- A 1% failure rate on an F1 car may in fact be considered acceptable, but completely insane in a production car that may sell millions of units a year. F1 drivers, despite all that we can do, also accept a much higher level of personal risk than most other everyday drives.
- F1 cars only have to worry about a single passenger, and due to the controlled nature of the race a more limited number of failure modes... which is to say the scope size of the problem is significantly reduced - and designing for F1 is already non-trivial!
- Carbon fiber is almost impossible to work with on a mass production basis - the nature of the material prevents high-precision parts from being made, which is hell on a production line, but easier when you're hand-building everything
And you're right - it's entirely possible for a limited-size team to build a car from scratch - assuming they had all the parts as it is. It's entirely impossible for them to attempt to build the drive train, engine, all panels, structural parts, suspension, etc etc, from scratch. It's also very difficult for them to afford the sort of testing required to get this car to market.
People (especially software types) simply do not appreciate the sheer, ridiculous amount of effort (and money) it takes to keep them safe. This is also why I'm always wary of cheap goods, because I have seen first hand the difference between well-designed, properly tested products vs. not.
What car do you drive?
- Avoid Pontiacs at all costs. The quality difference between, say, a Sunfire and a Chevy Cobalt is ridiculous. Even "higher end" cars like the G6 are somewhat questionable. Chrysler are built cheap. The radiator motor that I worked on while I was there was timed to stop working after about 4-5 years (brushes in the motor will wear out).
- Avoid anything built in Mexico. This is not a knock against Mexicans, but my company's Mexican plant was completely devoid of precision manufacturing. Where an American plant would laser-guide parts into their fittings, in Mexico they're literally eyeballed (and malleted!). A full 80+ percent of warranty returns were due to parts built in Mexico.
- The high-end Chryslers are actually pretty good in terms of quality, and when I was working there the Ford Fusion was still under development - and based on what I saw (of the radiator design) Ford seemed like they were willing to pay for good parts. A good sign I suppose.
- Nothing beats Mercedes and BMW in quality. You do get something for your money, regardless of what people might say. Whether the cost per unit-quality is low enough for you is your own decision though :) Where GM was running brush motors, BMW runs entirely brushless (e.g. theoretically infinite lifetime, will run as long as your solid state electronics and assemblies hold together), and they're powerful to boot.
The plant didn't make anything for Toyota, Honda, or Mazda at the time, so I can't comment on those.
Were you the engineer who designed radiator shrouds, fans, and motors?
I consider myself a hacker, but I am a mechanical engineer by training. I've worked in car plants, and I think I'm much more uniquely qualified to speak on this subject than the blog author.
People do not comprehend the kind of cash that a car company needs (and should need) to get into the market. Let me detail the process of producing your car's radiator motor (just the motor!):
- Come up with your awesome motor design (doesn't cost a lot, assuming you and your fellow startup partners are good engineers)
- Draft your designs (draftsmen around here are unionized, and paid a great deal for their work)
- Build a proof of concept. In fact, build several (requires access to a very extensive machine shop, and unless your founders are all exceptionally talented machinists, you have to pay unionized labour more money to get this done)
- Now build several hundred more prototypes. These will go through performance and reliability testing. This will require access to heat testing chambers, sound testing labs, strength testing machinery... you name it. Each of these labs can't really be rented, and cost tens of millions to build and staff. Each piece of equipment needs to be certified by the relevant safety and regulatory body ($$$), and you'll need several people on board full time just to document everything to the minutae required by the same bodies ($$$). Do you know how much a certified laser micrometer costs?
And that's just to get your first working prototype. You're already probably 8-digits invested by this point, and you haven't even yet built out the necessary infrastructure to mass produce your product. By the way, casting a mould for a single part (e.g. your motor casing) costs $200-300K by itself, so moving towards production is not cheap either.
Now multiply that by the tens of thousands of parts in your car.
People (especially software types) simply do not appreciate the sheer, ridiculous amount of effort (and money) it takes to keep them safe. This is also why I'm always wary of cheap goods, because I have seen first hand the difference between well-designed, properly tested products vs. not.
The car industry can never move back towards a tiny-startup model. Sure, there may be room for smaller players, but "thousands of startups" like the web model will never fly, unless you enjoy having your wheels fly off the axle as you drive.