AUTOMOTIVE ELECTRICAL SYSTEMS - THE POWER ELECTRONICS MARKET OF THE FUTURE John G. Kassakian Laboratory for Electromagnetic and Electronic Systems The Massachusetts Institute of Technology Room 10-172,77 Mass Avenue Cambridge, MA 02139 Phone: (617) 253-3448; Fax: (617) 258-6774 igk@,niit.edu Abstract - The automobile is undergoing a revolution in the design of its electrical system. This is the result of in- creasingly sophisticated engine and body controls, as well as the introduction of new, electrically controlled func- tions. The main electrical bus of the future will be 42 V, and it will be buffered by a 36 V battery [1,2]. As many devices and electronic control units require voltages different from 42, conversion from the 42 V bus to these other voltages will be necessary. Some anticipated features, such as electromechanical engine valves, will demand both conversion and sophisticated control at power levels in the 2 kW to 10 kW range. These, and other developments in automotive engineering, are promising to create a vital and challenging new market for power electronics in the next decade. industry, to develop an 80 mpg, “green,” automobile. Europe is pursuing an equivalent program to produce a 3 liter car [4]. The target vehicle class is the mid-size sedan, e.g., a Ford Taurus or Chevrolet Lumina. This partnership is now com- pleting its sixth year, and the highly probable design outcome for all three participating OEM’s is a variant of the hybrid. It has also become clear that one of the technologies dominat- ing vehicle cost is power electronics. The new requirements being placed on the electrical sys- tem, and revised criteria for cost, efficiency, packaging and expandability are providing the first opportunity for a radical redesign of the automotive electrical system since the change from 6V to 12V in the early 50’s [5]. One of the most impor- tant enabling technologies available to this redesign process is power electronics. In this paper we will describe the automobile’s present I. INTRODUCTION “It takes good tjuice’ and lots of it to run U modern auto; not the kind that Uncle Sum has put a ban upon, [but] the electric ’juice’. electrical system and the system developments that are ex- pected to create a very large and diverse market for power electronic controls, systems and devices. 2500 2000 - Putnams Automobile Handbook, 1918 It is only recently that manufacturers have begun to appre- ciate the electrical system’s influence on vehicle performance and price. The result has been an increasing interest in re- placing mechanical actuators by^ electrical, improving electri- cal system efficiency, finding alternatives to the present 14 Vdc system, and improving safety and comfort by introduc- ing new functions that are best controlled electrically. A fur- ther, rather interesting, motivation to electrify fhnctions is that electrically driven “things” provide packaging flexibility. Mechanically driven devices, such as power steering, water pumps and air conditioning, are located on the front of the engine and driven by the “front-end accessory drive” (FEAD). Automotive designers would love to have the styl- ing flexibility provided by eliminating the FEAD. The antici- pated increase in average electrical load in a high-end car is shown in Fig. 1. Further motivation for a critical assessment and redesign of the electrical system is provided by the US Partnership for a New Generation of Vehicles (PNGV) [3]. This is an aggres- sive program, co-funded by the US government and the auto 0-7803-5864-3/00/$10.00 0 2000 IEEE 1500 1000 500 04 I Model Year Fig. 1. Historical and anticipated average electrical load in a high-end automobile [6] 11. THE PRESENT ELECTRICAL SYSTEM The present electrical system employs a 12 V lead-acid bat- tery for energy storage and power capability, thus creating a 3 network with a nominal voltage of 14.4 V with the engine on. A conceptual diagram of the system,is shown in Fig. 2. Not shown in this diagram are the details of the electronic control units (ECUs) comprising microprocessors and MOSFET switches to control the engine, transmission and many other functions requiring a response to sensor inputs. Perhaps the most significant of these from a power electronics point of view is the antilock braking system (ABS). Hydraulic valves are rapidly actuated by power MOSFET switches in response to brake pedal pressure. Originally conceived to prevent brake lock-up, the ABS control system is now used routinely to provide traction and dynamic stability controls. It is an excellent example of how electronics, and particularly power electronics, has made possible significant performance and safety improvements in the car. A typical high-end car, e.g., a Lincoln Continental, con- tains over 50 fuses, a midsize car contains over 40, and a Jag- uar contains 80. These fuses are incorporated in two or more junction boxes placed in accessible locations within the en- gine and/or passenger compartments. In addition to fuses, these junction boxes often contain relays, and are an expen- sive and complex component of the electrical system. dashboard control s reactance, which can exceed 100 V under these conditions, appears at its terminals. Such a transient is shown in Fig. 3. The car's entire electrical system (excluding the now disconnected battery) is subjected to this transient. Some mitigation of the transient is provided by avalanche diodes in modern cars, but a maximum system voltage of 40-60 V is still assumed for design purposes. Power semiconductor devices, e.g., those used as relay drivers or in the ABS, are thus typically rated for 60 V even though the nominal system voltage is only 14 V. In Vsn de Load Oump (8000 rpm 58 AI I I I I I I NO Loads I I . ..... .: . . . . . . . . . 'o?O 0 20 40 80 time !mq 80 100 120 140 160 >BO Fig. 3. Typical load dump transient. The voltage is measured at a tail lamp. The initial spike is the alternator L(di/dt) voltage, while the longer transient is due to the decay of the alternator field 111. THE HISTORY OF 42 V By 1994 Mercedes Benz realized that they would need a I I higher voltage to support the electrical systems they were en- _I+ __ visioning for their future production vehicles. Table 1 shows A = alternatorlrectifier E =electronics some of the anticipated new electrical loads and their power L = lamp S = st art er requirements. Mercedes Benz also realized that they could H = heater M = motor not unilaterally decide which voltage to use, but needed buy- in by the international auto industry. MIT, which had already Fig. 2. Simplified diagram of conventional been doing electrical system research for Mercedes Benz, 14 V automotive electrical network. was asked to organize a working group of automotive OEMs The electrical environment in today's cars presents an ex- and suppliers to see if agreement could be reached on a new treme challenge to electronic devices. The maximum under- voltage. This group of 7 companies met regularly for a year hood ambient temperature is specified by different manufac- and a half, considering issues of safety, reliability, infra- turers as between 125°C and 1 50\"C, depending on underhood structure and transition costs. The result was a set of recom- packaging philosophy, while the minimum operating ambient mendations, principal among which was the proposal for 42 is i40\"C. All electronics has to survive a jump start attempt V, which would be the engine-on voltage of a 36 V lead-acid from a reverse connected 24 V battery, and this is guaranteed battery. These were made public through an article in the by a specification of -24 V for 10 minutes. And perhaps the August 1996 issue of IEEE Spectrum [l]. most debilitating characteristic of present cars is a phenome- A year after the creation of the MIT Working Group, Mer- non known as a \"load dump transient.\" This transient occurs when a fully loaded alternator has its load disconnected. The cedes Benz organized a meeting of German OEMs to con- assumed conditions are a battery being charged at its maxi- sider similar issues. The MIT Working Group recommenda- mum rate (full alternator output), and a battery terminal con- tions were presented to this German group, which immedi- nector suddenly coming loose. The alternator field current ately adopted them. The German group has now been ex- cannot be instantly reduced, so the alternator's voltage behind panded to include all European OEMs and many suppliers, 1-11 ---- 4 TABLE 1 New electrical loads anticipated by the model year 2005, except for e-m valves, which may not be available in production vehicles until 2010. The average powers reflect assumptions about use factors IV. THE 42 V ARCHITECTURES The structure of the 42 V electrical system is still fluid, and it is not clear that there will be an industry standard. Two practical alternatives being seriously considered by OEMs are shown in Fig. 5. Figure 5 (a) shows a dual battery system, while Fig. 5 (b) illustrates an architecture using only a single 42 V battery. Each of these designs has advantages and dis- advantages with respect to cost, reliability and flexibility, due mostly to the different requirements placed upon the power electronics in each. A common feature of both is the presence of a 14 V network. While some manufacturers are exploring the practicality of a comprehensive 42 V network, the present 14 V supply and service infrastructures present a formidable challenge. The philosophy behind the dual battery system, Fig. 5 (a), is that the starting function should be isolated from the storage function required for “key-off’ loads. These are the electrical loads drawing power when the ignition is off, e.g., keyless entry, theft alarm, clock. These loads are sufficient to drain the battery in today’s car when left parked for long periods, for example, at an airport for several weeks. In contrast, the single battery architecture of Fig. 5 (b) is based upon the desire to avoid the cost, weight and packaging problems created by a second battery. The philosophy here is that the energy management system will be smart enough to monitor the single battery and manage the key-off loads to prevent depleting the 36 V battery to the point where the car cannot be started. Load Electromechanical valves (6CYl@ 6 “) Water pump Engine cooling fan Power steering (all elec.) Heated windshield Catalytic converter pre-heat Active suspension Comm/nav/entertainment Peak (W) 2400 300 800 1000 2500 3000 12,000 Average 800 300 300 IO0 200 60 360 100 Total 2220 and is known as the ‘LForum Bordnetz”. Its work is organized and facilitated by a German company named SICAN, and it has assumed principal responsibility for refining the recommendations and turning them into IS0 standards. The MIT group has been transformed into the MITAndustry Consortium on Advanced Automotive Electrical/Electronic Components and Systems, and its membership expanded to 34 companies including 10 in Japan [7]. Thus there is now comprehensive international acceptance of 42 V as the system voltage of the future. Figure 4 illustrates a ‘‘logo’’ and “mark” which have been designed by the consortium and are being used by manufacturers and suppliers wishing to differentiate components and other products intended specifically for the new 42 V power network (“PowerNet”). The logo can be used in promotional material, etc., while the mark is intended to be used on the product itself to prevent its misuse in a 14 V system, Neither the logo nor the mark have been copyrighted, and thus are free for use by anyone. They may be downloaded from the MIT consortium website [7]. functions. dcldc converter T - 14V m* 42V PowerNet (a 1 Fig. 4. (a) 42 V logo; (b) 42 V product mark. (b) T* To42Vloads . T- distribution boxes containing both switching and fusing functions. dcldc converter +- If. - To14Vloads Fig. 5. Practical 42 V system architectures: (a) dual battery system; (b) single battery system 5 Both of these architectures require a dc/dc converter to tie together the 14 and 42 V subsystems. One difference between the converters is that the one in the single battery system must be rated for the peak 14 V load, while that in the dual battery design is only required to supply the average load, the peaks being provided for by the 12 V battery. The average 14 V load is between 750 W and 1 kW, while the peak load can be 2 kW, depending on the car and its accessories. These numbers will change with time as 42 V penetrates existing electrical functions. A perceived advantage of the dual bat- tery design is that reliability could be improved by providing for a transfer of energy for starting from the 12 V battery to the 36 V battery in the event of an accidentally discharged 36 V battery. In this case, while rated for only the average power, the converter is required to support bilateral energy flow. V. NEW ELECTRICAL FUNCTIONS Virtually all the new functions in development at the auto companies, as well as the electrification of present functions, e.g., power steering [8], require the application of power elec- tronics. Indeed, in many cases the cost of the power elec- tronics dominates the discussion of introducing such devel- opments. Reducing these costs is a major challenge for the power electronics industry, and whether the industry meets this challenge or not will determine its future in the automo- tive environment. Already power MOSFETs are replacing relays to control some functions in the car, and high intensity discharge (HID) lamps, requiring power electronic ballasts, are replacing in- candescent headlamps in high-end vehicles. Although the development of trench technology and its derivatives have considerably reduced the losses in MOSFETs operated at 14 V, the voltage requirements imposed by the load dump tran- sient makes them still substantially more expensive than re- lays. The new 42 V architectures will make semiconductor switching, and even active fusing, more economically viable. Among those new functions requiring substantial power electronics control are electromechanical engine valves, ac- tive suspension, electric air conditioning, electric power steering, electrically actuated automatic transmission clutches, and, perhaps most significant, a combined starter/alternator. Given the electric power requirements of future cars, the introduction of the combined starter/alternator creates what has come to be called a “lite hybrid.” One of its principle features is the ability to operate the car in “start/stop” mode, where the combustion engine is shut down when the car is idling, and restarted quickly to resume driv- ing. To illustrate the role of power electronics in these new sys- tems, we will discuss three of the most interesting concepts - electromechanical engine valves, electric active suspension, and the combined startedalternator. Electromechanical Valves The valves in today’s engines are controlled mechanically through linkages to the crankshaft. A camshaft is connected through a chain or belt (the timing chain) to the crank. The cams then act on pushrods, rocker arms or the valve stems themselves to cause the valves to open and close in synchro- nism with the position of the piston in the respective cylinder. Valve timing affects fuel economy, dynamic performance, and emissions, but since the cam design is fixed, these attrib- utes can be optimized at only a single engine speed and load , condition.’ If all the valves were controlled by an electrical actuator in- stead of a mechanical camshaft, each valve could be con- trolled independently for timing, profile and lift. This would make possible not only performance optimized with respect to fuel economy, dynamics and emissions at every speed and load condition, but would also permit the deactivation of cylinders to improve fuel economy at cruise, and potentially reduce starting torque requirements [9]. The former is achieved by closing all valves in a cylinder, and the latter by opening all valves in the engine. It is even conceivable that the engine could be “statically” started, that is, started with- out turning it over with a starter motor. The concept is to em- ploy direct fuel injection (DFI) to inject a small amount of fuel into the appropriate cylinder (determined by piston posi- tion) with its valves closed, then tire the cylinder to begin en- gine motion. Figure 6 shows a popular design concept for an electrome- chanical valve [lo]. It can be viewed as a resonant spring- mass system which oscillates between open and closed, being held at each state by the upper or lower electromagnet. Ideally no energy is lost in the transition, and because the magnetic circuit has no gap in the open or closed position, the required holding force can be obtained with very little MMF, i.e., magnet current. In reality the situation is more complicated because of thermally induced dimensional variability and the work required to overcome gas pressure on opening the exhaust valves. What is evident, however, is that whatever electromechanical mechanism is used, it will be controlled by power electronics according to a very complex algorithm involving inputs of engine speed, load, exhaust gas conditions, temperature and injection dynamics. Electric Active Suspension Active suspension is the singular function influenced by 42 V that will enjoy consumer notice. The idea is to employ * Some manufacturers have recently introduced engines with “variable timing.” These systems employ either a scheme to vary the phase of the camshaft with respect to piston position, or a dual cam design that shifts valve actuation from one to the other depending upon engine conditions. The two cams differ in their phasing and cam lobe designs. While im- proving performance over a single fixed cam design, neither scheme provides continuous timing and profile optimization. 6 Fig. 6. Conceptual diagram of an electromechanical engine valve employing an oscillating spring/mass system. Fig. 7. Conceptual diagram of electrical active suspension system using a linear actuator. electrical actuators in the suspension system to maintain the car on a level trajectory as the wheels undergo uneven verti- cal motion due, e.g., to a rough road. Although electro- hydraulic active suspension is available today, it does not possess the bandwidth of a fully electric system and thus does not achieve the anticipated performance potential. Figure 7 shows a schematic of an eIectric active suspension system using a linear permanent magnet motor as the actua- tor. Alternatives using rotary actuators have also been pro- posed. For example, Ford has developed a system employing ball screws to transform the rotary drive motor motion to lin- ear actuator motion. Cost made this design and implementa- tion impractical, and the power electronics was the major cost component, aggravated by the fact that the system was de- signed to operate from a 14 V network. Active suspension systems can be designed to recover en- ergy that is dissipated in the shock absorbers in today’s pas- sive systems. While energy recovery contributes to opera- tional efficiency, it complicates and makes more expensive the power electronics. Another challenge is maintaining sus- pension forces during turning. This does not require high bandwidth and can be accomplished with hydraulic or pneu- matic “biasing” devices. These will also be controlled by power electronics, Integrated Starter/Alternator generator system used in Oldsmobiles and Buicks, c. 1915. However, the advent of larger engines and higher compres- sion ratios widened the disparity between starting torque re- quirements and the torque capability of the generator. With the imminent introduction of alternators with higher ratings, and the benefits of “starthtop” vehicle operation, the integra- tion of the machines becomes a more practical concept. During the energy crisis of the ~O’S, Sweden experimented informally with a driving mode called “start/stop” in which drivers shut off their engines when stopped, at a traffic light, for instance. The hoped-for benefits are obvious, but the execution proved inconvenient and a major drain on the bat- tery, when, for example, many traffic lights were encoun- tered. Today’s interest in “start/stop” operation is motivated by emission control and noise abatement, especially in urban areas where congestion, idling and noise are common. In 1998, motorists in the United States spent an average of 34 hours stopped in traffic, which gives one a sense of the 1 11. potential impact of “start‘stop” [ Figure 8 illustrates a preferred implementation of the inte- grated starter/alternator (ISA). The flywheel in the transmis- sion bell housing is replaced with the machine, which then performs triple duty - flywheel, alternator, starter. The real challenge here is not the mechanical integration of the ma- chine with the transmission, but the cost of the power elec- tronics. Initial studies suggest that the cost of the power elec- tronics is between 5 and 8 times the cost of the machine. Still, manufacturers such as DaimlerChrysler, Ford and BMW are aggressively pursuing the ISA. Combining the starter and alternator has been a dream of electrical engineers in the auto industry for decades. In fact, the starter and generator were combined in the Delco motor- 7 42 V BUS 21 33 43 52 58 heel, alternator, starter, pulley, 8 possibly a belt ellmlnated low voltage limit for loads related to safety lower operating limit for all other loads maxlmum engine running voltage maximum steady-state overvoltage maximum transient overvoltage Fig IO Proposed voltage specifications for the 42 V PowerNet Fig 8 Combined startedalternator consisting of a switched reluctance permanent magnet (SRPM) motor integrated with the transmission VI. SEMICONDUCTOR DEVICES Nearly all the functions listed in Table I either require power electronic controls (e.g., electric power steering, elec- tromechanical valves) or benefit substantially by being inte- grated with them (e.g., controlling the water pump to match cooling requirements). In every case the power electronics either dominates the cost of providing the function, or con- stitutes a significant part of it. Figure 9 shows the dramatic effect voltage rating has on the silicon die area required to control a given power with a given loss. The curves are parametric in voltage margin above nominal and are normalized to the die area for 14 V (nominal) and a 30 V margin. Thus the detailed voltage specifications of the 42 V PowerNet will influence the power electronics costs. The proposed specifications are shown in Fig. 10. The maximum voltage is limited to 58 V by a combi- nation of alternator design and active or passive load dump suppression. Inspection of Fig. 9 shows that the combined benefits of the higher nominal voltage (42 V vs. 14 V) and the reduction in voltage margin (16 V vs. 30 V) is a reduction in die area of approximately 85%. Given both the desire (due to efficiency considerations) and necessity (due to thermal considerations) of reducing device loss, a smaller reduction would be used in practice. VII. ALTERNATIVE ELECTRICAL SOURCES A significant challenge posed by “start/stop” operation is supplying electrical loads while stopped. Under conventional idling conditions, the alternator continues to function, and the heavy mechanical load of the air conditioner (dc) is driven by the running engine. Under a “start/stop” driving mode, it is not reasonable to assume the air conditioner can be off when the car is stopped, e.g., stuck in traffic. One solution is to electrify the dc drive. By itself, this is a difficult challenge as the compressor requires between 2 and 3 kW of mechanical power, and the ventilating fans consume an additional 250- 500 W. But the problem is even more difficult during “stop” when the alternator is not supplying power. A battery of practical size cannot supply such a large electrical demand. This situation argues for a different means of generating electricity in the car. A number of generation technologies have been proposed, including thermophotovoltaic (burning fuel - creating light - generating electricity from photocells) [ 131, thermionic (thermal emission of electrons - collection on a cold anode) [14], and, of course, the fuel cell [15, 161. While each of these technologies has unique characteristics and materials requirements, they all share a need for a power electronic interface between the generating unit and the car’s electrical system. VIII. CONCLUSIONS 0 0 20 40 I 60 80 100 120 140 160 Nominal Supply Voltage V N M Fig 9 Die area as a function of supply voltage and safety margin for constant controlled power and device loss [ 121 . Without question cost is the greatest challenge facing the power electronics industry as it penetrates the automotive market. Equipment ratings in the applications discussed in this paper range from a few hundred watts to about 12 kW (peak). Today’s rule of thumb cost for converters in this power range varies from about $O.lO/W at the lower powers to perhaps $O.O7/W at several kW. And these numbers are for 8 high volume commercial applications such as PC power sup- plies. As discussed above, the automotive environment is much more challenging, and thus packaging (including the thermal design) will generally be more sophisticated. Should these challenges be met successfully, the automobile will prove to be a vital and exciting new market for power electronics. [4] “Automobile Industry Association Expects Production of Fuel-Efficient Cars to Begin by 2000,” The Week in Germany, German Information 1995, p. 4. Center, NY, Sept. 15, Volts Presents Its Case,” SAE Journal, Feb. 1954, [5] Terry, S.M., “12 p.29. [6] Miller, J. M., “Multiple Voltage Electrical Power Distribution System for Automotive Applications,” 3lst Intersociew Energy Conversion Engineering Conference, (IECEC), Washington, D.C., Aug. 1996. [7] http://auto.niit.edu/coiisortium/ [8] “Electric Power Steering Coming to Europe,” The Hansen Report on Automotive Electronics, vol. 9, no. 8, Oct. 1996. [9] “Camless BMW Engine Still Faces Hurdles,” Automotive Industry (AI), Oct. 1999, p. 34 ACKNOWLEDGEMENTS The author thanks the members of the MIT consortium for the careful deliberations that have resulted in the international acceptance of 42 V as the future PowerNet voltage. John Miller (Ford), Randy Frank (ON Semiconductor), Alfons Graf (Infineon), Vahe Caliskan (MIT) and Ed Lovelace (MIT) provided much of the data on which this paper is based. Special thanks are due Gary DesGroseilliers, Consor- tium Program Manager, for assistance with the physical preparation of the manuscript. [IO] F. Pischinger et. al., “Electromagnetically Operating Actuator,” US Pat- ent no. 4,455,543, June 19, 1984. [I I] “In gridlock’s grip: L.A. stuck at No. I,” USA Today, Nov. 17, 1999. [I21 Graf, A., D. Vogel, J. Gantioler and F. Klotz, “Intelligent Power Semi- conductors for Future Automotive Electrical Systems,” l 7Ih Meeting Electronik im Kraflfahrzeug, VDA, Munich, June 1997. [I31 Coutts, Timothy J. and M. C. Fitzgerald, “Thermophotovoltaics,” Sci- entific American, September 1998. REFERENCES [I] Kassakian, J. G., H. C. Wolf, J. M. Miller and C. J. Hurton, “Automo- tive Electrical Systems circa 2005,” IEEE Spectrum, Aug. 1996, pp. 22- [I41 Svensson, Robert, “The Use of Waste Heat from Otto Engines for Elec- tricity Generation with Thermionic Energy Converters,” Technical Re- 27 College of Applied Engineering and Maritime Studies, port VOI, Electrical Department, Chalmers Universtiy of Technology. Miller, J.M., et al., “Making the Case for a Next Generation [2] Automotive Electrical System” IEEE-SAE International Congress on [I51 “BMW to market sedan with fuel cell,” Los Angeles Times, May 15, Transportation Electronics (Convergence), Dearborn, MI, Oct. 1998. 1999. for a New Genera- [3] Review of the Research Program of the Partnership [I61 “BMW Developing Gasoline Fuel Cell in Cooperation with DELPHI,” tion of Vehicles, Fifth Report, National Research Council, National theautochannel.coni/news/press/date/l9990429/press02255 http://www. Academy Press, 1999. l.html 9