]\gt ; On the Measurement of Electric Currents of High Frequency .
By IPBBLL , B.A. , and D. W. DYE , B.Sc. ( Communicated by T. Glazebrook , F.R.S. Received June 18 , \mdash ; Read June 25 , 1914 .
) ( From the National Physical Laboratory .
) [ PLATE 10 .
] Both from the scientific and practical point of view the accurate measurement of alternating current , voltage and power at high frequencies is of considerable importance .
For example , in the domain of pure science the study of energy losses in dielectrics at high frequencies is most interesting , while on the practical side , in radiotelegraphy , there is a wide field for methods of measurement .
At frequencies from 100,000 to 2 , second by the methods at present in use it is comparatively easy to measure currents of the order of ampere , but many difficulties present themselves when we have to deal with currents of 1 to 50 or 100 amperea We have recently investigated several ways of measuring these larger currents , and the object of the present paper is to give a description of the more methods and the results obtained by them .
Thermal Instruments.\mdash ; The most common method of measuring a high frequency current is to pass it through a conductor and observe the rise of temperature of this as shown by its expansion or change of resistance , or by means of one or more thermojunctions in contact with it or near it .
The last method , that of a heater and separate thermopile , was investigated by one of us many years ago*and was used later to measure the voltages induced in small search coils by alternating magnetic fields .
The separate heater system was used by Mr. Duddell in his beautiful application of Mr. Boys 's radiomicrometer to current measurement ; and his thermo-galvanometer still represents the high-water mark in efficiency .
While , however , the contained thermojunction and moving cod no doubt give the highest efficiency , the use of a thermopile connected to a separate galvanometer has several advantages , especially in laboratories where economy of instruments has to be considel'ed .
A variety of heaters and thermopiles can be used.with * A. Campbell , ' Roy .
Soc. Edin .
Proc July , 1887 , p. 406 .
A. Campbell , ' Phil. Mag Jan. , 1899 , p. 2 .
'Phil .
Mag. , vol. 8 , p. 91 ( 1904 ) .
622 Messrs. A. Campbell and D. W. Dye .
Measurement of a single galvanometer , and both the sensitivity and the promptness oan usually be altered within pretty wide limits by adding resistance to the galvanometer circuit .
The thermopiles we used in general consisted of from four to ten pairs of thin wires about mm. diameter and 1 cm .
long .
The combinations used were various , but most frequently either iron-constantan or manganinconstantan .
The has two strong points in its favour ; it gives a voltage almost exactly proportional to the difference of temperature , and the temperature coefficient of its voltage sensitivity is very small .
On the other hand , the high thermal conductivity of the iron is not desirable , and , as will be explained later in this paper , its magnetic properties may trouble .
It was for these last two reasons that in many of the experiments manganin-constantan couples were used instead .
Perhaps the most commonly used thermal ammeter consists of a single thermojunction in metallic contact with a heater wire .
In our experience this is.not nearly so good as the separate heater system .
When the galvanometer circuit is in metallic connection with one point of the high frequency circuit , usually a small high frequency current will pass to the galvanometer due to the capacity to earth of the galvanometer and its leads , and this current may often cause a change of deflection by heating one of the thermojunctions .
It is often possible to obtain a deflection of the galvanometer when only one terminal of the heater is connected to the high frequency circuit .
These troublesome effects cannot be got rid of by screening the galvanometer or earthing a point of the circuits , but the use of a separate heater gives in generala complete cure .
Accordingly , in all that follows , unless the contrary is stated , it will be understood that the heater is completely insulated from the .
thermopile and galyanometer circuit .
When such a heater is used , the instrument can be calibrated with direct current , which is not always possible with instruments that have hermojunction in metallic contact with the heater .
In general , if the ammeter is to be accurate at different frequencies , the heater should be of such form and dimensions that its effective resistance shall remain appreciably constant throughout the range of frequencies used .
This condition is usually attained by making the heater of sufficiently thin wire ; a uniform tube of very thin wall would be anothetpossible form .
A convenient table has been given by showing the maximum permissible diameters of wires of various metals for 10 per cent. , 1 per cent. and per cent. variation in effec resistance at various frequencies .
The resistance of a constantan * Austin , ' der Drahtlosen Telegraphie , ' vol. 6 , p. 588 1913 ) .
See Zenneck , ' Leitfaden .
Drahtlosen Telegraphie , ' p. 352 .
J. H. Dellinger , ' Bulletiu of Bureau of Standards , ' vol. 10 , p. 92 ( 1913 ) .
Alternating Electric Currents of High Frequency .
wire of mm. diameter increases less than 1 in 1000 at a frequency of 3 per second ; the change iu similar wire of mm. diameter is less than ] per cent. The use of more thau two wires in parallel , in order to permit the measurement of larger currents , presents difficulties , * since exact symmetry is not easily secured .
Most of the heaters we have used were of short lengths ( straight , crinkled , or doubled back on themselves ) of constantan wire of diameters ranging from to mm. The thickest of these wires will not stand more than amperes in air , and of course , for proper working , could not be used for currents so large as this .
We have found , however , that the larger currents can be dealt with by immersing the whole combination of heater and thermopile in a large vessel of liquid such as paraffin oil .
Oit-cooled Eeater.\mdash ; The behaviour of a separate heater and a thermopile immersed in oil is very interesting .
The difference of temperature indicated by the thermopile depen not only upon the positions of the opposite junctions relative to the heater , but also upon the thermal expansibility , the viscosity and other properties of the liquid .
When the thermopile and the heater wire are both placed horizontal in the liquid , with one set of junctions above but projecting slightly beyond the heater , the readings of the galvanometer are in general by no means proportional to the square of the effective current in the heater , as they would approximately be if the surrounding medium were only air .
The relative positions of the heater and junctions can be adjusted by trial , so that over a large range of current the galvanometer deflections shall be almost exactly proportional to the current .
This unexpected result is very useful , as an almost uniform scale is so much superior to the square law scales of all previous thermal instruments .
typical is given in Table I , which refers to a heater of ohm resistance immersed to a depth of 8 cm .
in paraffin oil of boiling point 18 C. A current of amperes caused vaporisation around the wire .
The deflection attained a steady value in about 25 seconds .
The scale was straight and the galvanometer faced its middle point .
Table I contains the deflections corresponding to various currents in the heater ( for scale distance of 100 cm and also deflection/ current in each case .
The values are plotted in fig. 1 and the resulting direct reading scale is shown in fig. 2 .
ehaviour oheated wiquid iurther iustrated iAppendix t 624 Messrs. A. Campbell and D. W. Dye .
Measurement of Table From the Table and those diagrams it will be seen that the scale is uniform over a large .
We have not yet fully investigated the conditions for constancy or temperature coefficient in immersed heater ammeters of this type , but in the experiments described below they proved very useful when frequently calibrated with direct current .
In the earlier experiments a thermopile of iron and constantan was used with a straight wire heater .
With this combination an interesting effect was observed ; it was found that , as the frequency was raised , the same current gave higher and higher readings on the , errors of the order of 5 per cent. being noticed at a frequency of 00,000 per second .
It was found that the errors practically disappeared when the heater wire was bent back in a close narrow loop so as Alternating Electric Currents of High Frequency .
625 to reduce as far as possible the magnetic field produced by the current in it .
With a single straight heater the magnetic field near it due to a current of 1 or 2 amperes produces considerable magnetisation in the iron of the nearer thermojunctions , and at the higher frequencies the power dissipated by hysteresis and eddy currents accounts for the extra heating which caused the errors .
To avoid the possibility of error from this cause ( which has much more effect in the oil-cooled wires , running as they are at high current density ) , the iron-constantan thermopiles were mostly replaced by manganin-constantan ones .
While the system of oil-cooled heater gives trustworthy results when used with care , its range ( say , to 10 amperes ) is too restricted for the more extreme practical requirements .
The next method , which we now proceed to discuss , consists in the use of current transformers , and its range of application extends to the largest currents that can be produced .
Current Transformers.\mdash ; Some years ago it was proved by one of us*that a large alternating current could be measured with accuracy by the help of an air-core transformer in conjunction with alow-reading ammeter .
The proper conditions of working were investigated and the suitability of iron-cored transformers was also discussed .
Since that time current transformers have come into very wide use in the measurement of low frequency currents .
More recently Mr. T. L. Eckersley pointed out to us that the above-mentioned ation of air-core transformers was perfectly applicable to undamped alternating currents of high frequency , and he also showed from the theory that , even where there was considerable damping , good accuracy should be attainable .
At that he actually used such a transformer in high frequency experiments .
We have recently designed a number of current transformers specially adapted to meet the demands of modern high frequency practice ; before describing these and the tests made upon them , the theory of the current transformer with damped high frequency currents will be briefly stated .
Air-core Current with Damped Alternating Currents.\mdash ; Let the secondary circuit of the transformer be connected directly to the flmmeter ( heater or other system ) , this complete circuit having resistance and self-inductance L2 .
Let the mutual inductance from primary to secondary circuit be and the frequency be being written for * A. CampbeU , ' PhiL Mag Sept. , 1896 , p. 271 .
In the Donitz wavemeter ( ' Elektrotechnische Zeitschrift , ' vol. 24 , pp. 920 and 1024 ( 1903 ) ) a small variable current is used to connect the resonating circuit to the heater of the air thermometer , which serves as a current indicator .
626 Messrs. A. Campbell and D. W. Dye .
surement of and for .
Following Eccles , *when a long continued slowly damped train of oscillatory current of the form \ldquo ; passes through the primary circuit ( where times the logarithmic decrement per period ) , we may use the operator in place of the simple operator used for ordinary continuous alternating currents .
If and be the instantaneous values of the primary and secondary currents respectively , and the corresponding effective values , we have Hence For such a value of the damping that ( 2 ) we have If and are both small compared with 1 , we have .
If is small enough , then is very nearly equal to , and this value becomes more and more nearly exact as the frequency is raised .
When the oscillations are continuous , and equation ( 1 ) takes the well known form .
( 4 ) To find in what circumstances the approximation in equation ( 3 ) is sufficiently accurate , let us consider what values of and are likely to occur in practice .
In radiation telegraphy the -lengths commonly range from 300 to 6000 metres , the corresponding limits of being from 6 to .
Since meant ) , unless the damping is excessive , will not be greater than .
It will be found that , by suitable design of the transformer and heater , can usually be kept smaller than .
When these two limits.are not exceeded , it will be seen that equation ( 3 ) is a sufficiently near approximation for practical purposes ( within 1 in 1000 ) .
As will be seen from some of the actual transformers described below , the correction term in ( 3 ) cau generally be made negligible , and the ratio of currents is given by If on the contrary the damping is excessive , then equations ( 2 ) and afford a possible means of determining , by altering B2 until the observed current ratio comes L , in which case * Phys. Soc. vol. 24 , p. 260 ( 1912 ) .
Alternating Electric Currents of High Frequency .
of the Air-core Transformers .
Several air-core transformers were constructed with the windings adjusted by measurements at low frequencies ( per second ) to give xact values of , which may be called the " " nominal ratio Ratios of 20 : 1 , 50 : 1 , 100 : 1 were found convenient .
The secondary coil of highly stranded wire wound in single layer on one ( or two ) ebonite bobbin ( or bobbins ) , and the primary consisted of one or more turns of thicker stranded wire fastened in fixed position inside the bobbin .
The final adjustment of the nominal ratio was effected by slightly altering the position of the primary winding .
The bobbins were of 8 to 15 cm .
diameter , and of various lengths , the wire of the secondary circuits having 7 or 17 insulated strands of mm. diameter , or , in the latest type , 3 strands of mm. diameter .
The ends of the primary and secondary coils were brought out to a considerable distance at opposite ends of an ebonite base .
This was done in order to avoid bringing the leads too near the coils .
Except in the case of the astatic coils , readings were always repeated with the leads of the primary terminals interchanged ; by taking the mean of the two readings the inductive effects of the surrounding circuits were eliminated .
Tests on Air-core Transformers.\mdash ; Each transformer was tested at a number of different frequencies from 50,000 to 2 , per second , the corre- sponding wave-lengths being from to 150 metres .
From an oscillating circuit across a spark gap with large magnesium electrodes high frequency current was induced into another tuned circuit containing a selfinductance coil , an adjustable oil condenser , an oil-cooled heater , and the primary coil of the transformer , the secondary being directly mected to another heater ( in air ) .
The thermopiles of the two heaters were connected through suitable oesistances to reflecting moving coil galvanometers of 10 to 16 ohms resistance .
These galvanometers had phosphor bronze ( not silver ) suspensions ; their sensitivities were of the order of 70 mm. at 1 metre distance per microampere .
Heaters of various resistances were used ohms , etc The sensitivities may be judged from one example : a current of 50 milliamperes in a 1-ohm heater gave a deflection of 600 mm. at 1 metre , the movement of the light spot being quite prompt .
To ensure the highest accuracy , before and after the high frequency readings the two scales were tested at the same readings by means of direct current .
A number of readings of and ( the respective high frequency currents ) were thus obtained , and from their means the ratio was deduced .
The values of were of the order of 1 to amperes .
The steadiness was good , and the individual readings were very consistent among themselves .
Table II gives the results obtained with several of the transformers .
The VOL. XC.\mdash ; A. 2 628 Messrs. A. Campbell and D. W. Dye .
Measurement of last column but one gives the values of from measurements at low frequency ( per second ) , the including the -inductance of the heater circuit ( about microhen.ry ) .
It will be seen from these results that between very wide limits of frequency ( 60,000 to 2,000,000 per second ) the current ratios are remarkably constant , and with one exception explained below in good agreement with the values of determined at low frequencies .
As is shown by the last column , the observed variations in ratio are largely accounted for when the more exact formula ( 3 ) is taken ( being neglected ) , and when , instead of determining the at low frequency , its actual value at the high frequencies is determined by meter observations .
It was found that in transformers Nos. 375 and 376 the fell off by about and 2 per cent. respectively , when the frequency was raised to about 1,000,000 per second .
This diminution was evidently due to the subdivision of the stranding ( wires mm. diameter ) not being sufficiently thorough .
In transformer No. 411 the secondary coil has very much better stranding , and as a result the ratios are more nearly correct at the higher frequencies .
As its L2 is rather too small ( microhenries ) the resistance correction becomes important at when a 15-ohm heater is used , but the accuracy is satisfactory when the proper heater of 1 ohm resistance is used in the secondary circuit .
Table -core Transformers .
It should bo remarked that , due to the limitations of generating apparatus , the tests been made with comparatively small values of the primary current .
The propefties of air-core transformers depend , however , in no way Alternating Electric Currents of High Frequency .
upon the absolute values of the currents , as long as the coils have sufficient current carrying capability .
The 100 : 1 transformer had a well stranded primary circuit , capable of carrying 10 to 20 amperes , and the others had proportionate capabilities .
Thus a thermal ammeter reading to ampere may have its range extended with good accuracy to 10 amperes , and there is not the slighest difficulty in constructing similar transformers to extend the range to 50 or 100 amperes .
Possible Sources of Error.\mdash ; To test whether the distributed capacity , which is appreciable in most coils , has any influence upon the current ratio , condenser of 250 microfarads was put across the terminals of the secondary coil of one of the transformers .
As this scarcely affected the readings , it was concluded that , in general , the action of the distributed capacity is negligible .
The primary and seoondary windings , however , should have the relative position of their turns so arranged as to avoid currents due to intercapacity .
* To prevent the possibility of errors due to proximity to other parts of the circuit , it is desirable to build the air-core transformers with double primary and secondary coils , connected so as to give astaticism in the well known manner .
Iron cored Transformers.\mdash ; There is a very general belief that the use of iron in high frequency measuring instruments is almost impossible , since hysteresis and eddy current effects become so pronounced when the magnetic cycle is repeated at such frequent intervals .
In the case of current transformers , howeYer , this is quite a mistake , for , as will be seen below , most excellent results can be obtained by the introduction of iron cores , and it is not even necessary to use specially thin iron sheet or wire in making these .
One or two examples will illustrate this statement:\mdash ; Example 1 : A number of circular ring stampings from silicon-iron sheet about mm. thick were placed together ( with insulation between them ) , to form an anchor ring of section .about 1 .
cm : and mean diameter cm .
On this ring , well insulated with paraffined silk , was wound very evenly the secondary coil of 200 turns of thin insulated wire .
The primary coil consisted of two turns of stranded conductor arranged to pass centrally through the ring and return symmetrically at a fair distance from it , as shown in fig. 3 .
After slight adjustment of the relative positions of the primary turns and the ring at a wave-length of 160 metres , the current ratio was tested at various frequenoies , and also , for the lowest frequency , with several different values of the primary current .
The results are given in Table III .
* The capacity from primary to secondary was from 20 to S0 micromicrofarads .
630 Messrs. A. Campbell and D. W. Dye .
surement of FIG. 3 .
Table III.\mdash ; Iron-core Transformer ( A ) .
Resistance oferved r ' Ratio o ' From Table III it will be seen that this little transformer , although very simple in construction , gives excellent results , the ratio remaining very constant over the very extended range of frequency , and being also very nearly equal to the ratio of the secondary to the primary turns .
Tests at showed practically no variation with current .
The last line in the refers to a test made with a Duddell portable thermo-ammeter in the secondary circuit ; the result is identical with the others , which are for an 8-ohm heater with thermopile and galvanometer .
Example 2 : The same core with 200 turns was tested with a single primary turn formed by a stranded conductor passing centrally through the ring and up and brought back in two symmetrical loops .
For wave-lengths , 300 , 1500 and 6000 metres the current ratio was found to be , and respectively .
As might be Alternating Etectric Currents of High equency .
expected , it is not so easy to a single primary turn to give such a good result as with two or more turns .
The central conductor might with advantage be split up into a considerable number returning back cally in loops .
If the current to be measured is in a long straight conductor , the iron core with the secondary winding can be slipped over this in the way familiar with low frequencies .
Example 3 : Fair results were also got with a transformer having a core of thin silicon-iron wire ( mm. diameter ) .
The primary and secondary turns were 2 and 900 respectively , giving a nominal ratio of 450 .
Approximate measurements gave the following results:\mdash ; Table IV .
Measurement of Very Small Currents.\mdash ; Transformers may be used in measuring small currents to transform a small current up to a one , which a given instrument may be capable of measuring .
Of course , the effective resistance of the combination is larger than that of the instrument alone in a somewhat greater ratio than the .
In practice it is difficult to avoid also introducing considerable self-inductance , as the secondary winding is not sufficiently short-cirouited to annul all the primary effective self-inductance .
At the higher frequencies this self-inductance maybe objectionable .
If too extreme ratios , , are not attempted , a properly designed transformer .
may be very useful , as the following example shows .
Example 3 : A small transformer with iron core similar to that of Example 1 had and .
Working on an 8-ohm heater the ratio whioh should have been , showed the following variation:\mdash ; Table 632 Messrs. A. Campbell and D. W. Dye .
Measurement of In combination with a Boas vacuum thermojunction of ohms resistance , 5 milliamperes gave a deflection of 500 mIIi .
at 1 metre , the galvanometer circuit having a resistance of 17 ohms .
No doubt thinner sheets or wire for the cores would be an improvement , so as to allow the permeability to rise and give a higher value of , which is always desirable .
eneral Conclusions .
It is evident from our experiments that properly designed air-core transformers , when used with due care , in conjunction with thermal ammeters , afford a simple means of } currents of the order of 1 to 50 amperes with good accuracy at frequencies from 50,000 up to 2,000,000 per second .
Iron cored transformers , which have some advantages in ease of construction , can also be designed to fulfil the same purpose and to give very satisfactory results .
To a limited extent both types are liseful for measuring very small high frequency currents .
In conclusion , our best thanks are due to Dr. T. Glazebrook , our Director , for his kind interest and help in the work .
APPENDIX .
On the Behaviour of an Immersed Heater .
The behaviour of a thermopile ( or more simply a pair of thermojunctions ) near a heater wire oil has already been described for a particular relative position of the heater and junctions .
The general action seems to be that the upward stream of hotter liquid rises more quickly perhaps widening ; ) ) as the heating current is raised ; thus the temperature indicated by the thermojunction is not proportional to the square of the current , but is considerably less .
With various relative positions of heater and junctions very curious curves can be obtained , for example that shown in fig. 4 .
FIQ .
4 .
Campbell Dye . .
Soc. Proc. , , vol. 90 , Plate 10 .
FIG. 5 .
FIG. 7 .
FIG. 6 .
FIG. 8 .

