The target: Electronics 'sweet spot'

Demand for more electrical power in cars is being paced by inclusion of more and more electronic functions-tasks including communications, navigation, entertainment, telematics, and energy saving electrically activated systems.

One example of energy saving systems is electric steering, or drive-by-wire. This system would eliminate a power-and mileage-robbing hydraulic pump that continuously runs on current cars. Among other features, emissions can be reduced and mileage boosted using infinitely adjustable, electrically activated engine valves and variable geometry air intake systems. One particularly bright component under development to take full advantage of the higher power level is the integrated starter/alternator. Mounted to the engine bell housing, it would permit such functions as stop/start, to eliminate idling fuel consumption, and regenerative braking to recover energy. A similar device is used on hybrid cars to boost engine power as well.

Lucky number. An advanced automotive working group based at MIT, now the 42V Consortium (formally the Consortium on Advanced Automotive Electrical/Electronic Components and Systems), agreed on the 42V value in the mid-nineties. Organization members include most companies who build and supply parts and the automakers as well. The consortium does research, and has assumed a leadership role in standards development (http://auto.mit.edu/consortium). The latter includes concerns such as battery terminal size and shape selection to avoid confusion and accidents with 14V-system components and interfaces, as during a jump start.

While more power is going to be needed, higher voltage seems the way to go. Just remember from circuit theory that voltage is electrical 'pressure' that forces current through wires. The greater the voltage, the more power (voltage x current) that can be carried by existing diameter wires-thus avoiding an increase in weight of wiring.

But why 42V to push the added power around? Why not 28 or 56V instead of the 14V electrical systems standard on today's vehicles? The answer lies in the cost of silicon chips balanced against the potential for wire arcing. On the silicon side, in the mid-nineties, David Perreault, now assistant professor at MIT, developed the switched-mode rectifier (SMR) alternator. He essentially removed the triple-diode pack in a conventional alternator for producing dc, and substituted power metal-oxide silicon field effect transistors (MOSFETs) instead.

Simply put, the SMR acts as an electronic version of a transformer, chopping voltage so the windings "see" a more optimal voltage than the battery-level voltage. As its speed of rotation increases, the power produced in the SMR alternator still increases, rather than leveling off as in the standard alternator (see figure).

But power MOSFETs cost more, and cost is proportional to the area of silicon contained in the chips. Cost comes down as the supply voltage increases, however, but the curve levels out around 42V (see figure). Under diminishing returns, any further cost reduction sees a much greater increase in voltage, which in turn raises concerns about arcing from exposed or broken wires causing fires.

42V is not a magic number, however. Any auto electrical system does not operate at one voltage. In fact, 42V is only the nominal running voltage for the system architecture. Different off-nominal conditions dictate different voltages where the system must work or survive. MIT's Tom Keim, consortium director, notes that on a cold day, wire resistance goes down and "nominal" voltage may surge to 50V continuously. As currently envisioned by the consortium, 21V would be the lower limit for full safety systems functionality during start, with 30V the limit while running. Maximum dynamic voltage would be limited by alternator design and suppression methods to 58V during a load-dump pulse of up to 400 msec. A load-dump occurs when the alternator is cranking out maximum power and its load becomes disconnected (i.e. a battery cable comes loose).

Transition. While arcing may be the major concern in upping automotive electronics' voltage, other safety and safety- related issues have arisen. One concerns the transition to 42V systems, according to Joe Fadool, director of Siemens Auto-motive's Electrical and Electronics Distribution Systems Group (Troy, MI). Hybrid 14V/42V systems will come on line first, giving carmakers experience in limited applications. While one or a few systems can be accommodated with just up-converting voltage, as a greater fraction of electronics transitions to 42V dual batteries, will likely be installed, (a conventional 12V and a 36V system running voltage when the alternator generates power is slightly higher than the battery voltage level.) Such a dual architecture introduces added complexity and more failure modes as well.

Durability and performance of 42V components is an issue because of pushing higher currents through parts like switches and relays, says Shahram Zarei, technical specialist in research and vehicle technology at Ford. Redesign will be needed for components such as fuses and circuit protection devices. And, Zarei adds, a business case has to be made for 42V introduction. It has to have a high benefit/cost ratio for customers to accept and needs the economy of scale to push costs down.

Senior Sales Engineer David Goff of automotive electrical component supplier Omron Electronics (Schaumburg, IL) says the company is tackling power generation and distribution issues stemming from the higher voltage. Omron began prototyping relays in 1999 and switches in 2000 and looks to supply manufacturers with 42V electronic control units (ECUs) as well.

In relays, Goff notes coils need to be rewound to function at 42V. "Contact gaps have to be sufficient to handle arcing," he adds, "and dual, series contacts will reduce this to an acceptable level." Hybrid relays that combine electronics to dissipate transient arcing energy then break the mechanical relay contact may also be developed.

"For high-current switches, contact redesign to address arcing is similar to relay issues," says Goff. With switches, relays, and connectors in 42V systems, Goff says higher dielectric materials will be needed for insulating conductors from each other within the same dimensions.

ECUs use regulators to convert system voltage down to the logic-level value of 5V. To minimize the conversion losses over the greater step down from 42V, diode-based switching regulators can be used rather than linear voltage regulators that have higher heat losses. For dual voltage interim systems, the ECUs would stay on the 14V-bus system.

Ford's Zarei adds, "Many devices cannot operate at 42V, such as lighting incandescent lamps for example. Perhaps voltage could be modulated for these. While the coils in small electric motors may need to be rewound by a factor of three, the space may not be available to then house them. There is also the issue of motor brush life."

One concern that turns out not to be an issue is electromagnetic interference (EMI), notes MIT's Keim. "This doesn't occur in this voltage range, and while some effects scale with current, these are still low," he says. "42V will not cure EMI, but it won't make it worse."

Keim and Zarei both conclude that 42V systems will first appear on luxury cars where any price increase is a smaller fraction of total cost. "42V will be used if and only if it will work," Keim says. He stresses the potential fuel economy benefits of having electric-driven systems that only draw power when needed (power steering, braking) as well as new ones more precisely controlled for economy and emissions reduction (integrated starter/alternators and electrically-activated intake and exhaust valves). Good design engineering will probably see practical realization of many of these benefits.

Finding (arc) fault

Taking automotive electrical systems to 42V introduces greater risk of arcing faults. Arcing is not only an issue between exposed (i.e. chaffed) conductors and ground (parallel arc fault) or between the ends of a broken or cracked wire (series fault), but also for design of mating contacts in switches and relays.

Engelbert Hetzmannseder, group chief of mechatronics for EATON's Innovation Center (Milwaukee, WI) says while an arc fault can look spectacular, "It is not a showstopper, but a challenge. Arc physics has been around for a long time and there are solutions for switching and detecting" (see DN 9/4/2000, p. 76). Arc-fault protection devices based on microelectronics and detection algorithms for ac circuits are available for building wiring and under test for aircraft systems. "For automotive, the principle is the same," notes Hetzmannseder, "but the algorithms are different for dc" in determining normal and abnormal transients. With dc, the voltage doesn't change polarity. If an arc fault is detected, electronics will then trigger a relay, breaker, or switch.

He adds that EATON has prototype devices but is looking for consensus from the OEMs on switching, packaging, and basic architecture (i.e. mechanical or solid state). Questions on standards for given wire lengths, current loads, and voltage ranges will also affect architecture decisions, Hetzmannseder says. He adds the OEMs may be treading cautiously because they haven't had to do such current and voltage sensing before.
Other issues are traditional switch and relay concerns, such as interrupt-arc contact lifetimes. Here for example, going to lower-load, solid-state switching FETs (field-effect transistors), which can also current sense and do diagnostics, may be worth their added cost, offers Hetzmannseder. He concludes, the bottom line for any 42V developments will be whether any increase in system costs are offset by greater fuel efficiency over the life of the car.

42VDC Arc Faults: Physics and Test Methods
Dr. Engelbert Hetzmannseder, Dr. Joe Zuercher
EATON Innovation Center

Abstract

The new 42VDC automotive power system raises a number of challenges that were not applicable to standard 14VDC systems, not the least of which is arcing fault mitigation.

Arc physics teaches us that, for supply voltages larger than about 20VDC, an arc can be sustained theoretically indefinitely, while at lower voltages, i.e. 14VDC, the current can be interrupted with a very short arc.

An arcing fault, which can occur at any point in the vehicle's life, might not be detectable with present short circuit or overload protection devices. In the case of an arc fault in parallel with the load for systems with supply voltages larger than 20VDC, fuses or breakers may trip late or not at all due to the intermittent nature of the arc with significantly lower peak current. If the arc is in series with the load, a fuse will never open since the fault current is always lower than the load current.

Metal-to-metal short circuits occur fairly seldom. Unintended arcing faults are the norm at supply voltages larger than 20VDC. In a new 42VDC automotive power system, an unintended arcing fault may cause a reliability issue or even a fire hazard.

Several test methods have been developed to simulate various potential series and parallel arcing faults' causes at 42VDC and to investigate their effects.

The dangling wire test simulates a broken, dangling wire, such as may be found in the motor compartment. In the laboratory, an uninsulated automotive wire is brushed over a grounded metallic plate at about .50 m/s. The contact force, between the wire and the grounded metal plate, is achieved by the weight of the 0,3m long wire, only. Arc voltage and current traces are shown in Fig. 4.

Three possible circuit states can be identified in Fig.4:

Significant localized heating occurs during arcing when both current AND arc voltage are present. RMS responding monitoring devices "underestimate" local heating effects due to the lower duty cycle and high concentration of fault power.

The three circuit states above are also valid for the arc fault traces described in the following tests.

The loose connector and lug test reproduces loose connections utilizing automotive connectors and lugs. The loose contacts were opened and closed manually or by vibration. Disengagement of electrically "live" connectors also falls under this category of testing. Series arcs in a broken wire can be simulated and may occur when a broken wire is held in place by insulation or by a wire bundle.

The current and voltage traces, from the slow separation of connectors, are shown in Fig. 12. When a small gap (1mm) is maintained between conducting surfaces, a series arc can be maintained for many seconds significantly heating the metal. Series arcs usually are not as violent as parallel arcs since the fault current is always lower than the nominal current. While the heating effect is equivalent to a 100 watt light bulb glowing for 0,15 sec, a 15 joule event, it still represents a possible fire hazard due to the energy's concentration.

14VDC vs. 42VDC Arc Fault:

Only when the supply voltage is higher than the minimum arc voltage, can a stable arc be sustained. At supply voltages below 20VDC, e.g., 14VDC, usually only very short duration arcs occur in inductive circuits. These arcs extinguish once the arc has dissipated the inductive energy. At 42VDC, an arc fault may be sustained even though the current is significantly reduced below the bolted value since the voltage driving the circuit impedance is reduced by 30%-50%.

Parallel Arc Fault vs. Short Circuit:

During most fault conditions, the contact force is low, e.g. a "live" wire slightly touching the circuit ground. The contact spot melts and evaporates quickly forming an arc. This type of fault is defined as a parallel arc fault although in many cases this type of fault is wrongly identified as a short circuit. A short circuit, per definition, is a metal-to-metal fault. It can occur when the molten contact spot welds or when the contact force is high enough to carry the fault current preventing arcing. Since an arc limits the fault current, the arc fault duration is always lower than the short circuit duration. Additionally, depending on the arc cause and the motion of the arcing conductors, the arcing duty cycle can be significantly less than 100%. While the RMS value of the fault current is reduced, the arcing still represents a local heating hazard. Fuses or breakers trip late or not at all!

While fuses, circuit breakers, or conventional electronic fault protection devices cannot detect arc faults, they are still necessary for overload and short circuit protection!

A parallel arc fault is an issue for a supply voltage between 20VDC and about 120VDC (higher in AC systems). Below 20VDC, the arc is not stable (minimum arc voltage). Above 120VDC, the current limiting effect of the arc (V_SOURCE - V_ARC = voltage that drives the current) is small permitting the short circuit protection device to operate. For AC circuits this 120V limit is not applicable due to the current zero crossings allowing intermittent arcs.

Series Arc Fault:

In a series arc fault (the arc fault is in series with the load), the fault current is less than the nominal current preventing detection by a fuse or a breaker!

A series arc fault is an issue at any voltage higher than 20VDC becoming more severe the higher the voltage. Since less molten material is ejected by a series arc (reduced power), the total energy of a series arc fault event can still reach substantial levels before the arc is extinguished.

Both a series and a parallel arc fault can spread to adjacent circuits resulting in a significant reliability and fire hazard.

Testing:

Existing test standards for arc faults and wire harnesses and new test procedures have been adapted to reproduce close-to reality faults in cars.

Test set-ups for series arc faults:

Test set-ups for parallel arc faults:

The voltage and current traces for series and parallel arc fault tests show that the fault times were significantly longer when compared to short-circuit fault times, even though fuses were used in both cases.

The traces exhibit characteristics that differ substantially from residential and aerospace AC circuits' arc faults.

Many arc fault detection methods have been patented for AC applications; e.g. EATON Corp. has more than 50 patents issued and pending. Only a few patents have been published for DC arc fault detection methods, e.g. Hendry, EATON Corp. Eaton Corporation is the world leaders in arc fault detection and protection.