JOURNAL OF SCIENCE & ENGINEERING, MAY, 1965 VOL Il .PP
37～的
BLOCKING OSCILLATORS
Circuit Analysis, Applications
and Design Considerations
by
Meng Chien Kuo*
1. Introduct.ion:
The purpose of this paper is to give a general survey of the principles of
circuit operation, applications and design considerations of vacuum-tube blocking
-0scillators which are widely used in television an'.i radar works. Due to the
circuit complicity, although m'lny p'lp:lrs and books have been written purporting
to detail its operation, yet none of them have b~en completely successful. The
main difficulties are stated below:
1. More than a single mode of energy storage is always operative.
2. The various parasitics must be included; they establish the switching time
and thus place a lower limit on the pulse duratfon.
3. The tube operates in a region not previously defined, where its parameters
change drastically from the small-signal operation. In this region the values of
u (amplification factor,) rp (plate resistance) and re (internal grid-cathode resistance) are not even well known.
4. Because of the flux build-up and even possible core saturation, the timing
inductance becomes a function of the magnetizing current and the pulse width.
In this paper, no endeavour has been made to introduce an exact solution to
the various modes of operation. Only approximate calculations regarding typical
blocking oscillator circuits are included.
Most of the circuit analysis and cabulations are prepared with reference to
the following books and papers:
1. Benjamin, R: Blocking Oscillators, J.I.E.E., Vol. 93, 1946.
2. Moody, N.F.: Low-power Pulse Transformers, J.l.E.E., Vol. 的，
3. 0. S. P u ckle: Time Bases, Chapman & Hall Ltd, London, 1945.
4. M.I.T.: Principles of Radar, McGraw-Hill Book Co., 1946.
1946.
5. Fink: P rinciples of Television Engineering, McGraw-Hill Book Co., 1947.
6. Fink: Radar Engine ering, McGraw-Hill Book Co., 1947.
"' Associate Professor, Department of Applied Mathematics, Chung Hsing University.
一 37 -
n:i. 學 1/1>.. 第二期
7. Starr: Radio and Radar Technique, Sir Isaac Pitman & Sons, Ltd, London,
1953.
8. Samuel Seely: Electron-tube Circuits, 1958.
9. Jacob Millman & Herbert Taub: Pulse and Digital Circuits, McGraw-Hill
Book Co., 1956.
10. Leonard Strauss: Wave Generation and Shaping, McGraw-Hill Book Co.,
1960.
11. Lamb: Applied Mathematics for Engineers and Scientists, McGraw-Hill
Book Co., 1956.
2. Genera) Descr.iption:
The circuit of a blocking oscillator can be thought d as an ordinary tmi.ed
-grid oscillator circuit having an extremely high ratio cf inductance to capacitance, with the capacitance supplied by the distributed capacitance of the coil
and the interelectrode capacitance of the tube. The coupling between the grid
and the plate coils is made very close, and the turn ratio is such that the grid
drive is excessively large compared with that ordinary used in oscillator operation. A high resistance grid leak is employed, together with a small grid-condenser
capacitance, such that the time constant in the grid circuit is large compared
with the time represented by one cycle of the resonant frequency of the coil
system. This combination of _characteristics provides a口 extreme case of intermit司
tent oscillation, thereby giving relaxation oscillator behavior.
There are two general typ~s of blocking osciliators - the single swing type
in which the tube is cut off at the completion of one cycle, and the self-pulsing
type, in which each cycle of oscillation causes the grid to become more negative
until the tube is biased out of operation. The single-swing type oscillator usually
operates within the audiofrequency range, while the self-pulsing type produces
pulses of RF energy.
2-J. SINGLE SWING BLOCKING OSCILLATOR
The single-swing blocking oscillator is used primarily as a pulse generator,
either free-running or driven. The interruption of oscillation is made to occur
before a normal alternation has been completed. This is done by means cf an
excessively long time constant in the grid circuit, which results in a negative
grid voltage that keeps the tube cut off for a long period of time.
Figure 1 is a single-swing blocking oscillator circuit. It is effectively the
same as a tuned-grid oscillator , except that there is no physical capacitor across
the grid coil. The distributed capacitance of the windings along with the associated
inductance of the transformer determines its natural frequency. In order to
obtain a long time constant in the grid circuit, the value of the grid capacitor
(C) or the re sistor (R) is increased. Regenerative inductive feedback is used and
it must be ve ry strong. Since the coupling must be close in order t o obtain so
一 38 一
Blocking Oscillators
︱
︱
l
14
C
『一
---lll
R
B+
Figure 1. Single-swing Blocking Oscillator
B+
(B)
(A)
Polarity of voltage induced
in secondary at the start
Reversal of polarity
induced in secondary
Figure 2.
much feedback, an iron-:::ore transformer is normally used.
As plate current starts to flow, a m agnetic field is set up around the plate
windings of the transformer. This field builds up from zero to a maximum which
induces a voltage in the grid winding or secondary of the transformer. This
voltage is impressed upon the grid of the tube through the grid capacitor with
a polarity that drives the grid more and more positive (Fig. 2-A) as the field in
the plate winding is building up. The grid, when driven positive with respect to
the c'.lthode, draws current and electrons accumulate on the capacitor plate nearest
the grid. This action continues until the capacitor voltage reaches a maximum
value dependent on the secondary voltage and the resistance in the circuit.
The charging current of the capacitor does not vary exponentially, since the
voltage across the secondary is not constant and the grid-to-cathode resistance
is non-linear. Usually the charging current decreases a lmost linea rly for a period
of time . Then an abrupt change occurs and the current becomes zero almost
instantaneous ly. At this point the capacitor voltage is sufficient to cut the tube
off, although, the collapsing magnetic field reverses the secondary voltage which
adds an additional negative voltage to the grid (Fig. 2-B.) This drives the grid
well below c ut-off and the tube remains cut off until the charge on t he capacitor
一 39 一
’室主宰 11.. 第二期
has reduced sufficiently to allow plate current to flow again. As the magnetic
field collapses, the secondary voltage will go to zero, leaving only the capacitor
voltage to keep the tube cut off. The tube is usually cut off for a long period
of time compared to the conducting time (Fig. 3-A).
c／~==1:ι~v仁一一
(A)
Er
B+
。一一一一一一一一．一一一一一一片白一”，用戶
(B)
Figure 3. Blocking
E,
Oscillator 執raveshapes
Oscillation does not start again until the charge built up on the capacitor
has sufficiently leaked off through the grid resistor. The slanting top of each
grid wave is due to the grid c::tpacitor charging through a non-linear resistance
with a varying applied voltage. The time between each of the positive going
pulses is determined by the discharge time of the grid capacitor. Both the width
of the positive pulse and the time interval between the pulses may be varied by
changing the values of the components in the charge and discharge paths.
In Figure 3-B, while the tube is cut off the plate voltage is the applied voltage
(B-plus). At the time the tube conducting, there is a voltage induced in the
primary which will be in opposition to the force th-::it caused it. This reduces the
plate volhge to some value below B-plus. When the phte current is suddenly cut
off the plate voltage rises beyond B-plus. The collapse of the magnetic field
induces a voltage in the primary which accounts for the portion of the plate
voltage wave that rises above B-plus. The form and number of oscillations at
that point de pends on the Q of t he t ransformer circu it. Hig h losse s would cause
the wave to be damped out very quickly, which is highly desirab le in many
circuits.
If a resist or is inserted in the cathode circuit, as shown in Fig. 4, the current
flowing through this resistance will develop a positive pulse, which starts when
the grid goes above cut-off, and stops at the point where the gr id is again
一 40 ←
Blocking Oscillators
~11~！1 仙
01\
I\
I\
B.,.
Figure 4. Single-swing Blocking Oscillator
driven below cut-off. This positives pulse may be used to trigger other circuits.
Triggering voltages may also be o Mained from the grid or from a tertiary
winding, La in Fig. 4.
As shown in Fig. 3-A, the width of the pulse depends on the charging time
of the grid capacitor, and the resting time depends o 口 the discharge time. The
PRF (Pulse Recurrence Frequency) depends on both these factors. The charging
path of the capacitor is through the grid-to-cathode resis ’tance of the tube and
the secondary winding of the transformer. There will be a slight amount of
charging current through the grid-leak resistor, but since the resistance in this
path is usually much larger than the resistance of the main charging path, it has
a minor effect on the total charge time. It would be very difficult to calculate
the pulse width mathematically since the voltage applied to the capacitor changes
constantly and the grid-to-cathode resistance is non-linear. The inductance of the
secondary winding and the mutual inductance of the transformer will also have
an appreciable effect on the charging time. However, it is possible to vary the
pulse width by changing the value of C within certain limits. Reducing the
capacitance results in a narrow pulse (and also a higher PRF,) while increasing
the capacitance has the opposite effect.
Figure 5. Effect of Varying the Grid Capacitor
。－－
c/o
－一
Figure 6. Effect of Varying the Grid Resistor
一 41 -
n~ 曙紙第二期
The capacitor discharges through the grid resistor and the transformer secondary winding. Changing the size of C, the grid resistor, the transformer would
affect the discharge time of the capacitor, which would in turn change the pulse
width as well as the PRF. The usual method of varying the PRF is changing ·~he
size of the grid resistor. Increasing the value of the grid resis·;; or will increase
the discharge time and decrease the PRF. Decreasing the value of the grid resiscor
will shorten the discharge time and increase the PRF.
Changing the value of the cathode resistor of Fig. 4 would have some effect
on the pulse width and the PRF. Since this would also change the amplitude of
the positive pulse developed acro 田 the cathode resistor, such a method of controlling pulse width or PRF would not be used in most circuits. The amount of
variation that can be produced by this method is small compared to that of
changing C or R in the grid circuit. Two methods of changing both the PRF and
the pulse width are to change the turns ratio and the coupling of the transformer.
These methods are not used under ordinary design circumstances since the
oscillator might be made unstable or might cease to operate.
2-2. THE SYNCHRONIZED BLOCKING OSCILLATOR
The synchronized blocking oscillator shown in Figure 7 is employed as a timing
oscillator. When the oscillator is free running, the waveshapes are labeled "eg no
sync". The natural frequency for its operation is about 300 cps. However, a sync }1ronizing pulse can be introduced at the plate without making changes in its
circuit that will produce a PRE as high as 1000 cps. The synchronization pulses
shown occur at 700 cps. Their amplitude is such that one pulse makes foe. circuit
operate, the next pulse does not bring the grid abo.ve cut-off, but the third pul田
does make the tube conduct. Since the oscillator is synchronized on alternate pulses,
its PRF becomec 700/2, or 350 cps.
The amplitude of the synchronizing pulse has a great effect on the frequency
of this circuit. If the amplitude of the synchronization pulse shown is increased
505苔， the first pulse that occurs after the circuit begins operation will bring the
grid above cut-off and the circuit will be synchronized on every pulse. A low
amplitude may cau甜 triggering on every third pulse since the first two may not
have sufficient amplitude to make the tube conduct.
In each case, it assumed that the natural period of the circuit is longer than
the time between the pulses which cause the circuit to operate. If the period of
the blocking oscillator is shorter, it will produce long and short cycles because
the circuit w ill be tr igered on some pulses but w ill conduct on its ow n accord at
times.
Maximum frequency stability is achieved in this circuit by connecting the
grid load resistor to the plate voltage supply. The useful output fro m this circuit
is taken fro m a small (680' ohm) resistor in the cathode circuit. This resistor has
a negligible effect on the operation of the circuit but develops a voltage pulse
一 42 一
Blocking Oscillators
+250V
Input
Sync
Vol拉~
3.6M
Output
Positive Pulse
Sync 0
Pulse
+
0
e~
No c/o
Sync
,w,
戶”
1
p 甜的
-,
twez
....
一一一
+
oh
＋法沁
e11
何﹔仁F 一一－ 71_
一
w
。
e
Figure 7. Synchronized Blocking Oscillator
due to the plate current flowing through it.
2-3. THE DRIVEN BLOCKING OSCILLATOR
Some circu its r equire a blocking oscillator which operates only when trigered.
Some a circuit might be called a blocked oscillator since a permanent DC voltage
prevents the circuit from osc;ilhting of its own accord. Only a positive voltage
on the grid or a negative voltage on the cathode will cause the tube to conduct.
If the voltage is a pulse, th已 circuit will go through a cycle of ope ration, cut
itself off, and the DC voltage will keep it cut-off. Cut-off is maintained ‘ by a
-
J3 一
n:r.11 鼠第二期
vcltage divider from the plate voltage supply. With the values shown in Figure
8, the cathode voltage will be plus 50 vclts wHhτespect to ground. The grid is
grounded, and is therefcre 50 volts negative with respect to the cathode. The
grid wave shape shows .voltages with respect to the cathode since these are the
imporhnt voltages in unders·[; anding the circuit operation. Cut-off for most tubes
used in this type circuit occ urs at about -].8 volts, so ·i;he trigger pulse must be
more than 50-18, or 32 volts in amplitude to start a cycle of operation. In the
grid wave shape the normal discharge cf the grid condenser reduces the grid
voltage after a cycle of operation, but the discharge is to minus 50 volts, so the
circuit does not again conduct without another trigger pulse.
Trigger
lnput
200K
「一斗
間K
＋「一一－ τ一一一一一一一－一一一一一
Trigger
0
Or－一－一－－一一一一一一－一
e, to K
-50
e。
一一一 一一一
0 ，一一一－一一一一一一一一一一一一一一
Figgure 8.
Blocked Oscillator
The positive trigger pulse must be applied to the grid only. It is not practical
to supply a trigger pulse to the c'lth::ide be:;'luse the filter cond.enser is usually
found there. B ut a negative pulse m叮 be applied to the ph妞， where the transformer will invert the pulse to a p::isitive one on the grid. The output may be
taken from t he plate, where a large positive pulse is available, or fr om a third
transformer w inding where a pulse similar to the plate pulse can be obtained
with either polarity.
-44 戶
Blocking Oscillators
a.
Analysis of C.ircuit Oper:iticm:
In the c:nalysis cf the vacunrrA; ube
Figure 9 is t.'1ken as an example.
blockingιscillator,
Le
lOOK
一 20
L個＝ Magnizing
n:l
the circuit shown in
c
g
Ecc
Ideal
Transformer
inductance ; Le =Leakage inductance ; C •=Stray capacitance.
Ebb
Figure 9.
For the purposes of discussion, the operation of the blocking oscillator shown
in Figure 9 may be divided into four regions. These are characterized by:
I. Active region. The positive loop gain drives the tube into ” absolute ” saturation (that is, ec 二三 eb).
II. Tube saturation region. Here the circuit voltages build up toward their
peak values.
III. Timing. Current build-up in the transformer, core saturation, and/or the
charging of C return the tube to the active region and i;ermin<i·[;e the pulse.
IV. Switching and recovery. The regeneration of the circuit turns the tube
off, and the energy previously stored must now be dissipated. The output consists
of 丸 large voltage overshoot and either ringing or an exponential decay toward
the initial conditions.
Of all the parasitic elements present, only the two of major significance, that
is, La and C. will be included in the linearized models. The stray capacity, which
is distributed throughout the circuit, may be lumped into a single element. But
where should it be placed? If it is inserted from the grid to cathode or directly
across the magnetizing inductance, nothing constrains the plate voltage; once the
tube turns on, the full supply voltage appears across La and the plate immediately
buttoms at zero. This will not occur in the physical circuit. If the model permits
such a drop, the model is incorrect. To prevent the abrupt change in 恥，
the
stray
capacity should be connected from the phte to c-athode as shown in Figure 9.
Without actu-ally solving for the exact time r esponse, the b-asi-c b e havior of
the circuit and the role of the various parameters may be dis-cussed as follows:
Region I. Initial Rise. The voltage swept through during the initial portion of
the grid rise, from cut-off to zero, is a very sm-all percentage of the total voltage
change. More over, since the- grid will not load the plate circuit, the loop gain
will be much h igher than that measured in the positive grid region. T he rate of
一 45 -
n:i. 著版第二期
voltage change is very rapid, and the time contribution to the over-all pulse
duration significant.
In the positive grid region, the nature of the response would be found from
the model of Figure lOa.
Le
A=r背
R=rrllRL
(a)
n-;;:-f
-nE. 一
」」國
(b)
Figure 10. Models holding for Regions I and II of Blocking Oscillator of
Figure 9. (a) Incremental model holding within the!active
region; (b) Model defining the rise in the saturation region ;
(.c) Model used to solve for Eb.min a.nd Eo.max.
Applying Kirchhoff’s law to the two-loop network of Figure lOa, we get the
following equations:
1
.
1
個＋京；） i1-pc;i2=Aec
-*i1+(n2rc+PLe 十五忌，） ia=O
Subs叫ing ec ＝一辛＝ -i2nrc
the circuit poles would be ·given by the solution of the determinant
1
R 十一τ－
-pc.
Anre0 一一
pCa
--fc;
σr
n2rc+pLa 十五之
2γc
p2 十 CRC'. 十日 P 十
=0
R+n2re +nAro
R已 c.
c=O
For proper switching, one pole must lie in the right half plane. With the values
given in Figure 9, the two poles are located at
P1 ＝一 7.43x
107
P2=2.43 × 1Q7
In a relatively short time the effect of the negative exponential will have
damped out. The time response is primarily due to the positive pole, and we may
approximate the plate-voltage fall by
ebcEbb(2-eP2,t)'From the de finition that the rise time is the time required to charge from 10 to
一 46 一
Blocking Oscilla to耳；,a
90 per cent of the bounding voltage,
2.2
0. 091 u sec.
P2
The negative pole, along with the additional parasitic elements not included
in the model, will slow the rise by a factor of 1.5 to 2. Any increase in L. or
Cs would also adversely affect the switching time. This portion of the response
is finally bounded when the grid rise and plate fall drive the tube into saturation.
Reflecting the grid circuit voltages into the transformer primary results in
ti ＝－一一＝
Ebb+nEcc=ebi-eL 十 nee
where eL is the drop across the leakage inductance. Solving for
eb=ec=Eb1
we get
E
b1=
Ebb+nEcc-eL
n十1
This equation cannot be solved until eL is evaluated. But this is too difficult to
do. The particular voltage at which the tube saturates is not critical to the
circuit operation. Because L. is small, the voltage developed across it will not
be large even at the boundary. Assuming eL=lO volts, the circuit used as an
example yields
200-40-10
Eb1 ＝一一3一一一－＝ 50
volts.
Region II. Final Current Build-up. After the tube saturates, the circuit model
reduces to the one shown in Fig. 10b. Because RL is very much larger than the
285-ohm plate saturation resistance, the 3,000-ohm external load may be neglected.
The two poles in the left-hand plane are now located by the solution of
。， 1
n2r c,
r …牛 n2rr.
p ＂＋（一二一一十 一」） p 十」L一」＝ 0
、 rpsCs
L. "'
rpsL ”c.
or at
Pi=-9.51 × 101
P2= -1. 75 × 101
Charging continues from the boundary value of E01= Eb1 toward the steady-state
conditions, which can be found from the model of Fig. lOC. The pole located
closest to the origin is associ-3.ted with the longest time response, and consequently the duration of this charging interval is approximately
tz:.4-_1一＝
0.228
益
l P2 I
u sec.
From Fig. 10:; the peak voltages are given by
Eb,m1 戶一主丘r--(Ebb +nE cc)
Ips+u"Ic
E
nrc .－一（
’
]
＝一一一一－
Ebb+nEc
c]
rps+n"rc 、
J
Solving these equations in the circuit used as an example, we find t hat the plate
falls to 20 vo lts while the grid rises to +70 volts. In an actual circuit re would
decrease markedly when the grid voltage greatly exceeds eb. This lowers Ec,mu
and raises E b,m 』 n﹒
-47 一
n:i. 學 11. 第二期
c
r，.~~：~日Jff:Jl
-
Eb11
7
g
~cc
7
Figure 11. Model holding over the timing region
Region III. Pulse Timing. This,
·~he mos·~
important region, is defined by the
very simple model of Figure 11. Because Ln) Le, Cs is mu.ch less signifi-:;ant and
may even be omi·~ ：﹔ ed. The magnUizing curren-c begins increasing with the time
constant
t
Lill
crps! [n2rc
一－一－
toward the steady-s·l;:rte value
I ss·- Ebb
•ps
Meanwhile, the plate voltage rises from Eb,nln tcwc:rd Ebb 2nd the grid decays
一﹜
back toward Ecc· The two charging equations are
eb= Ebb 一（ Ebb- Eb,m!n)e-t I to
ec=Ecc
(Ecc-Ec,max)e-t/tc
Eventually the decay at the grid and the rise at the plate will permit the tube
to reenter the active region. Solving these two equations, the time elapsed until
eb=Eb2=Ec2 is
t泊＝ tcln
•
c
Ebb-Ecc+Ec,max
Ebb+Ecc
Eb,mln
The value of Eb2= Ec2 may be found by substituting the ta back into the two
charging equatiβns or by eliminating the exponential term between these two
equations. The simplest expression for the boundary value becomes
F.b 趴＋ nF.0 「
Ee＂＂。這
身＝ Eb2 ＝一一一一一一~
n+l
We might observe that the 53 volts at which the circuit leaves the saturation
region is slightly higher than the 50 volts at which it entered. The duration of
the pulse is
ta=20x 1。可 ln(l + 50/220) = 20x1。可 lnl.227=4.08 u sec
At the end of the pulse the magnetizing current has reached a peak Of
Im.(ta）＝~盟（1一♂叫“）＝
130
• ps
ma
J.f the transfo rmer cannot tolerate such a current without saturating, the pulse
would terminate proportion'ltely earlier.
Region IV. Termination of the Pulse. Once the circuit again beco mes active,
心regeneration
drives t.he tJJ.b.e toward and even beyond cutoff. As long as the tube
remains active, the poles defining the response will ba given by the equation
一 48 -
Blocking Oscillat-0rs
1
n2γ.
R+n2γc ＋~、 A
re
p2 ＋（芷江＋ τ.！＇.） p 十一Ri:-c:-」＝ 0
Below cutoff the time response is defined by the parallel RL, L血， and Cs circuit.
Depending on the degree of damping, the output may be either an exponential
decay or a damped sinusoid. The final decay is fr om a peak determined by the
energy previously stored in Lm. The external damping RL not only limits the
backswing to a safe value, but also prevents the ringing that may cause a false
retriggering of the circuit.
Ebb
Pulse duration
t
(tp)=t1+t2 十 ta
EG max
Eca
EGJ
E 。 L 一立一
..
Eco
t
II
Figure 12. Plate and grid waveshapes for the blocking oscillator of figure 9.
The magnetizing current of 130 ma flows through RL after the tube - is cut
off. This results in a peak voltage at the plate of
Eb ’回 ax=Ebb 十 Im(ta)RL=590
volts
At the grid the maximum possible backswing is
Ee.min= -20-390/2= -215 volts
The above calculation neglected the stray capacity, which will of course,
slow the switching and reduce the amplitude of the backswing.
4. Applicatfons 。f Blocking OseiJlators
There are t w.J tn'.lin uses of b locking oscillator, the production of very short
pulses of larg e energy and the production of sawtooth waveforms or time bases.
Besides, bloc king oscill泣。 rs are also frequently used in pulse sharpe ning, amplitude selecting, and frequency dividing or counting circuits. Another appliction
of blocking oscillators is to function as a low hnpedance switch to discharge- a
capacitor quickly.
_,9 -
n~ 學織第二期
4-1. PULSE GENERATORS
Either the free-running or the driven blocking oscillatcr is capable of generating a pulse of large peak power. For example, it is possible to obtain 0. 5 amp.
at 100 volts or 50 watts from a receiving-type tube. Of course, the average power
is small since the duty cycle (the ratio of period to pulse duration) is low.
Although either circuit can be designed for a given width of pulse, much greater
precision is achieved by the use of a pulse-forming line in shunt from the grid
winding of the feedback transformer to ground. A typical line-controlled blocking
oscillator driver is shown in Figure 13. Normally the oscillator is cut off by
the 140 volts bias. During the interval between pulses the line is charged to 140
volts since it is connected to the bias supply through the 27K resistor, the transformer secondary, S1 and Rk (the cathode resistor of the trigger tube). The
trigger pulse, which is about 150 volts in amplitude, is sufficient to raise the
grid of the blocking oscillator well above cut-off. It is coupled to the grid
through the capacitance of the pulse forming line.
+1200V
＋臼oov
Pulse .Forming Line
Open
Circ.uit
22K
___ , 715A Hard
·Tube Mod
Damping
Resistor
-850V
L
Figure 13. .Line Controlled Blocking Oscillator Driver
As in an ordinary blocking oscillator, the plate voltage drops when the tube
begins to conduct. This drop is coupled back to the grid in opposite phase due
to the transformer action and causes the grid to. go more positive.
This action
is rapid enough that it quickly drives the tube to saturation. In this condition
the voltage across the primary P of the transformer is about 1000 volts an,d that
across the secondary S1 is about 500 volts. Since the pulse line is charged to 140
volts and since it is in series with the secondary of 500 volts, it will charge to
the higher voltag e t hroug h the grid t o cathode resistance of the oscillator, the
secondary Si. and Rk. Further, the impedance of the circuit is matched to the
characteristic impedance of the line. Thus the volt♀g~ across the line goes from
140 to 500 volts in two steps as shown in Figure 14. The length of t he .step at
320 volts is the time it takes the voltage impulse to travel the length of the
line, be reflected and return to the sending end. At that time the line is fully
一 50 -
Blocking Oscillators
charged and current ceases to flow. The current flow
throu.εh
the grid to cathode
resistance is practically constant during this time and keeps the grid to cathode
potential steady insuring little change in the primary voltage. At the time the
line becomes charged, current flow ceases through the ch'lrging circuit and
causes a drop in the oscillator grid voltage.
This decreases the plate current
and causes a rise in plate voltage. Through the transformer this drop appears as
a drop in voltage at the grid and still further reduces conduction. The action is
cumulative, eventually resulting in the tube being cut off quickly. In this condition, the line is charged to 500 volts and as it has only 140 volts applied it starts
discharging. It discharges through the 27K resistor, S1 and Rk. As this path has a
much higher impedance than the Zc of the line, the discharge appears as a series
of steps which follows the general exponential curve. The output voltage which
is applied to the grid of the hard tube modulator is taken from another secondary winding S1 and is a 1000 volt pulse. Damping resistors shunt both secondaries.
soov
500
P伽 Line
V.olts
I
37且
I
lt40V
。 1-.l- ~一一一一一一一一一一一一一一一－
200 I一＿I 1aov一一一一一一一一一一一一一一一一一
Volts Across
Zc
0
o；~ma:or 們－
Grid Volts
J
I Pulse
I
I Duration
Figure 14. Blocking Oscillator Voltages
4-2. PULSE SHARPENING CIRCUIT
., The driven or blocked oscillator i弓 useful a弓 a pul 弓 e sh'lrp:ming circuit. The
input square wave becomes a peaked wave at the secondary of the differentiating
transformer as shown in Figure 15. Because of ’the inherent imperfections in
the trailling edge of the square pulse, the peaked wave due t::i the trailing edge
is not a sharp one. For the purpose of obtaining a sh'lrper pulse, t he peaked
wave is applie d to a blocked oscillator.
In place of the voltage divider from 十 260 volts used in the earlie r circuits,
a long time c o n弓 tant in the C'lthode circuit provides the cut-off voltag e for the
precediing circuit. After s~veral cycles of operation, the grid curre nt due to
positive grid v olhges charges the cathode condenser. It will discharge slowly
一 51 一
n:i11n. 第二期
between cycles and be recharged with each cycle. The RC in the grid circuit is
rather short and cannot hold the grid below cut-off for the several hundred
microseconds between pulses.
The negative pulse at the leading edge of the square pulse "'ill not affect
the already cut”。 ff tube. The positive pulse at the trailing edge will cause the
left tube to conduct. The current through the transformer and associated magn哺
etic field will induce a positive voltage at the grid of the right-hand tube. This
starts a regenerative action in the blocking oscillator associated with the right
-hand tube and it goes through a normal cycle of operation. Meanwhile the pulse
ends at the grid of the left-hand tube, and with no other voltage on the grid,
the tube is cut off again.
The left-hand tube is an amplifier and isolation circuit which causes the
blocking oscillator to start very soon after the trigger pulse arrives, then disconnect the triggering circuit from the blocking oscillator so it can go through
its normal cycle unaffected by the triggering circuit.
Wav:s~a~：＼一月一一！
叫＋︱一一十一－ 1 一一」－－
Plate
Waveshape
tnp:J\I
27K
z。﹔抖三才三羊
_
一」一一＿，＿：：＇ －斗一
Outout
I
I
- _1 一＿ j_ 一－~ 一－ i--
Figure 15. Blocked Oscillator as a Pulse Sharpening Circuit
一 52 一
Blocking Oscillators
The output is taken from a small resistor above the transformer in the plate
lead. A sharp negative pulse of low amplitude is produced across the resistor.
4-3. TIME BASES
4-3-1. The R.C.A. time Base
The Radio Corporation of America has developed a two-tube time base for
television reception. One tube ~cts as a blocking cscillator-pulse generator, while
the other tube acts as a trace generator. The oscillator, shown in Figure 16, is
conventional except that a low value of series resi弓 tance R4 is inserted in the
grid-cathode circuit by means of which a synchronizing potenHal may be impressed across the grid and cathode of the tube. It should be noted that the condenser
and the potential at the secondary terminal 弓 of the tran弓 forme r mu弓 t possess
such values ·i;hat the grid curren-i: chuges the condenser sufficiently to cut the
tube off after one half-cycle. Moreover, the charge must hold the tube cut off
for a period greater than the time required to scan one line of the television
raster. This implies that C1 should be small enough to permit grid current to
charge 缸， during one half-cycle, to a potenHal which is considerably greater
than that required to cut off the plate current in V1. The value of R1 is chosen
so th叫 C1 will discharge to the cut-off bias potential in a time slightly greater
than the duration of one scanning line.
The condenser C2 is charged through the high re 弓 istance R3 in series Vi ith
＇月， hen the grid i弓 driven positive by a rise in potential
on the grid of V1. The small resistance R2 is added to provide the required wave
-form for producing a saw-tooth current in a deflec·~ing coil composed of inductance and resistance in series. This inductance L is composed of two coils placed
Rz and is discharged by V2
about the neck of the cathode ray tube and C3 is made sufficiently large to have
no effect other than to block off the D.C. component of the current through L.
s+
B-
Figure 16. RCA Time Base
The effect of the resistance R2 is to add a component to the potential applied
across the inductance in such a way that the current through L be comes saw
-tooth in form. In order to provide a saw-tooth current through a pure resistanc~
a saw-tooth potential wave-form must be applied. In the case of pure inductance
however, a re ctangular potenfral wave-form, as shown in Figure 17a, is required.
When the indu -::tance includes a resistance, as it does in practice, t he applied
一 53 一
n~ 學織第二期
potential wave-form must be a combination of the saw-tooth and rectangular
forms, as shown in Figure 17b.
(a)
(b)
Figure 17. Potential wave-form required to produce a saw-tooth
current wave-form (a) a pure inductance (b) an
inductance possessing resistance.
This time base is capable of operation in the absence of a synchronizing
signal is impressed across the resistance 恥， the grid of V1 becomes more positive
and, provided this signal arrives when the charge on C1 has fallen to such an
extent as to permit the grid to rise to a potential just beyond cut-off, plate
current will commence to flow and will rapidly increase due to the cummulative
action of the tube and transformer. The actual duration of the flow of plate
current in the tube is determined by the circuit constants and not by the duration
of the synchronizing signal. It is, however, the commencement of the synchronizing signal which determines the instant at which the flyback commences.
It will thus be 甜 en that the blocking oscillator part of the circuit, i.e. ·the
tube V1. and the associated components, perform only one essential function, viz.
that of keeping the circuit operating in the event of the non-arrival of some of
the synchronizing pulses due to fading in th~ radio ch1nnel. Except for this, the
synchronizing signal might be applied directly to the grid of V2, but, in this
case, it would also be necessary to ensure that the condenser C2 could be discharged during the time the pulse is applied.
4-3-2 Kobayashi's Time Base
The Kobayashi’S Time Base, as shown in 'Figure 18, is essentially a blocking
oscillator in which the blocking action is obtained by a change in the plate
potential instead of in that of the grid. A tube, having the two win dings of a
transformer in the plate' and grid circuits respectively, is used to discharge a
condenser Cr whieh receiv_es_ its charge via a resistance R1. Whe n the plate
potential becomes sufficiently high to cause plate current to flow , a positive
pulse is applied via the transformer to the grid of the tube so that the plate
一 54 一
Blocking Oscilla t的 rs
current is cumulatively increased and the condenser is rapidly discharged. The
transformer is a high-frequency one. A later suggestion for the use of a low
-frequency transformer brought Gbcut an improvement and, with this change, a
modified form of the circuit is to be found in many television receivers.
Out
C1
B-
Figure 18. Kobayashi’s Time Base
In Kobayashi’s circuit, and in the modified version, the condenser C2 across
the cathode bias resistance R2, is made very large in comparison with C1 so that
a steady bias potential is developed across R2. This potential is proportional to
the average plate current and tends stabilize the
frequency.
日， for
example, the
repetition frequency increases as the result of an increase in the B+ supply
potential, the bias potential will also increase and will partly counteract the
change of repetition frequency.
The synchronizing impulse may be applied by means of a third winding on
the transformer on from a high impedance circuit to the grid of the tube. A third,
and perhaps the best, method is to connect a small resistance in series with the
grounded end of the grid winding of the transformer and to apply the synchron·
izing potential across it.
The greater value of the modified low-frequency transformer circuit by
compariso 位 with that of kobayashi is due to the fact that the grid of the tube is
swung through only one cycle for each discharge of the condenser. In this way
the condenser is discharged in one operation, whereas with kobayashi’s circuit,
the grid is swung positive and negative zt a radio frequency, the condenser is
discharged in a series of pulses.
4-3-3
Linear ﹔官e:f Bhukln、穹 Osclllr:tor
The mechanism of foe linearized blocking oscillator as shown in Figure 19 is
one of many circuits developed by Taylor; Just before the arrival of a synchron·
izing signal, the grid cf V1 i 弓﹜1巴 ld negative relative to to the cathode, by a
charge accumulated on C1 during the sweep, so that no plate current flows. When
the signal is applied, －~he grid is driven rapidly positive and plate current commenes to flow from the B+ supply rail. At the beginning of this stroke, the plate
potential across the tube is reduced to a low value but, as the curre nt increases,
the inductance of the choke La falls, the plate potential rises and the current
increases still further. The synchronizjng signal may be applied to t he grid via
-
55 戶
11• 司f :t4 第二期
an additional winding on the transformer if desired. The repetition frequency is
controlled, in the absence of a synchronizing signal, by the values cf Rr andC1. It
is possible to arrange, par<; ly by design ~nd par i; ly by experiment, that the plate
current variations shall provide a substanLhlly linear change of current through
the deflector coils.
B+
::l}
C1
Deflection
Coils
Ca
B『
Figure 19. Taylor's method of linearization
4-4. COUNTING CIRCUITS
The counting circuit shown in Fig. 20 consists essentially of a synchronized
blocking oscillator and a diode ” type integrator. The first blocking oscillator
operates at a frequency of f 1 pulses per second. Each pulse has an amplitude
e'qual to about 1.5 times the volbge of the B supply and charges capacitors C1
and Ca in series. Hence, capacitor Ca receives an increment of voltage, at each
input pulse, equal to
E
l.5EbbC1
一－
C2 一
C1+c;-
Since discharge is prevented by the diode, after n input pulses the voltage applied
to synchronize the second blocking oscillator is n.6Ec2· When this voltage exceeds
the cutoff limit, the second blocking ιscillator responds, and its frequency of
oscillation is
f
fI
fI
。＝百花E;~:o=n:「
Where Es is the
v。ltage
required to trigger the second blocking oscillator.
了~~工 ~Ill 但已之~II~ 于
Tf
~」色恐以比去斗可
Ini
J ﹒＼口／□了）
~t·
(' -1.
\.c.J
O~put fe p
cps
Figure 20. Blocking Oscillators and Counter Ciccuits as Frequency
Deviders
f,
'(
'f 吐 r
When the second blocking oscillator reacts, it completely discharg es capacitor
C:i (during t he grid-current surge), and C2 is thereafter ready to receive further
increments of chnrge from the pr 的eeding circuits. One of the advantages of
this circuit is the hct th'lt all criHcal volhge levels are proportional to the B
一喝一
Blocking Oscillatol"S
supply voltage, and heri.ce "the circuit tends to count the same number of pulses
when the B supply voltage varies. Properly designed counter circuits remain
stable when the pkte supply is varied from 200 to 300
vcl峙， but
if the number of pulses counted is small, say 7 or less.
this is true only
When the division
is
limited to a low factor, the successive values of grid vcltage are then sep;:rated
by larger steps, relative to voltage variations in the input circuit arising from
other causes.
Another method of acccmplishing pulse recurrence frequency division is to
use a blocking cscillator with parallel triggering. A circuit showing PRF division
by 4 is shown in Fig. 21.
卜To「／一－ Time
-1 一←－ Eco
_\_T-:=-tf/
HmM
-
S
h
-Ece
Er
－.~p卡
(b)
(a)
Figure 21. (a) Blocking oscillator with parallel triggering.
(b) Grid waveform· (except for oscillator pulse) showing
PRF division by 4.
In Fig. 21a parallel triggering is indicated hut, of course, several methods of
trigger injection may be employed. In Fig. 21b is shown the waveform at the
blocking oscillator ·grid. For simplicity, the pulse generated by the blocking
oscillator itself is not shown. When the blocking cscillator fires, grid current
charges the capacitor C, which is then left at an initial voltage E1. The oscillator
will fire again when the grid reaches the cutoff voltage Eco· The injected pulses
are now shewn super impcsed on the rising grid voltage rather than on the
critical firing voltage. The ill'.portant point is that a pulse (No. 4) occurs at a
time and has a sufficient amplitude to cause a premature firing of the cscillator.
The oscillator therefore fires at a moment dictated by the occurrence of a
pulse and is not permitted to terminate its cycle naturally. The synchronizing
characteristics for the thyratron sweep generator may be applied directly to the
blocking osc illator, provided only that the sweep amplitude E s is replaced by the
corresponding ampli:;ude Eco - E1 for the blocking
℃ scillator.
4-5. AMPLITUDE SELECTOR
A blocking
a即 lied
oscillator 己an
also be used as an amplitude selector. When the
voltage, which is increasing, reaches the reference voltage, a positive
一 5'7 一
n ：£ 學 It.. 第二期
pu] se reaches the grid and triggers the circuit. The output pulse can be taken
from the plate or grid, or across a cathcde resistance. The circuit arrangement
for accomplishing this purpose is shown in Fig. 22.
Reference
Voltage
Figure 22. Blocking Oscillator as Amplitude Selector
5. Theory and
D個ign
of Blocking-Os·cillator Transformers.
The pulse transformer acts as a coupling element in a blocking oscillator
circuit. The performance of the pulse transformer is closely related to the pulse
shape, duration and amplitude which are so much concerned in a blocking
oscillator. Since the number of factors affecting the response of a pulse transformer are too numerous, it is necessary to give a general survey about the
theory and design of pulse transformers used in connection with blocking
oscillators.
5-1 THE CIRCUIT EQUATIONS
The circuit equations introduced here are originally given by Benjamin, R.
published in the Journal of IEE Vol. 間， Part IIIA No. 7. With reference to the
circuit of Fig. 23 and using the definitions, notation and sign conventions
indicated theoron, let Va be measured with respect to B+ potential, and let Vg be
measured with respect to the cut-off bias of the tube when its plate is at B+
potential. Assume that C and Cg are very large and that the mutual inductance
孔11 is of such a sign that an increase in Va would tend to produce a decrease in
V. Let D==:d/dt. Then the following equations may be taken to specify the circuit
conditions:
Icm=CmD(V a - V)
Ica=C2DV
Ic1=-C1DVa
l g=CgD(V - V g)
Va=- L1Dh1 - MDiu= - RIR
V=L2DIL2 十 MDh1
Vg=lgRg
h1 + Ic1+ ll!=Icm +Ia
h1+Iea+IR=Icm
一 58 一
Blocking Oscillators
I .. +Ig=g ’血（Vg+Va ／µ’〉
g’血／μ’＝ r'a
La
C1
Ca
Ic2
v
-is
V Cut-off
Figure 23. Notation for Analysis
Whereµ ’ is constant and g ’m is a function of V g ang Va, but can, to a first
approximation, be considered constant. (Note that g ＇血， as defined here, is not
equal to the mutual conductance, but is the slope of a line joining the operating
point to the origin of the graph of cathode current versus (V g 十 Va/µ'). This
quantity is more nearly constant than the conventional mutual conductance. Also
Rg 至V g/lg, where V g is measured from cut off, not from
zero grid-cathode
voltage.
Eliminating first the currents and then the voltages between the above
equations leads to a 5th-order differential equation. Provided CgRg is rather
greater than the pulse length and M2 is not much smaller than L1L2 this equation
can be shown to be approximately equivalent to
T晶
1
1
I
1
叫百7 十 L1（ ~R－：－十~）一 M(g 血一 R; ）〕 D 十〔L1(C血十 C1）十 L2
一
1 ＋ 7-;)(C2+C
1
+2MCm]D2 十〔 C
R ：－旬，心＋〈τ
血〉〕（ L1L2-M2)D3
＋〔C血（ C1+C2）十 C1C2〕（ L1L2- M2)D4=0
(1)
Where the above expression operates on the voltage Va or V.
If L1L2=M2 (i.e. if the leakage inductance is negligible), and if L1=n2L2
if
n=step-down 酬。＝~~~﹔~﹔）
(iι
and hence M=nL2.
then：一
1
1
n+l
I
l+L2〔n2（頁一十~）＋~－ ng m〕 D+L2 ﹝（口十日2Cm+n2C1 十 C2〕 D2=0
Now put
and
1
1
1
n+l
-2一〔n2（五（＋王τ）＋ R-;-ng'm ]=g
(n+ 1)2C m +n2C 1 十 C 2""'C
In practice l < n<2.5, µ;>2.5 and
2.5
g’ m ：＞一百，
and both C1 a nd C2 are rather smaller than
Hence
n
•
gc2gm
一 59 戶
C 血﹒
(2)
n~ 學~第二期
and
C,,;,.(n+1)2Cm
Equation (2) can now be re-written as
_l_ __ 2 Ln 十 D2 = 0
(3)
LsC
C
The solution of this equation is
gt/Ce + i/(g2 /C2-l/L2C)t
gt/ce-1/ (g2 /C2-l/L2CH
(4-)
Va=A1e
+A2e
Where A1 and A2 are constants determined by the starting conditions. But Va=O
at time t=O, and therefore A1=-A2. Hence if g2/C 2>l/L2C. i. e . 汀，l(Lz/C)>l/ g
g2
1
(5再）
Va=-B1egt/C sinh v'（ 一 一一 －一－） t
C2 LsC
and if
g2/C2<1/L2C, i.e. if y(L2/C)<l/g
Va=-B2egt/C sin
v吋－~）t
(5b)
In either case, using the approxima te c onstant B, t he initial rate of fall of
plate voltage is
B
v'i 五三一」一︱
I
ca
(6)
LzC I
5-2- THE SIGNIFICANCE OF B JN EQUATIONS (5a). (5b) and (6).
It is proposed to show here that B is proportional to the rate of fall induced
at the plate by the trigger action, and that this initial slope has the effect of
making the blocking-cscillator operation start at a point a little way up the
exponential curve of the leading edge, and is approximately equivalent to a
trigger of smaller slope "cutting on ” the tube at an earlier instant.
Let the rate of fall of plate voltage due to the trigger be - r . Now this must
be equal to the initial value obtained from equation (6). Therefore
(7)
r=Bv /g2/C2-l/L2CI
But initially gm (and hence g) is very small and so r幸 B/y(L2C).
In equation (5b), the plate voltage is given by the product of an exponential
term and a sine, then it must change sign when the sinusoidal term passes through
zero. Hence the pulse length T, is equal to half the period of the sine wave. But
、 WU
U
rb
「
TL
。r
π7 ／
Therefore if
J
C
一
C京
T
的－L
the duration of a half-cycle of the term, and hence the pulse length, is
y(L2/C ） ~ 1/g ’
T 賓主πy(L2C)
l/y（山r 手
Substitute 平 for 1/y(L2C) in the equation 心／的LsC),
we get,
r 士 Brr/T
From equ ation (5a), if
咱哥l{.Lz/C)';;>1/g,
Va ＝一 ÷Bi(e2g1ι1)
一的一
Blocking Oscillators
dVa 一旦旦 Be2gtJc= 2旦 e2g(t ＋一旦logeB)/C
dt
C
一c
2g
Therefore the effect of B on the pulse waveform is approximately expressed
by the statement that the instant of triggering is advanced, as a result of an
induced rate of fa11, r, by an interval of the order of
主：：＿ (log. r-Iog. 立一〉
(8)
2gπ ’
But C/g is not likely to exceed 10-1 sec. Hence the slope of the trigger-wave
at the "cut-on ” point would have to alter by a factor of the order cf 7 (since
log. －挖去 7)
or more to make the resulting pulse ” jitter ’, by 0.1 microsec.
5-3. EFFECTS OF LEAKAGE INDUCTANCE
The above theory neglects effects of leakage inductance. This is permissible
in the design of blocking csci11ator tr盯isfcrmers for pulse length down to 0. 5
microsec. If shorter pulses are to be obtained, unless a special core of extremely
small dimensions is built for the purpose, the leakage inductance will become
a significant parameter. In that case the three terms of highest order, in the
differential equation (1), will tend to be dominant in the determination of the
behavior of the circuit. Consider, therefore, the terms in D2, D3 and D4 only.
This leads, after re-arrangement,
to ：一
1
1
C1月g 十 g'mCm 刊 C2+Cm)(-f「十一::r-)
〔 n2 十一一
扎• a
Cm(C1 十 C2）十 C1C2
D
(L1+L2+2M)Cm 十 L1C1+L2C
←一一一一五一一一一一 一← 主〕 D2步。
(L1L2-M勻〔Crn(C1 十 Ca ）十 C1C~j
(9)
Write this as
(9a)
(Da 十 2PD+Q2)D2.;0
Therefore
D主－ P± 》l(.P2-Q2 ）。r
D2=0
Now, in practical cases, unless the coupling coefficient of the transformer is
very small indeed, Q2~P2.
The solution of the equation, after substitution of boundary conditions, is
then
Va=-Be-Pt sin
〔Qt）帥
An additional term, varying linearly with time, would be mathematically admissible, but appears to have no practical significance.
If now L1=L2 =M (i.e. n=l and leakage inductance is small) and C1=C2三 mCm
then
L1L2- M2.;2Ml
(where 21 is t he total leakage inductance of the transformer).
P主旦旦旦±＿IL旦±＿（主立~~立B._±立~~al.
2Clm 十 2)
的
Qe:l/ v'OC1)
ω
This corre sponds 切，a burst of heavily-damped sinusoidal oscillation at a
very high freq uency, determined by leakage inductance and by stray c a pacitance.
一 61 一
n:a. 書版第二期
By suitably adjusting the value of the damping resistance, these damping oscillations can be made negligible after the first cycle, and so a single very short
pulse can be obtained.
5"4. SUBSTTUTJON OF PHYSICAL DJMENSIONS
Let
1t=pulse permeability of core
A=cross sectional area of core (cm2)
P=length of flux path in core (cm)
d=length of winding, measured along axis (cm)
k ＝二 effective
permittivity (dielectric constant) of the insulation
between the windings (see Note below)
A 二三 thickness of insulation layer (cm)
U =mean circumference of insulating cylinder (cm)
N2=number of turns on secondary
Note ﹔一 If
the added insulation layer had thickness d1> and permittivity k1
nnd the conductors (diameter D) are covered with an insulating sheath (generally
enamel) of thickness d2 and permittivity k2 then, provided d2<D
K」~~~豆豆〉
、三 μ
-A
E
一B
、、
J
，f 、3r、
z
-nu
N一叫
干i
LZ
一一
Then
針一唱一
~ k2dl 十 k1d2
-n
hua
v
Ycdw
TA
的
The capacitance between coaxial cylinders replacing the windings is
kUd
C~ ＝ 47l"60.9 × 1012farads
(14}
Now the energy stored in a condenser varies as the square of the voltage.
Therefore, if, in the above condenser, the voltage, instead of being constant
value, v (where v is the grid-plate voltage ] varies linearly with distance x,
along the axis of the coil, from αV to βV, the energy is
fβv
1
（←一 Co V2)(77.:)τ~I
（β 一 α ）
2
J av Vx2dv ]
=C--}-co V2）［÷（a2叫十的〕
＝（古巴Co V2) ｛÷的十β）2-afi〕﹜
If the transformer is so connected that the grid and plate voltage change in
opposite directions, and if both windings are in the same sense (i.e. both right
-handed or both left-handed spirals), this
C _ C.,
曲，一--y
yields：一
(n 十 1)2-n
(n 十 l)a
(15a)
and if they are in opposite senses,
C m=Cc/3 〔是 C ／《 n+l)2〕
(15b)
These two cas es are illustrated in Fig. 24 (a) and (b).
一位一
Blocking Oscillators
主1
。
。
(a)
(b)
±1
Figure 24.
As n 詣， in practice, not less than unity, the latter formula is always a good
approximation and will, hereafter, be exclusively employed. Thus, by equations
帥， ~and (15b）﹔一
N.2KUd
L.Cm-=道且；2. 76~ （"p/µA)l021
。。
and by equation T=1q/(L2C)
T ’= Ny'
KUd
A’言（ p/µA)
’
× 1.2 × 10-a•
開
’
where T is expressed in rnicrosec and A is expressed in mils ; N is the total
number of turns.
The leakage inductance can easily be estimated by equating the magnetic
energy stored in the cylindrical inter-winding space to that stored in an inductan凹，但a.
in series with winding number 2. This
yields：一
以暫且…
ω
5-5. DESIGN FORMULAE
Equation ~7) above neglects the damping of the oscillation by the tube, , i.e. it
corresponds to equations (5a) and (Sb). Hence the true pulse length is greater
than indicated by that formula. Experimeffl; shows that the pulse length is increased by about 40?6 leading
tc ﹔一
T’~主主主 I lS_立d_
- 1000
/_\’ S
09)
v
where S ＝~（p/µA ） ιreluctan~e.
The condition for starting exponentially is [ by equation (5a )]
that
v’( L2/C)>l/g
or, if n=l
y'(La/C) > 2/g 間，’， where g'm’，：：2g/n2,;,;g 血，
If the pulse-length is to be calculable, exponential operation should merge into
the sinusoidal type before the tube is driven too near to the condition of zero
~slope ’， mutua l conductance (at equal grid and plate potentials); there fo rs make
掛
R’皂、／（ L2/C)
Usually R，司S:8/gm
impedance of the plate load should be kept sufficiently large for t he current
一的一
'1:1. 夸 :Jl,.. 第二期
available to develope the required voltage.
Bt此， if
up to
n==l,
T 辛9i/(L2C ）〔by
equation T .S:π恥／（ L2C) and the
stateme 的 leading
equation 制J,
Combining this with equation (26) leads to:
Now, as
。n
LzcR'T/9
V2=N2A dB/dtx 10-s
N2=108 V2T/ AB
Where V2 is the amplitude cf a f'QU['.re pulse cf
flu 玄』 cend:y
and B is the highest
(in
gox~s),
(22)
vcJ －~age
ori winding number 2,
ccrerspιnding
to V2, reached in
time T.
Therefore
R'T
V22T2
T
--L2 －一 ----wAP－一
一一．一
2(8而× iυ－7)
by equations (21),
車司
and (22).
Therefore the magnetic energy stored in the core at the end of the pulse
32 × Ap × 1。一1
joules 一
V22T
(23)
一（ l/4.5)R’
王Ien…ny shu叫 resi
dissipate a relatively small portion of the total pulse energy and will not seriotl .
sly affect circuit operation during the generation of the pulse. Equations (22) and
帥 can
be combined to give:
’
B
R
V2 ''µll
-V
Ns
P
但也
一（一－
Now a satisfac·tory value of R’, for most cases, is 2800 ohms. Hence, if n=l,
equation (24) leads to ：一
B
250
N
于一－ x 一一~＝一土
V1
µ
p
.
Note that, if
L臼 N2A/p
(2j)
and docp, then as
U品加l(A)
Cmcc pi/(A)/ L0,,
But the leakage inductance, 11cc N2-i/(A)l'jp
Therefore
lee Lp/Cmcc Lp/C
For a given L and C, therefore, the leakage inductance is least in the transformer
of smallest overall dimensions.
5-6. DESIGN PROCEDURE
(1)
Choosethe smallest convenient core.
(2) Use equation (2.J) to find N, choosing B for absence of magnetic saturation.
(8) Use eq uation 個 to find insulation required.
Blocking oscillators designed along these lines, for pulse-lengths between 0.5
and 10 microseconds will give satisfactory service .
一缸，－
Blocking OsdHators
聞款才辰盪器之前討
郭孟堅＊
過要
本文編寫之目的，係就真空管間歇振盪器之作用原理，電路分析、主要﹔直用、以及設計重點等作
一般性主研討。此項振聾譚在草視體系、雷達設施及蓮子儀軍裝置中，均有其特珠用途。目前雖已有
多種書刊論著，致力於此項振盪器之車路分析及作用研討﹔但由於其毫路複雜，故向無一種著述獲得
完全滿意之結果。鼓符其主要直難前在，概述如火：
一、電路內 t是有多種儲龍類型發生。
二、電路內包括多種主附帶因素﹔此項因素對於轉換琦謂有直接影響，亦即壘的是服波之下極限有
限制作用。
三、真空管運用於通常不加利用主區域，在該區織內，電路參數較運用投徵弱﹝言號哼，有贖著之
變化，甚至放大因素，屏咀及棚陰間之內阻，均屬無法預知。
四、由於磁通建立以至攝芯飽和等關係，致使定持辜！盡量且是為磁化遺流及服浪寬度之函數。
鑒於上述主困難，本文對於此項草路各種類型主精確分昕，未加介紹，僅對投正是型間歇振聾單主
電路近以計算加以研討。在憲用方面則注重於通常使用之各主要草雄主分肝說1月﹔在議請方高珀介紹
特關重要之脈波變壓器之計算方式與設計程序。
本文之參致文獻名稱、著者及出版處等，均組詳列於緒論之後。
＊省立中興大學應用數學系副教授。
一的一