Turn your volume up to 11 and sit back and enjoy 90 seconds of pure, unadulterated noise in this Cosworth Formula 1 engine test as a CA2010 engine is pushed through its paces on Cosworth’s transient dyno at speeds of up to 18,500 rpm.

The Cosworth CA engine represented a pinnacle in the design of high-speed naturally aspirated engines for Formula 1 racing. Calling on over 50 years of experience that dated back to the legendary DFV, Cosworth created their most ground-breaking engine ever, capable of reaching 20,000 rpm in 2006.

Sadly, in 2010 the engine speeds were restricted by regulation to 18,500 rpm, but that didn’t stop the cars of that era sounding amazing.

We stock race-used parts from Cosworth’s TJ and CA engines that can be displayed for all to see. Just keep an eye on our Memorabilia section of our on-line shop to see what’s currently available. Each part can be bought directly from the website shop and delivered straight to your door.

If you enjoy listening to the roar of a Formula 1 engine being put through its paces then you can download it as a ringtone for your phone (Android only, sorry!).

The Need for Speed

In the ear-splitting ‘90s and 2000’s, engine speed was THE metric that Formula 1 engine manufacturers wanted to push to the limit. In a naturally aspirated engine, the goal is to get as much fuel and air into the combustion chamber, and the best way to do this is to run the engine as fast as possible. Unlike power figures, engine speed is something that rival competitors can measure from the side of the track, using audio recording equipment.

It is generally acknowledged that BMW were the first F1 engine manufacturer to break the 19,000 rpm limit in 2002. Toyota soon leapfrogged their fellow German rivals, reaching 19,200 rpm in 2005 despite the introduction of rules to increase engine mileage. These regulations caused a brief respite in the push for speed, but by the end of the V10 era in 2005 the entire grid was running at 19,000 rpm.

Switch to V8 Power

In an attempt to curb engine power, the FIA mandated a switch from 3.0 litre V10s to 2.4 litre V8s in 2006. However, this had little or no effect on the push for speed. Cosworth’s Bruce Wood explained how they did this to Race Engine Technology as part of their exposure on the CA . “To go faster you just have to keep making the bore bigger and the stroke shorter and sort out your valves.

“While developing the TJ we did tests on our single cylinder rig of 96, 97 and 98 mm bores – it was all about higher speed. We were considering a bigger bore and bigger valves and a compound valve angle before the mandatory switch to V8s was brought in. We had thereby established that combustion was OK with the 98 mm bore (the maximum permitted in 2006), so there was no reason not to move to it…”

In 2010 Cosworth returned to Formula 1 with the CA2010. By this time the rules had necessitated a cap in speed of 18,500 rpm. Nevertheless, the cars still sounded incredible, and drop in speed didn’t prevent engines like the CA2010 from reaching power figures of over 775 bhp.

Keeping It Together

20,000 rpm is a huge achievement – as this Formula 1 engine test video shows, even 18,500 rpm sounds terrific, but running at these speeds results in a number of engineering headaches. First of all, the inertia loads on the moving components roughly increase with the square of the speed. So a jump from 10,000 rpm to 20,000 rpm might be double the speed but will result in a quadruple increase in inertia loads.

Cosworth CA2010 F1 Engine

The second problem with increasing engine speed is that there is a corresponding increase in vibration. This was compounded with the switch from V10 to V8 engines in 2006. A V8 engine like the CA equipped with a flat plane crankshaft will naturally have out-of-balance vibration in a horizontal direction.

Bruce Wood gives a further insight into the problems that this horizontal shaking of the engine would cause on the CA. “When we first started running the CA, the scavenge pumps, which are held onto the sump with horizontal bolts, would fall off. Those are 8 mm cap screws, the heads of which snapped off because of the unbalanced force, which is why our scavenge pumps are now secured by Multiphase bolts!”

Torsional Control

One other problem to resolve before unlocking 20,000 rpm is torsional resonance of rotating components like the camshafts and crankshaft. Torsional resonance has always been an issue in Formula 1 engines – Keith Duckworth had to resort to flexible compound gears to keep the gear drive intact in the DFV in 1967. Fast forward 4 decades, and double the engine speed, and the torsional problems become a whole lot worse.

“In terms of the torsional vibration inside the engine, we knew what we were up against, which is why the CA has far more damping devices in it than our previous V10 engines.”, says Wood. “We have a ‘compliant’ gear train that has been in our Formula One engines for years, then in addition (to two dampened compound gears) the CA has compliant quill drives within each of its two auxiliary drives, a big viscous damper on the back of the crankshaft, viscous dampers on the back of each camshaft and friction quill dampers in the front of each camshaft. That means in total it has 13 dampers – 14 when fitted with KERS.”

Modatek provide parts and consultancy for a wide range of Cosworth historic engines, including the CA. Want more information? Just send us a message through our Contact Us page.

In our latest technical blog we take a closer look at pneumatic valve springs. They’ve been on Formula 1 engines for the last 30 years, but how do they work and why aren’t they in our road car engines?

Have you ever watched a Grand Prix and watched on in dismay as your favourite driver’s pit stop seemed to last for an eternity. Worst still, mechanics seemed to be stood around the car doing nothing, with the exception of one mechanic who is frantically trying to connect up an air hose to the side of the car.

Formula 1 car pitstop

Chances are that the engine of your favourite driver is currently suffering from what normally turns out to be a terminal failure of the pneumatic valvetrain system. The hapless mechanic will be trying to add air or nitrogen to the system to counteract a leakage somewhere deep within the engine.

So what is a pneumatic valve spring and is it really better than a conventional wire valve spring? Well, in a nutshell, a pneumatic valve spring is basically a cylinder of pressurised gas (air or nitrogen) which behaves in a similar manner to a wire spring by creating an upwards force on the inlet and exhaust valves.

All Formula 1 engines run some sort of pneumatic valvetrain and pneumatic valve springs are common on a few other premier motorsport categories like Moto GP. It goes by a number of different names – some engine manufacturers refer to it as the PVRS (pneumatic valve return system) or AVS (air valve spring), but they all work in roughly the same way.

Pneumatic valve springs were pioneered by Renault for its Formula One engine back in the later 1980s, and have long since been adopted by all the other manufacturers in Formula One. Cosworth switched to its own system in the early 1990s – remember the fuss when it supplied engines that were upgraded with pneumatic valve springs to Benetton and not McLaren, resulting in a frustrated Ayrton Senna?

Pneumatic Valve Spring Components

In a pneumatic valve spring air or nitrogen is fed into drillings in the cylinder head at a regulated pressure of (typically between 10 and 20 bar, depending on several parameters such as the volume of the body and the mass of the valve) from a supply bottle mounted externally on the chassis or from a compressor, and then introduced into the pneumatic valve body through a non-return valve.

Inside the body, the gas will be compressed to more than 80 bar as it is squashed by movement of a disc secured to the valve with cotters, called the reciprocating seal carrier. Around the outside of this carrier is a seal that runs inside a honed bore inside the body (sometimes the bore is in a separate sleeve that is inserted into the body). At the base of the body is another seal carrier which houses the stem seal.

Pneumatic Valve Spring Components

Both seals are usually energised by a combination of the internal gas pressures in the body and metal garter springs which forces the seals out even when there are lower gas pressures present.

The exact construction and materials that are used to manufacture the seals remains a closely guarded secret between the seal manufacturers and the engine manufacturers. This is even true on the seals that we supply to customers rebuilding historic Formula 1 engines.

If you want to see some pneumatic valve spring components up close then take a look at Brian Garvey’s excellent tear-down.

Controlling Oil

Both the reciprocating seal and the stem seal will be required to maintain a perfect seal whilst running at phenomenal speeds. It is imperative that the surfaces aren’t dry, otherwise the seals will overheat. The running surfaces of both seals have to be lubricated with oil, but the introduction of oil has to be carefully controlled.

Thankfully, in the environment that the pneumatic valve springs find themselves in, there is plenty of oil around. In the valvetrain chest, which is the space around the pneumatic valve bodies, there is always a mist of oil present. This oil finds it way onto the walls of the bores in each body, and can then lubricate the reciprocating seal.

The reciprocating seal has a shallow groove running around the middle that fills with oil, and when the seal moves the pressure in the groove increases until it reaches a point where the oil is forced into the body. Once in the body, the oil can then lubricate the static stem seal.

But too much oil in the body can be catastrophic. Unlike gas, oil is virtually incompressible. If too much oil gets into the body of a pneumatic valve spring then it will hydraulically ‘lock up’. This in turn results in massive forces on the seals and carriers, which can lead to catastrophic engine failure.

The amount of oil present can be maintained by using small pressure relief valves (PRV), which will open when too much oil is present, and if located correctly they will vent the oil out of the body. The operation of these valves is similar to the much larger PRV that are found next to the oil pressure pump.

Normally each pneumatic valve body will contain its own miniaturised PRV. The PRV consists of a tiny ball bearing which is pressed by a spring against a conical face to seal off the oilway. When the pressure in the body reaches a certain level the spring force is overcome and the ball moves, opening up the oilway. The installed length of the spring is carefully preset with a graded-length screw, which is selected during build to ensure that the PRV opens at the right pressure.

So, when the amount of oil present in the body gets too much, the pressure in the body opens up the PRV and the oil escapes out. However, it’s not always as simple as this. Sometimes gas will escape with the oil, depleting gas from the body, and so more gas will be required. Effectively, the body has to take a gulp of gas from the supply source.

If this occurs repeatedly or if the seals start to leak then the gas bottle will soon be exhausted and will have to be topped up, which brings us back to the lengthy pit stop, frustrated driver and exacerbated mechanic.

Normally this leakage doesn’t just go away, and if there isn’t enough replacement gas added during the pit stop then eventually the bottle will run out. In Formula 1 the emphasis is on saving engines for future use, so the team will normally elect to retire the car rather than risk a loss of valve control and ultimately engine failure.

Pressure Regulator

One important component in the pneumatic valvetrain is the pressure regulator. This device is spring-loaded and can be set to ensure that the supply of the gas going into the engine is at a consistent pressure regardless of the pressure of gas in the bottle.

Cosworth AVS Regulator

The pressure regulator on a race car is a more complicated version of one that might be found on a diving bottle. It’s normally mounted in a cavity in the sidepod, and as such is subjected to extreme vibrations and temperatures.

This regulator also typically has two pressure sensors installed – one will read the pressure from the bottle, and the other will record the pressure going into the engine. A deteriorating bottle pressure is a sure sign that the engine is consuming gas.

The Benefits of Pneumatic Valve Springs

Thanks to its ability to be able to cope with higher loads, a pneumatic valve spring offers two significant benefits over a wire spring. First of all, most wire sprung engines are limited to around 12,000 rpm because of the strength of the springs. If you want to go past that speed then you’ll probably have to switch to a pneumatic valve spring.

Secondly, most race engine designers want a profile that will give rapid opening of the valve, followed by the required duration of opening and then a rapid closing of the valve. Again, wire springs can be a limiting factor for the aggressiveness of the cam profile. If you want to run aggressive cam profiles then you’ll probably benefit from a pneumatic valve spring.

So how is a pneumatic valve spring superior to a wire spring? An esteemed engineer by the name a of Professor Gordon Blair wrote a series of articles for Race Engine Technology that examined valve springs in great detail.

In one of these articles (“Steel Coils Versus Gas”, RET 23) he included an in-depth comparison between a single-coil steel spring and a nitrogen-filled pneumatic valve spring. He analysed the amount of valve bounce from both systems and surmised that the valve control with a pneumatic spring was superior to that provided by the steel spring.

He highlighted several reasons for this improvement in valve control. First, the mass of the reciprocating seal and carrier was only around one-fifth of that of the steel spring. Also, the gas spring displays an inherent damping behaviour thanks to the hysteresis of the gas. And perhaps most important, unlike helical wire springs, pneumatic valve springs cannot suffer from surge problems.

Will We See Pneumatic Valve Springs on the Road?

The simple answer to this question is no. For a start, most road engines don’t need to run at the speeds seen in Formula 1. Moreover, most road engines don’t need to use aggressive cam profiles. So a conventional wire compression spring can do the job and there are plenty of metallurgical advances that make a wire spring incredibly robust.

The other problem is that, despite 30 years of research, pneumatic valve springs are still relatively unreliable, certainly when compared to a wire spring. There is an incredible amount of care and attention that is required when assembling a pneumatic valve spring. Just the slightest amount of dirt or debris can cause the seals to leak. In the OEM world of mass-production, it would be extremely difficult to assemble pneumatic valve springs on a production line within a sterile environment.

So it’s highly unlikely that you’ll ever see a warning sign flash up on your dashboard whilst you’re driving to the shops telling you that you need to come in for an unscheduled pit stop for a bottle top up.

Modatek can supply parts and consultancy to customers who want to rebuild the pneumatic valve springs in their historic engines. Get in touch to find out how we can help.

To the untrained eye, the skirt of a piston would appear to be perfectly cylindrical. To be fair, even to the trained eye that would seem to be the case. But if you were to roll a piston on its side down a flat slope, you’d notice that it gradually wanders off path – that’s because the shape of the skirt isn’t round! In this technical blog we uncover what shape it actually is, and why.

Cosworth PA0598 MAE Piston

Looking at the piston side on, the profile of the skirt is actually barreled, not straight. And looking from the top, the shape of the piston is oval, not circular. Actually, you’d be hard pressed to notice this by eye, as we are talking in terms of fractions of a millimeter, but its enough to make a difference when the piston is running in an engine.

This barreling and ovality is intentional – it’s there to take account of the deformation of the piston when it is loaded up and heated up, which has the potential to result in inconsistent bore clearance and possible seizure.

Bore Clearance

The clearance between the skirt and the cylinder bore’s inner diameter is critical, for many reasons. Too much clearance will allow the piston to tilt in the manner described above, whereas insufficient clearance can result in catastrophic seizure. Note also that tilting the piston can lead to increased movement of the rings and more wear of the ring grooves.

Generally, the running clearance is dictated for full-load operating conditions. The clearance is influenced by engine speed and piston temperature, so other operating conditions may also need to be considered. For example, the piston temperature and hence expansion can increase during lean running, which can lead to an unforeseen reduction in clearance.

Note that the recommended bore clearance is also affected by the materials of the piston, cylinder liner (if present) and cylinder block, specifically with regards to dissimilar thermal expansion rates. Even the two grades of the most prolific aluminium alloys in use these days – 2618 and 4032 – have different thermal expansion rates that mean the bore clearance needs to be different for each alloy.

Normally a piston made from a 2618 alloy will require more bore clearance than one made from a 4032 alloy. That isn’t an issue during running, as once the pistons have reached their operating temperatures the chosen bore clearances will be the same for both alloys, but it can cause a problem on start-up or when cold as the extra clearance required for the 2618 piston can create more noise.

Piston Skirt Loads

In a conventional reciprocating four-stroke engine the small end of the con rod will articulate forwards and backwards around the gudgeon pin as the crankshaft completes one revolution. As a result, the piston will be subjected to side loads that vary for each of the four piston strokes.

The largest side load by far will be during the power stroke, when a combination of inertia and gas loads will push the piston sideways as well as downwards. This is commonly referred to as the major thrust side load, and its direction is always on the side that is against the direction of rotation.

For example, if the crankshaft rotates clockwise when viewed from the front then the major thrust side will be on the left-hand side. The opposite side of the piston is referred to as the minor thrust side, and the minor thrust side is subjected to a lower sideways force during some of the other stages of the four-stroke cycle.

As well as deforming the shape of the piston, these major and minor side loads can have an extremely detrimental effect on the operation of the piston. For example, they can promote tilting or rocking of the piston in the bore, which can exasperate wear at the top of the lands and at the bottom of the skirt. This type of movement can also create vibrations that are then radiated through the engine and can even be detected and heard outside the engine, commonly called piston slap.

Piston Skirt Deformation

The external loads imparted to the piston from the gas pressure in the combustion chamber above the piston are immense, and will cause strain deformation of the piston shape and in particular the piston skirt area. The resultant shape of the skirt after this deformation is a complex one.

During the firing cycle the crown will be deformed inwards by the loads from the gas pressure in the combustion chamber, effectively splaying the skirt outwards. If the piston were shaped like a cup then the skirt deformation would be circular, but of course the piston incorporates the pin boss and structural ribs, so the deformation isn’t even.

The pin bosses provide support along the pin axis, so the skirt deformation at the piston’s open end can be broadly described as oval, with the diameter increasing along the pin axis and decreasing at the major and minor thrust directions. The shape isn’t completely elliptical though, as there are additional strain deformations in the skirt to consider. The pin bores themselves are subject to loads from the gudgeon pin, which can effectively flatten the skirt, and the skirt itself is deformed by the contact pressure acting on the skirt from the cylinder bore.

Superimposed on top of these strain deformations is the effect of thermal expansion. The temperature is highest at the crown, so at working temperature the diameter around the crown and around the lands will expand outwards more than the diameter of the skirt.

There will also be a gradual reduction in diametric expansion down the length of the skirt, decreasing as one goes down the skirt away from the crown. Like the strain deformation, the thermal expansion of the skirt isn’t even due to the presence of the pin bosses and structural ribs.

When the strain deformation and thermal expansion of the piston are combined, the resulting skirt topography ends up being a combination of broadly oval sections that increase and decrease in size down the length of the skirt, following what is best termed a barrel profile. Note that some manufactures refer to the barreling as cam or curvature.

Now, if one assumes that the bore in the cylinder block or cylinder liner is perfectly cylindrical during running (which it isn’t) then the shape of the piston skirt’s profile when it is machined has to be carefully designed so that it deforms (both by strain deformation and thermal expansion in the manner described above) when running to create a complementary perfect cylinder. As a result, piston designers will have to specify the necessary ovality and barreling of the skirt and lands that the machinist will need to achieve at room temperature.

Piston Skirt Design

When designing a piston, special attention is paid to the topography of the skirt. The shape of the skirt can have a large influence on engine performance, as it can control the tilting of the piston, lateral displacement, oil film thickness and frictional losses. As mentioned, the shape of the skirt is defined by barreling and ovality.

Piston Skirt Barreling & Ovality
Piston Skirt Barreling & Ovality

The simplest design of piston skirt will have the same amount of ovality along the entire length of the piston, and in some cases this will suffice. The designer ensures that the area of the skirt that will have the largest amount of expansion is catered for, but the remaining sections of the skirt will have a looser fit during running than might be desired, which can cause problems with reduced stability of the rings and poor guidance of the piston in the bore.

Going one step further to solve these issues, the ovality at staged sections down the length of the piston can be prescribed as a series of ellipses, which in turn are defined by their major and minor diameters, and which change in size following a barrel form. Commonly, this is done at discrete points roughly every 1-2 mm down the length of the piston.

The derivation of these ovalities would traditionally have been found from trial and error, requiring a series of engine tests to monitor the condition of the skirt. The pistons could only be inspected by removing them, hence such testing would require a number of rebuilds.

Too much contact between the skirt and the wall of the bore would be indicated by scuffing, which would indicate the areas that needed to be addressed. In the worst case, overly excessive contact would cause a catastrophic seizure, making it impossible to inspect the running surfaces left in the piston debris.

With the advent of FEA (finite element analysis), piston designers can calculate the combination of strain deformations and thermal expansion to derive the required skirt profile. Load cases for various stages in the operating cycle and different running conditions can now be considered to optimise a single profile for a number of scenarios. In addition, the piston designer can now take into account the effect of bore distortion.

The assumption that the bore is perfectly cylindrical during engine running isn’t necessarily true, even if the bore has been honed at temperature and with torque plates fitted to simulate distortion from the assembly and running loads induced by the cylinder heads, sump or mains caps, and crankshaft mains journals. There will still be some distortion to this circular section, which normally corresponds to the locations of the studs or fasteners next to the bore.

FEA can be a very powerful tool to predict the actual shape of the cylinder bore, and the results from it can be used to define the required piston skirt profile. Piston designers are turning to increasingly intricate computer-derived simulations, creating complex dynamic multi-body simulations based on the structural and thermal FEA of the piston and bore. The results of this analysis can create skirts that have the optimum contact conditions for a wide range of operating conditions.

So, the first objective of a good skirt design is to ensure that there is the correct clearance between all the parts of the lands and load bearing area of the skirt and the wall of the cylinder bore. As we’ve already seen, there are a number of factors that can influence bore clearance, and the chosen profile must take into account each of these factors.

Nowadays though, designers are also considering the dynamic behaviour of the oil that is trapped between the skirt and the bore. There has been plenty of published research into the effects of piston skirt profile on the lubrication of the piston and its subsequent influence on friction.

Such research concludes that the frictional behaviour of the piston is heavily influenced by the mode of lubrication (hydrodynamic, mixed and boundary), which in turn is constantly changing throughout the four-stroke cycle owing to the reciprocating motion of the piston. Further, it has been noted that almost 80% of the frictional power loss from the skirt occurs during the power stroke.

Piston Skirt Machining

The required shape of the skirt by the designer can be machined by mounting the piston on a rotating spindle, which is then moved in and out of a rotating grinding wheel whose movement is controlled by an eccentric cam. Manufacturers use specialist CNC piston turning centres for this, and they can be programmed by entering the required diameters and ovalities at the discrete sections down the length of the piston, as defined by the designer. Note that this type of machining will extend up above the skirt to the lands, and can include the ring grooves themselves.

More advanced machining centres now use modern diamond turning instead of grinding to achieve a tighter tolerance on the skirt profile. Some machines will need to be temperature controlled or will have an element of temperature compensation built into them to eliminate the potential problem of thermal expansion during machining – even a shift of 10 oC above room temperature can push the profile out of tolerance. In addition, as design techniques have advanced over the years the required profiles have become more complex.

As a result, skirt machining has had to be developed to allow for more complicated designs, such as the need for asymmetric skirts and even concave reverse-profile barrelling in certain sections of the piston. (On a piston with asymmetric skirts the bearing area of the major skirt is designed to be able to withstand the higher loads from the major thrust side, and as the deformation of the skirt is partly linked to side loading, it also means the skirt profile can end up being different on the major and minor skirts.)

Skirt Coatings

In theory, if the skirt profile has been designed correctly, and there is an adequate supply of oil, then there shouldn’t be any contact between the skirt and the bore, as there will always be a film of oil separating them. In reality though, that isn’t always the case – cold starts, oil contamination, overheating, extended high-speed running and many other problematic running conditions can lead to a breakdown of this hydrodynamic oil film.

Several piston manufacturers therefore apply a coating of some sort on their piston skirts, to allow the piston to survive the occasional skirt-to-bore contact. Although the major source of friction will come from the rings, there is some frictional saving that can be made with skirt coatings, and there is plenty of evidence to show that the application of coatings has helped with the durability of both the piston skirt and the cylinder bore surfaces.

Some manufacturers apply a dry-film lubricant that can be sprayed on after the skirt has been machined. For example, some of our Cosworth pistons have a fluoropolymer coating that has the tradename Xylan.

Cosworth TJ Formula 1 Piston

The thickness of the coating on the skirt can be quite thin, in the region of between 8 and 13 microns, and with such thin coatings there doesn’t need to be a manufacturing allowance when machining the skirt, as this coating is intended to wear (as is the case on the Cosworth TJ Formula 1 piston pictured above). Sacrificial coatings like these can also be extremely useful in identifying areas of high wear on the skirt – as long as the engine is stopped before metal-to-metal contact occurs!

This feature on pistons is based on an article written by Modatek’s Matt Grant for Race Engine Technology, issue 111. You can purchase a back copy of this publication here: https://www.highpowermedia.com/Product/race-engine-technology-issue-111

It’s now 57 years since the introduction of what would become the most successful Formula 1 engine in history. The Cosworth DFV engine won on its debut at the Dutch Grand Prix in 1967 and would go on to dominate in Formula 1, racking up 155 victories. No other engine has come close. Using notes that were prepared by company founder Keith Duckworth and chief designer Mike Hall, along with feedback from one of Cosworth’s long serving employees Malcolm Tyrrell, we take a look at the birth and technical spec of the early iterations of this incredible power plant.

Cosworth DFV in Lotus 49

Cosworth had already become an established name in the motor racing world with its successful Formula Junior, Formula Three MAE, Racing Lotus Twin Cam and Formula Two SCA engines, based on standard cylinder blocks with bespoke cylinder heads. But then, at the end of 1965, Cosworth embarked on a completely new project that would establish its name in racing for ever. In exchange for £100,000 from the Ford Motor Company, Cosworth would create a four-cylinder Formula Two engine, the FVA (Four Valve Type A), which would then be ‘doubled up’ to create a V8 for Formula One, the DFV (Double Four Valve).

For nine months in 1966, Duckworth entrenched himself in a room at his house, working every day from 9am until past midnight. Once a week he would visit the factory to hand over schemes and drawings, and get feedback on the progress of the manufacture of his designs. Chief designer Mike Hall joined Cosworth from BRM in October 1966, and was instrumental in the detail design of the major components and all auxiliaries, working from Duckworth’s layouts. The engine was completed in April 1967, winning first time out at the Dutch GP in June in Lotus’ new Type 49 chassis.

As per the contract with Ford, Duckworth and his team started with the FVA, incorporating the lessons learnt on this four-cylinder engine into the design of the DFV. Duckworth set a power output target for the FVA of 200 hp, reasoning that the DFV would need 400 hp to be a winning engine. As it turned out, both these performance targets were comfortably exceeded.

Engine Mounting and Layout

The oval shape of the Lotus 49 monocoque meant it would be difficult to extend the sides of the chassis down past the engine, resulting in the dictate from Lotus’ Colin Chapman that the engine should be a stressed member of the chassis from the outset. Part of the suspension would also mount directly to the engine; schemes and handwritten notes still at Cosworth show that the layout of the suspension went back and forth between Duckworth and Chapman before settling on the final positions.

Cosworth DFV in Lotus 49

The engine was bolted to the bottom of the rear of the monocoque bulkhead by a wide-based bracket on the sump, with the intention that these lower mounts would transmit the shear loads between the chassis and engine. Anecdotally, Duckworth would remark that the distance between the two engine mounting bolts at the front of the engine was chosen to match the width of Jim Clark’s posterior! At the top of the engine, the left- and right-hand cam covers were connected to the top corners of the rear bulkhead with thin triangular shaped steel plates, so as to take the tension and compression loads. The plates were also intended to deflect under the thermal expansion of the engine, calculated at the time to be 0.015 in (0.4 mm).

Duckworth was intent on keeping the engine as compact as possible, so to keep the front of engine flat and uncluttered he positioned the water, oil and fuel pumps along the sides of the cylinder block underneath the exhaust pipes. That would also allow for a lower centre of gravity, recorded at 4.6 in (116.4 mm) above the crank centreline. The left- and right-hand pump pulleys were driven at the front of the engine using a Uniroyal toothed rubber belt, driven from a pulley connected to the second compound gear running at half engine speed.

The overall dimensions of the Cosworth DFV engine showed that it was wider and higher than it was long – the height of the engine from the bottom of the sump to the top of the trumpets was 23.3 in (590 mm), the length measured at 21.6 in (550 mm) and the overall width between the extremities of the cam covers was 26.8 in (680 mm). With no clutch or starter motor fitted, the dry weight of the engine came out at 350 lb (159 kg).


With the rules stating that the DFV had to have a capacity of 3 litres, the individual cylinder capacity was reduced from 400 cc on the FVA to 375 cc on the DFV. The DFV copied the FVA’s bore diameter of 3.373 in (85.67 mm), with the smaller cylinder capacity achieved by a shorter stroke of 2.550 in (64.77 mm) to give a total capacity of 2993 cc.

Cosworth DFV Block & Sump

Like most of the other major castings, the cylinder block was created from fully heat-treated 7% silicon LM 25 aluminium. At the time, this material was considered to be the best commercially and readily available alloy in the UK, with the highest proof stress and Brinell hardness, and good casting and machining capabilities.

In the cylinder block were wet liners machined from centrifugally spun castings, made from chrome vanadium alloy iron. They were interfered into the block with two O-rings at the bottom of each liner to seal off the crank case chamber. The top of the flange on the liner had a recess for a Coopers sealing ring.

The sump was made up of box sections that ran from front to rear and across the engine to try to maximise stiffness. The box section at the front also carried water to connect and balance the two water pumps on either side, while the rear box section carried the scavenged oil to an outlet on the left-hand side of the engine.

Water Pumps

The first iteration of DFV engines had a single water pump, mounted on the right hand side, directly in front of the oil scavenge pump. Later DFV iterations featured two water pumps mounted on either side of the engine, so that each bank of the engine was cooled by its own water pump.

These later water pumps used 2.5 in (63.5 mm) diameter centrifugally bladed impellers contained within a volute spiral casing. Each pump had a maximum flow capacity of 45 gallons per minute (204 litres per minute), and the water flow from each pump was sufficient to restrict the temperature rise from the water pump inlet to the engine outlet to 7 C.

Fuel Pressure Pump

A mechanical fuel pressure pump was positioned at the front of the left-hand side auxiliaries. This gear-type pump was capable of delivering 40 gallons per hour (182 litres per hour) at maximum engine revs against a back pressure of 120 psi (8.3 bar). An electrically driven high-pressure pump was activated for starting purposes; this auxiliary pump was then switched off when the engine speed reached 2500 rpm.

The fuel pump was positioned as far out as possible so that it would be cooled by the air stream around the engine, in a bid to keep the fuel cold and prevent fuel vaporisation. A further measure to reduce fuel temperature included isolating the pump from its supporting bracket with a Tufnol insulator.

Oil Pressure Pump

At the rear of the left-hand auxiliaries sat the main oil pump incorporating an integral filter. The oil pump contained a 0.8 in (20.3 mm) wide Hobourn Eaton lobe-type rotor, which at maximum engine revs could displace 11 gallons per minute (50 litres per minute) against a 100 psi (6.9 bar) pressure relief valve setting.

A transfer pipe fed the filtered oil from the pump to the crankcase main oil gallery. From here the oil was fed to the five pairs of main bearings, and then through the cross-drillings in the crankshaft into the eight pairs of big-end bearings. It was estimated that at 11,000 rpm a minimum oil pressure of 68 psi (4.7 bar) was required to force the oil into the centre of the crankshaft due to the centrifugal head generated at this speed.

Oil Scavenge Pump

On the right-hand side of the engine were the scavenge pumps, two pairs of Hobourn Eaton form rotors with a separator between them, drawing oil in from the scavenge chambers in the sump. (Later pumps would feature Roots type rotors.)

Each cylinder head featured two head drains, one at the front and one at the rear, which fed down to the scavenge chambers. Finally, oil from the oil pump pressure relief valve was also fed directly into one of the scavenge chambers.

Early DFV engines suffered serious issues with bad draining of oil from the cylinder heads. The first engines only had a scavenge capacity of twice the oil pressure pump, so the scavenge pumps were unable to handle the oil and the blow-by gases. As a result, the blow-by gases would go up the head drains, preventing oil from coming down the other way, which caused the heads to fill with oil, eventually venting the oil to atmosphere and draining the oil tank.

Unable to increase the size of the head drains, a temporary sliding vane-type pump was fitted to take care of the blow-by. However, this in turned caused more problems, as the conventionally scavenged oil, which measured 11 gallons per minute (50 litres per minute), was now mixed with the equivalent of 40 gallons of air per minute (182 litres per minute). The resulting aerated oil led to the design of a centrifuge which ensured that the mixture returned to the oil tank contained 90% oil and only 10% air. This centrifuge was included in the new design of scavenge pump, which now had a capacity of 55 gallons per minute (250 litres per minute) – five times the capacity of the oil pressure pump.


Behind the front cover was a geartrain consisting of 14 gears. Assembled to the front of the crankshaft was the crank gear, which turned the first compound gear, an assembly of two gears. The smaller of these two gears drove another gear assembly, termed the second compound gear. This assembly was made up of three gears; a large central driven gear sandwiched between two smaller gears. To the left and right hand side of the second compound gear were the first of two head idler gears, driven from the outer gears of the second compound gear. On each bank, the first head idler gear turned the second head idler gear. Finally, the left and right hand second head idler gears drove the respective bank’s inlet and exhaust cam drive gear.

All the gears were made from forged vacuum re-melted EN39B steel blanks, case hardened to a depth of 0.020 in (0.5 mm). Considerable effort in production engineering and quality control was made to ensure that the teeth were accurately ground to ensure concentricity of their pitch circle diameters to the gear bearing bores, resulting in the required backlash and correct involute profiles.

However, despite the great attention to detail given to the design and manufacture of the geartrain, gear problems blighted the first race engines. At debut race of the Cosworth DFV engine, Graham Hill retired with cam gear failure, and broken gear teeth were found in the winning engine of Jim Clark. In addition, the gear failures were compounded by catastrophic valve spring failures when the engines were run at more than 9000 rpm. Cam lobes profiles were redesigned to bring down their maximum torque requirement from 36 lb-ft (49 Nm) to 26 lb-ft (35 Nm); however, even though the valve spring life improved, there were still gear failures.

Using strain gauges, instantaneous stab torques of 300 lb-ft (407 Nm) where recorded – far higher than the original torque calculations used for the gears. What was needed was a way to absorb the shock loading from the camshafts that was destroying the gears.

Cosworth DFV Compliant Compound Gear

In a typical moment of Duckworth ingenuity, the answer came in ‘cushioning’ the second compound gear, which up to this point had been a rigid assembly. The new design of second compound gear contained 12 small quill shafts that allowed the two side gears to rotate over a limited angular displacement relative to one another, thereby storing some of the huge energy from the cam loadings to successfully reduce the loading on the geartrain.


Although the first crankshafts were machined from billets, Cosworth quickly switched to fully forged blanks supplied by Smith Clayton Forge. The material for the crankshaft forging was heat treated EN40C 3% chrome molybdenum nitriding steel. After the crankshaft was machined, it was nitrided in an ammonia atmosphere at about 500 C, resulting in a hardened case depth of around 0.015 in (0.38 mm).

The main journals had a diameter of 2.375 in (60.325 mm) and the crank pins were the same as the FVA at 1.938 in (49.2 mm) in diameter, with the crankpins arranged to give a flat-plane layout. The crankshaft weighed 32 lb (14.5 kg), coupled to a flywheel weighing 8 lb (3.6 kg).

Calculations made during the design of the crankshaft showed that the maximum load would be on the centre main journal, creating a bearing pressure of 6600 psi (45.5 MPa) at 10,500 rpm. The big-end bearing pressures were estimated to be nearly 8000 psi (55.2 MPa). Both the main journal bearings and the big-end bearings were made by Vandervell, from a steel backing with a bronze intermediate layer and a lead indium overlay.

During the early 1970s a Holset-manufactured crank damper was situated on the nose of the crankshaft to reduce torsional vibration. Analysis by both Holset and Vandervell proved that the principal resonance peaks lay between 8,000 and 11,000 rpm. The largest of these peaks was the eighth harmonic excitation of the first order, which occurred at 8,594 rpm, unfortunately in the middle of the running range of the first engines of between 7,000 and 9,500 rpm. The maximum alternating torque was +/- 2,122 lb-ft (2,877 Nm) on the third crankpin, creating a maximum amplitude of +/- 0.95°. The crank damper was removed on later engines in the mid 1970s when the running range was increased away from this peak to 9,000-10,500 rpm.

The crankshaft proved to be an extremely reliable component, but in 1970 there were a series of widely known crank failures. Investigations pointed to a relatively simple grinding error. The corner radii of the crankpin journals were ground both before and after nitriding, but unfortunately the radius on the pre-grinding wheel was too large, which led to the post-nitride grinding wheel going through the nitrided layer, drastically reducing the life of the crankshaft.

Piston & Rings

DFV piston forging material was chosen to be RR58 aluminium alloy (developed by Rolls-Royce). Although this material had a slightly higher thermal expansion coefficient when compared with the high-silicon alloys used on production engine pistons of that era, it remained consistent from 20-200 C. The skirt profile featured tapering along the length of the skirt combined with ovality around the diameter, to provide a diametral skirt clearance of 0.003 in (0.076 mm).

Cosworth DFV Piston

Duckworth sought to minimise the weight of the cast-iron top compression ring, such that the thickness was only 0.030 in (0.76 mm) thick. This would ensure that the ring would stay seated on the bottom face of the piston groove under deceleration and thereby prevent gas leakage past the ring. For the record, the ring gaps were set at 0.017-0.022 in (0.43-0.56 mm).

The gudgeon pin was made from heat treated EN39, case hardened all over, with an outer diameter of 0.813 in (20.6 mm) and an inner diameter of 0.47 in (11.9 mm).

Connecting Rod

Both the rod and the cap were supplied as separate stamped forgings by Smethwick Drop Forgings, using re-melted EN24 steel. The cap was secured to the rod with a pair of 3/8 in (9.525 mm) 12-point big-end bolts, with location provided by two dowel pins. The small-end bush was a steel-backed bronze bearing supplied by Vandervell, finish-bored and honed after assembly. The rod centre distance was 5.230 in (132.84 mm), as Duckworth tried to keep the rod length as long as possible to reduce the secondary out-of-balance forces inherent in a V8 configuration.

Cylinder Heads

The FVA cylinder head featured an included valve angle of 400. Duckworth reduced this further on the DFV to 32° to give a shallower pent-roof chamber and hence a further reduction in surface area and hence less heat loss.

Both the left- and right-hand heads were machined from a common casting. These heads contained 1.32 in (33.5 mm) inlet valves and 1.14 in (29 mm) exhaust valves, both with 7 mm diameter stems. In the middle of the combustion chamber was a 10 mm spark plug. The valve seats and guides were made from aluminium bronze alloys, with 0.003 in (0.076 mm) interference in the head when cold.

Cosworth DFV Cylinder Head

The shape of the inlet ports was kept as straight as possible by Duckworth, with a diameter of 1.02 in (25.9 mm). The ports were fully machined; straight sections were bored, while curved and flared sections into the throats were copy milled. The exhaust ports were completely curved and so had to be copy milled throughout. After machining, the heads were given to the fabled finishing section, where the ports would be polished using a process known as ‘broddling’ within the organisation. It was not uncommon for the finishers to stamp their initials on the side of their cylinder heads so that they could compare dyno test results with each other.

Duckworth had a very rational approach to port design. “I have never believed that there is any point in having a gas flow rig and measuring the flow,” he once said. “I think it is possible to look at the shape of a hole and decide whether the air would like to go through it or not. A hole that looks nice and smooth and has no projections will generally flow easily. Most people start with something so horrible that to create an improvement should be very simple. I would claim that I could arrive at something close to their results from gas flowing just by putting my finger down the hole and seeing what it feels like.”


The camshafts ran in steel-backed white metal shell bearings, again supplied by Vandervell. The bearings were held between dowelled caps and one-piece cam carriers, which like the heads were made from a common casting. Oil was fed up from the main engine gallery into grooves in the middle bearing pairs and then into the hollow camshafts, where it would be directed through drillings in the other cam journals to lubricate the other bearings. The oil from the cam bearings also splash lubricated the tappets, after initial tests showed that feeding oil through holes in the cam lobes was not necessary.

The selected material for the camshafts was EN16T steel, which was liquid nitrocarburized (Tuftride) all over after machining to provide an anti-friction coating. Tappets were machined from EN40, fully ground all over and lapped on the tappet face. The tappets ran directly in the cam carrier, and were 1.25 in (31.75 mm) diameter by 0.9 in (22.9 mm) long.

Understandably, Duckworth paid a lot of attention to the profile of the cam lobes. The Cosworth DFV engine copied most of Cosworth’s other engines of that era and had a lift of 0.410 in (10.4 mm), and symmetrical valve timing with inlet valve opening at 58° before TDC and closing 82° after TDC, exhaust valve opening 82° before BDC and closing 58° after TDC, giving 116° overlap. During build, the tappet clearances were set to 0.010 in (0.254 mm) on the inlet side and 0.015 in (0.38 mm) on the exhaust side.

The ‘Bomb’

In the centre vee of the Cosworth DFV engine lived what Cosworth termed the ‘bomb’, a set of auxiliaries driven from the second compound gear. Within a magnesium centre casting was a Lucas rotating magnet alternator that produced 10 A at 12 V. Also in this assembly was the Lucas Opus ignition system (Oscillating Pick-up System), a plastic drum rotating at half engine speed into which was moulded eight ferrite rods running against a stationary pick-up.

Within the ignition system was a thyristor speed limiter set to 11,300 rpm. This was a standard Lucas product, which Cosworth would then wire into a rubber-mounted box before subjecting it to numerous rig tests to ensure consistent operation over the required speed range, with an overspeed test to check that the speed limiter operated correctly. The trigger disc for the ignition system was mounted on the nose of the crank.

Finally, also in the centre vee was the fuel injection metering unit, again supplied by Lucas. The unit consisted of a stationary hollow sleeve containing a series of radially drilled holes, some of which would feed fuel in from the high-pressure pump and some which would allow fuel out to the injectors mounted in the inlet trumpets. Within the sleeve ran a rotor that also had a corresponding array of radial holes, at selected angles to give the required timing of fuel delivery. Along the centre of the rotor were fuel metering shuttles that oscillated back and forth. A fuel cam that pivoted at the end of the unit controlled the length of the stroke of these shuttles, thereby determining the amount of fuel being delivered. The cam lever was driven by the throttle slides, so that when the throttles were fully open the shuttles could operate a maximum stroke and hence provide maximum fuel delivery to the injectors.

In some of the early races in the late 1960s the Cosworth DFV engine was plagued with fuel vaporisation issues, especially at races in South Africa such as those held at East London and Kyalami. The solution was the rerouting of the return fuel line from the fuel PRV to help cool the fuel.

The Launch of an Icon

The DFV was unveiled by Walter Hayes, public relations director of Ford, at a function at Ford’s Regent Street showrooms on 25th April 1967. In attendance were Keith Duckworth, Lotus founder Colin Chapman and Lotus driver Graham Hill.

Cosworth DFV Unveiling

Autosport had this to say about the launch of a new Ford Formula 1 engine: “The announcement this week of the Cosworth-designed and developed Formula 1 engine must be a cause of concern for Lotus’ rivals in the Grande Epreuve field. The compact V8 is extremely light and, as it is designed to do the work of the chassis frame at the rear of the car and carry the suspension, the new Lotus that has been designed around the engine can also be expected to be light and compact. With minimum weight, 400 bhp and Jim Clark and Graham Hill as their drivers, Team Lotus and Ford must be very strong contenders for World Championship victory once they have got their new car sorted out.”

Autosport’s words would prove to be very prophetic – Jim Clark won the first race that the new Lotus 49 powered by the DFV entered in 1967, and Graham Hill took the driver world championship the following year.

Lasting Legacy

The DFV would become the most successful engine ever in the history of Formula One. It went on to win 155 Grand Prix races, 12 driver world championships and 10 constructor championships. It spawned several other winning engine types – the DFW (Tasman Series), DFX (CART/IndyCar from 1975), DFL (Group C), DFY (Formula One from 1983), DFZ (Formula One in 1987), DFR (Formula One from 1988) and the DFS (CART/Indycar from 1988). And its DNA could be traced through subsequent Cosworth Formula One engines and even in the current generation of race engines.

Although numerous developments of the DFV were pursued by Cosworth – including shortened stroke, new cam profiles, changes to the ignition system, redesigned pumps and larger valve diameters – its basic layout never changed. Perhaps this is even more remarkable given that the DFV was the first entire engine designed by Cosworth.

The DFV is still very much alive in historic racing. We continue to support customers rebuilding these engines, supplying a wide range of genuine Cosworth parts along with parts that we’ve been able to reverse engineer and develop. You can find these parts in our on-line shop here: https://modatek.co.uk/product-category/dfv-parts/

This feature on the Cosworth DFV is based on an article written by Modatek’s Matt Grant for Race Engine Technology, issue 84. You can purchase a back copy of this publication here: https://www.highpowermedia.com/Product/race-engine-technology-issue-084

Every engine lubrication system needs pressure to move the oil around the myriad of galleries and drillings inside the engine. One or more pumps will provide the pressure, and the most common way to regulate the pressure is with an oil pump PRV (pressure relief valve) like the one shown below.

Oil Pump PRV Assembly

The oil pump is a pulsating heart that continually moves the oil so that it can lubricate running surfaces and remove heat from critical areas. If a typical high performance engine has an oil flow rate of around 60 litres per minute and an oil volume of around 6 litres, that means that in the 30 seconds or so that it’s taken you to read this far, the oil in the lubrication system will have already been pumped around the engine five times.

All engines need at least one oil pressure pump to create enough pressure to push the oil around the oil circuit. In addition, an engine with a dry sump will also have one or more scavenge pumps, whose job is to remove the oil from the various sections of the engine.

Positive-Displacement Pumps

Both oil pressure pumps and scavenge pumps tend to be of the positive-displacement type, whereby the internal elements of the pump move to create a void that opens up and expands, filling the void with oil. The elements continue to move in a manner that then compresses the void, forcing the oil out of the pump and into the oil circuit.

A high performance oil pump normally contains rotating elements such as rotors or gears, and they typically either run inside one another (classed as an internal gear pump) or side by side (referred to as an external gear pump).

These rotating positive-displacement pumps need a rotary drive, which on most race engines comes courtesy of the crankshaft or camshaft, either directly or from gears, chains or belts that in turn are connected to the crankshaft or camshaft. Consequently, the pressure delivered by the pump increases with engine speed.

In the most extreme circumstances, if the pressure in the oil circuit isn’t regulated then it could rise to such a level that could damage the engine. For instance, extreme pressures could rupture the oil filter or blow out any sealing plugs. Most oil circuits will therefore include at least one device that can control and regulate the pressure, to make sure it can’t exceed a predetermined maximum level.

Spring-loaded Piston PRV

The majority of PRVs are constructed from a spring-loaded piston, which moves when the oil pressure in the dead space above the piston reaches a certain level. When the piston moves, it reveals a port that allows the oil to vent out of the pump, thereby capping the oil pressure. While the theory behind such a device is relatively simple, as ever the devil is in the detail.

The operation of the oil pump PRV depends on a correctly defined spring (the one pictured above is our high pressure spring for the Cosworth YB oil pump). Some good old-fashioned engineering equations can give accurate results to define the spring required in the PRV.

Let’s use the following terms:

  • Required PRV opening pressure = Preq
  • Spring force = F
  • Piston radius = r
  • Spring rate = k
  • Length of travel of piston required to open PRV outlet port = x

The pressure on the piston is simply the spring force divided by the piston area, which is:

The spring force can be derived from Hooke’s Law:

Combining these two equations, we get:

However, we must also consider the back-pressure on the other side of the piston, downstream of the PRV. Although small, the back-pressure will affect the movement of the piston and should ideally be less than 10% of the required opening pressure of the PRV.

This back-pressure is a result of a huge range of variables, such as bearing clearances, oil viscosity and temperature, flow passage size and roughness, plus more. It is therefore hard to calculate but can be found from measurements taken when the engine is running.

The back-pressure will have an effect on the required pressure, and hence needs to be included in the calculation. If we term the back-pressure as Pback, then the calculation for the required pressure becomes:

So, if we’ve already decided on the piston radius and the length of travel of the piston (which is normally dictated by the space that is available for the PRV), we can rearrange this equation to choose the correct spring rate that will correspond with the required opening pressure:

Armed with this information, the spring designer can decide on the wire diameter, coil diameter, number of active coils and shear modulus to give the required spring rate, k, from this equation:

where d is the wire diameter, D is the coil mean diameter, N is the number of active coils and G is the shear modulus.

Spring Stress

The spring designer also has to check that the torsional stresses in the spring are within safe levels, so that the spring doesn’t elastically deform or break. FEA is one way to determine the stresses, but again some simple calculations can also be used to good effect.

The torsional stress τ in a spring under load F can be found from the equation:

KW is known as the Wahl factor, and is a corrective factor that takes into account the effect of direct shear and the change in coil curvature, and can be found from this equation:

where C is the spring index:

Checking the stresses in a PRV spring might seem unnecessary, but the effects of a broken PRV spring can be as serious as a failed valve spring. The PRV will continue to operate with a broken spring, but if it opens at a lower pressure than required, the oil supplied to critical components such as the crankshaft bearings and piston squirt jets will be at a lower pressure, and like a broken valve spring, the result can be catastrophic engine failure. Combining the above series of equations into a spreadsheet though can enable a PRV designer to quickly establish the link between required pressure and torsional stress in the spring.

Adjusting the Required Pressure

It is also possible to modify the required pressure of an existing PRV without resorting to changing the spring. If the spring has a fitted length of L1 and the length of the spring is L2 when it is compressed by x, then:

We can see from this equation that we can increase the required pressure by reducing the length of spring when it is compressed (L2). In reality, this can be done during assembly of the PRV by adding one or more shims to one end of the spring.

Oil Pump PRV Reliability

One would think that being constantly flushed with oil, the piston would be free to move up and down the cylindrical bore in the sleeve without any issues. However, one of the biggest problems with spring-loaded piston PRVs is that they can tend to jam. That is especially true if dirt or debris gets trapped in the gap between the piston and the bore.

For this reason, suitable filtration of the oil is vital, and usually the oil pump PRV is located downstream of the oil filter so that it receives the oil in its cleanest state.

Also, the clearance between the piston and the cylinder is kept as low as possible, to keep any contaminants out. It’s not uncommon for the piston and the sleeve to be machined as a matched pair. Normally the piston diameter will be measured, then the bore in the sleeve will be machined to the correct size to give the required clearance for that particular piston.

Some PRVs have clearances as low as 5 microns, so it’s essential that this machining is as accurate as possible. Usually, the piston is ground and the bore in the sleeve is honed, as both of these manufacturing methods result in an extremely low dimensional tolerance of just a few microns and can give exceptional levels of circularity and run-out.

The material for the piston and sleeve tends to be as hard as possible, so that any debris doesn’t scratch the walls of both parts. Also, the leading edge of the piston is normally kept as sharp as possible – even a small chamfer can trap debris that will then find its way into the radial gap between the piston and bore.

Also, it is important that the piston and sleeve are demagnetised, otherwise small magnetic forces can cause the piston to stick.

Oil Pump PRV Bypass Return

The choice of where to route the oil that comes out of the PRV bypass seems to be a matter for debate. The two options are to feed the oil back to the oil pump’s inlet or to ‘dump’ the oil back into the engine, usually into the sump or oil tank. Returning the oil from the PRV to the oil pump inlet is the more popular option on race engines.

One benefit of returning the PRV flow back to the oil pump inlet is that it can help to prevent the onset of cavitation. Briefly, cavitation is the damage caused to a surface by the formation of tiny bubbles in the oil. The bubbles are created when the pressure in the oil drops below the oil’s vapour pressure. The oil will boil, instantaneously creating thousands of these tiny bubbles. When the pressure in the oil rises above the vapour pressure again, the bubbles instantly collapse.

This rapid movement of oil leads to small zones of highly pressurised oil, which when combined with the shockwaves from the collapsing bubbles can result in pitting of any nearby metallic surfaces.

There are a number of tricks that can stop cavitation, and most methods are aimed at increasing the pressure at the inlet to the oil pump. For example, if there is a filter on the pump inlet then the mesh size could be increased, or the inlet port to the oil pump could be contoured to help the oil flow more easily into the pump.

Oil Pressure Requirement

I have designed numerous oil pump PRVs in my time, and I’ve found that the most difficult part of the process is actually deciding the pressure the PRV should open up at. Ask a group of engine designers how much oil pressure an engine needs and you’ll probably receive a number of conflicting answers.

Some will say the pressure needs to be high enough to keep highly loaded bearings lubricated or to feed the piston squirt jets. However, others will say the demands of creating too much pressure can increase the parasitic power losses, so it needs to be as low as possible.

In truth, the exact oil pressure required will normally be decided only after multiple tests, either in the car or on a dynamometer. It is important therefore to make sure the designed PRV has room for adjustment, as mentioned already by the use of shims or via an external adjustment device.

An old rule of thumb used to be that an engine needs 10 psi of pressure for every 1000 rpm of engine speed, so for example an engine that revs to 8000 rpm will need 80 psi. In reality, that is probably an over-cautious estimate for modern race engines. Developments in both lubrication and bearing technology mean that the higher grades of oils used in a race engine can withstand more extreme pressures in the plain bearings of the crankshaft (the area in which the oil is typically the most stressed), and enhanced additives allow the oil to behave better for longer.

One train of thought is that the required oil pressure is closely related to the clearance of the crankshaft bearings combined with the viscosity of the oil, because oil pressure can drop if the viscosity is reduced or if the bearing clearance is opened up. Given that reduced bearing clearances can have a positive effect on power, some engine builders will offset the increase in required pressure that might arise from reducing bearing clearance by running lower viscosity oils.

In truth, one has to consider the various sources of pressure drop along the entire lubrication circuit when specifying the oil pump’s required pressure. Some pump designers will group these sources together to come up with an ‘effective orifice area’, which is the equivalent flow area of all of the holes and gaps that the oil has to flow through. It will be a combination of the bearing clearances, plus clearances to other mating parts such as camshaft followers along with small holes such as those in squirt jets.

Broadly speaking, oil pressure is proportional to both the effective orifice area and the oil viscosity: the oil pressure rises if either the effective orifice area or the oil viscosity is increased. This can be observed when starting a cold engine – the lower temperature means the clearances are small and the oil is thicker. Both of these effects combine to give higher oil pressure.

As the engine gets hotter, the clearances in the engine begin to open up, and the oil gets thinner. The increase in both the effective orifice area and the oil viscosity will result in a drop in oil pressure, so it is vital to make sure that the chosen oil pressure requirement is optimised for the required range of temperatures the engine will experience.

This feature on oil pump PRVs is based on an article written by Modatek’s Matt Grant for Race Engine Technology, issue 136. You can purchase a back copy of this publication here: https://www.highpowermedia.com/Product/race-engine-technology-issue-136

OK, so this might not be the most exciting of subjects, but understanding the conventions of Cosworth part numbers can save a lot of time and effort. If you’re trying to identify a part then the part number that might be marked on it can give you several clues as to what engine it came from.

To understand the history of Cosworth part numbers, let’s go back 65 years to when Cosworth was first formed. Cosworth’s first foray into engines centred on modifications to the Ford Anglia engine, producing components like camshafts. They soon progressed to building cylinder head assemblies, followed by complete engines that were designated Mk I, Mk II, Mk III, Mk IV, etc, followed by engines such as the MAE, SCA, FVA and DFV.

One of Cosworth’s early strengths was that every part was identical from batch to batch, so it was easy to swap parts between engines. There were minimal modifications required prior to fitting each part, and this made engine build a lot quicker and far more reliable. They achieved this through very detailed drawings and technical specifications that would give the manufacturer all the information they needed to produce the part. Cosworth would then book the part into stores when it was either manufactured in-house or bought in. Then all Cosworth had to do when building the engine was to book out the required parts and assemble them into the engine.

DFV Parts List

The production and tracking of components for these engines necessitated a system to keep track of the parts that would be required – the parts list. Hence each part needed its own unique part number, one that could be used to track the life of the part through design, manufacture and assembly. Cosworth quickly realised that random part numbers wouldn’t work, and that they would need to have a system to follow to generate each part number.

The Project Codes

The system that Cosworth devised was fairly simple (remember KISS – Keep it Simple, Stupid!). Every part number consisted of two letters that designated the project code followed by four numbers, for example YB1429. With the amount of engine projects rapidly expanding as the company developed, this system meant that one could instantly identify which engine the part was intended for.

There were over 50 different project codes, and some of these never saw the light of day, but here are some of the more common ones, arranged in alphabetical order:

Project CodeEngineYear
BABD series1969
CACA series (Formula 1)2006
CKCK series (Ford CR series Formula 1)1999
FAFVA series1966
HBHB series (Formula 1)1989
JDJD series (Ford Zetec-R Formula 1)1996
SASCA series1964
TALotus Twincam1963
TJTJ series (Ford CR & RS series Formula 1)2003
VJVJ series (Ford Zetec-R Formula 1)1998
WAWA series (Mercedes-Benz)1984
XBXB series (Indy Car)1992
XDXD series (Indy Car)1996
YBYB series1984

Cosworth also used other project codes in their part numbers, especially for parts that might be used on one or more projects. These included:

PP/PR – ‘proprietary parts’, these were for parts that were bought in

LL – ‘liner length’, these were bought in parts that were supplied by length, such as O-ring cords

DE – electronics parts

PA – this code was used for pistons

KK – these were kits of parts, like piston rings

The codes were supposed to be unique for each project, but occasionally the same code was used for two different projects. For example, CA was meant to be the code for the Cosworth 4WD Formula 1 project in 1969, but it was also used for the V8 Formula 1 engine in 2006.

The Casting Codes

Cosworth also had another set of project codes, called the ‘casting’ code, that were specifically for parts like castings, forgings and billets. Again, these codes were linked to the engine project, and Cosworth instigated a clever way of defining the casting code. The first letter was the same as that for the project code, and for the second letter, just go 13 letters along in the alphabet. So, for example, the casting code for YB was YN.

There were some exceptions, and there was also a rule that letters like I and O had to be skipped as they could be confused with numbers, but on the whole the casting codes followed this pattern.

This brings us to an important point. Many people incorrectly identify a part by the casting number, but this only identifies the part when it is in its part-finished form. For example, in the photo below, YN0627 is the part number of the YB head casting, not the finished machined component.

YB Cylinder Head

There will be a number of different types of cylinder heads that are machined from the YN0627 casting, so knowing just the casting number doesn’t completely identify the head. Here are the part numbers for the heads for the different types of YB engine:

Engine TypeCylinder Head Part Number

Another example of a common misconception comes with pistons, which had a forging code of PM. Cosworth would imprint the forging part number into the forge tool, and this would be visible on the piston. But the PM part number only referred to the part number of the forging, not that of the machined piston. Given that the same forging could be used for a variety of different pistons, it is the finished part number that is required.

The Four Numbers

As mentioned, the part number consisted of two letters followed by four numbers. For most projects, four numbers would be enough, as it would be extremely unusual for a project to need more than 9,999 part numbers.

At first these four numbers were sequential, starting with 0001. Usually each project had its own folder that listed these numbers so that there could be no duplication. Over the years, Cosworth started to instill some ‘intelligence’ into these four numbers. The four numbers would begin with an 8 for assemblies, and numbers beginning with 05 were reserved for schematic drawings.

At one point Cosworth also introduced a rule that said that the last number would odd for left hand components and even for right hand components.

Certain projects stipulated more rules for the four numbers, such as reserving 0001 for the cylinder block, 0002 for the LH cylinder head, 0003 for the RH cylinder head and 0010 for the crankshaft. However, this rule was fairly short-lived.

YB1429 Head Gasket Drawing

Cosworth used the part number on all documentation, including drawings (as per the example of a drawing excerpt above for our YB1429 WRC head gasket), purchase orders and invoices.

SAP & Sequential Part Numbers

These fairly simple rules for defining the part number with the project code and four numbers ran fairly smoothly for a number of decades. However, all this came to an end when Cosworth launched SAP at the beginning of 2007 as its new ERP (enterprise resource planning) system.

SAP was essentially a giant database that contained virtually all of the company’s records. It enabled Cosworth to be able to keep track of the entire life of a component, from design through to manufacture, assembly and usage. But one of the problems of SAP was that it was no longer possible to allocate part numbers that followed Cosworth’s rules. (Actually, this proved not to be true, but by the time a solution had been found, it was too late.)

Instead, the company switched to sequential numbers starting at 20000000. There was a central computerised database that would supply the next number available, but without the intelligence that the old part numbering had provided.

To make matters more confusing, for the first few months Cosworth used numbers starting at 10000000 to identify raw materials and kits. You’ll notice that some of our piston sets follow this pattern, such as 10001487 for our BDG Hoyle piston sets.

BDG Hoyle Piston Label

If you’ve survived to the end of this article, then congratulations. It is definitely not the most riveting of subjects, but will hopefully help to shed some light on how to identify Cosworth components.

The humble valve spring might appear to be a relatively simple piece of engineering, but in reality it can be the most highly stressed component in a race engine. The move to faster engine speeds and more aggressive cam profiles can often result in terminal failure of the springs leading to catastrophic engine failure, prompting spring replacement earlier than preferred. Engine builders and manufacturers are therefore focusing more and more efforts on ways to increase the life of the spring to avoid costly engine rebuilds.

Cosworth DR4601 Valve Springs

In essence, the prime function of the spring is to provide a force that will keep the reciprocating movement of the poppet valve under control throughout the entire cycle of the engine and at all operating speeds. Loss of this valve control can lead to valve-to-piston contact, extreme loading of the seat in the cylinder head, bouncing of the valve on the seat and damage to the tip of the valve.

A compression spring like those found in engines provides a reactive force when its length is reduced, predominantly owing to a twisting motion of the coiled wire. Under just a static load, it can be assumed that the load in each coil is identical, and if the coil geometry is uniform along the spring axis then the stiffness of each coil is also identical. The highest stress will occur on the inside diameter of the spring, which is where one could expect failure to originate from.

Spring Surge

Of course, when the engine is running, the linear motion of the valve imparted by the rotation of the cam lobe causes a continuous compression and expansion of the spring. As a result, the dynamic loading on the spring has to include the inertia of the spring, which is not considered under static loading. It is the dynamic loading from the inertia that will result in the most common of spring issues, namely spring surge.

Spring surge can be described as vibration of the spring that occurs at a harmonic of the spring’s natural frequency. When describing surge, the movement of each particle of the spring needs to be considered, and in a way this movement of each coil can be seen visually with a child’s Slinky toy spring.

During the initial opening phase of the lift curve, the spring is compressed and the spring coils accelerate. The uppermost spring at the camshaft end will see the entire inertia of the spring, but each successive coil sees less inertia loading thanks to the lower mass below it and lower acceleration due to a smaller deflection. As a result, as we move down the spring away from the camshaft, each coil accelerates and moves at lower values than the one above it. This begins the first compressive wave of the coils, with the camshaft end closing up quickest.

At around the mid-point of the travel of the valve, its acceleration is zero, and at this instant the coils all move at the same speed. Then, as the valve begins to decelerate, the coil furthest from the cam starts to close up more than those above it, creating a compression wave going in the reverse direction. It is this continual cycle of the compression wave that creates a vibration in the spring and is referred to as surge.

Preventing Surge

Whilst there are numerous tricks to reduce or even eliminate surge, the most obvious one is to create a spring with a natural frequency well outside of the running range. The movement of the valve is dictated by the profile of the cam lobe, and can be mathematically broken down into a series of sinusoidal curves with Fourier analysis, from which the harmonics of the profile can be derived, which are expressed as multiples of the camshaft rotational speed.

When one of the harmonics coincides with the spring’s natural frequency, the effects of surge will be pronounced and can result in the compression wave spiralling out of control, leading to loss of contact between the spring and the retainer and spring seat at either end, plus the build-up of excessive stresses in the coils. As the amplitude of the lower harmonics are larger than those of the higher harmonics, some spring designers will recommend that the natural frequency of the spring is at least eight times the frequency of spring operation, whilst some technical publications quote 15-20 times.

Wire Geometry

The cross-section of the wire is usually circular or ovate; the latter term actually means egg-shaped, but in the case of springs this can also be any elliptical shape made up from a number of radii, and can be either symmetrical or non-symmetrical. An ovate spring will typically have the major axis perpendicular to the spring axis, which helps to reduce the stress on the inner diameter as the maximum area of the wire is at the point of highest stress, and can lead to a shorter spring length owing to the wire being slightly flattened.

However, some of the higher grades of steel do not work well with the dies used to make the ovate wire shape, as the additional carbon can extrude the special dies needed to form the more complex cross-section. Consequently, the round wire can be made with more carbon content than an ovate wire. It is also harder to control the orientation of the ovate profile when coiling the spring, as it will have a tendency to twist down the length of the spring during coiling. In fact, changing the external shape of the spring can have a far stronger effect on the life of the spring than using an ovate cross-section.

Spring Shape

Nowadays there are a myriad of options available for the external shape of the spring’s helical coils, although they can be broken down loosely into three categories: straight, conical and beehive.

Conical and beehive springs are termed as progressive, as the stiffness will vary with length. This is also a case for springs with an unequal pitch between the coils along the length of the spring. In progressive springs, each coil has a different stiffness, which means that when the spring is compressed, the coils with a lower stiffness will deform more than those with a higher stiffness. Eventually the less stiff coils become coil-bound (where adjacent coils come into contact), which reduces the number of active coils available as the load is increased, increasing the overall stiffness of the spring in a progressive manner. As the stiffness is varying, so too is the natural frequency, hence in theory progressive springs are less susceptible or even immune to the problem of spring surge.

Both beehive and conical springs also have an advantage over a straight spring in that the retainer can be smaller, which in turn can lead to reduced valvetrain reciprocating mass. The springs themselves can also be lighter, and so the inertia of the upper coils will be lower. One final observation made by a spring supplier is that conical springs can also have a natural alignment action that is very beneficial for very small valve stem diameters.

Nested Springs

Another solution to surge is to use nested springs, where two or three springs are used in parallel, with smaller springs packaged inside larger ones. As with progressive springs, the stiffness of the combined spring varies with length, and each spring will have a different natural frequency, again helping to avoid surge.

Cosworth DR4601 Valve Spring

The outer diameter of the inner spring and the inner diameter of the outer spring are usually chosen so as to create a small amount of interference between the two springs (as is the case with our DFV valve springs). It is essential that the direction of windings is different between the springs, otherwise they will get caught up in one another. The interference will provide a means of damping, allowing unwanted energy to be converted to heat from the friction between the two springs.

When run for extended periods, this interference will of course wear the surfaces of the springs and reduce the life of the nested spring assembly. Owing to titanium’s inherent nature to gall (the macroscopic transfer of material between metallic surfaces) when in contact with other titanium surfaces, titanium nested springs cannot be run with interference.

Where the designer wants to avoid contact between the inner and outer springs, stepped retainers and spring seats can be used. This will separate the two springs to eliminate friction and hence damage to the surfaces.

The size of the spring is determined by multiple factors: the cam profile and associated lift, acceleration rates and opening, closing, flank, nose and seating velocities; valve train masses; the moment of inertia of the rocker (if present); the operating speed of the engine. All of these parameters will give an idea of the required amount of spring travel, spring force and spring rate that is going to be required to control the valve at speed.


Because of the high stresses that the spring will experience, special care has to be paid to the cleanliness of the raw material, the surface finish and any methods that can be used to leave compressive residual stresses in the surface that would counteract the stresses from running.

When choosing a material, spring manufacturers will look at its torsional modulus of rigidity and torsional yield strength, as well as the more practical requirements such as cost and availability. Most compression springs are made from steel or titanium drawn wire, with the former being more common.

Cosworth PR8121 BD Valve Springs

The actual chemical composition of the steel alloys used by spring manufacturers remains a closely guarded secret, which is understandable given the intense amount of research and testing they carry out to find the perfect mixture of elements. However, what is known is that most steel alloys used in spring manufacture will contain chromium and silicon. Vanadium is also included at small levels to increase the strength of the material, along with manganese, molybdenum and nickel in some cases.

Titanium springs are usually made from Beta-C and LCB (Low Cost Beta) titanium alloys. Titanium can offer the benefit of lower density and higher strength, plus better resistance to corrosion when compared with steel. However, its cost can often mean that that titanium springs are about five times more expensive than their steel counterparts. Also, some of the steel alloys used nowadays have proven to have higher fatigue limits than the titanium alloys available.


The manufacture of a steel valve spring starts with the material being rolled into rods by a specialist steel mill, that is capable of producing the level of cleanliness required for racing.

Some steel mills will take extra steps to ensure that any inclusions in the microstructure are pushed into the centre of the wire, where the stresses during running will be lower. Also, great care is taken to make the inclusions smaller (it is the inclusions that can make the wire brittle), giving rise to the term ‘superclean’ chrome-silicon as used by some steel mills.

The wire is drawn down to size by pulling the rod through a series of dies, all the while being tested and scrutinised in line with the manufacturer’s quality standards. Attention is paid to tensile strength, surface inspection and chemical analysis, and eddy-current testing is used to verify the surface (an electromagnetic field is created around the wire to allow for microscopic identification of any surface defects such as pitting, cracks and corrosion).

Some spring manufacturers will also use scanning electron microscopes (SEMs) and X-ray diffraction machines (XRDs) to check for material integrity. An SEM allows metallurgists to view the surface topography and composition, while an XRD can measure compressive residual stresses. Such methods can be incorporated into various stages of the manufacturing route of the spring to ensure no degradation in quality.

The wire is then shaped into the designed helical coil pattern using CNC machines to control the winding of the wire onto mandrels, with the wire being either hot or cold. The use of such complex machines allows for better repeatability between batches of springs and improved accuracy in the pitch and diameter of the coils. Next, the spring has to be stress relieved. The coiling stage permanently deforms the wire, creating harmful residual stresses, and so a heat treatment operation at a relatively low temperature is needed to remove them. Note that coiling is more complicated when the wire’s cross-section is not circular.

With the shape of the spring now complete, attention is then paid to the ends of the spring. It is imperative that the end faces of the spring are perpendicular to the axis of the spring and parallel to each other so that the forces will be evenly distributed in the valve stem. As a result, the ends are ground. As the grinding can leave sharp edges, a finishing step is carried out to remove uneven areas on the surfaces of the ends. Without this final operation, the edges could break away into the cylinder head chamber or dig into the retainer and spring seat, creating fatigue crack initiation sites.

Surface Treatments

At this stage, there are numerous processes that spring manufacturers will carry out to increase the life of the spring. We will look here at shot peening, nitriding, polishing and cryogenic treatment, but there are numerous other techniques that manufacturers are less willing to reveal, for obvious reasons.

Owing to its cost-effectiveness and practicality, shot peening is a relatively common technique to impart a compressive stress in the surface. Here, small spherical beads made from steel, glass or ceramic are fired onto the faces of the spring in a controlled manner. The impact of each bead will create a dimple in the surface, stretching it, and below the dimple the movement of the material creates the compressive stresses required.

There are three parameters that can be varied to alter the magnitude and depth of the compressive stress – bead size, intensity and coverage. In general smaller size beads will yield a more polished surface. The intensity is the amount of energy used to project the beads, while the coverage is the amount of area that is impacted by the beads (note that this is always more than 100%). All these variables will depend on the material of the spring and any subsequent processes.

Nitriding is a heat treatment procedure that will diffuse nitrogen into the surface of the spring to give a case-hardened surface and can also impart a compressive stress into the surface. A harder surface is especially useful in a nested spring design, where there is interference between the springs.

In gas nitriding, the spring is placed in an oven at temperatures of about 500°C for a period of time while ammonia is flowed around the spring’s surface. The alternative to gas nitriding is plasma nitriding. Whereas gas nitriding relies on a high temperature to create a reaction with the surrounding gases, plasma nitriding uses intense electric fields to create ionised molecules of the gas (usually nitrogen) around the spring’s surface. Note that if a spring is to be both shot-peened and nitrided, the nitriding step is carried out first, otherwise the high temperatures during nitriding would relieve the compressive stresses induced at the shot-peening stage. The added advantage of nitriding first is that the substrate is harder, so the compressive stress from shot peening is increased.

One or more refinement procedures are also usually carried out to remove any remaining surface defects and imperfections, both between certain operations and at the end of manufacture. Electro-polishing is one such method that has been proven to be beneficial, although it can lead to hydrogen embrittlement, to which the high-strength alloys can be susceptible. However, a combination of chemical and mechanical isotrope finishing is becoming more common, and this creates a polished mirror-like surface without the issues seen with electro-polishing. Some manufacturers will go even further after polishing by adding a final peening operation with minute beads (often referred to as micro-peening or nano-peening).

A final operation carried out after all or some of the above is to pre-set the spring. Here, a relatively large load is applied to the spring, such that while the centre of the wire is elastically deformed, the surface of the wire undergoes plastic deformation. This procedure will set the free length of the spring, as the plastic deformation means that the spring will not return to its original length. Pre-set springs are less likely to relax over time, and if the pre-setting is carried out at a controlled elevated temperature than the spring will be more capable of withstanding service in hot environments too.


Despite every effort in design and manufacture to increase the reliability of valve springs, spring suppliers see a surprising number of failures due to improper installation. One of the more common issues they see is incorrect design of the retainer. The spring needs to be correctly contained in the retainer to stop it from excessive lateral movement, but not overly constrained such that it is forced into the retainer.

Cosworth DFV Valve Spring, Retainer & Seat

Handling the springs also has to be done with care to avoid damaging the surface. They should never be placed in a vice or pliers, and plastic tooling should be used when separating interference-fit nested springs. Also, springs that have been delivered with a rust preventative coating should not be cleaned with acidic or evaporative cleaners, as this can cause rapid drying and promote the formation of rust on the surface of steel springs.

A static spring testing machine can be used when selecting and fitting valve springs to confirm the rate of the load versus deflection; such machines can detect the onset of binding of the coils.


The life and maximum operating envelope of many race engines is restricted by the valve spring. While it is possible to extend the life of the spring by reducing engine speed or compromising on cam profiles to lower the acceleration of the valve, there are numerous methods available in the design and manufacture of the spring that should be considered – wire cross-section, the geometry of the helical shape and a combination of nested springs can all be exploited to reduce or even eliminate certain failure modes. Numerous manufacturing processes exist that will create beneficial compressive stresses at the surface.

This feature on valve springs is based on an article written by Modatek’s Matt Grant for Race Engine Technology, issue 89. You can purchase a back copy of this publication here: https://www.highpowermedia.com/Product/race-engine-technology-issue-089

Cosworth CA and DFV Formula 1 Engines
Cosworth CA and DFV Formula 1 Engines

The Cosworth DFV and the CA engines represent bookends of nearly half-a-century of normally aspirated Formula One engines. These two engines allow for a discussion of how Cosworth was able to double both engine speed and power output, as well as guarantee a staggering threefold increase in engine life. In this special tech feature we’ll compare these two engines, taking a closer look at some of the critical components.

An Icon in F1 History – the DFV

Very few companies can survive more than 60 years of trading and still follow the founders’ original mission intent. In the cut-throat world of motor racing this is an exceptionally rare feat, and one only has to look at a race report from half-a-century ago to see that most of the teams and engine suppliers of the time have since disappeared, falling foul of financial setbacks, changes in regulations and necessary diversifications away from motorsport.

Yet one company has flourished since its inception in 1958 and can still claim to be true to its roots. That company is Cosworth, built on the foundation that “it must be possible to make an interesting living messing around with racing cars and engines”, as decreed by founders Keith Duckworth and Mike Costin.

Cosworth DFV Engine Cutaway

In 1967 Cosworth launched the DFV into Formula 1, initially for Team Lotus, and it instantly became a race winner. The following year the DFV was made available to other teams, and it went on to become the most successful engine ever in the history of Formula 1, scooping 155 wins, 12 drivers titles and 10 constructors titles.

The first DFVs might only have been able to delivery just over 400 bhp and reach 9,000 rpm to start with, but as each failure mode was methodically overcome (starting with the valve spring, then torsional gear drive problems), power and speeds gradually rose. By the time the DFV had finished active service in 1985, peak speed had topped 11,000 rpm and power was in excess of 500 bhp.

Working in conjunction with Lotus, the DFV was the first successful Formula One engine designed to be a fully structural member of the chassis, with mounts on the heads, cylinder block and sump connected to the chassis and gearbox bulkheads.

The DFV had a V8 configuration, with the two banks separated by a vee angle of 90 degrees, and for various reasons had the air intake to the two cylinder heads in the centre vee and the exhausts on the outside of each bank, running down either side of the engine below the heads.

The auxiliary water, oil and scavenge pumps were housed on either side of the cylinder block, tucked away underneath the exhausts and driven by a belt that in turn was driven by the nose of the crankshaft. The drive to the camshaft gears was via a series of gears at the front of the engine, again driven from the crankshaft nose. In the centre vee sat the alternator and fuel pump assemblies.

The Need for Speed – the CA

Some 40 years later, Cosworth launched what would turn out to be their final V8 Formula 1 engine, codenamed the CA. Formula 1 engine regulations had gone through many changes since the advent of the DFV, with brief interludes of turbocharging in the mid-80s, followed by a return to normally aspirated 3.5 litre engines up until the end of 1994, and then 3 litre engines in various configurations.

Cosworth CA Engine Cutaway

When the CA was first launched in 2006, and with no regulations capping speed, a mind-boggling and class-leading 20,000 rpm was attainable in qualifying, and by the end of that same season the CA could run up to that speed over an entire race distance. Frustratingly though, regulations aimed at reducing soaring development costs had capped the maximum engine speed to 18,000 rpm when Cosworth returned to Formula One with a modified CA in 2010.

Even so, peak power of 780 bhp was attainable, almost twice that of the DFV when it was first launched, despite a reduction of 14% capacity. Perhaps more impressively, there was a huge jump in engine mileage. The DFV competed at a time of unlimited engine changes, and hence only had to be capable of completing one race distance. Fast forward to the 2010-2013 era, and drivers were only allowed 8 engines for the entire season, which meant that the CA had to be capable of completing around 1,500 miles between rebuilds.

Comparing the DFV and CA

What is perhaps surprising is that the CA’s overall architecture was almost identical to the DFV’s, save for the removal of belt drives and the repositioning of the alternator and fuel pump (in subsequent Cosworth Formula One engines the alternator was relocated to the back of the left-hand auxiliaries, while the fuel pump ended up submerged in the car’s fuel tank and driven by a quill shaft from one of the drive gears on the front of the engine). While Duckworth would certainly never claim to have pioneered this layout, it is interesting to note that most Formula One engine manufacturers later followed a similar approach.

A comparison of DFV and CA engine weights wouldn’t necessarily be fair, as in addition to the reduction in capacity, the CA’s weight was mandated by the regulations, which specified a minimum dry weight of 95 kg. However, it is worth noting that the corresponding weight of the DFV would have been around 168 kg when it was first launched.

The regulations that governed the design of the CA also defined a minimum height of the centre of gravity from the bottom of the sump at 165 mm. As this figure was easily achievable on the CA, any weight-saving requirements were rendered unnecessary, hence extra material in the heads and cam covers could be used to improve the engine’s overall stiffness.

Cylinder Block Comparison

Cosworth CA and DFV Cylinder Blocks
Cosworth CA and DFV Cylinder Blocks

The most striking difference between the cylinder blocks is in their respective sizes. The DFV block stands almost twice as high as that of the CA, due in part to the height of the sump. The distance from the crank centreline to the bottom of the sump for the DFV was more than 133 mm; for the CA that fell to just 58 mm (the minimum allowed by the regulations).

Some of this reduction was made possible by gradually miniaturising the bottom-end geometry over successive engine designs, such as shrinking the crank counterweights following the shift to bolt-on tungsten weights. Of equal significance was lowering the piston stroke (64.77 mm for the DFV versus 39.77 mm for the CA) thanks to an increase in cylinder bore size and a reduction in capacity.

There was a marked increase in bore size from the DFV to the CA, 85.67 mm for the DFV compared with the CA’s 98.0 mm, as dictated by the regulations. Despite this, the CA block is slightly shorter in length, as both engines have almost the same bore spacing. The distance between the walls of the bores could be reduced on the CA because it ran with coated parent metal bores, whereas the DFV block was an open-deck variant containing cast-iron cylinder liners.

Both blocks were cast from aluminium alloys, LM25 TF for the DFV and a similar in-house derived aluminium alloy for the CA. Cosworth actually experimented with magnesium cast blocks and cylinder heads for the DFV during the 1970s, but that was soon abandoned owing to the increased complications that came when using magnesium. In essence, the magnesium blocks and heads were problematic because of the large difference in thermal expansion coefficient values for the magnesium material, the steel main bearings and the nickel-aluminium bronze alloy valve seats.

Cylinder Head Comparison

Cosworth CA and DFV Cylinder Heads
Cosworth CA and DFV Cylinder Heads

Duckworth famously remarked that the DFV was the first race engine to incorporate a narrow included valve angle (the angle between the inlet and exhaust valves). At the time, rival engines had valve angles of around 60 degrees, but Duckworth sought to reduce this to give a shallower pent-roof chamber. The DFV was designed with an included valve angle of 32 degrees – compare this then to only 18 degrees on the CA.

Actually, the CA had compound valve angles, with both the inlet and exhaust valves also inclined 6 degrees apart along the crankshaft axis. The switch to a compound valve angle improved the shape of the combustion chamber on the CA, plus an opportunity to make a small increase in inlet and exhaust valve diameters.

Like the DFV, the CA cylinder head featured a separate cam carrier – in the case of the CA, this was necessary to be able to package the compound valve angles. The CA heads featured a pneumatic valve return system instead of conventional wire springs, which allowed the engine to run at such high speeds.

Piston Comparison

Cosworth CA and DFV Pistons
Cosworth CA and DFV Pistons

Thanks to the tightening up of the included valve angles in the cylinder head, the CA piston crown didn’t need deep valve pockets, as seen on the DFV. Instead, the crown was almost flat, with shallow pockets to clear the valve heads. The lack of sharp edges and pockets in the crown had a huge beneficial effect on good flame propagation and the elimination of detonation.

Piston material was largely unchanged from the DFV to the CA, thanks to the CA’s regulations banning the use of exotic materials such as aluminium beryllium and metal matrix composites. The DFV piston forging used a grade of RR58 aluminium alloy that had originally been developed by Rolls-Royce. By the time the CA was designed, Cosworth had already defined a confidential proprietary blend of aluminium alloy.

Another change that was quite noticeable when comparing the two pistons was the undercrown design, which on the CA was an elaborate arrangement of highly polished ribs and buttresses, optimised using various design techniques. The net effect was a piston that, although bigger on bore size, was almost half the weight of that from the first of the DFVs.

Cosworth Formula 1 CA Piston

One other note was the amount of oil cooling supplied to the CA piston, via an array of squirt jets fed from the main oil gallery in the block. The correct cooling of the piston was a critical factor in achieving the required durability while also withstanding increased gas pressures and speeds.

DFV & CA Continuation

Whilst both engines might have ceased active service, they can still be heard roaring around race tracks today. The DFV is a popular engine for historic Formula 1 race categories, and some of the Formula 1 cars from the end of the V8 era are still being run by privateers with power from the CA.

Modatek actively supports customers with rebuilds of both engines, supplying genuine parts such as pistons, bearings, seals and countless other critical components. We even have an on-line shop for DFV parts.

We also sell race-used Cosworth CA engine parts in our Memorabilia section of our on-line shop.

This feature on the Cosworth DFV and CA engines is based on an article written by Modatek’s Matt Grant for Race Engine Technology, issue 100. You can purchase a back copy of this publication here: https://www.highpowermedia.com/Product/race-engine-technology-issue-100

There is nothing ‘plain’ about a plain bearing, especially when it comes to fluid film journal bearings, like those found on the crankpin and mains journals of the crankshaft. There have been continuous advances in crank bearing material, which enable bearings to last longer between rebuilds.

The first examples of journal bearings were made from Babbitt metal (sometimes also referred to as white metal), named after Isaac Babbitt who in 1839 developed an alloy to be used as a bearing surface in plain bearings. The exact composition was kept a closely guarded secret for a number of years, but was eventually disclosed as an alloy of lead, tin, copper and antimony. These days we tend to refer to any lead- or tin-based alloy material as a Babbitt metal.

Babbitt metals tend to be extremely soft, especially when compared to the hardened crankshaft journals. However, the composition of a Babbitt metal is a matrix of small, hard crystals contained in a softer metal. The aim is that, as the bearing wears, the softer metal yields and creates routes for the oil to pass through, improving lubrication and giving the bearing some degree of conformability.

Originally, a thin layer of Babbitt metal would be applied directly to the bore of the substrate material, but the need to periodically refurbish this surface led to the introduction of interchangeable steel or bronze shells onto which the Babbitt metal could be applied.

Thin Walled Bearings

The most famous example of the replaceable bearing shell that was first developed for high-performance engines was the Vandervell ‘thin-walled’ shell bearing. As the name suggests, the shell comprised a thin strip of steel that was rolled to create a semicircular shell, and the bearing material could then be coated onto the inner diameter, creating what was termed a bimetallic bearing. The steel backing was made from a high-strength steel alloy that allowed the bearing to be interfered into the rod and cap halves without yielding.

Today’s bearings still rely on a high-strength steel alloy backing in most cases, but here the similarities with their distant ancestors ends. Bearing manufacturers discovered that a steel-backed bearing combined with a thin Babbitt metal layer was prone to wear, reducing the time between bearing change intervals. If they wanted their products to last longer and be more reliable, they had to find a way of adding strength and wear resistance into the bearing. That thinking led to the introduction of a multi-layered bearing, which could provide the compromises needed between all the functions the bearing had to provide.

Bimetallic bearings were therefore usurped by trimetallic bearings, which are composed of a backing, a substrate (or lining) and an overlay. The substrate layer gave the bearing its load-carrying capability and provided resistance to wear and cavitation.

Bearing Material

The most common material for the substrate layer is a copper and lead alloy, which replaced the previous versions of tin-based Babbitt metals a number of years ago. Such alloys tend to consist of 20-40% lead, with the rest made up of copper and sometimes small amounts of tin, silver or nickel.

Cosworth GB0047 Rod Bearing

Vandervell (now part of the MAHLE group) created its own unique specifications for crank bearing material, characterised as strip-cast leaded bronze cast onto a steel backing plate, most notably VP1, VP2 and VP10, all of which have differing amounts of lead, tin and iron elements.

Overlay Material

Overlays used to consist of lead-tin or lead-indium Babbitt metals, but as with the substrate, there is now a plethora of different overlay material options available.

MAHLE state that the overlay provides three critical characteristics – conformability, compatibility and embeddability.

Conformability – in an ideal world, the bearing housing and the journal shaft would be infinitely stiff and the alignment of the shafts would be perfectly true. Of course, in the real world that is not the case: the cylinder block and con rod assemblies will flex as load is applied, and the crankshaft journals will twist and bend, albeit by imperceptible amounts in most cases. But it means the bearing material must be able to distort elastically in response to distortions of the mating parts.

Compatibility – it is almost impossible to prevent boundary and mixed lubrication conditions during low-speed running, so there will sometimes be metal-to-metal contact between the bearing’s running surface and the crankshaft journal. The chosen material for the surface of the bearing must therefore be compatible with that for the crankshaft journal. A poor combination of materials can lead to galling or even seizure.

Embeddability – even with the best filtration systems and oil additive treatments, there will always be minute foreign particles in the oil that the bearing will have to cope with. As a result, the overlay material of the bearing must be able to absorb or embed this debris in the surface, otherwise the particles will eventually score the surfaces of the bearing and the journal at high loads.

Bearings from Modatek

The original crankshaft journal bearings that were used by Cosworth on their early engines like the DFV were developed with Vandervell. Over subsequent years these bearings have proven to be reliable in a wide range of applications including Formula 1.

We continue to use bearings from Mahle, who took over Vandervell in 2007. We now stock a wide range of Mahle bearings, including rod and mains bearings for the Cosworth YB engine.

MTK0016 YB Rod Bearings

This feature on crank bearing material is based on an article written by Modatek’s Matt Grant for Race Engine Technology, issue 114. You can purchase a back copy of this publication here: https://www.highpowermedia.com/Product/race-engine-technology-issue-114

Have you ever wanted to take a peek inside a Formula 1 engine? Up until now the insides of a Formula 1 engine have remained a guarded secret – until now! Take a look inside one of the best engines from the mid-2000s, the Cosworth TJ.

We stock race-used parts from Cosworth’s TJ and CA engines that can be displayed for all to see. Just keep an eye on our Memorabilia section of our on-line shop to see what’s currently available. Each part can be bought directly from the website shop and delivered straight to your door.

The Cosworth TJ

The TJ was the first used by Jaguar Racing in 2003. The team continued to use this engine when they morphed into Red Bull in 2005. It also saw active service powering the Jordan and Minardi teams in 2004 and 2005, respectively. Its final year of racing was in 2006, when Torro Rosso (formally Minardi) ran the engine against the V8’s.

It was a clean sheet design, learning on the lessons from the successful CK engine that had been used by Stewart Grand Prix / Jaguar Racing from 1999 to 2002. There were several important concepts copied from the CK, such as the unique central-beam cylinder head philosophy.

The TJ received an extensive amount of development, eventually allowing the engine to reach a maximum speed of 19,000 rpm and peak power in the region of 900 bhp.

Last of its Era?

When it was introduced in 2003 the TJ was the epitome of the “anything goes” regulations from that era. Devoid of rules that restricted the number of engines that could be used by each driver, the emphasis was very much of extracting the maximum out of each engine in just one session or race.

It wasn’t uncommon back then for teams to get through over six engines in one weekend (and that’s not allowing for unscheduled engine changes!). As a result, engines like the TJ were designed with performance over reliability in mind.

Midway through the life of the TJ, rules aimed at extending engine life started to creep in. In 2004 the FIA introduced the “two race” requirement that each engine had to attain. Engines like the TJ that were designed for outright power and performance had to be re-engineered to last longer.

In addition, restrictions in later years on low density materials plus minimum weights and centre-of-gravity heights meant that the lightweight features seen on the TJ became redundant in subsequent engines.

In its final year of service with Torro Rosso in 2006, the TJ’s maximum engine speed was capped at 16,700 rpm amid fears that it would outpace the newly-introduced V8 engines (including the 20,000 rpm CA from Cosworth).

However, the TJ lives on, and can be seen (and heard) running in the back of numerous historic Formula 1 cars.

Modatek are pleased to be able to help keep these engines alive for historic race series and demonstration events. If you need assistance then get in touch.

Have you ever wondered what the green coating is on the skirts of some of our Cosworth pistons? If you have, then you are not alone! The coating is actually a fluoropolymer material that has the tradename Xylan, and Cosworth have been using this coating for a number of years on a wide range of parts, and not just the piston. Oil pump and scavenge pump bodies also received the Xylan treatment to help prevent wear from rotors.

Not all Cosworth pistons have this coating – pistons from earlier engines like the BDA, for example, were designed without any skirt coating. But the increases in piston speed in their race engines meant that some form of coating was necessary to help prevent scuffing of the skirts against the cylinder bore walls.

In the early ’90s Cosworth started to coat the skirts on pistons from several different race engine categories, including those destined for Formula 1 like this one from the race-winning HB engine (the coating also looks great when etched with the driver’s name!). Cosworth soon started Xylan coating nearly all of its race and high performance engine pistons, including  the ones we sell today for the YB engine.

What is Xylan?

Cosworth PA2062 Piston Xylan Coating

Xylan was developed by DuPont in 1969, primarily for kitchenware utensils as an alternative to Teflon. Its excellent wear properties meant that it was soon adopted by the automotive industry, where it found its way into numerous different applications.

Xylan is in effect a composite material comprising of a dry film lubricant contained in a matrix with high-temperature organic polymers. This creates what can be termed as a plastic alloy that has excellent surface characteristics and is easy to apply. Xylan has a very low coefficient of friction, so its perfect for the interface between the skirt and the cylinder bore. It also provides exceptional wear resistance, and quite often Cosworth would strip engines down and see that the Xylan coating was completely unmarked. This is in part due to another benefit of Xylan – it has excellent surface adhesion.

The Cosworth Process

As with any coating, good preparation of the substrate surface is vital. In the case of pistons, the skirts need to be completely clean and free of any oil. Once cleaned, the skirt area is masked off with special tape which can simply be peeled off after coating.

The piston xylan coating is then sprayed on in one of Cosworth’s special custom-made spray booths onto the skirt. Each pass of the spray adds around 5 microns, so the thickness can be carefully built up to the required level. Most piston skirts only need a couple of passes to get the thickness up to the required level of between 6 and 12 microns.

The pistons are then left to dry – normally this can take 24 hours. Once dry, the pistons are good to go, with no other treatment necessary.

Cosworth YB 4WD 8:1 PA2062 Piston

If you’d like to know more about the Xylan coatings that Cosworth use on piston skirts then get in touch.

All of our Cosworth production pistons start life as a piston forging to ensure that they are as strong as possible.

Put simply, the forging process involves pushing a bespoke die under great pressure into the billet. As a result, the billet material flows into the required shape of the piston. The forging process gives us the finished machined shape in the undercrown. This reduces the amount of machining that is required. Further, the orientation and distortion of the grains in the material is optimised to give superior mechanical strength. Aligning the axes of the material grain in a certain way can have a significant impact on the structural properties of the piston.

Temperature Effects

Our pistons begin their life as a billet of extruded aluminium alloy bar. This billet is heated to a pre-determined temperature to ensure that the billet is soft, but not at melting point. When the billet is at the correct temperature, it is placed into the forge and shaped by the die. The forging temperature needs to be carefully controlled. This temperature will have a significant effect on the homogeneity of the microstructure. If there were localised changes in the billet temperature then this could lead to inconsistencies in the material properties of the finished piston. For example, if the die was cold then the outer surfaces of the piston would cool rapidly. This would lead to a varied grain structure in the finished piston.

Hence the die is heated to the same temperature as the billet, creating a process known as isothermal forging. This process  that keeps the billet at its maximum elevated temperature throughout the entire forging operation. During the forging process, any cooling at the interface between the die and billet is eliminated, which can greatly improve the flow characteristics and hence the grain structure of the finished item.

Cosworth YB 4WD 8:1 PA2062 Piston

If you’d like to know more about our genuine Cosworth pistons and how they can provide you with the performance and reliability that you need from your engine then please get in touch via our Contact page.