Solid-state switching foraircraft electric systemsThe entire concept of aircraft electric systemsmust be revised when designing for solid-state control.Not only must the design protect the wiring,it must protect itself and operate with high efficiency

Lee D. Dickey, Clyde M. Jones LTV Aerospace Corporation

An aircraft electric system designed around a solid- aircraft electric systems that gather and display informa-state approach offers many potential advantages over tion and distribute the flow of power to equipment. Thean electromechanical counterpart. These include application of such technology to aircraft electric-dis-higher reliability, longer life, more ruggedness, more tribution systems, however, has been slow to receive ap-versatility, and better compatibility. To achieve these probation, because of the applications approaches thatadvantages, however, it is not merely sufficient to generally have been made to date.replace electromechanical components with their Figure I is a block diagram of an aircraft electricsolid-state equivalents. Here is one approach that has system showing the area of primary concern of thisbeen tested in the laboratory and in aircraft during article. Figure 2, representative of present-day electricthe past ten years-with success. Included is a re- systems, is included to highlight the fact that logic func-port on an advanced system undergoing evaluation. tions are performed at the power level. Such present-day

systems are inherently heamy, require complex harnessThe trend in advanced high-performance aircraft is to routing, and are growing larger and heavier with each

more sophisticated missions, which require more complex new requirement. These systems also require large quanti-electric and electronic systems. This complexity hascreated an exponentially increasing quantity of switches,circuit breakers, and wires. Moreover, complex digitalavionics systems require clean (transient-free) electric FIGURE 1. Block diagram of an aircraft's electric system.power-emphasizing a need for electromagnetic and inter-face compatibility between signal systems and digitalcomputers.The increase in complexity and compatibility require-

ments cannot be compensated or fully met by existingelectromechanical devices that are simply the product of(two decades of) miniaturization. This miniaturization,without a comparable reduction in system voltage, hasresulted in such high electrical-stress levels that misappli-cation of the devices has become all too easy. Major ad-vances must therefore be made to optimize future aircraftsystems-and these advances cannot be achieved by usingconventional approaches and techniques.A new approach is necessary-one that will simplify

circuitry, comoine functions, separate signal- and power-switching requirements, and eliminate unreliable featuresto control switching and protect circuits and devices. developfent

Taking a cue from other applicationsVery rapid advances have been made in the electronics

field, especially in digital computers, through the develop-ment and application of solid-state switching techniquesthat utilize semiconductor devices. It is evident that suc-

cess might be achieved by applying like techniques to

IEEE spectrlirn NOVEMBER 1970 73

ties of power-handling, electromechanical devices that and small size. If properly applied, they could provideare subject to wear, create large voltage transients, have the urgently needed improvements in aircraft electric-limited cycle life, and have low cycling rates. Besides, distribution systems.they do not lend themselves to automated checkout and The number of solid-state components that can be con-thus create complex checkout-maintenance problems. nected in series while complying with MIL-W-5088, orA number of solid-state devices (relays and circuit allowing high power-control efficiency. is limited by the

breakers) have been placed on the market with the idea of voltage drop through the solid-state devices (typicallyproviding a solid-state device to replace an electromechan- 0.5 to 1.5 volts per device). Ordinarily, aircraft power-ical one on a part-for-part basis. This tactic has some control circuits require several relay contacts in series.obvious drawbacks considering the voltage level at which The use of solid-state relays in such circuits, therefore, islogic functions are performed in most aircraft electric not permissible. Nor is it practicable to combine solid-control circuits. Because of the voltage-drop and power- state and electromechanical components since electro-dissipation characteristics of semiconductor devices (even mechanical circuit breakers cannot protect solid-statewhen operated in deep saturation), the approach pro- switches. (Admittedly, the purpose of circuit breakersduces a very ineflicient system and makes it impossible to in present-day airplane electric systems primarily is tomeet the voltage-drop requirements of MIL-W-5088 protect the wiring; but solid-state relays, flashers, timers,(Aircraft Electric Distribution System Specification) in and other such components are more likely to be de-many circuits. It is possibly for this reason that many de- stroyed as a result of overload or fault condition thansigners have concluded that solid-state switching for their electromechanical counterparts when operatingpower-distribution systems is not practical. Were high- under the same conditions.)voltage drop an insurmountable impediment, their con-clusion would be justified. However, semiconductor A new approachdexices have the advantages of high reliability, long life, The overall effort to develop an advanced electric sys-

tem for aircraft has been given several names, amongwhich are: contactless switching, solid-state switching,and SOSTEL (solid-state electric logic). All represent an

___ @application of semiconductor technology to the manage-

Fr A ment and control of aircraft electric systems. Our studiescentered on separating power switching from signal

- T T switching, and on switching power through a minimumT___ __Bus_T_ number of semiconductor devices. This procedure mini-

Circuit n\ K mizes the voltage drop between the source of power andprotection I the equipment supplied. More important, it provides

Manual switch .- efficient power control since power dissipation is held to/ Manual in cockpit a minimum. (Also, low power dissipation necessitates

switchless sinking and, therefore, reduces size and weight re-Flap positionswitch in wing quirements.) In addition, the separation of power from

Generator signal switching provides high system efficiency and makes/ Landing gear position it possible to meet the voltage-drop requirements of

switch, etc.MIL-W-5088. All factors considered, the separation

L<I1 concept makes maximum use of semiconductors' ad-ELg vantages and minimizes their disadvantages.

FIGURE 2. Present-day electric system. The solid-state electric system shown in Fig. 3 repre-sents a basic model of its sort and is composed of threebuilding blocks: (1) the digital signal sources (such as

FIGURE 3. Solid-state electric-system concept. temperature and pressure transducers); (2) the controllogic (including bus monitoring and built-in test); and

Signal Control Po.ier Switchiing (3) power controllers (bus-switching and load controllers).Sources Authority S The source signals are fed into the control-logic unit.

-t There, they are correlated in a prescribed manner to

Bus generate a signal that controls the power controllers.>I1; controller The logic switching is performed by standard integrated-

circuit NAND/NOR gates that provide maximum reliability> | __ACbu s with minimum space and weight.

The fanout capability ordinarily providedb multipoleswitches and relays in an electromechanical system isnow provided by integrated circuits and is performed at

p the signal rather than the power level. This techniqueconsiderably reduces the number of wires needed togather intelligence from the various controlling functions.And since the intelligence is gathered at the signal level,

;ffigL egend: what wire there is may be sized small. Power flow from

P <1 -sgaevel source to bus to load is controlled by power controllers.,4 f -Power level This sort of system is so flexible that a designer can

incorporate an automratic bus-monitoring system to

74 IrTE spectruni NOVEMIIWR 1970

I. Comparison of electric systems (A-7 aircraft)System Reduc- ~

Conven- Solid tion,tional State percentb

Weight, kg 262 143 4.Volume, 10' cm, 45.3 24.6 45 q 4 1 CReliabilitysource a Ic(faelureS/106 h) 498 116 77p.wet al emer-

MaintainabilityMan-hours per

flight-hour 0.014 0.0028 80..

provide optimum loading of the power source. And byassigning load priorities, it becomes possible to operate an FIGU RE 4. BITE control panel.emergency source at optimum capacity. Although theemergency source must first supply power to all emer-gency loads, automatic bus monitoring permits theplane's pilot to operate lesser priority loads during periods of each signal source through the multiplexed data-trans-when emergency loads are not "on call." The system also mission line.is amenable-unlike electromechanical systems-to ready TheMCU decodes the received data, solves the Booleancheckout and a technique was developed to perform pre- switching equations according to instructions perma-flight go/no-go tests on the low-level-voltage portions of nently stored within a nondestructible, read-only memory,the system. (See box.) and then properly transmits coded output data over a

multiplexed data transmission line to the addressed re-Advanced control logic mote output terminal. At each remote output terminal,As previously noted, Fig. 3 represents a basic system the data are decoded and routed to the particular channel

of its sort. A more sophisticated control system-one addressed by the MCU. The output of this channel is thenthat is now under investigation-uses multiplexing tech- "hard-wired" to a specific bus, load, or indicator powerniques and consists of the components shown in Fig. 5. controller. The output signal remains constant-eitherThe control-signal-sensing and the power-switching ap- high or low-until the next updating by the MCU.proach used in Fig. 3 remain essentially unchanged. The MCU obviously is the heart of the data-handlingChange occurs in the method of handling and processing system and controls the entire data-handling process.the control data. Under this more advanced system Changes in its control logic can be accomplished by re-concept, remote, multiplexed input terminals are located programming the MCU with instructions from paper orin proximity to large groups of signal sources and thus magnetic tape, thereby eliminating the need to makeprovide a significant reduction in the length of the wires wiring changes in the aircraft. (For reliability, two inter-that gather control information. These terminals monitor changeable MCUs can be used per system-one opera-each signal source and, upon command from the master tional, the other standby. Parity check and monitor cir-control unit (MCU), code and serially transmit the status cuits within the MCU provide a continuous built-in test.)

Go/no-go testThis go/no-go test is intended to check out the will be off at the conclusion of each test. A digital

low-level portions of the solid-state system. Bus indicator identifies the faulty signal source duringcontrollers and power controllers are not checked the "short" and "open" tests. It also pinpoints theand are actually inhibited while the test is in opera- faulty input-buffer card during the input-buffertion. (They, however, are checked as part of the tests.subsystem functional checks.) The logic test checks the logic circuits throughThe test is divided into four parts: logic, input all possible input combinations to determine

buffer, signal-source short, and signal-source open whether they will perform when required. Theas indicated by four positions of the rotary selector input-buffer test checks the input buffers to deter-switch on the built-in test-equipment (BITE) control mine whether a signal will pass through them. Thepanel shown in Fig. 4. When the control switch is signal-source-short test checks for short circuitsin the FLY position, the BITE circuits are disabled between the signal leads and the power and groundand the bus-power and load-power controllers are leads, both in the signal sources and the logic unit.enabled. Before each test is performed, the reset The signal-source-open test checks the signalbutton is depressed. At the start of each test, both source's output circuits. The built-in test circuitsthe GO and NO-GO lights illuminate; one of them should be packaged as a part of the logic.

Dickey, Jones-Solid-state switching for aircraft electric systems

Signal Data Handling Power SwitchingSources (multiplexing and

programmable logic)

transmission lines

~~~~~~~Legend: # lr

_ISignal level 1 .X F0 SgaXlPower level

FIGURE 5. A proposed advanced solid-state electric system.

The remote terminals --in their ability to connect with line can operate either shorted or open. Under this con-all circuits-are totally redundant. For our particular cept, one transmission line can be totally disabled, anddesign, the remote terminals have been optimized at 64 the remaining line either open or shorted, and the systemchannels each. The package configuration and the cir- will still operate.cuitry of the remote input terminals are identical; thus The signal sources and control logic can be imple-these units are interchangeable. The same standardization mented with minimal development effort using the solid-and interchangeability exist for the output terminals. state circuitry generally in use today. Requirements forThe multiplexed data transmission line, indicated in the power controllers, however, are many times more

Fig. 5, consists of two pairs of twisted, shielded wires, severe and make the job of applying semiconductoreach forming a closed loop. The transmission lines are technology to electric circuit control a difficult one.redundant and are best routed separately from each Therefore, the remainder of this article dwells on theother to improve system survivability. Each transmission requirements and characteristics of power controllers.

FIGURE6.Transient-surge, dc-voltagestep-function locus FIGURE 7. Transient-surge, ac-voltage step-functionlimits for category A (per MIL-STD-704A) equipment. The locuses for same category equipment pertinent to Fig. 6.abbreviations ESSL, NSSL, and ASSL in the illustration Abbreviations have same meanings as in Fig. 6.denote, respectively, emergency steady-state limits, nor-mal steady-state limits, and abnormal steady-state limits. 20

90 Ci)80 140

70 - 12 0 ||7050; EL80i4060>50~~~~~~~~~~~~~~~

0 40~~ ~ ~ ~ ~ ~ ~ ~ 4

100LV500 0.01 0.02 0.05 01 0.2 0.5 1 2 5 10

500 0.01 0.02 0.05 0.1 0.2 0.5 1 2 5 10 VsVus Time, seconds Time, seconds

76 IE1rE spectrUmn NOVEMBER 1970

Power controllers 30.0Power controllers must switch electric power in a

system satisfying MIL-STD-704 requirements and react 2.0to a 5-volt, 10-mA dc signal from the control-logic unit.

Bus-switching controllers are used only to connect 1.6power sources to the bus. Load-switching power con- -trollers are used between the bus and the loads to provide ccontrol, and current-limiting and circuit protection.The power controllers have the following operating - s

features:* Good isolating characteristics between control and

0

power circuits.0.* Operating efficiency: 95 percent.* Voltage drop: 0.5 V dc maximum, 1.5 V ac maximum. 00.4 0.6 0.8 1. 0.8 0.6* Leakage current: device rating X 10-4ampere at maxi- Lagging power factor Leading power factormum temperature. FIGU RE 8. Power-factor limits for utilized equipment.

* Operating temperature range: -54 to +850C.* No requirement for an external power supply. 28 VMaintaining power controller operation while aircraft

bus-voltage level varies is a major problem. Noticein Fig. 6 that the voltage level swings between 10 V dcand 80 V dc, and that in Fig. 7 the ac voltage varies be-tween 60 V rms and 180 V rms. The ac units must alsocontrol loads with power factors between zero laggingand 0.4 leading, as shown in Fig. 8. On-off

DC-load power controllerThe dc power controller contains a power switch, a Trip est

regulator, a driver, and circuitry for limiting current and indicatorprotecting itself (Fig. 9). FIGURE 9. DC-load-switching power controller.Power switch. The power-switching transistor (for our

device, an n-p-n silicon power transistor) must have suf-ficient voltage rating to withstand MIL-STD-704 voltage- 1.4transient stipulations and sufficient power dissipation tobe compatible with current-limit and circuit-protection 1.2modes of operation.

Reguilator. The internal regulator is used to provide abuffer effect so that the voltage to the internal circuitry ois independent of bus-voltage variations. A Zener diode 8may be used for low-current power controllers, but ashunt regulator-consisting of a Zener diode, transistor.and resistor-must be used to provide better packaging .6

efficiency for the higher-current units.Drive circuit. The power transistor must be driven .4

deep into saturation ifthe voltage drop and power dissipa-tion are to be held to a minimum. This requires a large 0 10 20 30 40 50 60 70base current, which typically is one tenth the load-cur-rent value. Also, the base-to-emitter and base-to-collector Supply. voltsjunctions must be forward-biased for saturation. A dc- FIGURE 10. Overall current-limiting characteristics of dc

to-dc converter can be used to provide signal-to-load power controllers.isolation and the necessary voltage for saturating thetransistor. This concept will permit switching power to FIGURE 11. Trip characteristics for short circuit and/orthe load at efliciencies in excess of 95 percent. overvoltage for dc power controllers.

Current-lilmit and circuit protection. There are two --- X .-- -reasons for incorporating these features: (1) to protect Minimum trip timetfor rated load (reterence)7the unit itself and (2) to protect the circuit it is controlling.The power dissipated in the switching transistor is heldto a minimum during normal operation by providingthe proper base drive current to cause the transistor tosatuirate and thus incur a small voltage drop.During an overload, short circuit, or voltage transient.

there is a tendency for the load current to increase, caus- 20

ing the transistor to fall out of saturation and increase its 0power dissipation. Since the input voltage cannot be con- 0.01 0.02 0.04 0.08 0.1 0.2 0.4 0.8 1 2 3 4trolled by this device, current limiting must be provided to = No-trip area Time. seconds

Dickey, Joncs--Solid-state switching for aircraft electric systems

hold power dissipation within limits and prevent the semi- the tripout time may be between I and 3 seconds. (Theconductor junction from rising above rated value during tripout time must be long enough to prevent nuisanceconditions ofmaximum voltage on the input with the out- tripping caused by short-lived transients when using aput terminal shorted to ground. Figure 10 shows the MIL-STD-704 power source [see Fig. 11].)current-limiting characteristic specified for the dc power The +600-V-dc-spike voltage specifications listed incontrollers. Notice that at rated voltage (29 V dc), the MIL-STD-704 (Fig. 12) require that special attention becurrent is permitted to be limited to: 100 percent rated for given to both input and output terminals of the device.a short-circuited device (Rload/R-ated < 0.667) output and130 percent rated for an overload (RtouJ/Rr,te,, > 0.667).* AC-load power controllerThese limit characteristics, called "foldback," are incor- The ac power controller contains a power switch, aporated to provide current values greater than rated for power supply, a driver, circuitry for zero-crossover turn-starting surges, yet allow current reduction to protect the on, current sensing, and circuit protection (Fig. 13).switching transistor during circuit-protection action (cur- Power swvitc/. The power switching elements are back-rent limit and tripout). It would be impractical to design to-back silicon controlled rectifiers (SCRs). SCRs arethe power controller to dissipate such a large amount of ideally suited for this application since they automati-power on a continuous basis, and so a tripout circuit is cally commutate with alternating line voltage.provided to turn off the power switch and indicate with SCRs also have low leakage currents (1.0 mA or less)a trip signal that the action has occurred. Remote reset is in the "off" condition, which enhance their usefulnessaccomplished by application of a 5-volt, 10-mA signal for power-switching circuits.(reset) from an external source. Notice that at 29 V dc The voltage drop of SCRs is higher than that of satu-

rated power transistors, but this is not serious consideringthat the supply is at 115 volts. The primary requirements* The resistaince ratio is used to define an overload for a normal of an SCR are that it has sufficient voltage rating to meet

operating-voltage range. The device should not limit current ortripout for ratio values between 1.0 and 0.78. MIL-STD-704 voltage characteristics and an P2t rating

_ _ S__ _St~~~~~~~~~~10 1000l il.._1_=V60c0

400~~ ~ ~ ~ ~ ~ ~ -0

0~~~~~~~~~~~~~~~~~~~~~~~~~0

0~~~~~~~~~~~~~~~~~

°300 1 1000 0 0 1 10

-600 -6 10-5 10-4 r-3 10-2, Tme, seconds

Time, seconds FIGURE 14. Trip characteristics for ac power controllers.

FIGURE 12. Envelope of spike voltages for dc equipment.

FIGURE 15. Load-switching power controller.

FIGURE 13. AC-load-switching power controller. _ -Silicon wafer

-Metal case

Power switch

On-off_ 3 k t 1 11811i \X\ g I - ~~~~~~~~~~~~~~~~~~~~~InsulatorReset _ X t -- -Copper plate

T-rip Laindicator <_

indicatorConnections - ~ t--- ---- Mounting stud

78 IEEE spectrum NOVEMBER 1970

to withstand the high fault current that can occur in the circuit within the bus power controller provides addedgenerating system. In a typical 20-kVA system. fault cur- protection. The design concept for low distortion andrents of 350 ampere-seconds/phase have been recorded. zero-voltage-crossover turn-on is basically the same asSCRs rated for 35 A rms have an 12t rating of approxi- for the ac-load power controller.

mately 85. Therefore, very large semiconductor chipsmust be used or external protection provided by a device Package design(such as a crowbar) on each phase of the bus. Our present Although packaging efficiency is an important factor inplan calls for a crowbar circuit that senses bus current, determining the size of the power-control module, theturns on when the current approaches 140 amperes (peak), power dissipated within the module is of paramount im-and diverts the excess current to ground, allowing use of portance. To prevent the junction temperature of devicesa 35 A rms SCR chip in the load power controllers. within the module from exceeding maximum ratings, theApplication of ac power controllers is limited by half- heat generated within the module must be conducted

cycle current rating-particularly in systems using con- away. The low-current devices present no problem, butventional generators-because of their high transient- high-current devices must be attached to an external heatcurrent capacity. Use of these devices with variable-speed, dissipator. Since the junction of the main switching deviceconstant-frequency (VSCF) systems is not limited be- (power transistor and SCR) is most critical, the thermalcause transient-current amplitude is controllable. resistance between this device and the module surfacePower supply. A power supply is required to provide must be low. To obtain a small package size, integrated

regulated dc excitation for the driver, current-limit, and circuits, transistors, SCR wafers, and thin-film techniquesprotection circuitry. The power supply consists of a trans- must be used (as shown in Fig. 15). The objective is toformer, full-wave rectifier, filter, and series regulator. minimize weight by allowing the system designer to treatThis power supply prevents line transients from affecting the power controllers as he would a power transistorthe internal circuitry. and to provide heat sink for the particular application.

Drive circuit. To provide isolation between the control The advantages cited for solid-state controllers arecircuit and load circuit as well as for power-transfer emphasized by the results of a study made for the A-7efficiency, a dc-to-dc converter is used. This concept also aircraft in which conventional and solid-state systemprovides a continuous gate signal to the SCRs and en- approaches were compared (Table I).sures triggering for any load power factor during steady-state operation. This article is based on work, initiated in January 1960, directedZero-crossover sensing. Zero-crossover sensing is re- toward the development of an advanced electric system for air-

craft. The effort has been sponsored during the intervening decadequired for zero-crossover turn-on. The output from this by the Navy, Air Force, NASA, and industry.circuit commands the driver circuit to turn on at zerocrossover for starting loads within the device rating. Lee D. Dickey (M) received the

Circuit protection. As in the dc-load controller, circuit B.S.E.E. degree from Texasprotection must be incorporated into the unit to protect Technological College, Lubbock,itself as well as the external circuit. SCRs inherently do Tex. in January 1951 and then,not limit current as do power transistors. Therefore, it until December, worked for J. B.is essential that circuitry be provided to sense time and Payne and Associates. He has

been with LTV Aerospace Corp.currents, and to trip out to protect the SCRs. As for the ever since (it was Chancedc-load controllers, the time duration should be long Vought Aircraft, Inc., when heenough to prevent nuisance tripping as a result of line joined), working on the designtransients (see Fig. 14), or switching loads (such as lamps of electric and electronic sys-and motors that allow an inrush of current into the cir- tems for aircraft. Between 1960

and 1963 he had technical re-cuit). This can be accomplished by sensing load current sponsibility for all R&D programs for the Product Designand actuating a control circuit that determines the time Section. Since 1963, he has been responsible for the super-before tripout. Once tripped out, the device can be reset vision and technical management of all design and R&Dby applying a reset signal. activities of the Electrical and Electronic Group. Mr. Dickey

has been chairman of the IEEE Power Distribution, Condi-tioning, and Control Subcommittee since 1967. He has

Bus-switching power controller written several articles and presented several papers onThe bus-switching controller performs a relay function solid-state aircraft electrical systems. Mr. Dickey is also a

only. Transformer rectifier units are used to provide member of the SAE.dc power in many systems; therefore, it is important onlyto discuss ac units. Clyde M. Jones (M) received the

The ac-bus-witchin pe b k dB.S. degree in electrical engi-The ac-bus-swttchas g power-controller block diagram neering from Texas Technologi-

is identical to that shown in Fig. 13 except that it does not cal College in 1949. Between 1949contain the current-limiting and protection circuitry. and 1956, he worked for South-But it does include a lockout feature that is not included western Public Service Co. He

in the ac-load controller. High-current SCRs are used. then joined Sandia Corp. andwas employed there until 1962.

Therefore, 12t ratings are not a problem. Circuit protec- He has since been with LTVtion need not be incorporated since circuit protection is Aerospace Corp., engaged inprovided by the load controllers, which are connected the design of electrical systemsbetween the bus and the loads. The system concept uti- for aircraft. For the past four

lizes a single bus, which makes it imperative not to connect r years, Mr. Jones has beenrespnsile or dvacedelectrical-systems design and

two asynchronous sources simultaneously to the bus. R&D activities. He has written and collaborated on manyThis feature can be designed into the logic, but a latching papers in his field and is a member of the SAE.

Dickey, Jones-Solid-state switching for aircraft electric systems