The tale of CEM, Alfa Romeo’s in-house electronic engine management system, which redefined what was ‘state of the art’ in engine technology, outdoing Bosch with a fraction of its research budget. To no avail.
The history of tailpipe emissions regulations started, as many may know, with the USA’s Clean Air Act of 1966. Alfa Romeo’s share of the US market was minuscule, but the engineers at the Milan HQ could see the writing on the wall: it was now just a matter of time before similar measures would be enacted in Europe as well.
The Italian company needed to find ways to preserve the power and torque characteristics that Alfa’s customers had come to expect from its engines, into a future of pressing societal and regulatory demands. Alfa Romeo’s first response to the American regulations was a mechanical fuel injection system it co-developed with the controlled supplier company Spica. Still, the Italian engineers also worked with the eponymous German supplier Bosch, the pioneer in electronic fuel injection systems.
Yet the Alfa Romeo’s proud, competent, and passionate engineers weren’t entirely satisfied with Bosch’s systems, considered too crude to achieve the engine performance they were after: more of a necessary evil than an elegant engineering solution. As those who read my book about the Alfa 6 may remember, that’s the reason why Alfa’s flagship initially used a set of six carburettors: European legislation at the time still lagged considerably behind the American’s.
Alfa Romeo’s research and development budget could only stretch so much, though, given the Italian company’s small size and its steadily declining fortunes. Enter the CNR (Consiglio Nazionale delle Ricerche), Italy’s largest public research institution: in 1979, it awarded a massive 23 Billion Liras contract to the CRF (Centro Ricerche Fiat), the research branch of Fiat, for the development of advanced technology for the transportation sector. Alfa Romeo (remember, still state-owned at the time) saw an opportunity: it became a subcontractor for the CRF, charged with the development of electronic engine management systems and hybrid vehicle technologies.
One of the secrets of Alfa Romeo engines was the Alimentazione Singola, or the use of one throttle body for each cylinder, as it allowed more overlap in the engine’s timing. This wasn’t deemed possible with the Bosch fuel injection systems then available, so Alfa Romeo set out developing its own, the CEM (Controllo Elettronico Motore). The engine used as a base platform was the classic 1962cc inline-four twin-cam engine then used on the Alfetta.
Its pair of double-choke side-draught Weber 40 DCOE carbs and distribution ignition was replaced by a fully integrated electronic engine management system that utilized one throttle body for each cylinder and allowed “modular” operation. The CEM-equipped Alfetta 2.0 could run on just two cylinders at light engine loads and smoothly transition back to four cylinders, saving fuel yet retaining the power (130HP) and drivability of the standard carburetted engine.
Alfa Romeo then took the technology out of the laboratory to place it on Milan’s roads: ten Alfetta 2.0 CEM were put into the hands of local taxi drivers for six months during 1983, during which each car clocked around 40,000 closely monitored kilometres. The small fleet performed admirably, and the modular function proved to reduce fuel consumption by around 12%.
The success of the taxi experiment led to a wider one, involving a fleet of 1000 Alfetta 2.0 CEM that were sold to actual Alfa Romeo customers, picked among the marque’s faithful. This was no small commitment, though: 1000 cars meant the CEM became a production reality, one that involved Alfa’s supplier network and cost-benefit evaluations for the future. That’s where things began to unravel.
It was calculated that, if fitted in series production to 10,000 Alfetta 2.0 each year, the production cost of the CEM added around 50% to the cost of the 1962cc engine, which would then translate in a showroom price difference far too high for the customer to accept. Alfa Romeo had redefined what was state of the art in engine technology, outdone Bosch with a fraction of its research budget, but the commercial viability of the CEM was questionable, at best.
Bosch was a specialized company that supplied everyone in the business, including Alfa Romeo itself, while the Italian company was a loss-making specialist manufacturer that built less than 200,000 cars each year: the economies of scale simply didn’t work. It’s unclear whether all the planned 1000 Alfetta CEM were built, but the car did become available in 1983, with quite a few surviving today in enthusiast’s hands.
The last hurrah of the CEM would be the Alfa 90 2.0 V6, an Italian-Tax-Break-Special made from 1985 to 1987: its engine was a CEM-equipped 1996cc version of the beloved Busso V6 engine, but devoid of the ‘modular’ function. Around 1500 cars were produced. Meanwhile, in late 1986 Fiat had taken over Alfa Romeo, and profitability was to take precedence over costly feats of engineering, no matter how clever. But that’s another story.
Driven To Write 
Fascinating, Matteo.
I had experience of the contemporary Lancia and BMW fuel injection systems. I’ll try to remember how it went, but I’m rather busy, so it won’t be for a few days! Other liquids, closer to the human alimentary system, take precedence.
Is it reasonable to say that the same thinking formed the basis for the MultiAir principle? A quest by Alfa to precisely control intake flow on a level “normal” VVT-systems couldn’t provide.
I must say the uptake of the MultiAir engines is quite good, although the same could be said for other newish engines that utilise vvt, often in combination with twin scroll turbo’s.
Enjoyed learning something I never knew, thank you.
Really great contribution, Matteo, thank you. Every days a school day with DTW
Could it be possible that AGIP’s “Stazione sperimentale combustibili” (Experimental station for fuels) at S . Donato Milanese was in some way involved in this project? I vaguely recall someone in S.Donato telling me something about the Alfetta CEM (and if I remember correctly they had one on the ramp for testing), but it was forty years ago…Both were state-owned and both based in Milan, so the collaboration would have been easy.
Before someone blames me for not asking more, I probably did, but I was there mainly because of a research on gas analyzers, so it may have been just a conversation over a coffee cup.
Interesting – I’d never heard of the system myself.
There is a difference though, surely, between fuel injection itself, and pollution reduction, and engine management systems. The three are separable. It’s completely unclear to me from this article whether this CEM refers entirely to digital ECUs, or the separate fuel injection system it controlled, or the feature of individual intake runners per cylinder, or catalytic converters for pollution control or any combination thereof. By 1982 there were a lot more companies than Bosch making fuel injection components in Japan, and the American and Japanese companies had VERY advanced ECUs as well, like the Ford EEC IV.
My impression has always been that the development of the Bosch lambda oxygen sensor in the middle-late 1960s to help determine the oxygen content of the exhaust was the first breakthrough in pollution control, because it allowed dynamic control of the intake mixture, or amount of fuel. 1969 VWs and Volvos had primitive electronic fuel injection control in Canada using the lambda sensor.
Then came the catalytic converter in 1975 to convert carbon monoxide and excess HC in the exhaust. Unleaded petrol was required. This was followed by three-way catalytic converters in the early 1980s which also had a go at NOx. The Americans ran feedback control carburettors from lamba sensors for years on cheaper vehicles, employing cats to convert the nasty stuff along with company ECUs. And were far ahead of pollution control measures in the EEC even at that.
The BMW M1 had Kugelfischer mechanical injection with individual cylinder throttle bodies in the 1970s. I agree that the K-Jetronic Bosch with that big intake pivoting flap to measure airflow was a bit of wet blanket if one got one’s kicks from quick throttle blips in neutral. I owned four of them. So that’s why mass air flow “measurement” utilizing hot wire cooling became the norm, no moving parts and quicker response. I had myself used hot wire anemometers as a student in the late 1960s to study the air velocity profile in tubes of varying cross section, and that Danish instrumentation was fantastically expensive being all analogue. The breakthrough was making it literally 100 times less costly.
In that same timeframe, Honda was developing the clever cheap-trick two-stage VTEC for further intake mixture timing and volume control depending on revs along with their PGMFI fuel injection, while BMW struggled away developing Vanos for variable cam intake valve timing, and as it turned out nearly everyone was onto that same basic idea too, including Porsche Variocam, and Toyota too. Now it’s been standard for two decades and also includes the exhaust cam since about 2008. Some systems have completely variable valve lift as well and no throttle butterfly, the most elegant in 2007 from Infiniti on a 3.7l V6 that wasn’t rev limited like BMW’s Valvetronic.
The fuel injection systems themselves came in many flavours over the years, from cheap single-point, to continuous flow to sequential, but all relied on lamba O2 sensors and catalysts (two sensors on three-way catalysts) and the controller ECU for pollution control using unleaded fuel, in North America and Japan anyway – before leaded fuel was phased out in the EEC in the early 1990s.
So what was so special about CEM beyond, one presumes, throttle response? And how specifically did it beat Bosch? Or the other giants in the US and Japan who worked independently of Bosch? The claim of “redefining the art” seems a bit lofty to me; e.g. electronic sequential port fuel injection and an advanced ECU came standard on Ford 2.3 litre turbos in the autumn of 1982, and Bosch had nothing to do with it.
…And as much as I admired BMW’s Valvetronic upon its introduction, I would argue that Fiat’s MultiAir system is the most elegant approach to throttling the engine via continuously variable valve lift. (And somebody must agree with me, because I seem to recall it winning Best New Engine of the Year in 2011.)
I’m not especially familiar with the CEM system, but the big feat would appear to be viable cylinder deactivation, which no one else managed for a couple decades. It would also be the ‘90s before anyone else offered individual throttle bodies on an electronically fuel-injected production car, but that had more to do with cost than technology.
And “engine management system” in this context likely refers to the electronics and fuel delivery, but not exhaust gas aftertreatment, various options for which are merely enabled by a given engine management system. (Speaking of which, three-way catalysts were introduced in the ‘70s—I would say “in conjunction with exhaust gas oxygen sensors,” but you’re evidently aware of a Canadian application of lambda sensing without a catalyst, which is news to me.)
And, of course, it should also be noted that Alfa Romeo introduced variable valve timing in like 1981, nearly a decade before Porsche followed suit and Honda introduced their novel solution that actually switches between two entirely different cam profiles.