Five thousand prototypes and 15 penniless years later, Dyson's Dual Cyclone model was launched on the market. It now outsells all of its rivals by at least ten to one and the Dyson factory in Wiltshire has a turnover of #163;100 million a year.
Did he strike lucky or is he a genius? Sitting next to a massive drawing board in his factory office he starts to explain by holding up a Canon Elph, one of those cleverly designed compact cameras with a lens that retracts into the body, leaving a camera small enough to fit into any trouser pocket. "Look, the Japanese are not brilliant people making quantum leaps," he says, referring to the camera's designers. "They work bloody hard making prototype after prototype. And the end result is clever because the hours of testing have pushed it to the limits."
It's a message he hopes will be driven home to children studying design and technology at school. He says they have the perfect vehicle to learn in the form of the national curriculum requirements, which he says mimic the way his company works. And he is doing his bit to help by ploughing #163;100,000 into an education resource centre at London's Design Museum, which was opened by Culture Secretary Chris Smith on Tuesday. "D amp; T courses in schools are outstanding and very much better than courses I have seen at art schools and university art departments," he says.
He points to the fact that pupils are asked to come up with an idea, sketch it on paper, make crude models out of wood or cardboard and tape - "we do a large amount of that here" - then, having proved the principle, make as good a working model as you can.
"That's brilliant. It exactly matches what should be in the design process. And it gives the proper meaning of design - not window-dressing, like the retail chains in the 1980s, but about much more fundamental issues, the technology the product uses, how well it works. And then there is how long it lasts, whether it is eco-friendly and its final form."
The design process for Dyson's vacuums started in 1978, when James got fed up with the inefficiency of his old Hoover at home and began to investigate. He tested it and found the suction quickly fell away even when a new bag was put in. As soon as a thin layer of dust formed on the inside of the bag it clogged up the pores in the paper and reduced the airflow.
Dyson didn't think of a solution until he had to build a 30-foot cyclone filter, the sort used in sawmills, in a factory producing Ballbarrows - one of his earlier inventions. It was needed to prevent inhalation of the epoxy powder sprayed on the barrows and baked to toughen the metal frame. It was then that he realised filtration did not have to involve a physical barrier, it could be achieved through centrifugal force.
He rushed home, tore the bag off his Hoover, got out the cardboard and gaffer tape, put together a model of the factory cyclone and connected it up to the machine's outlet hole with a piece of piping. The concept of the vac without a bag was born.
That, though, was only the start of the story. It took years of modelling,testing, testing and more testing to develop the Dual Cyclone, with an outer cyclone rotating at 200mph removing large debris and most of the dust, and an inner cyclone rotating at 924mph driving the finest dust out of the air before it leaves the cleaner.
He started with a team of four design engineers making prototypes in his coach house but now employs 150 engineers and scientists, most of them working in new product development teams at the Wiltshire plant.
They still, however, use cardboard and tape, as design manager Simeon Jupp explains. "At the product conception stage we knock out our ideas together on a block of paper, sketch in
3-D and then get to work with a glue gun and a stack of cardboard. It's a shorthand method of communicating. We make something as quickly as we can, out of cardboard or Styrofoam. The beauty of cardboard is you end up with walls. And with foam you can carve out the bulk and form of the product."
Each new product team has ten designers, and the cross-fertilisation has to be managed to ensure consistency of detail and aesthetics. "We are detail freaks. If we learn a good detail we won't be scared of using it everywhere. It could be the way a certain seal works, the way the lip on a moulding might mechanically interlock, " says Jupp.
Typically a new product team spends two months developing the concept but up to six months producing detailed drawings to the point where every micron of geometry is defined, every draft angle shown, so the toolmaker makes exactly what the designer wants.
Jupp, 34, has been there since the coach house days, but still relishes every new discovery about the materials he is working with. And he has inherited his mentor's belief that any obstacle can be overcome if you keep trying out and testing new options - you have to take risks to move on.
In fact, he believes the teams are getting better at exploiting the properties of the materials to provide a range of functions - using the elasticity of plastics, for instance, to create a snap fit, rather than screwing parts together. This increases the potential for recycling because most of the product can be thrown into the crusher for regrinding without having to be disassembled.
The Dyson staff have also borrowed a trick from coopers, who in the days of horse-drawn carriages would make wooden and metal wheels by heating up a metal rim, dropping on to the wheel's edge and allowing it cool, so it compressed and formed a tight bond. "Two plastic parts that snap together when they are cold make a better fit if they are hot," says Jupp. "It gives it more strength."
On several occasions the toolmakers and welders have said it is impossible to make the moulds they wanted (they use toolmakers in Portugal because of their history of glass making, and injection moulders in Italy because of their history of pasta-making). They told Dyson, for instance, that the spine of the first upright model could not be made.
"We forced them to try it. We said we'll risk it, " says Jupp. "Until you put in the pressure and squeeze in the plastic you don't know what it will do. It's not a complete science. The first time we ran it that tool broke the inside of the cores. But we believed in the overall design and refined it. James will never accept that something won't work if he feels it will."
Dyson, 50, is defiantly anti-conservative. Tall and slim, at work he wears clothes as colourful as a children's TV presenter - bright blue shirt, silvery-grey belt, red socks and rust-coloured loafers - and encourages his staff to do the same. He doesn't want them to be constrained by a suit-wearing businessman's mentality and deliberately employs graduates straight from college, undulled by the norms of the outside world.
Even his product designs have turned the vacuum cleaner from a dowdy household object into a colourful appliance - his latest stair-hugging, see-through cylindrical version, due out this month, looks like a psychedelic see-through snail.
The simple conceptual change had huge engineering implications, because transparent material is much more crystalline, much stiffer and more difficult to mould than opaque plastic. New compounds had to be especially formulated. But all that effort wasn't made just to achieve a sense of style. It's an evolution of the idea in the original Dyson model - that if you can see when the bin is full you know if it is doing the job. "The way a Dyson product works has always been integral with its style," James Dyson says. "This shows how design and technology can be one entity, and in a fun way. It uses its style to show how it works."
But above all, Dyson wants budding design engineers in our schools to understand that design isn't principally about style and finish - it's about how a product works. "Performance is the most crucial aspect," he says.
James Dyson, Against the Odds: an autobiography, #163;18.99, Orion Business Books
Dyson Centre for Design Education and Training, Design Museum, Shad Thames, London. Tel: 0171 403 6933