilowatts and an acceptable efficiency forcontinuous operation.

7.1. Electron-velocity controltubes

Tubes which operate by modulating the velocity ofan electron beam have been developed for micro-wave use. Such tubes are mainly of two types:

– klystrons and travelling wave tubes, where theelectron beam flows linearly;

– magnetrons, where the electron beam follows acurved path under the action of orthogonal elec-tric and magnetic fields.

Klystrons and travelling wave tubes find a wideapplication in broadcasting and microwave tele-communication systems. Both types of device usean axial magnetic field to “collimate” the electronbeam, i.e. to prevent the beam from spreading outas a result of small initial divergent angles, and therepulsion effect of the negative charges whichmake up the beam. The speed of each movingelectron is controlled by an electric field and iscalculated from the following equation for kineticenergy:

Ve� 12

where: V = the accelerating voltage acting onthe electron

e = the charge on the electron(–1.6 x 10–19 C)

v = the velocity of the electronm = the mass of the electron

(9.1 x 10–31 kg).

From this:

v � 2Vem 107� (cm�sec)

And hence:

v � 5.93� 107 V� (cm�sec).

7.1.1. Klystron

The Klystron was invented in 1937 and basicallyoperates as follows (Fig. 10). Electrons are firedfrom an electron gun which contains a cathodeheated by a filament, and a voltage source whichproduces acceleration of the beam. The micro-wave input signal at the first cavity creates volt-ages across the gap traversed by the beam, causingthe electrons in the beam to accelerate or deceler-ate. As it travels towards the collector, the electronbeam is thus “velocity modulated” by the inputsignal; groups of electrons bunch together to form“electron packets”. When these electron packetsarrive at the output cavity, they generate a micro-wave resonance at an increased power level rela-tive to the input signal, due to the energy suppliedby the electron beam.

Additional “intermediate” cavities are providedon some klystrons to improve the bunchingprocess, resulting in a higher gain and improvedefficiency. By varying the size of a cavity, it ispossible to adjust its resonant frequency. Thus ina multi-cavity device, if all the cavities are tunedto the same frequency, the gain is maximized butthe bandwidth is reduced. On the other hand, ifthe cavities are tuned to slightly differentfrequencies, the gain is reduced but the bandwidthis increased.

Klystrons with two or more cavities are used asnarrow-band linear power amplifiers. They cangenerate about 10 kW under continuous operatingconditions, at a frequency of several GHz. Themaximum gain achieved by a klystron is some tensof decibels and its efficiency ranges from about30 % to 50 %.

Electron gun

Input cavity Output cavity

Collector

Waveguidecoupling

Coaxial linecoupling

Inputsignal

Figure 10Schematic diagram of

a two-cavity klystronamplifier.

Another version of the device, known as a reflexklystron, is used as an RF oscillator (Fig. 11). Inthis design, the output cavity has been eliminatedand an electrode (repeller), at a suitable negativepotential, is used to repel electron packets back tothe two grids of the sole cavity.

7.1.2. Travelling wave tubes

In a travelling wave tube (TWT) (Fig. 12), theelectrons generated by the heated cathode travelalong the axis of the tube, constrained by focussingcoils, until they reach the collector. Spaced closelyaround the electron beam is a helix which iscapable of propagating a slow-moving electro-magnetic wave.

The RF signal enters the tube via the input wave-guide and travels along the helix wire at velocities

close to the speed of light. However, along the axisof the helix, the velocity of propagation (the so-called phase velocity) is much lower, typically 0.1to 0.3 times the speed of light; it is equal to theproduct of the velocity of light and the ratiobetween the pitch and circumference of the helix.

The electrons generated by the TWT are acceler-ated by a high voltage such that, by the time theyreach the helix, their velocity is similar to that ofthe axial phase velocity of the RF input signalalong the helix. When the RF signal enters thehelix, the longitudinal part of its field interactswith the electron beam, causing some electrons toaccelerate and others to decelerate such thatelectron packets are formed in a similar manner tothat of the klystron. The result is a progressiverearrangement in phase of the electrons in relationto the RF wave. In turn, the modulated electronbeam induces additional waves on the helix andthis process of mutual interaction continues alongthe length of the tube. The outcome is that the DCenergy of the electron beam is transformed intoRF energy in the helix, and the RF wave isamplified.

In order to avoid self-oscillations, the amplifiedRF wave should not be reflected back to the inputof the helix; any reflected portion of the signalmust be attenuated along the helix. To achieve this,the wire of the helix is made of, or coated with, amagnetic material. As a result, the potential gainof a TWT is reduced but a good compromise be-tween gain and stability can usually be reached.

Unlike the klystron, a travelling wave tube has nospecific resonating elements. It can be fabricated

Electrongun

Cavity Interactiongap

Repeller

Focusing coilsHeater Cathode

Gun anode

Collector

Inputwaveguide

Outputwaveguide

Figure 12Schematic diagramof a travelling wavetube.

Figure 11Schematic diagram ofa reflex klystronamplifier.

Another version of the device, known as a reflexklystron, is used as an RF oscillator (Fig. 11). Inthis design, the output cavity has been eliminatedand an electrode (repeller), at a suitable negativepotential, is used to repel electron packets back tothe two grids of the sole cavity.

7.1.2. Travelling wave tubes

In a travelling wave tube (TWT) (Fig. 12), theelectrons generated by the heated cathode travelalong the axis of the tube, constrained by focussingcoils, until they reach the collector. Spaced closelyaround the electron beam is a helix which iscapable of propagating a slow-moving electro-magnetic wave.

The RF signal enters the tube via the input wave-guide and travels along the helix wire at velocities

close to the speed of light. However, along the axisof the helix, the velocity of propagation (the so-called phase velocity) is much lower, typically 0.1to 0.3 times the speed of light; it is equal to theproduct of the velocity of light and the ratiobetween the pitch and circumference of the helix.

The electrons generated by the TWT are acceler-ated by a high voltage such that, by the time theyreach the helix, their velocity is similar to that ofthe axial phase velocity of the RF input signalalong the helix. When the RF signal enters thehelix, the longitudinal part of its field interactswith the electron beam, causing some electrons toaccelerate and others to decelerate such thatelectron packets are formed in a similar manner tothat of the klystron. The result is a progressiverearrangement in phase of the electrons in relationto the RF wave. In turn, the modulated electronbeam induces additional waves on the helix andthis process of mutual interaction continues alongthe length of the tube. The outcome is that the DCenergy of the electron beam is transformed intoRF energy in the helix, and the RF wave isamplified.

In order to avoid self-oscillations, the amplifiedRF wave should not be reflected back to the inputof the helix; any reflected portion of the signalmust be attenuated along the helix. To achieve this,the wire of the helix is made of, or coated with, amagnetic material. As a result, the potential gainof a TWT is reduced but a good compromise be-tween gain and stability can usually be reached.

Unlike the klystron, a travelling wave tube has nospecific resonating elements. It can be fabricated

Electrongun

Cavity Interactiongap

Repeller

Focusing coilsHeater Cathode

Gun anode

Collector

Inputwaveguide

Outputwaveguide

Figure 12Schematic diagramof a travelling wavetube.

Figure 11Schematic diagram ofa reflex klystronamplifier.

with a pass-band of 600 – 800 MHz and can yielda gain of some tens of decibels. The maximumpower levels produced by a TWT are of the orderof several hundred watts at 10 GHz, and its effi-ciency is around 30 %.

8. Transistors and othersemiconductors

Although we can refer to the cat’s whisker crystalas the earliest ancestor of modern semiconductordevices, reference should also be made to thesilicon point-contact diode, which was developedduring the Second World War for use in microwaveradar receivers, and the germanium junction diodewhich was developed shortly after.

However, the new electronic age began in 1948when W. Shockley, J. Berdeen and W.H. Brattainof Bell Telephone Laboratories announced thefirst transistor (a name abridged from “transferresistor”). This important electronic device hada strong impact on the field of electronics, in bothengineering and economic terms. Transistorsand, in general, semiconductors are based uponthe controlled presence of imperfections inotherwise nearly-perfect crystals. The materialsgenerally used are germanium and silicon.

8.1. Semiconductor physics

The germanium or silicon atom may be consideredto consist of four valence electrons surrounding anucleus which has a positive charge of fourelectron units. The atom as a whole is thus neutral.This symmetry is broken if we introduce somechemical impurities, the so-called donors oracceptors, into silicon and germanium crystals.The presence of these classes of impurities causesan excess of electrons or holes, thus making thecrystal a conductor or rather a semiconductor.Typical donor elements for silicon and germaniumare antimony and arsenic which each have fivevalence electrons. Consequently, one electron forevery atom of impurity becomes an excess elec-tron and wanders throughout the crystal (in then-type semiconductor).

Aluminum and gallium are typical acceptors forsilicon and germanium. These elements have threevalence electrons rather than four. As a result, theycannot complete the paired electron structure ofthe four bonds surrounding them, when theyreplace an atom in the crystal lattice. If an electronfrom a bond somewhere else is used to fill in thishole, the acceptor atom will acquire a localizednegative charge and a new hole is created in the

atom which had given up its spare electron. Thecrystal becomes a p-type semiconductor because apositive particle wanders throughout the crystallattice.

It is evident that in a single crystal there may beboth p-type and n-type regions. The boundarybetween such regions is called a p-n junction. Suchp-n junctions have interesting electrical andoptical properties. Due to the different charge onthe particles which wander throughout the tworegions (electrons and holes), an electric field isestablished across the junction and no current willflow across the junction under conditions ofthermal equilibrium.

The application of a negative voltage to thep-region and a positive voltage to the n-region(reverse bias) will increase the electrostatic poten-tial difference between the two regions, so that nocurrent due to the majority carriers flows into thedevice. If the reverse voltage increases to thebreakdown value (Zener voltage), the currentflowing into the device will rapidly increase to thepoint of reverse saturation. This is due to theminority carriers which are present in the crystal asa result of breaking the covalent bonds. Whenforward bias is applied, the current increases expo-nentially to the point of forward saturation.

8.2. Junction transistor

The junction transistor consists of a single crystalof germanium or silicon divided into three regions:these regions are alternately n-type, p-type andn-type and are called the emitter, base andcollector respectively. When an n-p-n transistor isused as an amplifier, the positive bias is applied tothe collector region in order to bias the p-n junctionbetween the collector and the base regions in thereverse direction. If a positive signal is appliedbetween the emitter and the base, a large numberof electrons will be able to diffuse into the base.

Because of the thickness of the base region, wecan consider that all the electron current enteringthe base region from the emitter arrives at thecollector. Only a negligible fraction of the emittercurrent flows in the base circuit. All changes inthe emitter current appear in the collector currentand, thus, the junction transistor behaves muchlike an ideal triode but requires no cathode powerand offers the hope of practically unlimited life.

8.3. Other transistor types

In addition to the junction or bipolar transistor(so called because the conduction involveselectrical charges of both signs), unipolar tran-

with a pass-band of 600 – 800 MHz and can yielda gain of some tens of decibels. The maximumpower levels produced by a TWT are of the orderof several hundred watts at 10 GHz, and its effi-ciency is around 30 %.

8. Transistors and othersemiconductors

Although we can refer to the cat’s whisker crystalas the earliest ancestor of modern semiconductordevices, reference should also be made to thesilicon point-contact diode, which was developedduring the Second World War for use in microwaveradar receivers, and the germanium junction diodewhich was developed shortly after.

However, the new electronic age began in 1948when W. Shockley, J. Berdeen and W.H. Brattainof Bell Telephone Laboratories announced thefirst transistor (a name abridged from “transferresistor”). This important electronic device hada strong impact on the field of electronics, in bothengineering and economic terms. Transistorsand, in general, semiconductors are based uponthe controlled presence of imperfections inotherwise nearly-perfect crystals. The materialsgenerally used are germanium and silicon.

8.1. Semiconductor physics

The germanium or silicon atom may be consideredto consist of four valence electrons surrounding anucleus which has a positive charge of fourelectron units. The atom as a whole is thus neutral.This symmetry is broken if we introduce somechemical impurities, the so-called donors oracceptors, into silicon and germanium crystals.The presence of these classes of impurities causesan excess of electrons or holes, thus making thecrystal a conductor or rather a semiconductor.Typical donor elements for silicon and germaniumare antimony and arsenic which each have fivevalence electrons. Consequently, one electron forevery atom of impurity becomes an excess elec-tron and wanders throughout the crystal (in then-type semiconductor).

Aluminum and gallium are typical acceptors forsilicon and germanium. These elements have threevalence electrons rather than four. As a result, theycannot complete the paired electron structure ofthe four bonds surrounding them, when theyreplace an atom in the crystal lattice. If an electronfrom a bond somewhere else is used to fill in thishole, the acceptor atom will acquire a localizednegative charge and a new hole is created in the

atom which had given up its spare electron. Thecrystal becomes a p-type semiconductor because apositive particle wanders throughout the crystallattice.

It is evident that in a single crystal there may beboth p-type and n-type regions. The boundarybetween such regions is called a p-n junction. Suchp-n junctions have interesting electrical andoptical properties. Due to the different charge onthe particles which wander throughout the tworegions (electrons and holes), an electric field isestablished across the junction and no current willflow across the junction under conditions ofthermal equilibrium.

The application of a negative voltage to thep-region and a positive voltage to the n-region(reverse bias) will increase the electrostatic poten-tial difference between the two regions, so that nocurrent due to the majority carriers flows into thedevice. If the reverse voltage increases to thebreakdown value (Zener voltage), the currentflowing into the device will rapidly increase to thepoint of reverse saturation. This is due to theminority carriers which are present in the crystal asa result of breaking the covalent bonds. Whenforward bias is applied, the current increases expo-nentially to the point of forward saturation.

8.2. Junction transistor

The junction transistor consists of a single crystalof germanium or silicon divided into three regions:these regions are alternately n-type, p-type andn-type and are called the emitter, base andcollector respectively. When an n-p-n transistor isused as an amplifier, the positive bias is applied tothe collector region in order to bias the p-n junctionbetween the collector and the base regions in thereverse direction. If a positive signal is appliedbetween the emitter and the base, a large numberof electrons will be able to diffuse into the base.

Because of the thickness of the base region, wecan consider that all the electron current enteringthe base region from the emitter arrives at thecollector. Only a negligible fraction of the emittercurrent flows in the base circuit. All changes inthe emitter current appear in the collector currentand, thus, the junction transistor behaves muchlike an ideal triode but requires no cathode powerand offers the hope of practically unlimited life.

8.3. Other transistor types

In addition to the junction or bipolar transistor(so called because the conduction involveselectrical charges of both signs), unipolar tran-

sourcegate

Psubstrate

sourcegate

Nsubstrate

Typical N-MOS Typical P-MOS

sistors have also been developed. In thesedevices, conduction is due to electrical chargesof only one sign; the junction field effect tran-sistor (JFET) and metal oxide semiconductorfield effect transistor (MOSFET) are perhapsthe best-known examples.

JFETs are active devices with three terminals(gate, source and drain) which come in twoversions: p-channel and n-channel. Both versionscan operate in either a linear or a non-linear mode.Between the channel and the gate there is a p-njunction, the polarization of which controls thecurrent flowing between the source and the drain.In order to operate in the linear mode, the gate-channel junction should be reverse-biased. Theoutput and transfer characteristics of a JFET arevery similar to those of a triode.

The MOSFET transistor (Fig. 13) consists of ap-type or an n-type substrate. The gate is insulatedfrom the substrate by silicon oxide, while thesource and drain are reverse-doped regions withrespect to the substrate. The source and substrateare innerly connected so that they are at the samepotential. The channel is formed by the freecharges in the source region which are attracted tothe gate by its polarization voltage.

FET transistors are widely deployed in the inte-grated amplifiers used for microwave applica-tions up to frequencies of around 14 GHz. The so-called monolithic microwave integrated circuit(MMIC) is fabricated by using gallium arsenidetechnology.

For higher frequency applications, parametricamplifiers with negative resistance are used. Atypical negative-resistance element is the galliumarsenide varactor.

Further important semiconductor devices are theluminescent diodes (LED: light emitting diode)and the opto-electronic components (photodiode,phototransistor, etc.) which have been developedespecially for use with optical fibres.

9. Microchips

By the end of the 1940s, as electronic systems grewin importance and size, more power was con-sumed, their weight and costs increased while theirreliability decreased. The cold war and the spacerace both demanded cheaper, smaller and morereliable electronic systems. Following on from theinvention of the transistor in 1948, J.S. Kilbycompleted the first integrated circuit in 1958; itconsisted of a phase oscillator formed entirely ofsemiconductors (transistors, resistors and capaci-tors) in a single block called a microchip.

The use of large computers to design ICs, to-gether with rapid developments in the chemicaland photo-optical technologies applied to semi-conductor electronics, has led to:

– greater purity of the silicon wafer used as thesubstrate;

– more accurate control of the doping of thedifferent substrate regions which provide thebasic components of the electronic circuit(transistors, diodes, resistors, capacitors, etc.);

– the highly-accurate fabrication of microscopicmaterial layers (junctions, connecting tracks,connection soldering);

– continuous improvements in the performanceof chips due the use of new materials, such asgallium arsenide and new manufacturingtechnologies (e.g. planar process, MOStechnology);

– a higher integration capability to reduce the sizeof the circuits so that a small chip can includea greater number of functions and components.

Nowadays, integrated circuits are found in everyfield of electronics; they have entirely replacedcircuits built from discrete components.

10. Digital signal processing

The processing of an electrical signal essentiallyinvolves:

– signal processing both in the time and fre-quency domains;

Figure 13The basic structureof a MOSFET.

sourcegate

Psubstrate

sourcegate

Nsubstrate

Typical N-MOS Typical P-MOS

sistors have also been developed. In thesedevices, conduction is due to electrical chargesof only one sign; the junction field effect tran-sistor (JFET) and metal oxide semiconductorfield effect transistor (MOSFET) are perhapsthe best-known examples.

JFETs are active devices with three terminals(gate, source and drain) which come in twoversions: p-channel and n-channel. Both versionscan operate in either a linear or a non-linear mode.Between the channel and the gate there is a p-njunction, the polarization of which controls thecurrent flowing between the source and the drain.In order to operate in the linear mode, the gate-channel junction should be reverse-biased. Theoutput and transfer characteristics of a JFET arevery similar to those of a triode.

The MOSFET transistor (Fig. 13) consists of ap-type or an n-type substrate. The gate is insulatedfrom the substrate by silicon oxide, while thesource and drain are reverse-doped regions withrespect to the substrate. The source and substrateare innerly connected so that they are at the samepotential. The channel is formed by the freecharges in the source region which are attracted tothe gate by its polarization voltage.

FET transistors are widely deployed in the inte-grated amplifiers used for microwave applica-tions up to frequencies of around 14 GHz. The so-called monolithic microwave integrated circuit(MMIC) is fabricated by using gallium arsenidetechnology.

For higher frequency applications, parametricamplifiers with negative resistance are used. Atypical negative-resistance element is the galliumarsenide varactor.

Further important semiconductor devices are theluminescent diodes (LED: light emitting diode)and the opto-electronic components (photodiode,phototransistor, etc.) which have been developedespecially for use with optical fibres.

9. Microchips

By the end of the 1940s, as electronic systems grewin importance and size, more power was con-sumed, their weight and costs increased while theirreliability decreased. The cold war and the spacerace both demanded cheaper, smaller and morereliable electronic systems. Following on from theinvention of the transistor in 1948, J.S. Kilbycompleted the first integrated circuit in 1958; itconsisted of a phase oscillator formed entirely ofsemiconductors (transistors, resistors and capaci-tors) in a single block called a microchip.

The use of large computers to design ICs, to-gether with rapid developments in the chemicaland photo-optical technologies applied to semi-conductor electronics, has led to:

– greater purity of the silicon wafer used as thesubstrate;

– more accurate control of the doping of thedifferent substrate regions which provide thebasic components of the electronic circuit(transistors, diodes, resistors, capacitors, etc.);

– the highly-accurate fabrication of microscopicmaterial layers (junctions, connecting tracks,connection soldering);

– continuous improvements in the performanceof chips due the use of new materials, such asgallium arsenide and new manufacturingtechnologies (e.g. planar process, MOStechnology);

– a higher integration capability to reduce the sizeof the circuits so that a small chip can includea greater number of functions and components.

Nowadays, integrated circuits are found in everyfield of electronics; they have entirely replacedcircuits built from discrete components.

10. Digital signal processing

The processing of an electrical signal essentiallyinvolves:

– signal processing both in the time and fre-quency domains;

Figure 13The basic structureof a MOSFET.

– deleting unwanted information (e.g. filteringout noise);

– extracting specific information from a signal;

– deleting redundant and/or irrelevant infor-mation;

– adding extra information (e.g. encryption);

– storing, manipulating and transmitting signalswith the best fidelity and reliability.

The processing of analogue signals (amplification,modulation, etc.) is very difficult and is restrictedby real limiting factors that can sometimes be im-proved upon, but never removed entirely:

– the circuits need to be very linear with lownoise;

– a special design is needed for almost everyapplication;

– the circuits depend on sensitive components(resistors, capacitors) which need adjustment;

– circuits tend to drift in both the short-term andthe long-term;

– the signals are difficult to record and store;

– the signals are not easy to compress.

Digital signals, on the other hand, use devices (suchas amplifiers) which have only two operating states:saturation (on) and cutoff (off). It is thus possibleto obtain better use, reliability and efficiency fromeach device in the chain. Futhermore, the on-off

signals can be regenerated many times throughoutthe chain, with no losses whatsoever.

In 1938, Alec H. Reeves patented an inventionconcerning the transmission of voice signals byusing coded pulses of constant amplitude, similarto those used in telegraphy. This invention wascalled pulse code modulation (PCM). However, itwas not until 1962 that such an invention found anapplication in the field of telephony, with the pro-duction of the first 24-channel PCM system byBell Systems of the USA.

In a PCM system, the conversion of the analoguesignal into a digital signal is carried out throughthree different steps: sampling, quantizing andcoding.

In the sampling step, the analogue input signal istransformed into a timed discrete signal, by pulseamplitude modulation (PAM) of a carrier formedof equally time-spaced pulses. The modulation iscarried out by changing the amplitude of the trans-mitted pulses in sympathy with changes in theamplitude of the input analogue signal. Theresultant PAM signal is then “quantized” byassigning to each sample a discrete level of ampli-tude, selected according to a definite level scale(e.g. 256 quantized levels). The amplitudes of thePAM samples are then “coded” according to asuccession of elementary bits (8-bit words in thecase of 256 quantized levels). The sampling mayalso be carried out by either modulating theduration or the position of the sampling pulse,instead of its amplitude.

Michele Lemme graduated in Electrotechnical Engineering from the University of Naples in 1952 andinitially stayed on as a Research Assistant at the University.

In 1956, he joined Vatican Radio at a time when the St. Maria di Galeria transmitting centre was in theprocess of being set up.

Since then, until his recent retirement, Mr Lemme has devoted his whole working life to St. Maria diGaleria, where he was the Director and the prime motor behind all its developments.

Rolando Menicucci graduated in Radiocommunications from the G. Galilei Technical Industrial Instituteof Rome in 1958.

After completing a special course in Nuclear Physics at the National Research Council (CNR), he joinedthe Selenia company – where he was granted a scholarship – and worked on the development of radarsystems in the project laboratory.

Mr Menicucci also worked in a company which produced car radio receivers before joining Vatican Radioin 1964. He now heads the telecommunications section of Vatican Radio.