I worked on continuously variable transmissions while a student engineer working at NASA's Glenn Research Center in Cleveland, Ohio, then called the Lewis Research Center, or LeRC. It was over a period of years in the late 1970s.
A Lithuanian mechanical engineer named Algirdis Nasvytis was the genius who invented CVD. We machined models of his designs many times, assembled them, and many of us STILL couldn't figure out how he'd ever been able to create such a beautiful yet ultimately simple design. I've never seen a mechanical linkage more beautiful than a Nasvytis-designed CVD.
We were under contract to the US Army to develop a CV transmission for helicopters, and to the US Navy to modify them to fit inside torpedos. The challenge with helicopters was to allow the transmission to operate for sixty nautical miles (if memory serves me correctly) without transmission fluid. This operational requirement came from the Vietnam War era, when rotor transmissions were shot and the helos had to descend immediately - usually into enemy territory - rather than beable to fly back to the friendlies, because loss of transmission fluid led to a sudden heating of the mechanical gears, which led to catastrophic melting and seizure of the entire assembly. The NASA engineers were able to meet those requirements using a clever combination of hardened metals inside the Nasvytis CVD and solid-lubricating coatings of the metallic components that were originally developed for the Apollo space program.
The Navy required a quiet transmission system. Geared teeth in conventional trannies made an awful lot of racket. The Nasvytis CVDs were eerily quiet. At the 2000-4000 rpm rates of turn of the Mark-48 class torpedoes, the Nasvytis drive was almost silent. By contrast, the conventional geared torpedoes made so much noise we had to put them in soundproofed rooms when we showed the Navy the difference. Sound level meters showed at least 30 or 40 dB of difference in acoustical noise.
The Nasvytis drives comprise a main shaft providing input torque at one end of the drive, and a main shaft turning at a different rate of speed at the other end of the drive. The mechanics in between consisted of an odd assortment of fixed conically shaped cylinders which transmit torque gearlessly by means of idler wheels to other conically shaped cylinders. The idler wheels move up and down the cones, so that the rotation rate depends on where the wheels touch the cone(s). The whole mechanism was interlocked with various internal rings and other devices which held the idler wheels in place relative to the cones.
The points of osculation required very special fluid which was the opposite of engine oil. You didn't want this fluid to be slippery; instead, you wanted it to be 'stiff' so that it would transmit as much of the shearing force as possible, without a lot of shear, which resulted in waste of energy and conversion to heat. Dr. Nasvytis found NASA Lewis engineers to be an almost perfect mix of mechanical engineers who understood linkages as well as lubrication engineers, who were world class experts in lubrication fluids, who understood stresses and shears in fluids, turbulence, Reynolds number, Navier-Stokes equations... One group usually didn't talk to the other group, and yet at NASA Lewis they were in the same building, within the same department, and Nasvytis's designs were soon seeing wide usage in Defense Department applications.
A word about Dr. Nasvytis. He was a tall, stately gentleman who knew all manners of technical knowledge in an incredibly wide range of fields. His manner was polite. He was courtly in the manner of many European men with technical degrees. Although many of us were pocket protecting engineers who were uncomfortable when outside our areas of expertise, Dr. Nasvytis enjoyed the arts community of Cleveland tremendously. He especially enjoyed listening to George Szell's conducting of the world class Cleveland Orchestra. His cultured manner did not go unnoticed by the rest of us. Many of us aspired to become the sort of worldly-wise man he was.
(He may have been Latvian. I am not certain of his nationality, but it was one of these two. Thanks to JudyT for pointing out the linkage between the spelling of his last name and the link to Lithuanian nationality.)