In the past gearboxes for mid-engined, rear-wheel drive racing cars had the gear shafts mounted longitudinally behind the differential. This design still predominates in formulae below F1 and anyone interested in such units should refer to Hewland’s excellent website.
The first changes of note from this approach were Ferrari’s dominant 312T range of cars that began winning with Lauda in 1975 and culminated in Scheckter and Villeneuve’s dominant 1979 championship season. These cars had the gearshafts mounted transversely and in front of the differential. The obvious benefit to this layout is the reduction in the yaw moment of inertia due to the mass being within the wheelbase.
The change was not universally adopted due to ground effect aerodynamics dictating narrow gearboxes. Pete Weissmann’s fantastic transverse designs used by Brabham are worthy of note however, and are discussed by Van Valkenburgh.
Throughout the 1980s the Hewland style gearbox was still used by the majority of teams. Reasons for this include simple serviceability, and with turbo and composite chassis technology an off the shelf Hewland FGB was a reliable component that wasn’t going to go wrong.
Interestingly it was the end of the turbo era that signalled the full-scale adoption of the inboard transverse gearbox. Faced with heavier atmospheric V8s and a desire to maintain the weight distribution of the turbo cars Enrique Scalabroni at Williams and Dave Wass and Paul Crooks at Benetton created a new generation of transverse gearbox that would soon become state-of-the-art.
The use of transverse inboard gearboxes continued until the mid 90s, but following the Imola tragedies the FIA restricted the diffuser aerodynamics dramatically.
To optimise the diffuser it became necessary to narrow the bodywork in front of it to ensure that the now critical central section was adequately supplied with air. Narrowing the gearbox was once again a crucial design target, perhaps more so than in the era of the tunnel cars of the late 70s and early 80s.
So it was that longitudinal gearboxes were back but this time in an inboard guise. Important developments in case construction cemented the change and in modern Formula One every team runs an inboard longitudinal gearbox.
The following diagram from Xtrac shows the basic layout. The input shaft from the clutch has six ratios (or seven in some cases) permanently fixed to it and these drive the six corresponding ratios on the output shaft. The output shaft ratios aren’t locked to the shaft but run on bearings. To lock an output gear to the shaft and thus engage that gear a toothed ring is slid by the selector fork and the teeth engage with corresponding features on the gear itself. This ring is called a dog ring and the teeth are called ‘dogs’.
Once one gear is engaged the output shaft will rotate and transfer drive to the differential. Obviously the drive must be rotated through 90 degrees and this is done with a conventional bevel gear set. A further spur gear reduction serves the dual purpose of allowing more freedom to position the differential and allows the primary ratios to rotate closer to engine speed, which reduces the torque load they are required to take. It also allows the greater reductions required with 18,000rpm+ engines without increasing the size of the gears.
Lubrication is provided by Gerotor type pumps, which are driven off the input shaft.
To understand this system of parts we will look at each one in turn.
Before the 800-odd horsepower of a modern F1 engine reaches even the gearbox it must pass via the clutch. A modern F1 clutch is quite astonishingly small in size. The most documented unit is the 110mm diameter Sachs unit used by Ferrari.
The Sachs clutch uses a wire eroded titanium cage and multiple carbon fibre reinforced carbon plates (the same material as carbon brake discs and pads). The drive cage is now welded directly to the flywheel and the drive plates are just 97mm in diameter. The overall mass of the unit is just 830 gramms!
The latest generation of F1 clutches are pull operated rather than push operated. This allows the removal of a fulcrum plate which further saves mass.
The gears are straight cut. The main reason for this is that it is the most efficient way of transmitting power. Excessive friction due to the axial load generated by helical gears is avoided.
The teeth of the gears are of an involute form. This ensures that the tooth to tooth contact gives a conjugate action so the velocity ratio remains constant as the angle between the teeth changes.
The design of involute spur gears is actually one of the first things you learn in mechanical engineering design. As a result there are many resources available. The most useful I have found online is the gear design manual on the Boston Gear website.
The material properties required in a gear like this are unsurprisingly high strength, high stiffness, toughness and good fatigue resistance. The most suitable material is a heat treated modified form of SAE 4340 nickel-chrome-moly steel often called 300M. The shafts are also be made of this material.
Straight cut gears create no axial forces so the input shaft only requires positive location from a radial point of view. This is done using cylindrical roller bearings with the hardened gearshaft acting as the inner race.
The output shaft is more complicated because it has to accept a large axial thrust due to the bevel transfer gearing. Generally a double row taper roller or angular contact ball bearing next to the bevel gear is used which takes all the axial load. A floating cylindrical roller bearing provides radial support at the other end.
While this float allows axial expansion as the temperature goes up, the requirement to keep the double row bearing from rotating in the case means that a very large press fits are required. These fits remove all the play in the bearing, which can damage it and is the reason F1 gearboxes are preheated before use.
A solution to this problem being pioneered by the German company Cerobear is the use of hybrid bearings with flanged outer races. The figure below shows the new installation. The outer race of the bearing is a flange that is bolted to the gearbox casing. This provides the rotational location of the bearing without the press fit. The axial preload required by the bearing can be accurately controlled by tightening the nut that secures the inner race onto the shaft.
These bearings are referred to as hybrid because they use silicon nitride ceramic balls and PEEK (poly-ether-ether-ketone) cages in conjunction with conventional steel races. Similar units are used for the wheel hubs.
Another advantage of hybrid bearings is that they actually operate better without a direct oil feed. The oil mist created by the rotation of the gears is generally adequate and this could allow reduced oil usage (currently a box uses about 1.5 litres).
The gearbox casing of an F1 car has an important dual function. Firstly it must positively locate the gearbox parts so as to ensure their correct operation. Secondly it must be stiff both in torsion and bending because it is an integral part of the vehicle’s chassis and must accept aerodynamic and suspension loads.
Both of these tasks must be achieved at elevated temperature, i.e. above 100°C. Magnesium was the material of choice for many years due to it’s low density and good castability but it’s modulus is lower than that of aluminium for example.
Aluminium is perhaps a better material than magnesium due to it’s stiffness particularly with the requirement for narrower longitudinal casings. Unfortunately the state-of-the-art in aluminium casting (sand casting) didn’t allow thin enough wall sections for an efficient case to be manufactured. This changed in 1993 when Sauber and Tyrrell exploited new die-casting developments to create a thin walled aluminium case.
In terms of performance at elevated temperatures composite materials and titanium alloys are attractive for this application. With the advent of high temperature epoxies carbon composites can operate at high temperatures and the low coefficient of thermal expansion is desirable. Titanium is very stable at high temperatures but is difficult to cast and weld.
Ferrari started innovating with both these materials. Using TIG welded titanium sheet and machined plate sections John Barnard created a stiff and light main case and used moulded carbon fibre for the end case and bellhousing. Problems arose due to the bolted interface between the titanium and CFRP sections. These joints were in the vertical plane and caused a reduction in the torsional stiffness of the case as a whole. Modifications have been made to make these joints on angled planes and the hybrid CFRP/Titanium case has successfully carried Ferrari to driver’s and constructor’s titles.
The other current technology is the die cast titanium case used by Minardi (see diagram). This case would not exist without the technology of stereo-lithographic rapid prototyping. Complex shapes with undercuts and re-entrant features can be cast for the first time and the Minardi case is a both a neat and structurally efficient design.
The information available on the Arrows and Stewart carbon fibre cases is very sparse. It is clear that the original Arrows case has a full carbon fibre skin, but the elements that held the gearshafts and bearing in pace were machined from titanium.
Semi-automatic and automatic shifting
In an ultra narrow formula one car the need to get a mechanical gear linkage from the cockpit to the rear mounted transmission has been a problem for designers for years.
The profoundly inelegant and difficult to manufacture solution of creating a cylindrical tunnel through the fuel cell to accommodate the linkage clearly irritated John Barnard who’s detail design is legendary. When he arrived at Ferrari in 1987 he decided, having seen Lotus’s use of electro-hydraulics for active suspension, to apply the technology to gearshifting.
Using the now well known arrangement of paddles behind the steering wheel Barnard linked these to a servovalve (MOOG Type 35) and four actuators. The four actuators were used on the three separate selector rods and a separate actuator for reverse. This meant that any gear could be selected in any order.
Initially the system also ran with fully automated upshifting, i.e. the rev limiter was hit and the upshift took place. This automation was actually removed at the test driver’s request to allow him more flexibility to short-shift should he require more control of the engine.
The downshift on these early semi-auto boxes was just like a manual in that the revs had to be matched by the driver and the clutch used manually. The actuator was a simple switch as opposed to a looped control system.
Ferrari’s new technology won first time out in the 1989 Brazilian GP piloted by Mansell who also won in Hungary later in the year.
When other teams began to develop similar systems, the shift speed was such that they felt a sequential mechanism like the ones found in motorcycles could be more compact and require only one actuator. Williams made this change on the FW14 and this was the direction semi-autos then took.
The final step was a means of controlling the synchronisation on downshifts. The arrival of electro-hydraulic throttles allowed computer controlled blips to replace the heel-and-toe technique.
At this point the FIA stepped in and stated that the throttle and clutch could only be taken out of driver control during a shift and that each gear change required a discrete control input from the driver. At this point in time the up- and downshift in a semi-automatic gearbox obeyed the following general patterns:
An upshift is carried out without the clutch and at full throttle. The engine management system is used to retard the ignition to slow the engine as the dog ring is disengaged.
With the speed of the shift in the region of 0.02 seconds the engine cannot actually slow down enough to synchronise properly, the drop is in the region of 600rpm against the required drop of between 1500 and 2000rpm.
The excess energy present as a result of this is transferred to the wheels, aiding acceleration. However this can destabilise the car so the speed of shift must be traded off against vehicle stability.
It would be logical to assume therefore that upshifts in the wet are deliberately slower than the transmission is capable of.
A downshift is carried out with the throttle closed and with the clutch. The throttle isn’t closed by the driver lifting his foot, but by the drive-by-wire system. Downshifts are therefore full throttle in terms of what the driver does directly with the pedal.
When the shift is requested the rpm that the engine would rotate at given the road speed and the new gear is calculated, if this exceeds the rev limit the shift is rejected and the driver must flick the paddle to re-request a shift.
The throttle is closed and the clutch is actuated to unload the dog ring. To synchronise the gearshaft speeds, the throttle is automatically blipped and then the selector drum is rotated. Downshifts are slower than upshifts at around 0.05 seconds but the prevention of rear wheel lockup is a more important goal than shift speed.
The paragraphs above described the systems used by all F1 teams until the Spanish GP this year. The regulations now allow the total control of gearshifting, throttle and clutch actuation by the computer control systems on the car.
The changes won’t have much effect on gearshifting In my opinion because even the worst drivers could get the engine speeds in the correct ballpark, such that the system wouldn’t reject a shift. The only change will be that the driver’s reaction time is removed from a downshift sequence.
To allow the freedom to automatically shift to a desired gear by skipping gears will require the reversion to individual actuators and shift rails to replace the sequential arrangements. I don’t think this will occur from the point of view of system mass and packaging considerations.By Ben Michell