Summary

The first production electronic fuel injection system was the Bendix Electrojector, offered briefly by Chrysler for 1958. It was dropped due to unreliability, but it inspired Bosch to develop the similar but more successful D-Jetronic system, introduced by Volkswagen for 1968 (which required a patent licensing agreement between Bendix and Bosch) as a way of meeting U.S. emissions standards. In the seventies, both Chevrolet and Cadillac offered a second-generation Bendix electronic injection system, which was very similar to D-Jetronic and used some Bosch parts.

Once considered exotic technology, electronic fuel injection has been around a surprisingly long time. In this installment of Ate Up With Motor, we review the origins of EFI and examine the relationship between the pioneering Bendix Electrojector, Bosch D-Jetronic, and the second-generation Bendix system that introduced GM to electronic injection in the 1970s — a complicated web of technology, business, and politics.

IMPORTANT AUTHOR’S NOTE

Please note that this is a history, NOT a repair or maintenance guide for fuel injection systems. I CANNOT to tell you how to repair Bosch D-Jetronic or Bendix fuel injection, and I CANNOT help you find parts! If you have maintenance, repair, or troubleshooting questions, please consult a qualified mechanic familiar with older fuel injection systems. (I am NOT a mechanic and am NOT qualified to give troubleshooting or repair advice.)

The Bendix Electrojector

Introduced in 1957, the short-lived Bendix “Electrojector” was the first commercially available electronic fuel injection system for passenger cars. It may be surprising to learn that the Electrojector was also one of the earliest production passenger car fuel injection systems of any kind, arriving right on the heels of the Rochester “Ramjet” continuous injection system offered by Chevrolet and Pontiac, and only about three years after the Mercedes-Benz 300SL became the first production car with a fuel-injected four-stroke gasoline engine. (It was preceded by a number of small fuel-injected two-stroke cars from Goliath and Gutbrod, introduced around 1951.)

There had been experiments with fuel injection going back to the 1890s, and it had been used with some success in aircraft engines, stationary applications, and race cars (and of course for diesel engines). However, passenger cars presented much greater challenges for fuel metering: A fuel injection system viable for street use needed auxiliary systems to provide for starting, engine warm-up, and idle, like a carburetor, and also needed to be able to maintain something close to a stoichiometric air/fuel ratio through a broad range of operating speeds, at part-throttle as well as full throttle, and respond immediately to changes in throttle position. Metering systems that were adequate for industrial or aircraft engines were simply not versatile or flexible enough for passenger car use on the street.

The Electrojector system had been in development since about 1951. It was the brainchild of Robert W. Sutton, assistant chief engineer of fuel systems engineering in the Bendix Eclipse Machine Division in Elmira, New York. Sutton was an aviation buff, and the injection system appears to have been at least in part a synthesis of certain recent developments in aircraft fuel systems, albeit intended to meet the far more complex metering needs of passenger cars. By 1953–1954, Bendix had a working prototype of this new system installed in a 1953 Buick for preliminary demonstrations to automakers, although the system — whose electronic control unit still used vacuum tubes until its fourth design iteration — was not yet ready for production. Nonetheless, Bendix found some auto industry interest, and in late 1956, American Motors Corporation (AMC) announced that the Electrojector system would shortly be an option on the new Rambler Rebel. In January 1957, Sutton and his boss, Albert H. Winkler, presented a paper on the new system at the annual meeting of the Society of Automotive Engineers (SAE) in Detroit.

Like the Rochester Ramjet or the later Bosch mechanical systems, the Electrojector was a port injection system, delivering fuel through separate injectors mounted just above each intake valve (rather than into the combustion chamber as with the diesel-derived Bosch gasoline direct injection system on the 300SL). Unlike the Rochester system, the Electrojector was intermittent, delivering fuel to each cylinder in timed pulses. Fuel for the injectors was supplied by a common fuel rail, which an electric fuel pump maintained at a constant pressure. The amount of fuel delivered to the intake port therefore depended on how long each injector valve remained open. Since the injector valves were controlled by solenoids, the open time of each valve and the duration of each fuel pulse depended on the pulse width of the electrical signal that energized the solenoid, also known as the pulse duration and usually measured in milliseconds. This duration was proportional to the total resistance in the circuit, which could be varied based on different control inputs.

The idea of electronically controlled injector valves wasn’t a new idea even in the mid-fifties. Back in the early thirties, California engineers Harry E. Kennedy and Brooks Walker had devised a similar electronically controlled injection system with magnetically operated port injection valves, intended as part of an ambitious but somewhat half-baked scheme to allow spark ignition engines to run on cheaper diesel fuel. The Atlas Imperial Diesel Engine Co. (Atlasco) of Oakland, California, installed this system, which also included a high-tension arc ignition system and an alternator rather than a generator, on marine and truck engines, but it never went into mass production, probably for reasons of cost as much as reliability concerns. A few years later, Milanese engineer Ottavio Fuscaldo, then working in the experimental department of the Italian aircraft company Caproni, developed a similar electronic system, adaptable for either port or direct injection. The Caproni-Fuscaldo system was installed on an Alfa Romeo 6C 2500 that ran in the 1940 Mille Miglia, and was at least considered for aircraft applications, although we don’t know if it ever actually saw any aviation use. It would have needed considerably more development make it suitable for passenger car duty, but the basic principle was the same as the Electrojector and subsequent timed electronic injection systems. One of Fuscaldo’s patents (U.S. Patent No. 2,332,909) is among the references cited in the core Bendix electronic fuel injection patent (U.S. Patent No. 2,980,090).

Like the Atlasco and Caproni-Fuscaldo systems, the Electrojector initiated each fuel delivery pulse using what Bendix called a triggering selector, essentially an additional set of distributor breaker points. Depending on the application, the triggering selector could either be incorporated into the ignition distributor as an additional “sandwich” section or else carried in its own housing, driven by gears off the distributor shaft and rotating with it at one-half engine speed. The triggering selector acted as a mechanical speed sensor, and also made it possible to time the injection pulses to coincide with the engine’s combustion cycle, delivering fuel during the intake stroke of each cylinder.

Each time the triggering selector breaker points opened, they sent a spike signal to a multivibrator circuit in the transistorized electronic control unit (ECU), also known as the modulator. The modulator transformed this spike signal into a square pulse, whose duration varied with the resistance in two control circuits connected to the system’s various sensors. The ECU then sent the modified pulse back to a distribution commutator in the triggering selector, which would energize each injector at the appropriate time, with the pulse width determining how long the injector remained energized.

There were two control circuits, also called sensor circuits: the main control circuit, which had an intake manifold pressure sensor and an acceleration sensor switch, and an auxiliary control circuit with sensors for cold start and warm-up, idle enrichment, and intake air temperature, plus a de-loading device for cranking on wide open throttle in the event of a flooded engine. There was also an altitude compensator in the ECU that adjusted the modified pulse width to compensate for variations in barometric pressure. All of these sensors were purely analog. For example, the manifold pressure sensor was a potentiometer controlled by a crank and a spring-loaded piston that reciprocated in a cylinder open to manifold pressure. (Sufficiently high manifold vacuum would also cut off fuel flow during deceleration or overrun.)

The Electrojector used speed-density metering: Engine speed, signaled by the triggering selector, determined the frequency of injection pulses, while manifold pressure determined the basic pulse width, and thus how much fuel each pulse delivere. Auxiliary sensors triggered additional enrichment (i.e., increased pulse width) under certain operating conditions. All that sounds very impressive — indeed, it wasn’t until the early eighties that pundits and automotive journalists overcame their awe at the Hugo Gernsback wonders of the electronic “brain box” — but mechanical fuel injection systems for passenger car applications necessarily incorporated all or most of the same types of sensors, for the same purposes, even if they performed their metering calculations with vacuum signals or the motion of cams and levers rather than with electrical pulses.

Ironically, one of the big selling points for the Electrojector was that it promised to be cheaper than mechanical fuel injection systems. As Winkler and Sutton explained in their 1957 SAE paper, the Electrojector operated at much lower pressures than mechanical injection systems, needed no power-consuming pump drive, had fewer moving parts, didn’t demand costly precision machining, and had less critical fuel filtration requirements. Some of the anticipated cost advantages presumed greater production volumes and improvements in electronics manufacture that didn’t materialize until decades later, but if you compare the relatively unstressed mechanical components of the Electrojector with the daunting complexity and enormous cost of the diesel-derived Bosch six-piston jerk pump used on the Mercedes-Benz M198 engine — which had to precisely meter very small quantities of fuel while withstanding fuel rail pressures up to 1,700 psi (117 bar) and injection pressures of up to 682 psi (47 bar) — the idea that the Bendix system might eventually be cheaper to manufacture doesn’t seem altogether far-fetched.

Indeed, when AMC announced the Electrojector as an option for the 1957 Rambler Rebel, the quoted retail price was $395 — a lot of money in 1957, but a substantial $89 cheaper than the Rochester mechanical system on a ’57 Chevrolet. AMC canceled the fuel injection option option before it actually went on sale, but in late September, Chrysler announced that it would offer the Electrojector on its hottest 1958 Chrysler, Dodge, DeSoto, and Plymouth models. List price was $400, still cheaper than Chevrolet or Pontiac charged for the Rochester unit. Whether the Bendix system would have retained any price advantage had both systems reached large-scale production is hard to say, but the Electrojector initially appeared competitive from a cost standpoint.

Bendix claimed that on a “typical” V-8 engine, the Electrojector offered more power, more torque, and better fuel economy than a four-barrel carburetor across the entire operating regime; eliminated hesitation on abrupt throttle changes; and provided excellent starting and warm-up performance. AMC announced that the 327 cu. in. (5,354 cc) V-8 in the 1957 Rambler Rebel would produce 288 hp (214.8 kW) and 350 lb-ft (475.6 N-m) of torque, 33 hp (24.6 kW) more than the carbureted version. Chrysler quoted a more modest 10 hp (7.4 kW) advantage for Electrojector-equipped cars, although as we’ll see, this rating was largely speculative.

Like any fuel injection system, the Electrojector offered a number of potential performance advantages (like facilitating the use of tuned intake runners for resonance supercharging, as Mercedes did on the 300SL), but achieving that potential required more substantial engine changes than Chrysler (or apparently AMC) was prepared to make. In fact, the Chrysler cars equipped with fuel injection were assembled with carbureted engines and then taken to the DeSoto plant, which removed the carburetors and installed the Electrojector system. Consequently, the main performance benefit was a small power increase due to the elimination of the manifold heat riser, which a port injection system didn’t need.

Chrysler Engineering Excoriates the Electrojector

Although it sounded great on paper, the Bendix Electrojector system was an almost unmitigated disaster in execution. We haven’t seen any AMC engineering evaluations — AMC didn’t officially comment on its reasons for abandoning the Electrojector option — but surviving Chrysler Engineering documents paint a very grim picture of the system’s problems and design flaws.

The most immediate of these was the grievous unreliability of the solenoid-activated injectors. In principle, the injectors offered very precise control of fuel delivery, but in practice, fuel flow was so wildly inconsistent that Chrysler test engineers were never able to obtain actual gross or net power and torque ratings for the injected engines. Early injector valves were too slow to open and much too slow to close, which meant that the actual effective duration of each fuel pulse could be several times the calculated value. Injector leakage also caused excessive fuel consumption at cruising speeds, and in at least a few cases dumped liquid fuel into the cylinders, which could have engine-ruining consequences as well as presenting a potential fire hazard. Chrysler engineers had to completely redesign the injector valves and divide the ECU’s multivibrator circuit into two separate channels before the injectors would open and close with anything approaching the necessary speed. Even then, injector fuel flow variation remained about twice that of contemporary carburetors — ±6 percent versus ±3 percent — and obtaining any semblance of consistent fuel distribution required installing or replacing the injectors in carefully matched sets.

Another issue affecting fuel consumption was high evaporative losses. Today, cars have evaporative canisters that capture fuel vapor so that it can later condense and return to the fuel tank, but in the late fifties, evaporative emissions were simply vented to the atmosphere. Because the Electrojector was constantly returning excess fuel from the pressurized fuel rail to the tank, Chrysler found that injected cars lost significantly more fuel to evaporation than did carbureted cars.

Driveability of Electrojector-equipped cars was further marred by cold starting problems and by the fact that the system’s capacity for acceleration enrichment was very limited. The acceleration sensor was a vacuum diaphragm switch responsive to manifold vacuum, positioned in series between the power supply and the manifold pressure sensor, which when open would increase resistance on the primary control circuit, increasing injection pulse width. This worked only in hard acceleration, was only momentarily effective, and could not vary the amount of enrichment based on operating conditions. Bendix explored other acceleration enrichment systems with more progressive action, but they weren’t used on the production Electrojector.

On top of that, for reasons Chrysler was never able to determine or correct, the ECU suffered output decay at higher engine speeds: At speeds over 4,000 rpm, fuel delivery would fall off when it should have been increasing, causing the engine to run lean. The only way Chrysler engineers found to compensate for it left the engine running too rich at lower speeds, further exacerbating the system’s excessive fuel consumption.

The Chrysler documents we’ve seen make no mention of the oft-repeated contention that the Electrojector system was susceptible to interference from outside radio sources, and after examining the system’s wiring diagrams, Chrysler 300 Club member John Grady, an electrical engineer, dismissed those anecdotes as very unlikely. However, Grady felt that the lifespan of the trigger contacts in the distribution commutator was likely very short, and noted a distinct lack of voltage regulation. Some other reports indicate that the system’s waxed paper capacitors would “drift” over time and could eventually fail completely in hot, humid conditions (something that may not have come up in Chrysler engineering tests, but was potentially very troublesome for owners in warm climates).

Interestingly, Chrysler engineers found that timing each injection pulse to correspond with the opening of the intake valve, was an unnecessary complication. Such cyclic timing was important with direct injection, but with port injection, it made little difference whether the intake valve was opened or closed when fuel was delivered to the port.

To make a long story short, the Electrojector system was oversold and grossly underdeveloped. The Rochester Ramjet system was also finicky, especially early examples, but the Chrysler reports make clear that the Electrojector as it existed in 1957–1958 should never have been released to actual customers. AMC apparently reached that conclusion early on, but Chrysler buyers were not so fortunate, although Electrojector sales were few.

Exactly how many systems were sold is unclear. Automotive journalist Karl Ludvigsen later reported that AMC bought about 50 Electrojector systems and Chrysler about 300 (a figure that Bendix project engineer Jerome G. Rivard repeated in a 1974 SAE paper on the development of electronic injection), but those totals appear to reflect preliminary orders rather than actual deliveries.

AMC installed a few Electrojector units in engineering test or preproduction cars, perhaps as many as six. It doesn’t appear that any cars so equipped were actually sold to customers, although at least one may have eventually ended up in the hands of a private owner. As for Chrysler, the engineering report indicates that production changes that were slated to take effect with the 85th system were canceled in May 1958 because Chrysler had decided not to order that many. Since some early systems were likely reserved for testing and service replacement purposes, the number installed in production cars was undoubtedly lower still. Comments by the superintendent of the DeSoto plant that installed the system indicate that the Electrojector went into 54 cars in early 1958, although the actual total may have been as few as 35. Almost all later had their injection systems replaced with dual four-barrel carburetors by dealer service departments; it doesn’t appear that there was a formal factory recall to remove the Electrojector equipment, but the division service groups were authorized to do so in response to repeated owner complaints. Very few cars survived with their original injection systems intact.

Chrysler Engineering eventually recommended against further fuel injection development, either of the Bendix system or of an in-house system, originally developed in 1956 and shelved following the decision to adopt the Electrojector. (The in-house designs, developed under the A-632 and A-664 programs, were mechanical, not electronic — low-pressure continuous injection systems.)