]\gt ; Prof. J. A. of Btectr " " On the Conversion of Electric Oscillations into Currents by means of a Vacuum Valve By J. A. FLEMING , M.A. , D.Sc .
, F.RS .
, Professor of Electrical Engineering in University College , London .
Received January 24 , \mdash ; Read February 9 , 1905 .
An electric oscillation being an- alternating current of very high equency , cannot directly affect an ordinary movable coil or movable needle galvanometer .
Appliances generally used for detecting electric waves or electric oscillations are , therefore , in fact , alternating current instruments , and must depend for their action upon some property which is independent of the direction of the current , such as the heating effect or magnetizing force .
The coherer used in Hertzian wave research is not metricaJ , since the action is merely catastrophic or accidental , and bears no very delinibe relatio1l to the energy of the oscillation which starts it .
Even the demagnetising action of electric oscillations , though more definite in operation than the contact action at loose joints , is far from being all that is required for quantitative research .
It is obvious it would be an advantage if we could utilise the direct current mirror galvanometer for the detectio1l and measurement of feeble electric oscillations .
This can be done if we can discover a medium with perfect unilateral conductivity .
Some time ago , I considered the use of the aluminium-carbon electrolytic cell with this object .
It is well known that a cell con- taining a plate of aluminium and carbon , immersed in some electrolyte which yields oxygen , such as dilute sulphuric acid or an aqueous solution of any caustic alkali , or salt yielding oxygen , has a unilateral conductivity within limits .
An electric current under a tain electromotive force can pass through the cell from the carbon to the aluminium , but not in the reverse direction .
This action has been much studied and is the basis of ma1ly technical devioes , such as the Nodon electric valve .
The action by which this unilateral conductivity is produced involves , however , a time element , and after much experimenting I found that ib did not operate with high equsncy currents .
My thoughts then turned to an old observation made by me in 1889 , communicated to the Royal Society , amongst other facts , in a Paper in 1889 , and also exhibited experimentally at the Royal Institution in 1890.* This was the discovery : that if a carbon filament electric ' Roy .
Soc. Proc vol. 47 , p. 122 , 1890 , " " On Electric Discharge between Electrodes at different Temperatures in Air and High Vacua by T. A. eming , communicated ])ecember16 , 1889 ; see also of the Institution , ' OsciUations Continuous rrents .
lamp contains a pair of carbon filaments or a single filament and metallic plate sealed into the bulb , the vacuous space between possesses a unilateral conductivity of a particular kind when the carbon filament , or one of the two filaments , is made incandescent .
I have quite lately returned to this matter , and have found that this unilateral conductivity exists even with alternating currents of high frequency and is independent of the frequency .
Hence , in a suitable form , it seemed possible that such a device would provide us with a means of rectifying electric oscillations and making them lmeasurable on an ordinary galvanometer .
The following experiments were , therefore , tried:\mdash ; Into a glass bulb , made like an incandescent lamp , are sealed in the ordinary way two carbon filaments , or there may be many filaments .
On the other hand , one carbon filament may be used and a platinum wire may be sealed into the bulb terminating in a plate or cylinder of platinum , aluminium or other metal surrounding the filament .
It is preferable to use a metal plate carried on a platinum wire sealed into the glass bulb , the plate being bent into a cylinder which sul'rounds both the legs of the carbon loop .
The diagrams in fig. 1 show various forms of the arrangement .
Diagram shows a bulb with a single carbon filament surrounded by a metal oylinder , shows one with two carbon filaments , and a carbon filament and two insulated metal plates .
ends of the carbon filament which is rendered incandescent are marked \mdash ; and the terminal of the other electrode of the valve is marked .
The bulb must be highly exhausted to about the pressure usual in the of carbon filament incandescent lamps , and the metal cylinder or plate must be freed from occluded air .
Suppose that we employ bulb colltaining one carbon filament surrounded by a metal cylindel .
The filament may be of any voltage , bub I find it most convenient to employ filaments of such a length and section they are brought to bright incandescence by an E.M.F. of 12 volts .
The voltage and section of the filament should be so arranged that the temperature of the filame1lt corresponds with an " " efficiency as a lamp-maker would say , of or 3 watts per candle .
The filament is conveniently brought to incandescence by a small insulated battery of cells .
circuit is then completed through the vacuous space in the bulb betweell the cylindsr and the filament ] ) another wire which joins the external terminal of the metal cylinder and that terminal of the carbon filament which is in connection with the negative pole of the ting battet.y .
In this last circuit is placed a sensitive mirror ) nometer of the movable needle or movable coil type , and also a coil which may vol. 13 , Part LXXXIV , p. 45 , Friday evening discourse on , 1890 , Problems on the ysics of an Electric Lamp when this ) was experimentally shown .
Prof. J. A. Fleming .
Conversion of be the secondary circuit of an air core transformer in which osoillations are set up .
As is now known , the vacuous space the bulb permits negative electricity to move in it from the filament or cathode through the vacuous space to the cylinder or and back through the and coil , but not in the direction , as long as the cylinder is cool and the carbon filament not at a temperature much above the melting point of platinum .
To illustrate the action of the bulb as an electrical valve , the following experiments can be shown:\mdash ; Electric oscillations are set up in a metal wire circuit by the of a Leyden jar , as usual .
This circuit takes the form of a thick wire of one or more turns , bent into the form of a circle or square .
Some distance from this , we place another wire , of several , say eight or ten turns , also bent into the form of a circle , and connect this last wire into the circuit of a galvanometer and vacuum bulb made as describsd , so that it is a circuit having unilateral conductivity .
On exciting the oscillations in the primary circuit by an induction coil we have an alternating high frequency magnetic field produced , which affects the secondary circuit at a distance .
The oscillations in this last are , , ever , able to flow only in one direction .
Hence , the galvanometer is acted upon by a series of intermittent but unidirectional electromotive forces , and its needle or coil deflects .
Since the field is a high frequency field , we call show the screening effect of a sheet of tin foil or silver paper in very simple and effective manner by the ffect it produces in cutting down the galvanometer deflection when the metal sheet is interposed the primary secondary ults .
Also , 1 we move the secondary coil away from the pJ'imary coil or turn the two coils with their planes at right angles to one another , then the galvanometer deflection diminishes or falls to zero because the induction is decreased .
Accordingly , we have in this vacuum valve and into Continuous Currents .
mirror galvanometer a means of detecting feeble alternating currents or oscillations .
Another method is to employ a galvanometer and two vacuum valves .
These must then arranged , as shown fig. 2 , one circuit of the differential galvanois in series with one valve and the other circuit with the other , but so joined up that currents flowing through the valves in opposite directions pass round the two galvanometer wires in the same direction as regards the needle and , therefore , their effects are added together on the galvanometer needle .
Each valve must then have its own separate insulated battery to ignite the filament .
Also , it is necessary that the connection with the oscillatory circuit must be in both cases to the hot filament by that terminal which is in connection with the negative pole of the local battery used to ignite the filament ( see fig. 2 ) .
This arrangement of a differerltial galvanometer two valves transforms , of course , more of the alternating oscillation into direct current than when one valve alone is used .
It provides us with a means of deteoting electrical oscillations not merely in closed circuits but in open electrical circuits .
When so using it , it is nec.essary to associate with the oscillation valve and galvanometer an oscillation transformer raising the voltage .
The resistance of these valves , when in operation , may be anything from a few hundred ohms to some megohms , depending on the state of incandescsnce of the filament and upon the eIectromotive force employed to drive the current through the vacuous space , as well as upon the size of the filament and the plate .
This resistance VOL. LXXIV .
2 Prof. J. A. Fleming .
Conversion of Electric does not obey Ohm 's law , but the current increases to a maximum then slightly decreases as the voltage progressively increases .
form of oscillation transformer employed with the device is as A small air-core induction coil has a primary circuit , which consists 52 turns of gutta-percha covered wire , wound in a helical groove on an ebonite rod inch in diameter and 6 inches in length .
primary circuit is made of a No. 20 or No. 22 S.W.G. copper The secondary circuit consists of-36,000 turns of fine silk-covered No. 36 , wound in six coils , each having about 6000 turns , and joined in ssries .
This secondary circuit has ons terminal connected one common terminal of the vanometer and the other to the terminal of the two oscillation valves ( see fig. 4 ) .
The primary coil this osoillation transformer has one terminal eonnected to earth and other to a long insulated rod which acts as an aerial or electric wave collector .
To prevent the direct action of the transmitter upon the secondary coil by simple electromagnetic induction , it is best to wind the secondary coil in two equal parts in opposite directions and to wind the primary in a corresponding manner .
If an electric wave sent out from a similarly earthed transmitter falls upon the rod , then an electrical oscillation is set up in the receiving circuit and therefore in the primary coil of the oscillation transformer inserted in series with it .
This oscillation is raised in voltage by the secondary coil of the transformer , and by reason of the unilateral conductivity of a vacuum valve , plac.ed in series with the coil , one part of the oscillation , , the positive or the negative current , passes round the galvanometer coils and affects it .
If we employ a dead beat galvanometer of the type called by cable engineers a ' Speaking Galvanometsr then intelligible signals can be sent by making small and larger deflections of the galvanometer corresponding to the dot and dash of the alphabet ; anyone who can " " read mirror\ldquo ; can read off the signals as quickly as they can be sent on an ordinary short submarine cable with this arrangement .
The arrangement , although not as sensitive as a coherer or magnet detector , is much more simple to use .
Also it has one great advantage , , that it enables us to examine the behaviour of any particular form of oscillation producer .
By means of it we can detect changes in the wave-making power or uniformity of operation of the transm arrangement , by the variation of the deflection of the Thus , for instance , if a spark-ball transmitter is being employed an the deflection of the galvanometer in association with the rece aerial is steady , if we put the slightest touch of oil upon the spar -a -ball of the transmitter , their wave-making power is increase an and the deflection of the galvanometer at once increases .
Since the current through the galvanometer is the result of the groups of oscl llations Oscillations into Continuous Currents .
are created in the receiving circuit , and since in the ordinary transmitter these oscillation groups are ssparated by wide intervals of silence , it is obvious that we can increase the sensitiveness of the above describsd arrangement by employing a very rapid or interruptor with the induction coil .
If , for instance , we employ a Wehnelt break with the induction coil or a high speed mercury break or alternating current transformer , we get a far better result as indicated .
by the deflection of the galvanometer than when the ordinary low frequency spring or hammer break .
The point of scientific interest in connection with the device , however , is the question how far such unilateral conductivity as is possessed by the vacuous space is complete .
The electrical properties of these vacuum valves have accordingly been studied .
A bulb containing a 12-volt carbon filament rendered brightly incandescent by a current of about to amperes was employed .
The filament was surrounded by an a , lumini cylinder .
The length of the carbon filament was cm .
, its diameVer mm. , and surface 70 square mm. The aluminium cyhnder had a of 2 , a height of 2 cms .
, and a square .
The filament was shaped like a horse-shoe , the distance ) etween the legs bein mm. This filament was rendered incandescent to various by applying to its terminals 8 , 9 , 10 , and 11 volts respectively .
Anothel insulated battery of secondary cells was employed to send a current through the vacuous space from the cylinder to the filament , cotlnection being with the negative terminal of the latter .
The current throngh ths vacuous space and the ltial difference of the cylinder and negative end of the hot carbon filament were measured by a potentiometer .
The effective resistance of the vacuous space is then taken to be the ratio of the so observed potential difference ( valve P.D. ) to the cuYrent ( valve current ) through the vacuum .
The following table records the observations .
The column eaded .
gives the potential difference between the hot filament the cylinder , that headed A gives the current throngh the vacuous space in milliamperes , that headed the resistance of the space in ohms , that headed K106 is 100,000 times the conductivity .
The result is to show that the vacuous space does not posscss a constant resistance , but its conductivity increases rapidly up to a maximum and then decreases as the valve potential difference ) gressively increases .
If we plot the current values as ordin and potential difference of the valve electrodes as abscissae , we find that the current curve quickly rises to a maximum value and then falls again slightly as the potential difference increases steadily .
The condnctivity curve also rises to a ximum and then decreases ( see fig. 3 ) .
The facts so exhibited are well-known characteristics of gaseous Prof. J. A. Fleming .
Conversion of Electric Table Variation of Current th ough , and Conductivity of , Vacuum Valve with varying Electromotive Force , the being an Incandescent Carbon Cathode and Cool Anode .
conduction in ified gases .
It may be noted that there is in bhese current-voltage and voltage-conduction curves a general resemblance to the magnetisation and permeability curves of iron .
To examine further the of this conduction , the following experiments were made .
If a vacuum bulb , as described , is uP in series with a galvanometer and an electrodynamometer and an alternating electromotive force applied to the circuit , the two ]nstrumentS will both be affected .
The galvanometer is , however , affe only by J. J. Thomson , ' duotion of Electricity through Gases , ' Chap. YIlI .
Prof. J. A. Fleming .
Conversion of h'lectric one half of the alternating current will be cut out .
The will read the true mean ( T.M. ) value of the remanent current , and the dynamometer will read the root-mean-square value .
If the conductivity in one direction is not zero , galvanometer will read the T.M. value of the difference of the and negative currents , but the dynamometer will read the R.M.S. of their sum .
In the case , the current through the valve may be considered be a continuous current superimposed upon an alterrzating current .
If we call I the maximum value of the nearly si11oidal current in direction , and I ' the maximum in the opposite direction , then we may say that the dynamometer reading ( D ) expressed in true current value is equal to where is the amplitude factor , and also that galvanometer reading ( G ) in true current value is equal to ( I-I ' ) where is the ' of the currsnt .
Hence\mdash ; , or The fraction say , expressed as a percentage may be called ths rectifyingpower of the valve , for it expresses the percentage which the actual unilateral electric flow or continuous current ough the valve is of that continuous current which would flow if the unilateral dnctivity were perfect .
Perfect rectifying power , however , does not exist .
There is not an infinite resistance to movement of negative electricity from the metal cylinder to the hot filament through the vacuum , although this resistance is immensely greater than that which opposes the movement of negative electricity in the opposite direction .
This point was examined , as follows : A very sensitive electrodynamometer was skilfully constructed by my assistant , Mr. G. B. Dyke , the fixed coil having 2000 turns of No. 47 silk-covered copper wire and the movable coil 1000 turns .
The suspension of the movable coil was by a fine flat phosphor-bronze wire at top and bottom .
, deflection was observed by a mirror and scale .
is the instantaneous value of a periodically varying current with maximum value I periodic time I , then the root-mean-squars value ( R.M.S. value ) of is defined to be and the true mean valtle ( .M .
value ) of is defined to be form amplitude factor ars the names given by the author periodic curve , and to the ratio of the R.M.8 .
of the ordinates to the maximum value during the period .
Oscillations into Contimlous This dynamometer was placed in series with a shunted movable coil galvanometer of Holden-Pitkin pattern , and the two together placed in series with a variable section of an inductionless coil through which an alternating current was passing .
A vacuum valve as above described was in series also with the galvanometer and dynamometer .
The alternatin current was derived from alternator giving a nearly true sinoidal electromotive force .
form factor of the electromotive force curve of this alternator was determined and found to be that for a true sine curve being The vacuum valve sifted out the alternating current flow and allowed the eJlts in one direction to pass , but nearly stopped those in the opposite direction .
The indicationsof the electrodynamometer were proportional to the root-mean-square ( R.M.S. ) value of the sum of the two opposite currents , and that of the galvanometer to the true mean value ( T.M. ) of their difference .
The galvanometer and dynamometer were both calibrated by a potentiometer by means of continuous current , and curves constructed to convert their scale readings to milliamperes .
Then with various ating current electromotive forces , their readings were taken when in series with a vacuum valve and recorded in the following tables .
The letter denotes current in milliamperes as read by the so calibrated dynamometer and that read by the galvanometer .
The ratio is denoted by , and the rectifying power , viz. , by The table shows that the value of is not constant , but for each state of incandescence of the filament reaches a maximum which , however , does not greatly differ from the mean value for the raIlge of currents used .
If we set out the mean values of in a curve ( see fig. 4 ) , in terms of the power expended in the carbon filament , we see that the rectification is less complete in proportion as the temperature of the carbon filament increases .
This is probably due to the fact that as the filament gets hotter , it heats the enclosing cylinder to a higher temperature and enables negative electricity to escape from the Hence , I feel convinced that if the metal cylin der could be kept quite cool by water circulation the rectification would reach 100 per cent. or be complete .
An ideal and perfect reetifier for electric oscillations may , therefore , be found by enclosing a hot carbon filament and a perfeotly cold metal anode in a very perfect vacuum .
With a bulb such as that used for the above experime11ts all we can say is that the CUl.rent passed through the vacuum is from 80 to 90 per cent. uous , ) cent. implying that vacuum is perfectly -conducting in o1le direction and permits the flow of negative electricity only from the hot to the cold electrode .
The necessity for keeping the cathode cold is shown by the following experiment : alternating-current arc was Prof. J. A. Fleming .
Conversion of Eleciric mometer ( D ) to Galvanometer Form Factor of E.M.F. Curve formed between oarbon rods , and an iron rod was placed so that its end dipped into the arc .
An ammeter was connected in between either carbon and the iron rod , and indicated a continuous current of nsgative electricity flowing through the ammeter from the iron rod to the carbon pole .
This current was , however , greatly increased by making the iron rod of a piece of iron pipe closed at the end and Oscillations into Continuous Currents .
kept cool by a jet of water playing in interior .
In this manner I have been able to draw off a continuous current of 3 or 4 amperes from an alternating-ourrent arc using 15 alternating-currenl amperes .
Returning , then , to the vacnum valve , we may note that the curves in fig. 3 show that the vacuous space possesses a maximum conductivity corresponding to a potential difference of abont20 voIts between the electrodes , for the valve used .
The interpretation of this fact may , perhaps , be as follows : the incandescent carbon there is a continual production of electrons or negative ions by atomic dissociation .
Corresponding to every temperature there is a certain electronic tension or percentage of free electrons .
If the carbon is made the negative electrode in high vacuum these negative ions are expelled from it , but they cannot be expelled at a greater rate than they are produced .
Therefore , there is a maximum value for the outgoing current and a value for the ratio of current to electromotive force , that is for the conductivity .
This fact , therefore , fixes a limit to the utility of the device .
The current th ough the vacuous space is , to a very large extent , independent of the electromotive force creating it , and is at no stage proportional to it , or least only within a narrow range of electromotive force near to the maximum conduotivity .
Whilst , therefore , the device is useful as a simple means of detecting electric oscillations , it has not that uniformity of conductivity which would make it useful as strictly metrical device for measuring them .
It can , however , perform the useful service of showing us how far any device for producing electric oscillations or electric waves produces a uniform or very irregulal train of electric oscillations , and what changes conduce to an improvement or reduction in the efficiency of the transmitting device .

