Keith Duckworth
—the designer of engines which win Grands Prix
KEITH DUCKWORTH has the sort of talent which inspires others to work for him. On top of that, he has been responsible for designing a tenet of racing engines, all of which have achieved victory either in their very first race or soon afterwards, starting with a fairly straightforward conversion of a production Ford unit, followed by Formula Two engines of various types. It is the Cosworth Formula One engine which currently rules the roost. Graham Hill’s latest book “Life on the Limit” contains the following credit: “Keith Duckworth deserved tremendous praise for producing an engine which could vanquish the opposition so masterfully and for designing such a beautiful looking engine. It really was a . . . joy to behold; so simple, which is the essence of good design”.
The four men who run the small Northampton factory that houses Cosworth Engineering Ltd. are Keith Duckworth, Mike Costin, Bill Brown and Benny Rood, of whom Duckworth fondly remarks “there are four of us involved in this outfit—all rather above average in various ways”. Many are the firms which have had small beginnings in the racing game, with different personnel drifting in and out but Costin and Duckworth, both former Lotus employees, started Cosworth in 1958 and their names have been associated with it ever since.
Not yet 37, and with a rapidly-greying mop of unruly hair, Duckworth has an irresistible sense of impish humour which makes him an ideal spokesman for his firm, despite what he calls a fundamental cynicism about writers. Racing engine designers are a rare breed indeed, but there’s nothing of the prima donna about Duckworth: if one of his engines breaks, he goes back to the drawing-board if the fault is shown to be a design deficiency. It must be utterly galling for rivals that things so seldom break, for Cosworth engines won their first Formula One and Formula Two races of 1967 and have achieved unprecedented successes which show no sign of flagging. Because they were financed by Ford money it has become fashionable to shout “money bags” at Cosworth, but no amount of money can make an uncompetitive engine win, as Ford USA found out the hard way in the 1968 and 1969 Can-Am seasons; in fact, Cosworth’s success has been won on a remarkably low budget and Cosworth has become an essential part of the highly specialised light industry which constitutes British racing-car manufacture these days. No fewer than 57 Formula One DFV (Double Four Valve) engines have been made, while the initial run of 40 Formula Two FVA (Four Valve A-series) has now been more than trebled. It’s quite a contrast with the days when one Grand Prix engine was likely to differ in several subtle respects from an outwardly identical unit from the same factory.
Born in 1933 in Blackburn, the son of a mill owner, David Keith Duckworth had family motoring connections via his grandfather, who took part in motorised competitions before the Great War in a car whose name cannot be established. The young Duckworth took an early interest in model aircraft, making his first rubber-powered model at the age of eight before moving on to more complicated radio controlled designs, for which he also made the electronic equipment. He had a motorcycle as soon as the law permitted and in due course he became the owner of several rapid two-wheelers, among them three water-cooled two-stroke Scotts.
Faced with National Service on leaving school (Giggleswick) in 1951, Duckworth opted for the RAF, hoping to pilot the “real thing”‘ in place of those models, but although he was accepted for aircrew training and did 540 hours (some of it in twin-engined Oxfords), the Air Ministry found him surplus to requirements and switched him to navigating. These days he is able to combine both types of aviating activity in an American Brantley helicopter, which is his pride and joy and enables him to make quick flips all over the country, although the helicopter principle offends his engineering science. Its 5.9-litre four-cylinder Lycoming “stands on its car: helicopters are basically unstable, and feel it”.
The Duckworth career took a decisive turn when he left the RAF, for not only, did he start a successful degree course in engineering at London’s Imperial College, but he also became the owner of a Lotus 6 which he intended racing. It had one of the first 25 1,100-c.c. Coventry-Climax engines to be manufactured, making it pretty competitive, but after three events its disillusioned driver decided (a) that it was going to take an awful lot of hard work before he became any good at it, and (b) that he couldn’t afford it. The Lotus went, probably to the benefit of study, but some valuable lessons about tuning Coventry-Climax engines had been acquired.
They were not needed during Duckworth’s first employment. He asked Colin Chapman for a job with the budding Lotus organisation, having worked there during college vacations, and he was set to work on transmissions. Chapman was in the throes of trying to make his own design of 5-speed gearbox work properly, a target which Duckworth reluctantly decided could not be achieved without a complete redesign. There was a sudden parting of the ways after only two months in Lotus employ, which resulted in Duckworth setting up Cosworth Engineering in October 1958 with Mike Costin, a friend from Lotus. Costin, in fact, was unable to take part in Cosworth activities until he joined hill time three years later, having completed a Lotus contract which he signed shortly after the new business got under way.
With Costin working out his time at Lotus, Duckworth and his wife Ursula were left in charge of the foundling concern, their aims: being to continue Coventry-Climax development and also to build a complete Cosworth car. The latter project got as far as the chassis stage, but only in 1969 did the actual car carrying the firm’s name get to a racing circuit. Duckworth very rarely approaches a problem from the “classic” viewpoint, attacking each new project as though no one had done it previously. That there is very little “book” knowledge of racing engines goes without saying, and anyway he questioned most of what had been set down in print, considering it outmoded or downright wrong.
Duckworth prepared Climax engines were quite good enough to give private owners in Elvas an edge over works Lotus drivers, so it was not long before he decided to go ahead and convert a Fiat engine for the newly-introduced Formula junior. An early examination of the Fiat revealed that it was not going to prove an ideal base, so instead he decided to use the then new Ford 105E unit. Today, in Formula Three, no one would contemplate any engine other than the small Ford, but in those days, this little unit was giving the existing tuners some nasty headaches, mainly due to camshaft design problems. Duckworth, having been taken in by book reading, learned all the problems of valve gear operation painfully—but it made him think about the subject, and he re-designed cams which then worked, and Cosworth have since become the pace-setters in this field.
The Cosworth FJ engine made its first appearance at Brands Hatch on Boxing Day 1959 (where a sump plug fell out during practice, the resulting loss of oil ruining the engine) and went on to power the vast majority of the winners in that and subsequent seasons. The company had been faltering until that time: thereafter it grew and outgrew its premises, first at Friern Barnet, then at Edmonton, and today at Northampton, where the factory has already been extended twice, but where there is room for plenty of expansion, even if it does mean cutting down the size of the helicopter park.
The first true racing engine from Cosworth was the 1-litre Formula Two SCA (Single-Cam A-series), introduced in time for the 1964 season. There was £17,500 of Ford support, so it was natural once again to utilise the Ford block: “it was available and we had FJ experience, so it seemed worth a go”. The SCA utilised a single overhead camshaft, while the rival BRM and Honda units both had two, but it was not until the third and last year of the Formula (1966) that the Honda caught up and passed the Cosworth. Of his own engine, Duckworth says simply “its combustion was poor: the Honda was good”. Yet the SCA is ‘still being used today in American Club racing and in those rather less arduous conditions continues to win races.
Ford’s association with Cosworth became even closer in 1966, when a joint project was announced, not only for a new Formula Two engine designed to the 1,600-c.c. limit (a production block being required by the regulations this time), but also for a Formula One unit. This time Ford’s investment was an announced £100,000 and there is the additional bonus of what Duckworth calls “encouragement”, not to mention the use of metallurgical laboratory facilities.
The F2 engine was soon running in an ageing Brabham which Cosworth had purchased for Mike Costin to drive in Club races as a mobile test bed, and he had already won several Club races as well as cracking some lap records by the time 40 favoured customers took delivery of production FVAs. In its first and third years, the FVA won every Formula Two race, the sole exceptions being five victories for the Ferrari Dino V6 towards the end of 1968 in circumstances that have been described as peculiar. If the opposition had not been quite so vociferous in their power claims, Duckworth says that he might not have coaxed quite so much power out of the FVA, so the others would have stood more of a chance! He says the FVA outperforms other designs because of “honesty and a reasonable power curve”, assets which rivals don’t always have.
The FVA has been an unqualified success, and there will be a sports-racing application for 1970, when a new European Championship for cars conforming with the latest Group 5 and 6 rules will be run. The new engine is the FVC (Four Valve C-series, the B-series having been an experimental 1,500-c.c. version), which has a stroke dimension increased from 2.72 in. to 3.06 in. Power is only marginally more, but torque is greatly improved. Some early problems with piston materials have been overcome and already 70 units have been ordered.
An even bigger future is forecast for the productionised FVA, known as the BDA (Belt Drive A-series), which may well herald a whole new series of Ford engines. Although not produced directly by Cosworth, the BDA is being developed at Northampton. As suggested by its name, the BDA has belt drive to its two overhead camshafts, operating four valves per cylinder. The four-valve arrangement offers a good combustion chamber space and the smaller valves show less tendency to burn out than the large single valves fitted, for instance, to a normal Lotus-Ford twin-cam unit. There is also a petrol consumption advantage, Duckworth’s experience with a BDA fitted into a very ordinary looking Cortina being that it is difficult to push consumption below 25 m.p.g. Intended for fitment initially to the Ford Capri, there have been some piston material problems with the BDA, but these have been overcome and customers should be getting this very exciting power plant in the near future.
The Formula One DFV engine was a joint effort with Ford, an effort to restore British prestige in racing at what seemed to be a low ebb. Foreign manufacturers looked like scooping the pool, many of them announcing a formidable effort for the 1966 3-litre F1, and only BRM to oppose them on behalf of Britain. As a Ford statement said at the time, “in November 1965, the future offered a lot of reasons for pessimism. British racing-car constructors had relied until then largely on Formula One engines from BRM and Coventry-Climax. At the end of 1965 Coventry-Climax said they would make no more Grand Prix engines. And Honda was coming in with more and more money. Ferrari was still fighting for supremacy, while in France de Gaulle himself was on the point of offering £500,000 for anyone who would develop a French Grand Prix engine.”
The Ford requirements were for a V8, essentially comprising two FVA cylinder heads mounted on a common block. “Our still undesigned engine would not be the most powerful on the track: how could we make it the fastest?” asked the Ford release.
“We decided that weight was the answer. A car weighing around 1,000 lb. would be competitive against cars weighing 1,250 lb. even if the heavier car had more power. Moreover, those of us in Competitions believed that racing machines were becoming too complicated.”
Duckworth was present at the first Ford negotiations. It took him only five months to complete the design stages of the unit where Ford personnel had said that they would require two years. There is the oft-quoted remark that although £100,000 may sound like a lot of money, it had cost Ford £1m simply to put synchromesh on bottom gear in the Cortina range. The cost of Mercedes’ Grand Prix efforts in 1954/1955, or of the Eagle GP car more recently, must have been considerably greater than £100,000.
That the engine went on to win its first race—the famous Dutch Grand Prix of 1967 in the hands of Jim Clark—must surely go down as one of the greatest motoring achievements of this or any other decade. The parameters between which Duckworth had designed the Cosworth V8 included not only Ford requirements, but also the idea (shared with the BRM H16) that the engine should increase the lightness of the car by being a stressed member of the chassis. This is one of the more significant advances to be seen under the 3-litre F1, at a time when Grand Prix racing does not undergo vast numbers of annual technical changes like it used to do. By dispensing with chassis side-members, it became possible to mount the oil and water pumps on each side of the crankcase. The sump itself is exceptionally shallow and actually carries the main bearings, which are thicker than those seen on any previous racing unit, a theory of Duckworth’s which— as usual—contravened established practice. With no pumps cluttering up the front of the engine, and the electrics tucked away in a box buried between the vee of the engine (where they can be completely unplugged via a 3-pin socket), the overall length of the unit was kept to a staggeringly short length of 21.5 in.
Of course, there have been problems. One of the most difficult, says Duckworth, is “getting the oil and air out of the engine separately”. The timing gear layout was redesigned for the 1969 season, when different camshafts and other minor alterations enabled the rev. limit (governed by an electrical cut-out) to be raised 500 r.p.m. from the previous 9,000 mark. The engine, which had only been available to Lotus in 1967, then to McLaren and Matra in 1968, went on general sale in 1969, and of course in 1969 it has had only two regular opponents, the Ferrari and BRM V12s. The circle had come all the way round: it was the other teams, not the British, which were feeling the draught. Duckworth does not like this state of affairs, which he says Cosworth has achieved through “a practical approach to fuel cams, etc.”. He adds that “people try to knock us because we do things a bit more efficiently than the others”, leaving one to believe that he would like some more opposition. Whether the V12 brigade, to be joined by Matra in the coming season, will have found that solution at last is a subject about which Stewart has already made his decision: he’s going Cosworth V8.
In 1969, there were several camshaft failures which caused most teams some consternation. Duckworth divides them between those suffered by one of the two Lotus 63 4-w-d machines (which had an inadequate oil system—its camshafts showed signs of failing because bits had been knocked off by the valves, which had obviously been thumped by the pistons) and various others, where the camshaft invariably failed “for reasons we have never understood, either behind or just in front of the rear cam”. Doubtless there will be a suitable cure in time for the 1970 season.
The future of the Cosworth 4-w-d Grand Prix car is less certain. Readers will recall that this machine, the first complete car to come from Cosworth, was designed by former McLaren employee Robin Herd, prior to his joining the ambitious March project. There was talk at the beginning that the car was an entirely Ford-backed project and a great air of mystery surrounded it for a long time before it made its first public appearance. Once it had been made clear that the car was Cosworth’s responsibility and not Ford’s there was a set-back with the introduction of wings, which (Herd said) would nullify the advantages which the four-wheel-drive system was expected to confer. When the car was eventually revealed to the public gaze, shortly before the 1969 British Grand Prix at Silverstone, much was expected of it.
But testing revealed some totally unexpected problems. The most important of them was the sheer physical effort which is required to drive a 4-w-d car with wide tyres around a circuit, and this difficulty was also encountered by the Matra, Lotus and McLaren 4-w-d machines which by that time had also been revealed to the public. With refreshing candour, Duckworth says that “it was something we should have thought about before the car set a wheel on a track!” A good idea of the difficulty can be gained by imagining the forces with which a driver must cope when braking hard in a powerful single-seater. In this condition, there are severe forces corning back through the steering, forces which are inevitably very enervating for the driver, although in a conventional 2-w-d car they are only reached under comparatively brief periods of braking. In the 4-w-d machines, with today’s wide tyres, these forces also occur under acceleration: with a great deal of power on tap, as with a modern F1 racing car, this means that the driver must cope with a tremendous physical strain and that conducting a 4-w-d on the limit is a spectacular sight. In the case of the Matra MS84 and Lotus 63 designs, a compromise was reached by racing the cars with a vastly reduced amount of drive to the front wheels, culminating in the complete disconnecting of the drive mechanism to the front wheels of the Matra at the Mexican Grand Prix, which is of course naturally defeating the object of 4-w-d in the first place. The McLaren and Cosworth cars were both withdrawn, the Cosworth without ever actually racing. In spite of Herd’s departure to March Engineering, Duckworth is persevering with the car and alleady has some modifications in hand, although he will not say what they are. As he says, “its future depends on the success or otherwise of the experimental bits”. Herd’s new car and the 1970 Lotus are expected to be of conventional 2-w-d layout, so the future of four-wheel-traction rests to a great extent in the achievements of the Cosworth.
The car itself is unquestionably the most radical-looking Grand Prix contender of the decade, its high-sided chassis pontoons and angular nose being designed for the maximum in negative lift without resorting to wings of any kind. The Cosworth V8 itself, now becoming almost universal in Grand Prix racing, must currently be the most important single factor in the future of top level motor sport. Yet Duckworth is cautious: he says there is still an “infinite” amount of development remaining in it, but qualifies this by adding that the “degree of success will be shown by future results”.
Seen among the spectators at the penultimate Can-Am race of 1969, at Riverside, Duckworth’s presence naturally gave rise to wide rumours that he was thinking of turning to this form of racing for his next project. To an engineer whose career has been spent with pure racing units, the American scene must be one which seems ripe for take-over. Not only do Can-Am cars utilise basically standard V8s, but the Indianapolis “500” and USAC National title have become a battle between singleseaters powered by turbo-charged versions of the Meyer-Drake Offenhauser engine which was first seen at Indianapolis 35 years ago. Duckworth’s record would suggest that a Cosworth Can-Am or USAC Formula engine would be an all the way winner. But the Americans are wont to change their engine regulations at short notice, as Andy Granatelli found to his cost with the famous turbine car of 1967. “Constant changes of engine capacity make (the USAC formula) impractical”, says Duckworth, and is hardly more confident in the future of Can-Am. With an unlimited capacity engine permitted by the present regulations, his own solution to the Can-Am requirements, bearing in mind the necessary package size of the unit, the power per pound and fuel consumption, would be a lightly stressed V8 of between 10 and 12-litres. The result, a racing engine producing in the region of 800 b.h.p., could be sold initially to one or two favoured teams, as was the case with the F1 V8, before being put on general sale. If it worked as well as one has come to expect of a Cosworth engine, then there would be a large number of teams knocking on Cosworth’s front door. Whether the American regulation framers would be quite so enthusiastic is quite another question, one about which Duckworth has his doubts. A restriction on capacity could make a very expensive engine into so much useless metal.
Where they are made
The design of a Cosworth engine is not just calculated to conform with capacity regulations and the required outside dimensions. Equally important is that it should be possible to manufacture each component on the machines which are available. Cosworth’s equipment would make any engineer envious and Duckworth—who previously had no knowledge of machine tools—has a self-taught knowledge of the machinery which fills his new machine shop. He is particularly proud of some of the tape-controlled machines like the newly-acquired Spiromatic 2B36 jig mill with tape dial-in, programmed via a specially punched tape to drill and tap castings without any attention from the operator other than to remove finished parts and mount fresh ones for the process to re-start. Castings are bought-in and even some turning sub-contracted because “we can get it done well and economically outside, so that we can concentrate on more important things”. But “we do do pretty well all the machining apart from cams, cranks, pistons, etc., which are made outside”. Production blocks for the FVA (they are manufactured for the heavy-duty requirements of the Lotus twincam engine) are checked with an ultra-sonic wall thickness tester, half of them being rejected as a result. A fatigue-testing machine “suffers from a remarkable lack of use”, but an argon-arc welder is busy fabricating pipes, fuel injection trumpets and sumps.
In complete contrast with the tape-operated automatic equipment, there is still no substitute for the human eye when it comes to “fettling” cylinder heads and inlet ports. Duckworth himself modifies the first of the chamber designs from his own knowledge, after which a man with a steady eye and a hand tool makes each one just like the last one. A Tracemaster copy milling machine is another recent arrival, capable of copying a pattern to plus or minus less than one thou’. Duckworth says they’re “hoping to rough combustion chambers out” with it, although it doesn’t mean the end of the hand-work. Rebuilds of existing engines and the assembly of new units take place in the same shop. Both the DFV and the FVA are complicated engines and if a competitor runs into trouble at the track, by far the best thing to do is install a fresh engine and send the troublesome unit back for its rebuild. Because of very heavy seasonal pressure there are very few FVAs being sent back to Cosworth for rebuilds these days, most of this work being handled by outside concerns. However, all the Formula 1 units are dealt with exclusively in Northampton. It used to be rumoured that certain drivers received “special” engines, built to produce slightly more power than “run of the mill” units, but this simply is not so. On the other hand, some drivers imagine that one or other of their engines is a “flyer”, while they suspect others of being “rogues”. Duckworth says this is not the case at all, for customers’ engines are rebuilt from the same components, without any switching of parts and there is no question of favouritism. What does tend to happen is that an engine may give marginally better or worse results when new, because of variations in the setting of the fuel cam (which is notoriously difficult to set correctly on the test-bed). Thereafter, regardless of how many times that particular engine has been rebuilt or repaired, the driver tends to think it had the same characteristics as when he tested it the first time.
For the moment, Duckworth is busily engaged on getting the BDA into production and solving the problems of his Grand Prix car. He has also been involved with the new Formula Three which comes into force on January 1st, 1971. The FIA has settled on a 1,600-c.c. limit, and any engine which has received a Group 1 homologation (5,000 produced within a 12-month period) will be admitted, regardless of whether or not it has overhead camshafts. The current Formula Three requires a 36-mm. inlet restrictor, which has done nothing to restrict power outputs for the simple reason, as Duckworth points out, “that it is far too big”. The new 1600 F3 will involve a 20-mm. inlet restrictor, which Duckworth believes will keep the horse-power down to 520 b.h.p. at engine speeds of less than 8,000 r.p.m. This should keep expense down by permitting the reliable use of standard parts, unlike the currently very expensive adaptations of the basic Cosworth MAE with its special steel cranks, rods, pistons and valve gear.
One of the complaints about the current F3 is that wheel rim sizes increase every year, permitting less skilled drivers easily to exploit their car’s roadholding to the full. This makes for excellent racing but makes it difficult for truly outstanding young drivers to shine unless they have the very best equipment. Duckworth has suggested that there should be a limit on tyre tread width: this would reduce the cornering power and introduce an element of throttle control, which is so important in a training formula. Like the existing Formula Three, the British Formula 5000 adaptation of the American Formula A requires that engines be based on standard units as far as cylinder blocks and heads are concerned. The FA/F5000 engines have proved unreliable because they also use many standard internal parts which are (not surprisingly) prone to break in a racing application. There was an American V8 block of the Boss Mustang variety idly sitting on a bench during MOTOR SPORT’s visit to the Cosworth factory, but then we understand that Duckworth has already had a close look at the Chevrolet. One thing is sure: if he does get around to doing a proper racing conversion of either one of the American V8s, it will be far more expensive than the prices asked by the small American tuning concerns which concentrate on this work at present. Racing is not only competitive from the driver’s aspect, but competition naturally makes it necessary to increase engine revs, something which in turn makes standard parts (as opposed to components designed and manufactured with a racing application in mind) subject to sudden and expensive failure.
Duckworth’s co-directors have all been connected with him since the firm’s earliest days. Mike Costin, the Development Engineer, was with De Havilland before moving to Lotus and becoming co-founder of Cosworth; Bill Brown, General Manager, is responsible for the intricate task of chasing up suppliers and is the person best known to customers on the telephone, while Benny Rood was once a motor-cycle racer whose engineering business was absorbed by Cosworth in early days. Yet the great success of Cosworth has been attributed to the lack of “committee” work and to Duckworth’s innate “feel” for cutting across already established engineering principles. British racing owes him a debt which history will record in record books for years to come.—M.G.D.