AMPLIFICATION OF LOW ALTERNATING VOLTAGES WITH THE MAGNETIC AMPLIFIER

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP78-04861A000400030018-7
• Release Decision: RIPPUB
• Original Classification: K
• Document Page Count: 11
• Document Creation Date: December 20, 2016
• Document Release Date: June 6, 2006
• Sequence Number: 18
• Publication Date: August 1, 1955
• Content Type: REPORT

Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
AMPLIFICATION OF LOW ALTERNATING VOLTAGES
WITH THE MAGNETIC AMPLIFIER
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
 Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
AMPLIFICATION OF LOtV ALTERNATING VOLTAGES VVJ I TH
THE MAGNETIC AMPLIFIER
11
mmel
F. Ku
ETZ-A, No. 11 t 1st June 1954) 367-372
(From German)
INTRODUCTION
The magietic amplifier has hitherto been used almost exclusively for the amplification
of currents which vary only slowly in relation to the time occupied by a period of the mains
frequency. In the following remarks, the conditions will be Investigated under which pre-
magnetised chokes are also suitable for the amplification of alternating voltages. In this
connection we shall be thinking mainly of its application as zero current amplifier in a.c.
bridges, though it can equally well be employed as a measured value amplifier in control circuits.
COMPARISON OF a,c. VOLTAGE INDICATORS
The accuracy of a bridge measurement is critically dependent upon the sensitivity and
zero point precision of the zero indicator. The most widely used zero apparatus for a.co
bridges is the vibration galvanometer. Its properties correspond in many respects with the
requirements of an indicator apparatus in the case of a.c. bridges. The indication is
proportional to the amount of the bridge voltage and therefore independent of the phase. The
moving system can be tuned simply to the working frequency. As the mechanical vibration system
composed of re-setting force and mass has a steep resonance curve owing to its extremely low
damping, the harmonics of the measured voltage scarcely affect the balance. A further
advantage is the low input resistance of the instrument. This fact is important in relation
to the interference field sensitivity. The higher the resistance of the Input circuit, the
more difficult It is to protect the indicating instrument from the strong leakage fields, which
are always present in the heavy current test station, by screening or other measures.
In spite of these advantageous features, there has been no lack of attempts to replace
the vibration galvanometer by other measuring devices. These efforts were mainly due to the
requirements of the works test.station, which regularly carries out control and acceptance
tests of cables etc. Conditions of space are here usually very unfavourable. For example,
neighbouring machinery sets up vibration to which a vibration galvanometer is particularly
sensitive. Moreover, in view of the large number of measurements to be carried out in one day
the low-brightness of the screen picture Is a disturbing factor. A further disadvantage Is
the low resistance to overload of the vibration galvanometer, which If the electrical values
of the cable to be measured differ considerably from the intended value, due to some defect of
construction, may easily lead to destruction of the vibration system.
In the endeavour to replace the sensitive vibration galvanometer by the robust rectifier
instrument it is an obvious matter to connect an amplifier between the latter and the bridge.
Appropriate electronic amplifiers were constructed for the purpose but for several reasons they
were used in practice only to a slight extent. In order to suppress the disturbing Influence
of the harmonics, the amplifier had to be constructed in the form of a resonance amplifier.
.It is very difficult to build resonance circuits for 8:) c/s with adequate selectivity so that
the indicating instrument mainly shows the harmonic voltage, especially in the neighbourhood
o.f the balancing point. Another disadvantageous effect is that the thermionic valve represents
a high ohmic amplifier component and is consequently highly sensitive to disturbing fields.
The expedient of increasing the tuning frequency for the subsequent resonance amplifier stages,
on the principle of the heterodyne receiver, by mixing with a fixed intermediate or high
frequency oscillator frequency, does not introduce any substantial advantages.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
 Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
The rectifier instrument alone is also useless as a zero current indicator when a sensitive
galvanometer is being employed instead of the normal moving coil instrument, since below a certain
threshold voltage practically no rectifier effect occurs in the metal rectifier. This difficulty
can be overcome by the aid of special modulation circuits. Figure i shC ws one of the many circuit
modifications. Eight resistances, of which Ri = R4 = R5 = Re and R2 - R3 = R6 = R7, were connected
to form a bridge. At the points (x, b they are connected to the measured voltage and at the points
C, d to an auxiliary voltage Uh which is taken off the voltage Um feeding the measuring bridge.
Uh and the resistances are so selected that the rectifiers carry rated currents. No current flows
through the indicating instrument, since the bridge and all the voltages are symmetrical. If the
measured voltage Um is applied to the points b, the potentials of the points a and f are displaced
by the additional voltage drops across the resistances, so that a current flows through the indicating
instrument. Since the rectifiers carry approximately rated current, independently o;r U, rectification
is linear even in the case of very low signal voltages. The circuit may also be extended by the
rectifiers shown in faint lines, so that the same behaviour is obtained in both half-waves of the
alternating voltage.
With this arrangement, approximately the sensitivity of a vibration galvanometer may be
attained. The deflection of the indicating instrument, for constant Uh, is proportional to the
bridge voltage and to the phase angle between Jh and Um. This property makes balancing of the
bridge extremely difficult, as the phase of the bridge voltage is liable to undergo mich alteration
In the vicinity of the point of balance and thus the deflection of the indicating instrument may
increase in the course of balancing, although the absolute value of the measured voltage decreases,
and vice versa.
The uncertainty of the indications does not occur when the frequencies of Uh and Um differ
by a small amount, e.g. i c/s. The indicated current fluctuates in this case with the differential
frequency. Quite apart from the fact that it is difficult to keep to a given small frequency
difference, the sensitivity of the zero indicator is reduced, since the natural frequency of the
sensitive indicating instrument 1-s in general not high in comparison with the difference in frequency.
Balancing is also hardly any easier than In the case of the vibration galvanometer, because the
amplitudes of oscillation have to be read.
VOLTAGE AMPLIFICATION OF THE REACTOR CIRCUITS
WITHOUT FEEDBACK
The disadvantages of the indicating methods described suggest investigating the feasibility
of using the magnetic amplifier for the purpose under consideration [i]*. The fact that this
amplifier is extremely insensitive to interference voltages is a favourable circumstance in this
connection. It also suppresses of its own accord the higher harmonics of the measured magnitude
owing to its natural low-pass characteristic, and can be made so robust as to withstand the
roughest treatment. The operation of the various magnetic amplifier circuits of interest in this
connection cannot be discussed within the scope of the present paper, but reference must be made
to the extensive literature on the subject [2,3].
The simplest amplifier arrangement (Figure 2) consists of two identical reactors each with
two windings, a load or working winding and a control winding. The working windings Wb are
connected to the working voltage U with the frequency fb. The windings are connected together
in such a way that no current can flow in the short circuited control circuit if there is no pre-
magnetisation.
The parallel reactor connection (Figure 2a) is out of the question for the amplification
of alternating currents. If'in fact an alternating voltage is applied to the control winding,
alternating voltages are also induced in the working windings and these Voltages are in phase
with each other. Parallel connection of the working windings thus represents a short circuit
for the control alternating voltage and neither of the two reactors is pre-magnetised. If
the alternating control voltage is replaced by a d.c. voltage, the short circuit causes a slow
increase in the control current Is and thus also of the working current Ib.
In the case of the series reactor connection (Figure Pb) there is no short-circuit, so
that a variation of the control current will not induce any compensating current in the working
circuit. The dynamic properties of this circuit are readily understood with a 'few assumptions.
It will thus be assumed that the ma gnetlsation curve is rectangular and that the frequency of
,,d is small in comparison with the working frequency fb of
the control alternating current (f,
the amplifier. If the control magnitude Is a direct current, only the ohmic resistance Rs of
the control circuit acts as Input magnitude. The attainable voltage amplification Is then
b IbRb
b IS S
V,
Flo
* For references, see end.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
 Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
3
owing to the high permeability of the reactors, the alternating current, in the absence of
pre-magnetisation, is negligibly small. With pre-magnetised reactor, the alternating current
set up is such that the mean value of the alternating flux equals the pre-imagnetisation flux.
Consequently
% Is = Wb lb .
...... (2)
s I4-~S Rb
V. UO wb . _. _ ...... (3)
S I4 RS '')b RS
If an alternating voltage with the frequency f8 is applied to the control winding the input
resistance also contains an inductive component. The voltage amplification is now:
VU = _.._ WS Rb 2 Tr fg ...... (4)
According to vale and Atkinson[4], on the assumption of a rectangular magnetisation characteristic,
the control circuit inductance Is
u
'S U~ 46
Die to the periodically varying permeability, LS is not a constant but a mean value which is
associated with the mean flux 'M of the reactors. If equation (5) is substituted in
equation (4) and taking equation (1) into account, we have after a few transformations:
v Vu,O
fS SV +1
...... (6)
If the abbreviation fb i fs WS = !1 is introduced, where Al denotes the inertia faactor, we have
V
I V, 2
1L2 M,
Figure 3 represents equation (7).
It supplies the factor by which the voltage amplification for a.c. control Is less than for
d.c. control. in order to be able to estimate the amplification to be obtained at fs = 80 c/s,
values such as the following, which may also be attained in practice, are assumed:
R = 10, fb X.
RS 3
"b /0.28 (WS[[tY4 + 1
As shown in Figure 4 the greatest amplification is obtained when b)SlWb = 1.3 but even then it is
less than 10. For comparison, the voltage amplification in the case of direct current control
Vi is included. While VU, can be increased by a large number of turns ratio, the a.c.
amplification drops steeply for a large b)S/tdb. The series reactor connection without feedback
is not, therefore, suitable for the amplification of alternating voltages, since its
amplification factor is too small.
According to equation (2) control flux and working flux are equal for the series reactor
connection. If the control and working numbers of turns only differ, as in the present case,
to a slight extent, the control current is approximately equal to the alternating current and
no current amplification takes place. The voltage and thus the power only are amplified owing
to the fact that on account of the small control frequency fS in comparison with fb for the same
current a lower voltage need be applied to the control coil than that which drops across the
load resistance Rb.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
 Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
VOLTAGE AMPLIFICATION IN THE CASE OF REACTOR
CONNECTIONS WITH FEEDBACK
The control flux demand may be substantially reduced when a large part of the pre--magnetisation
flux is covered by the a.c. circuit. Figure 5 shows the two most usual feedback circuits. In
the case of circuit (a), the working current flowing through the load resistance is rectified and
fed to a special feedback winding 4r In circuit (b), a rectifier is connected in series with
each of the two parallel connected reactors. As the rectifier suppresses the current in one
direction, only half of the magnetisation characteristic curve will be used. It the whole of the
magnetisation characteristic curve has to be used, an additional flux is required which displaces
the lower limit induction up to the opposite saturation Induction. Since the series reactor
circuit, owing to the additional feedback coil, requires more winding space, it has in the course
of time been increasingly replaced by the saturation circuit (b).
Figure 6 shows the variation of flux and current in the auto-saturation circuit when the control
flux as or. the flux $s pre-magnetises the reactors. The instantaneous value of the flux (~, 4II)
cannot be less than 4s or greater than the saturation flux c5g,. For the case of fully compensated
feedback (circuit (a): Wk = circuit (b): ideal rectifiers) and assuming a rectangular
magnetisation characteristic, the whole of the pre-magn etising flux is provided by the a.cd circuit.
An infinitely small control flux Is sufficient to alter the degree of saturation of the reactors.
In reality, the magnetisation characteristic departs more or less from the ideal form and the
control current source has to provide a flux which is to some extent the equivalent of the flux
along the rising section of the magnetisation characteristic. Accordingly, the magnitude of the
amplification attainable depends to a considerable extent on the properties of the core material,
though the latter play a subordinate part in the circuit without feedback.
In order to ascertain voltage amplification the control circuit inductance L3 must again
first be determined. As Hedstroem and Borg [3] show, it will be sufficient for this purpose to
consider the mean flux 46 If the frequency of the control current fs is small in comparison with
the working frequency fb, the control current I. and hence the mean flux during one period of the
working voltage can be regarded as approximately constant. In the case of a control current
discontinuity 8 Is which varies the mean flux by 8 an inductance
LS = 2 ws S /(8 IS)
...... (8)
is on the, average effective. The flux fluctuates before the control instant between ~81 and ,
(mean value INi = 0,5 (~Sj + tsi)) and afterwards between 2 and 4s4 [mean value (r2 = 0.5
( + s2)]. The mean value therefore varies by
8 'k = 0.5 (%1 - sz)-
...... (9)
The variation of the mean voltage at the load resistance during half a period of the working
'frequency [8 t = 1/2 f& is:
8 UM = U41 - U~y2 = U '" "b 1/(2 fb) _ [U y wb 1/(2 fb)1 _
(8 2- 8 Ii) = 2 &)b fb (Isg - tS2 - ~sg + psi),
z 1/(2 fb)
..... (10)
SUN = 2twb fb(~S1- ,) = BIbRb, ..... (11)
-- 2 = 8 Ib Rbl(2 wb fb) . ..... (12)
Equation (12) substituted in equation (9) and the latter in equation (8) give
Ls.= a-2 b ..... (13)
Is kb
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
 Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
5-
The voltage amplification in the case of alternating current control is then, as in the case
of the series connection without feedback
8 Ib Rb
S Iq /(2 it fs L3) 2 ;
...... (14)
Equation (13) substituted in equation (14), If VUO Is the voltage amplification in the case
of direct current control, gives:
= 1 ,~~7.rVU01 +
...... (15)
= 1//(7T Vj'O1T) 2 + 1
...... (16)
If V,, is large in comparison with m, equation (16) may be simplified as follows:
m,
Vt0 7TV '
M fb Z$ 1
=
Y =
u 7r fs WS 7
...... (18)
i.e. In the case of a.c. control the voltage amplification is independent of the voltage
amplification In the case of direct current control and proportional to the inertia factor IT!.
Equations (16) and (17) are represented in Figure 7.
Finally, we have still to bring voltage amplification in the case of direct current
control into relation with the reactor data.
V = _ = 8 UJ = S URb. WS
.
S US 8-18 RS RS S O
The working voltage U applied to the reactors must be measured in such a way that in the
absence of pre-magnetisation, the magnetisation characteristic will be rectilinear from one
saturation inflection to the other. Then
7
U = -~ . fb wb 13, , ( = saturation flux) (2))
/2If, for $5 = 0, the voltage URb applied to the load resistance is also 0, then
U = 2/ZU~
~- . ...... (21)
Irs~
Equation (20) substituted in equation (21) and the latter in equation (19) give
V 4b w S S ~
s . ...... (22)
8 O
In order to obtain high a.c. amplification, the number of control turns W. must be made low.
This will of course reduce Vim? but this loss can be compensated by Increasing the number of
working turns 11b or the working frequency fir There are upper limits
for both filb and fb, on the one hand due to the available winding space and on the other to the
eddy current losses occurring in the core. The latter also smooth out the steep section of
the magnetisation characteristic curve, so that for high working frequencies the quotient
sS QS drops sharply.
In contrast with the properties of thermionic valves, amplification with the magnetic
amplifier decreases with increasing control frequency. The harmonics of the control current
are amplified less than the fundamental wave. If it is desired to avoid these linear
distortions, it will be necessary to arrange for the inertia factor M, applicable to the control
frequency range concerned, to be substantially greater than V.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
 Approved For Release 2007/10/23: CIA-RDP78-04861AO00400030018-7
- 8 -
THE RESONANCE AMPLIFIER
The foregoing section showed that voltage amplification depends to a great axtent on the
inertia factor M. As m will nearly always be less than 100, it appears advisable to consider some
other means of increasing Vi.
The sensitivity of the parallel reactor connection with self saturation can be considerably
increased if, as Figure 8 shows, a capacitance Is applied in series with the reactors. We then
obtain a series resonance circuit in which inductance varies in rhythm with the control current.
The fact that each of the two reactors Is connected in series with a rectifier is not important
so tar as resonance Is concerned, since the rectifiers only regulate the current distribution
between the two reactors and do not In the least affect the inductive character of the parallel
reactor connection. Owing to the series capacitance, the alternating voltage applied to the
reactors is no longer constant but varies with the output. The stability of the anplifier
therefore becomes decidedly more critical. The danger of Instability is, however, diminished by
the fact that the load resistance damps the series resonance circuit.
The measurements described below were carried out with ring-core reactors. Iron cross-
section 1 sq.cm.: mean length iron path 18 cm.: core of mumetal, 'ob = 3000, I's = 500, fb - 500 c/s.
Figure 9 shows the working characteristics E-b = f(Q.J or URb = f(V of the resonance
amplifier. Only the steep section of the characteristic is shown. Owing to the series capacitance,
the shape of the characteristic Is retained: only the steepness increases with diminished C. in
this case there is no point in reducing the capacitance beyond 0.7 ?F, since voltage fluctuations
or variations of the control circuit resistance may then set up natural oscillations.
Figure 10 shows very well how sensitivity is increased by series capacitance. in this
figure, the slope of the characteristics of Figure 9 has been plotted against series capacitance.
In this example, therefore, the resonance amplifier is approximately 2.5 times more sensitive than
the amplifier without series capacitance.
Between the stable range and the instability limit there is a transition state in which the
arrangement is excited to natural oscillations. The origin of such oscillations, which have their
source in the curvature of the magnetisation characteristic of the reactors, is extremely complex
on account of the non-linear relationships. In the case of the resonance amplifier, natural
oscillation expresses itself In periodic fluctuations of the total flux, and thus of the load
current tb of the reactors. The load current appears to have superposed on it a lower frequency
which may be designated as fsub' In the oscillogram of Figure 11 the working frequency and
subharmonic frequency are as 1:10.
The natural oscillations are unwanted in the resonance amplifier. It is therefore
Important to know under what limit conditions they occur. In the present amplifier, therefore,
the subharmonic frequency fsub vas measured in its dependence upon the individual parameters, such
as 0, Rs. Rb and ~s the measurements being made by means of a cathode-ray oscillograph and valve
generator, using Lissajous figures, in order to eliminate any reactions in the amplifier.
The frequency of natural oscillation is lower, the smaller is the series capacitance C.
Figure 12 shows the dependence of the frequency fanb upon the load resistance Rb arut the control
circuit resistance R. Both resistances Influence it in the same way. If Rb and R9 increase,
fanb also increases. The dependence on the control flux (% is also important. In Figure 12
the latter influence 1s, expressed by the fact that for each curve fsub = f (Rs) = constant two
frequencies are always shown. The upper frequency is related to the lower limit flux, i.e. the
smallest control flux at which oscillation first just sets in. The lower frequency accordingly
relates to the upper limit flux. Consequently, the greater the control flux, the lower is the
frequency of the oscillation excited. The control flux now represents only a sma LL part of the
entire pre-magnetisation flux, the greater part of it being supplied by the self saturation
rectifiers. All factors influencing the feedback ratio also influence fanb. Hero the control
circuit resistance Re is of particular importance. If it is reduced an additional flux may be
formed across the control circuit, Increasing the pre-magnetisation and thus varying fsub In the
same way as with an increase In the control flux.
Since the load resistance determines the magnitude of the current in the saturation phase,
it is also a decisive factor in charging the capacity in the separate half-waves. Since the
aiterra Ling current tb is inversely proportional to the load resistance for a given pre-magnetisation,
a reduction of Rb must have precisely the same effect on the frequency fanb as a reduction of RS
or an increase of 4
Approved For Release 2007/10/23: CIA-RDP78-04861AO00400030018-7
 Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
Figure 13 indicates the oscillation range limits. Measurements were carried out
for two series capacitances C = 0.7 ?F and 0.6 AF. The voltage LJRb across the load
resistance which gives a measure of the output of the amplifier, is plotted as abscissae.
It is clear that the oscillation range limits depend only to a slight extent on the control
circuit resistance Rs unless the control circuit is opened. The influence of the load
resistance Rb is all the greater. With Increasing Rb, the oscillation range becomes
steadily more restricted. It is therefore expedient to operate the, amplifier with the
maximum possible load resistance. A comparison of the oscillation ranges for the two
capacitances shows that with decreasing series capacitance, the oscillation range widens
considerably.
CONSTRUCTION OF THE AMPLIFIER
With the ring-core reactors described, a two stage zero-current magnetic amplifier
?was'constructed with the circuit shown In Figure 14. The signal voltage is fed to the
control coil of the first reactor pair. The alternating current, after rectification in
G11, passes through the load resistance Rbi. Between Rbi and the control coil of the
following stage lie a low-pass filter and an additional series capacitor which keep all
frequencies over 400 c/s, as well as the direct-current component ava y from the control coil
of the second stage of the amplifier. Suppression of all higher frequencies (especially
the 1000 c/s voltage) is necessary, as these have an incomparably greater amplitude than the
signal voltage, and would considerably overload the second stage and lead to instability.
A band pass with high selectivity, which contains chokes instead of resistances, is also
connected to the load resistance of the second reactor pair. The upper limit frequency is
here considerably lower, since the indicating instrument only serves to detect the 50 cycle
components. The auxiliary control circuit, which gives the reactors a constant basic
magnetisation, displaces the working point to the centre of the linear part of the working
characteristic. The voltage amplification of a stage with direct current control, amounts
to VUO = 260. On the other hand, the inertia factor is small: 1 = 60. The simplified
equation (18) can therefore be used to ascertain the voltage amplification. From this
Vl~ = 19.2. Having regard to the voltage loss occurring in the intermediate circuits, the
result is a total amplification of
Vu total = 19.2 x 19.2 x 0.35 = 129.
So that the indicating instrument of the amplifier will show a clearly readable deflection
corresponding to 5 /IA, a 50 cycle voltage of 0.15V must be applied to the load resistance
of the second amplifier stage. The corresponding amplifier input voltage is equal to
1.2 mV when Vii total = 129. For the case of the resonance amplifier, pre-calculation of
the amplifier factor gives rise to considerable difficulties, and it must therefore be
measured. Figure 15 shows the dependence of the instrument current on the input voltage
of the amplifier. The control voltage source had an internal resistance of 2000 Q. With
an initial voltage of 0.25 mV a deflection appropriate to 10 /.PA Is obtained. A 50 cycle
voltage of 0.19 V must therefore be available at the load resistance of the second stage of
the amplifier. The voltage amplification is now Vii total = 760. This figure lies far above
that attainable with amplifiers without series capacitance.
Figure 16 shows the complete set. To facilitate balancing, the sensitivity can be
varied by series resistances in the first control circuit. To provide optimum matching of
the input resistance of the amplifier with the measurement bridge, the first reactor pair
was made with three different control windings, which can be selected by a switch. A small
instrument permits control of the correct working point of both stages of the amplifier.
SUMMARY
The object of this paper is the investigation of the conditions under which pre-
magnetised reactors are suitable for the amplification of small alternating currents, for
example in place of the vibration galvanometer. The parallel reactor connection with self-
saturation proves to be the most suitable for the purpose. Simple expressions are derived
for the voltage amplification VU on the assumption of a rectangular ma gletisation characteristic.
The amplification factor may be increased by connecting a capacitance in series with the
reactors. A two-stage amplifier is then described, which operates on a working frequency
of 500 c/s. A voltage of 0.25 mV can be measured with this apparatus. At the amplifier
output, the set includes an ordinary moving coil instrument with point bearings.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
 Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
- 8-
REFERENCES
[i] F. K L. The application of magnetic amplifiers for zero-current indic=ations at
50 cycles. Dissertation Technische Hochschule Darmstadt, 19b.''.
[2] T. Buchhold. Theory of the magnetic amplifier. ' Arch. Elektrotechn., all (1943) 197-211.
[3] S.E. HEDSTFOEM and L.F. BDRG. Transductor fundamentals. Electronics,
^. (1948) 8893.
[4] H.M. GALE and P.D. ATKINS)N. A theoretical and experimental study of the series--
connected magnetic amplifier. Proc. F.E.E., U (1949) 99-124.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
 Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
U ? measured voltage
Uh a auxiliary voltage
Uh >> Um
Fig.2: Basic circuit of the magnetic amplifier
without feedback.
(a) Parallel reactor connection
(b) Series reactor connection.
Fig-l: Rectifier-modulation
circuit. The deflection
of the indicating instrument
is proportion to U and to
the cosine of the angle
between Us and Uh.
flg.3: Ratio of voltage amplification for a.c.
output V to voltage amplification for d.c.
output %0, plotted against the quotient
Vuo/m (a inertia factor) for the magnetic
amplifier without feedback.
Fig.S: Basic circuits of the magnetic amplifier with
feedback. (a) Series reactor connection with separate
feedback winding, (b) Parallel reactor connection with
self-saturation. The output voltage drops across the
load resistance Rb (UM).
Fig.4: Voltage amplifications Vu and Vuo plotted against
ratio of number of terms wm/wb. The maximum voltage
amplification Vu - 9.5 is obtained for wm/wb ? 1.3.
M - mean reactor flux
Fi.g.6: Parallel reactor connection with self-saturation.
The alternating flux of the two reactors I and II
fluctuates between t4e values of the control flux I
and saturation flux (129. s
Solid curve: Equation (16)
Dashed curve.-Equation (lf
Fig.7: Ratio of voltage amplification with a.c. output
V to voltage amplification with d.-c. output as as a
function of the quotient V. /m for the magnetic
?plifier with self-saturation.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
 Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
.
Be
control flux. 9b - working flux, URb - voltage drop across load resistance
Fig.9: Linear range of working Fig.10: Slope of working characteristics
characteristics of the resonance in the linear range plotted against
amplifier for different series series capacitance.
capacitances.
Flgg 8: Assonance amplifier
with parallel reactor
connection and self-_watur-
atlon. The cQpaclor C
forms a series resaneneo
circuit with the amplifier
ranrtnrs.
Flg.ll: Natural oscillation of the resonance
aaplifier.e sgbharmonic frequenccyy fsub
is to the working frequency fb as 1:1U.
Lower Upper
limit limit
SDI Eat 1
~1 4 ~
sl
I ioacill- ) ? 6 i
?
'ation $ ? Oaeill ' ?
is range 1 " ration la
y 1 1 V) 3; y range '.?.
C-4- F I7
v
1 S
1
i .-81?s
I
%
VJV
17g.13: Oscillation range limits of the resonance
amplifier for two series capacitances. The
amplifier is stable in operation outside the limits.
$00 c/ a
L? _
Fig.14: Complete circuit
zero-voltage amplifier.
? R, `4.0 L'? for R An cis
R, 1'} !. For R, ... Rr.. fl' C/a
Fig.12: The natural frequency f b excited in the
resonance amplifier against eon rol circuit
resistance () and load resistance (Rb).
ulnpu t
Fig-15: Instrument current I plotted
against alternating Input voltage
Utaput of the two-stage zero voltage
ampli fl or.
Flg.18: View of final set. (a) Font view, (b) side view with
cove removed. Between the ring core reactors are the rectifiers'.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030018-7
1. L_