Vous êtes sur la page 1sur 4

in addition to the power-factor control

is indicated. In some cases it is possible


that a current relay can be used to pro-
History an eve opment o t e
vide "cut off" with reasonable accuracy
and little danger of hunting or pumping
action of the power-factor control.
ectronic ower onverter
E. F. W. ALEXANDERSON E. L. PHILLIPI
Appendix. Determination of FELLOW AIEE NONMEMBER AIEE
Optimum Unswitched Capacitor
Kilovolt-Amperes to Give Mini- THE TERM "electronic power
verter" needs some definition. The
con- duced at the elbows of the side arms and
Circuit Losses connected to the control circuit. Direct
mum Circuit Losses object may be to convert power from current power of 9,000 volts was in this
direct current to alternating current for way successfully converted into alternat-
J. W. BUTLER d-c power transmission, or to convert ing current and then again into low-
MEMBER AIEE power from one frequency into another, voltage direct current. A new electrical
or to serve as a commutator for operat- industry was immediately foreseen where
Figure 7 shows the variable portion of the ing an a-c motor at variable speed, or for power would be transmitted long dis-
daily kilovar load pattern of a circuit. Let transforming high-voltage direct current tances by high-voltage direct current and
l, represent the unswitched capacitor into low-voltage direct current. Other railroads would be electrified by locomo-
objectives may be mentioned. It is tives carrying electronic converters from
thus evidently not the objective but the a high-voltage d-c trolley to low-voltage
|0
IO 7- .-I f(t)
means which characterizes the electronic
power converter. Other names have
d-c motors. After another 20 years this
new industry is now well under way on a
been used tentatively but have not been modest scale but comprising installations
o ¢ _ , ________X__ accepted. The emphasis is on electronic up to 20,000 kw.
means and the term is limited to con-
TIME- T version of power as distinguished from Evolution of the Electronic Converter
Figure 7. Daily kilovar load pattern of a electric energy for purposes of communi-
cation. Thus the name is a definition. The first tests of the inverter were
circuit made with a mercury-pool tube and the
History latest installations have mercury-pool
kilovars or current. The square of the rms cathodes. Between these points there is,
corrected current in the circuit is given by Within the scope of the definition a however, a significant development. The
mercury-arc rectifier is an electronic structure of the tube itself can assume a
(Ihm)2 = -TJ0f {Iof(t) -Ic dt= power converter. From this point of number of successful forms, such as the
view we must trace its history back more thyratron, the ignitron, and the excitron.
1 rT 2 rT than 40 years. In the first 20 years of But all these forms of tubes have certain
IC2+ T Iof(t) }2dt TIcio/
- f(t)dt this period the development was rather characteristics in common which force
Jo slow. An important discovery however the electrical engineer to accept a new
(1) was made by Langmuir in 1913 when he point of view in the methods of its ap-
Differentiate equation 1 with respect to found that a negative electric potential plication. This has perhaps been the
I,, equate to zero, and solve for value of IC introduced between the cathode and reason why the practical development
to minimize the loss, and obtain anode prevented the start of current has up to the present been rather slow.
flow through the anode. There has been an unwillingness to face
d(Irms)2 2I - f(t)dt 0 (2) At this time the initial development the facts or what so far appear to be the
dIf T 0 took place of the high-vacuum radio tube facts, that high-voltage tubes are subject
or which was an evolution of the deForest to momentary faults known as arc backs
audion. These tubes were soon used and loss of grid control. The hope that
Io T successfully at high voltages to convert the tubes could be so improved that these
IC j1f(t)dt (3) direct current into alternating current of faults could be eliminated has so far been
audio and radio frequencies and serious wishful thinking. But a much more con-
fT thought was given to their use for power structive point of view has been adopted
oJ f (t)dt =area of the load pattern, conversion. The high internal voltage which accepts the electronic faults as
drop was however a serious obstacle. inevitable while measures are taken which
andwhenrdvidedby Tminusothe
when divided by T
and
width, the
minusthewidth,t Attention therefore returned to the mer- render the faults perfectly harmless so
any kilovar pattern, the minidum circuit cury arc and ten years after Langmuir's long as the incidence of the faults is kept
loss with a fixed kilovarsupply obtained
is discovery, tests were made which proved within reason. Installations of d-c power
by using a value of I, equal to the average that the mercury-arc rectifier could be transmission and electronic frequency
value of the requirement. inverted so as to convert direct current changers are at present operated on this
into alternating current. The tests were principle with great success. Electronic
made with a standard high-voltage recti-
Reference fier of the type used for series arc lighting.__,;_..
It had a bulbous head and two upturned ,I
1. THE EFFECT OF KILOVAlI SUPPLY ON THE side arms. Carbon electrodes were intro -- -)A - LINK OR ) )L
DESIGN OP SYSTEMS FOR LOAD GROWTH, T. W. Paper 44-143, recommended by the AIEE commit TRANSMISSION
Schroeder, J. W. Butler, N. H. Meyers. ELEC- tee on electronics for presentation at the AIEE LINE
TRICAL ENGINEERING, volume 63, February 1944, summer technical meeting, St. Louis, Mo., June 26- ,,
--
pages 69-76. -30, 1944. Manuscript submitted April 25, 1944;j
i)e-i-
-- made available for printing May 18, 1944. I __!r__
J. W. BUTLER is sponsor engineer, central-station en-
gineering department, General Electric Company,
E. F. W. ALEXANDERSON and E. L.
with the General Electric Company,
PHILLIPI are Fiur 1.Eetoi ovetr ulcn
Schenectady,Fgr .Eetoi cnetr
Schenectady, N. Y. N. Y. version ulcntype

654 TRANSACTIONS A lexarnderson, Phillipi-Electronic Converter ELECTRICAL ENGINEERING


Figure 2a. Wave shapes of single-phase Figure 2b. Wave shape of six-phase inverter Figure 2c. Wave shape of six-phase inverter
inverter with low a-c reactance with high a-c reactance
faults occur several times a day. They inverter. By change of the grid controls, If, on the other hand, the ionization left
can, however, be observed only by deli- the power flow can be reversed. The rec- over from the arc back causes the grid to
cate instruments and do not in any way tifier and the inverter are however not lose control in the next half cycle the re-
interfere with power service. independent devices like a generator and sult is a shoot-through. This occurrence
a motor but function together as a whole is, however, more the exception than the
Suppression of Electronic Faults -the electronic converter. Electronic rule. It seems to be one of those se-
faults which occur in the inverter are quences of events characteristic of elec-
An electronic fault may or may not suppressed by the rectifier. The recti- tronic phenomena which follow a law of
result in a short circuit, depending upon fier suppresses its own faults but even chance over which we have no control.
what means are used to suppress it. In this statement must be accepted with But as engineers, we are not much dis-
low-voltage rectifier installations the some qualifications. The rectifier can turbed about this so long as we can avoid
faults occur so seldom that it has proved suppress its own faults because the in- any harmful consequences.
satisfactory to disconnect the fault by a verter cannot feed current back into the Besides arc backs there are two other
high-speed circuit breakler. In high- rectifier but if several rectifiers feed the causes which may initiate shoot-throughs
voltage installations the faults occur same circuit in multiple they can feed in an inverter. These are not electronic
so often that this would not be permis- current into each other's faults and there- faults in the same sense as arc backs, and
sible because power delivery must be con- fore fault suppression depends upon inter- there is a fair chance that they may be
tinuous. In the original installation of action between the several rectifiers. entirely eliminated. One of these causes
d-c power transmission for the New York The oscillograms of Figure 2 show a which has been observed in our earliest
Power and Light Corporation between typical performance of a single-phase and installations is voltage transients of very
MechanievilleandSchenectady,thissitua- a six-phase inverter. The rather com- short duration which make the grid
tion was first met by the use of constant plicated wave shapes of the six-phase momentarily positive at a time when it
current as described in a paper in 1935 inverter can be easier understood if the should be negative. Faults from this
by Willis, Bedford, and Elder. In a con- records of the single-phase inverter are cause have been eliminated by the use of
stant-current system a momentary short examined first. The three recorded suitable filters. A third cause of faults
circuit is of no significance. Such a sys- curves represent the wave shapes of the is the occasional failure of an igniter to
tem is therefore adapted to the character- current in the a-c line, the voltage be- fire. This difficulty has not yet been
istics of the electronic converter. Opera- tween a-c lines, and the voltage between entirely overcome but the rate of progress
tion at constant current was successfully cathode and anode of one tube. The com- in this technique promises that it will
continued for four years but when plans mutation period is marked by the abrupt eventually be entirely eliminated.
were made for future installations it was depression of the alternating voltage While thus some causes for faults are
found that the constant-current system near the end of each half cycle. The brief gradually being removed we must face
was not well adapted to existing practice recovery of the voltage before it passes the conclusion that shoot-throughs in
of power distribution. The reactance- through zero is the margin of commuta- high-voltage inverters will occur from
capacity networks used for conversion tion. This same margin is also indicated time to time and we must be prepared to
between constant current and constant by the brief negative swing of the voltage meet this situation in a practical way.
potential were therefore removed and between cathode and anode. If the load From this starting point there are
the system has been operating at con- is increased to the point where this mar- several possibilities. The inverter may
stant potential for several years. In this gin is used up, the inverter goes into be adjusted so as to work with a good
connection new means were developed short circuit. With this in mind we can margin of commutation and the power
for suppressing the electronic faults. examine the six-phase wave shapes and supply may have a short-circuit current
These methods have proved entirely suc- interpret them as the composite of three that does not greatly exceed the normal
cessful and have been adopted in later single-phase wave shapes. Each phase load current. In that case the inverter
installations. The sequence of events wave shows its own commutation period may have enough margin to commutate
is illustrated by oscillograms taken dur- and also secondary reactions from the the increased current and the fault
ing practical operation. Samples of commutation of the other phases. clears itself. It is then difficult to dis-
these oscillograms are shown herewith. The electronic fault may be of the cover that it has occurred and the opera-
Figure 1 shows a simplified diagram of nature of an arc back or a loss of grid tor would seldom notice it. Faults which
an electronic converter of this type used control known as a "shoot-through." clear themselves are characteristic of
for d-c transmission and frequency con- A simple arc-back has so little effect upon circuits containing several rectifier and
version. It should be noted that the the operation of an inverter that it is inverter units in scries. It was the nor-
power circuits of the rectifier and the difficult to discover whether it has taken mal performance of a 24-phase develop-
inverter are identical and that it is the place. It wipes out the negative voltage mental set tested a number of years ago
grid control which makes one part of the that marks the commutation margin with four rectifier units and four inverter
whole into a rectifier and the other into an but this in itself causes no disturbance. units in series. This is worthy of note

SEPTEMBER 1944, VOLUME 63 Alexanderson, Phillipi-Electronic Converter TRANSACTIONS 655


___________________

-3.
SW 4,h. .,~~~~~~~~~M
t - A,

A. Fault involving opposite inverter tubes in B. Fault including two legs of one inverter C. Fault including two legs of one inverter
both inverters. Both coils of d-c reactor in. transformer. Both coils of d-c reactor in. transformer and two legs of one rectifier
Bottom trace is secondary current of current Bottom trace is secondary current of current transformer. Both coils of d-c reactor in.
transformer in direct current transformer in direct current Bottom trace is secondary current of current
. ~~~~~~~~ Figure 3. Oscillograms of inverter Faults. transformer in direct current
Arrows indicate made Fault-igniter miss
tubeX changer used with the thyratron motor
VVVVVV %~~~~ ~ in one tube does not disturb the sum of described in a paper by Alexanderson and
electromotive forces in the system enough Mittagin 1935. Thiselectronicconverter
~~1I~~vvvy~~~~Tw~to snterfere with normal operation. has not yet been given a name by which
~*Returning to the sequence of events it can be readily identified. It is pro-
that may follow an electronic fault we posed here that it be known as the single-
may assume that in some cases the cur- conversion frequency changer. The
rent which follows the fault is so highinvted by electronic
cgthat the inverter cannot commutate it. only once whereas in the dual-conversion
s is the case in the d-c power trans- type of frequency changer the power is
msssion from Mechanicville -to Schenec- first converted into direct current and
tady and in the frequency changer de- then back to alte lating current. Both
cass it ie cnecessary to clear the fault by synchronous tie between two a-c systems.
ctntrol oftthe rectifer. The oscllograms Besides this the single-conversion type
of Figure 3 show such a typical sequencer has many desirable characteristics, such
The sudden rise in current acts upon a as being adapted to feed a stub-end load.
fast relay controlling the igniter and grid The motor for which it was originallv
- ~~circuit, osqenl current which
h

has been built up in the power circuit is


developed is a stub-end load and start-s
from standstill with a good torque like a
no longer maintained by any electromo- d-c motor. The single motor may be re-
tive force and decays at a rate dependent placed by a synchronous condenser and
I.

~a ~~~~~~~~~~on
the time constant of the circuit. It is
important that the rectifier should not
an independent distribution systemn in-
cluding a multiplicity of motors.
D. Wave shapes of phase voltages and phase go into action again until the current has The problem of suppressing electronic
currents during faults decayed to a value which the inverter can
commutate. A time relay delays the
faults is solved in this type of electronic
converter in a different way. When for
when we contemplate d-c power trans- action accordingly. In about one-fifth of a one reason or another a loss of grid
mission at high voltages. Such installa- second the normal operation is resumed, control occurs in one of the tubes the
tions will necessarily have many units in The Single-.Conversion Frequency resulting disturbance lasts only one-half
series and they can be so designed that Changer cycle because each tube is a rectifier as
electronic faults will clear themselves, well as an inverter and the rectifier func-
Each unit in such a system operates at Another type of electronic converter is tion commutates the fault current out of
constant potential but an electronic fault shown in Figure 4. It is the frequency the tube where the fault occurred. The

656 TRANsACrioNs Akexauderson, Phillipi-Ekctronic Conrverter ELECTRICAL ENGINEERING


Figure 4 (left).
Single - conversion-
lightning may result in circuit faults.
Whether such faults can be counted on to
type Frequency
changer
clear themselves automatically depends
upon the characteristics and stability of
the systems that are involved. An elec-
tronic converter has a definite load limit
like an inductance motor or a synchro-
Figure 5 (below). nous motor. In the grid control we have,
D-c transFormer however, excellent means for regulating
the system so that those load limits are
.AAARAAAt-..AASA,,\I Low- never exceeded. Such regulating systems
practical result is that this type of elec- voLTAGE VOLTAGE have been incorporated in the instala-
tronic converter is entirely free from the ~MISSIOsN CIRCUIT tions now in operation as described in
problem known as shoot-through in the companion papers.
simple inverter. Thus it will continue to It is important that the application
function with slightly reduced power engineer should understand the character-
even if several tubes are disconnected, istics and limitations of the several types
whereas the simple inverter becomes in- of electronic converter.
operative if a single tube fails to function. The rectifier-inverter combination may
be described as a frequency changer or a
The D-C Transformer d-c power transmission depending upon
whether they are located in the same
One of the first objectives in the devel- station or far apart and joined by a d-c
opment of the inverter was to provide a transmission line. The Mechanieville-
way for transforming direct current from Schenectady installation is a frequency
high voltage to low voltage. D-c trans- changer as well as a transmission because
formation has not yet found application the power is generated at 40 cycles and
in any large scale. But the experience delivered at 60 cycles. Similar installa-
gained in d-c power transmission indi- tions may fill a demand in the future as
cates that it is from a technical point of Figure 6a. Wave shapes oF inverter in d-c nonsynchronous links between 60-cycle
view entirely practical. The electronic transFormer systems which are not operated in syn-
industry is waiting for a demand where chronism. This type of installation is
this experience can be applied. The primarily adapted to exchange of power
questions that are likely to be asked are: between independent systems. In its
How will it start? Will it require moving original form, it is not adapted to start
machinery? How will it ride through a dead load. When needed, however,
faults? suitable arrangements will undoubtedly
In order to give a practical answer to be made. An electronic d-c transformer
these questions, tests have been made rep- fed from a d-c line may be used to start
presented by the diagram, Figure 5, and a dead load as shown on the accompany-
oscillograms, Figure 6. The power source ing oscillograms. The power thus made
is assumed to be a rectifier station like the available may be used to start a syn-
one used in the d-c transmission from chronous condenser so that it can be fed
Mechanicville to Schenectady. The re- from another electronic converter. This
ceiving station has no moving parts. It will serve to illustrate many possibilities
has a step-down transformer from the of using the electronic power converter in
high-voltage inverter to the low-voltage Figure 6b. Wave shapes oF rectiRer in d-c one form or another. It can safely be
rectifier. A capacitor is in shunt with the transformer said that with the installations of the
transformer. The load is a d-c motor. electronic power converter already made,
One oscillogram shows how the operation this new tool of the electrical industry
is initiated from a dead start by applica- has come out of its infancy.
tion of the grid control to the high-voltage
inverter. A second oscillogram shows the References
characteristic wave shapes. 1. THB DIRECT-CURRENT TRANSFORMER UTILIZ-
ING THYNATON TUBES, D. C. Prince. General ERkc-
Adaptability of the Electronic ric Review, volume 31, pages 347-50.
Converter Converter ~~~~~~~~~~~~~~2.
HOT-CATHODE TEYRATONS,
General Electric Review, volume
A. W. Bull.
32, number 4,
pages 213-23; number 7, pages 390-9.
Three types of electronic converters 3. CONTROL OF AN ARC DISCHARGE BY MEANS OF
have sodiscussed
have so far been farbeendiscussed and it has been ~~~~~~AGRID, A. W. Hull, 1. Langmuir. Proceedings of the
National Academy of Science, March 1929.
shown how means have been provided for C
suppressing electronic faults. In practical Brown.GRnTralElecD RCUYARC RECTIFIeRS,
H. ME

operation there are also apt to be circuit Figure 6c. D-c transformer starting a motor pages 439-44.
faults. The d-c transmission line is for load 5. GRID-CONTROLLED RBCTIFIKRS AND INVERTERS,
instance sometimes struck by lightning. C. C. Herskind. AIEE TRANSACTIONS,
9
volume 53,
The short circuit following a lightning 6. Te MOtoR,
p
. .
stroke is cleared by the control system the insulators. In the design of the elec- A. H. Mittag. AIEE TRANSACTIONS, volume 53,
which is provided for suppressing elec- tronic converter a virtue has thus been 1934, November section, pages 1517-23.
tronic faults. The transmission line is made of a necessity and a real improve- 7. CONSTANT-CURRENT D-C TRANSMISSION, C. H.
therefore immune to lightning strokes so ment in practical reliability has resulted. Willis, B. D. Bedford, F. R. Elder. AIER TRANS-
ACTIONS, volume 54, 1935, January section, pages
long as permanent damage is not done to Disturbances from other causes than 102-08.
SEPTEMBER 1944, VOLUME 53 Alexanderson, Phillipi-Electronic Converter TRANSACTIONS 657

Vous aimerez peut-être aussi