Continuously Variable Transmission (CVT)
by François Dovat
(© François Dovat)
In the early days of automotive technology, the engine's power was low and a simple solution could often be applied: a friction variator (today named CVT for Continuously Variable Transmission) by means of a flat disc driving a wheel at right angle. By sliding the wheel on its shaft, it was possible to vary gradually the transmission ratio, either in forward motion or in reverse... Moreover, a neutral was obtained by pulling back the plate on its shaft to interrupt its contact with the wheel. The device accomplished then the four functions of clutch, variator, inverter and bevel gear! It was used in particular in the USA by Carter, Lambert and Metz (the latter's car was deliverable in a kit of 14 boxes), in England by GFK and in Switzerland by Turicum during the years 1906 - 1920. It was not an automatic system because its regulation was entrusted to the driver.
Obviously, such a transmission is completely out of question as soon as the engine reaches some real power. Various kinds of friction variators were then imagined and tested. One of the most promising is due to Dr. Josef Beier; it comprised a great number of slightly conical male discs stacked on a splined shaft. Another parallel shaft carried female flat discs whose periphery was chamfered at the same angle as the male discs, in order to produce short contact lines. A large and strong spring end-loaded the pile of female discs the ones against the others, trapping between their rims the coned male disks. The transmission ratio was varied by bringing closer or moving away the two parallel shafts. Oil jets were directed between the discs so that the drag force was obtained by fluid shear of the oil film at the contact points. This transmission was tested before WW II on two Volkswagen and it appeared reliable. In 1954, Napier adopted it for its turbocompound aircraft diesel "Nomad" to vary the reduction ratio between the turbocharger and the crankshaft. Three piles of discs were set at 120° around a central shaft, transmitting up to 195 hp. The slip varied over the operating range between 2% and 5% and there was a constant loss of some 20 hp throughout the whole range of operating conditions but the engineer responsible for the development of the Nomad, Ernest Chatterton, estimated that these losses could be reduced by further improvement of the system.
An alternative consists to use a pack of discs with internal contact. A kind of clutch engagement in direct drive is then possible when both shafts are concentric. Meshing on the splined shaft at maximum reduction ratio would also be possible if the rimed discs are cogged at their inner circumference. Between 1973 and 1976, Citroen developed and tested on a GS car the system designed by engineer Flichy, a concept which included two external sleeves providing double variation so that the input and output shafts were coaxial. In the figure, the input shaft 1 drives the male discs 4, the output shaft 2 being driven by its female discs 5. The ratio continuously varies between the direct drive and the maximum underdrive by offsetting the sleeve 3.
Another "friction" system, known as toroidal and invented in 1877 was improved by Frank Hayes in the Twenties and used again around 1930 by Austin. 700 Austin Seven "York" equipped with this transmission hardly gave satisfaction. Nevertheless, research and development on this principle did not cease in Great Britain (see our file). http://www.histomobile.com/histomob/tech/2/107.htm
Meanwhile, another concept of mechanical CVT became widespread: the V-belt and conical pulleys of variable effective diameter. Patented in 1897 in the USA by H. C Spaulding, this device was initially used on machine tools, then as from the Fifties on motor cycles (Mobymatic of Mobylette) and scooters (DKW Hobby) as well as for the drive of centrifugal compressors (Mac Culloch VS 57). It was then popularized by DAF under the name Variomatic in 1958, with a ratio span of 3.72. This transmission remained in production until 1990 for the Volvo 340 with the Renault 1.4 L engine – in spite of a longevity of the two belts hardly reaching 40000 km. Working in the open, it was also fitted successfully on two Brabham-Cosworth of Formula 3 which gained 2 victories in 1967. It had been possible to tune their engines to extract the maximum of power without being concerned about their torque back-up. Some F3 Tecno were then also equipped with such a CVT.
This system has been further sophisticated lately since it is able today to transmit torques of more than 350 Nm, thanks to a single chain or metal belt working in oil inside a casing. Since 1987, the fragile rubber belts had seen the competition of a metal version mass-produced by Van Doorne Transmissie (VDT) and composed of two parallel endless steel rings enclosing 300 to 400 "push plates". The sides of these push plates, in V at the angle of the pulleys flanks, ensure the contact with them. The push plates press against each other while being constrained to follow the path dictated by the rings. That's why this belt is known as a "push-belt". A Formula 1Williams FW15C using this new type of belt would have semi-officially set a new elapsed-time track record at Silverstone in 1993[1 ]. The CVT were then banished by the FIA for the following year.
Williams-Renault FW 15 C - CVT
In Europe, the preconceived and prevalent idea remains that automatic transmissions are intended for the quiet and slow driving characteristic of the elderly and not so sporty guys, although the opposite was clearly shown in racing by JimHall's Chapparals in the 60's and by some other racing drivers, amongst them the engineer and great journalist Paul Frère.
In order to give a better image of the product by launching it again on the race tracks, Emery Hendriks of Van Doorne Transmissions contacted in 1991 Patrick Head, technical director of the Williams Formula 1 team. After a simple gentleman's agreement, it was decided to go forward. VDT built a test bench able to handle CVT powertrain prototypes with Formula 1 engines of 800 kW and more. The transmission was planned for a ratio span of 2.5 and a lifespan of 4 hours. Its development took one and half year, Williams taking care of the design and manufacturing of the casings, gears and conventional starting clutch controlled by the usual pedal. A sophisticated electronic control system was conceived in collaboration. Several small oil pumps were set, each one affected to a specific task. This made it possible to increase the transmission efficiency up to 95% instead of the much lower percentage of the marketed versions. A car was ready in 1993 and entrusted to David Coulthard, then development driver at Williams.
The results were promising, accelerations being uninterrupted and the maximum power permanently available. Moreover the pilot had his two feet available to accelerate with one and to brake with the other (advantage inherent to every automatic transmission as soon as the driver is accustomed to this practice) and its attention could be focused on driving itself. It was possible for him to adjust the engine revs according to the needed power, in fact to reduce the rpm if his advance on the other competitors would have been sufficient.
Being given that a CVT contains intrinsically all the conceivable ratios, it wouldn't have been necessary any more to transport for each race some 600 kg of gears in order to adapt the transmission ratios to the track's characteristics. However, the FIA banned any driving assistance such as ABS, ESP and CVT, this probably in a bid to maintain the spectacle and the popularity of the Grand-Prix. The Williams FW 15C then finishes quietly her days in a museum and documentation relative to these developments takes the dust in the drawers of the two companies.
Yamaguchi Jack: CVT' S rapid expansion, Automotive Engineering, March 1996
Without author's name: CVT, Transmission à Variation Continue ou Transmission Viable Commercialement
in 'Ingénieurs de l’Automobile N° 690, Juin-Juillet 1994'
The FW 15 C is visible behind a DAF Tecno F3 CVT of 1968.
Naturally, these transmissions require a regulation system, a clutch and an inverter. The solutions of the different manufacturers largely vary. DAF then Volvo used a centrifugal clutch as well as a centrifugal regulation supplemented by a system linked to the depression in the intake manifold. Some, including Nissan, use a hydrodynamic converter with a lock up clutch. Subaru and Fiat (Selecta) exploit a magnetic powder clutch and a gear train mechanically shifted by a synchronized sliding sleeve for the reverse gear. The regulation and control of the variator are hydraulic whereas the clutch is electronically controlled. As Ford (CTX) did previously, Audi prefers two multidisk clutches and a planetary reverser, the whole transmission being controlled by a hydraulic unit. This unit also regulates the pressure exerted on the pulleys flasks enclosing the belt: the clamping pressure increases according to the transmitted torque in order to limit the mechanical load and avoid useless losses. The ratio span of the CTX was 5.8 whereas the recent Multitronic of Audi reaches 6.05, which allows low engine's rpm when the need for power is small. (see our file)
The drivers are so much accustomed to hear the engine revs soaring linearly along with the car speed and then fall abruptly at each gearshift that the constant noise emitted by an engine revving at continuous - and optimal – rpm whereas their vehicle accelerates frustrates them. They have the wrong impression that the clutch or the belt slips. This fact has always been an important obstacle to the CVT's market penetration, just as the need for automatically increasing the revs at the time of the hammering down on the accelerator. If the system of control lets the aforesaid revs go up first before increasing the force of traction, the driver feels a lack of spontaneity in transient response. On the contrary, if the engine continues to rev slowly, it won't often be able to develop all the power desired by the driver. To avoid this dilemma, the manufacturers choose a compromise between the two solutions.
To avoid this problem, electronics also makes it possible today to program "virtual ratios": the continuous variation that heaps of inventors, researchers and engineers ardently sought since more than a century is by-passed whereas the technology finally makes it competitive! One can program any number of discrete ratios, but 6 to 8 are set in order not to exceed the discerning capacities of the average driver. In so far as the continuous mode remains available in alternative, these lock-ups of the variator are not useless because they can be used for engine braking. Nevertheless, a more elegant solution for engine braking (and racing) which does not exclude the previous one, would be to leave also to the driver the choice of the rpm, the revs being regulated to a determined value independently of the speed variations of the vehicle and, possibly, independent of the engine load.