Hacker News Reader

Comment by omginternets | Hacker News Reader

> death by cad

How does this work in practice? I mean, what is it about CAD that gives this kind of result?

A system prototyped by hand rapidly finds the bolt or pump that can only be removed or fitted if you have an extra elbow and third arm. Or see round corners. Special tools were to be avoided or used for the rare exception. So plan B before lunch! The old approaches of an access panel over the timing chain or sprockets, or a removable plate fixing a bearing in place barely adds weight and complexity, nor does a grease point on a swing arm or steering joint, but needs a different understanding. Understanding that there's a life beyond a warranty that global manufacturers are trying to forget.

CAD is great for making stuff fit - once. In combination with the progression toward write once manufacturing, the only consideration is the cost and complexity of assembly. Can our factory robots and CNC do this? Can we reduce the component count? Can we use one-time moulded clips in place of those machine screws? Some is driven by cost saving, much by the purely mental visualisation capability of the new process.

Maintenance, even regular expected and necessary maintenance, increasingly takes a distant back seat. What once took 20 minutes to change now takes half a day. It now fits in a difficult corner or got combined with a dozen other systems making it now a £500 replacement.

It's visible top to the tiny fractions at the very bottom - saving fractions of a penny on self-tappers or clips self moulded into the single injection plastic where once (in ancient times like the 1990's) there'd be a small metal captive nut or bracket to bite into, or real machine screws and nuts.

Any and all of these can take years off expected lifespan mainly for the saving of pennies or a production step.

There are a lot of opinions about what could or could not be built better in cars.

The most interesting trend, obviously, is that people are just buying fewer of them, so it will soon be a moot point.

Doing maintenance never has mass-market appeal.

The best approach, in my opinion, is actually the opposite of what most mechanics (or MechE/EE people) will say: we should try to achieve as much as possible in software, with the absolute simplest hardware.

Consider that an electric car, the fastest growing product in the category, has a considerably simpler engine and drive train. The software rules the roost there. It's just a coil. The engine and batteries are not really maintainable (they're certainly recyclable).

There's a lot of knowledge tied up in what you're saying, which is a really interesting point of view. But optimizing ICE vehicles and their attendant maintenance is sort of a dead end.

Then again, I'm agreeing with the general idea: CAD is bad. But what should the right response be? Don't buy things that need CAD designs at all!

I believe it refers to assembling things digitally without thinking about how an actual human would work with the same things in the physical world.

One particular instance of this I pay attention to is the placement of oil filters on motorcycles. My brother bought a Honda CBR600RR, and he wanted to replace the oil filter; I watched him spend an inordinate amount of time removing the cowling, clip by clip, screw by screw, from the bike, in order to expose the filter (which you could not get a normal filter wrench onto, because of the awkward positioning).

After that I pay a lot of attention to whether the oil filter is visible and accessible from the outside of the bike without removing a cowling, Kawasaki seems to do a good job of this.

To be fair, the CBR600RR is a race replica bike, and it's designed for the track (despite having lights and a license plate). The full fairings are there for aerodynamics, the same sort you'll find tested and proven in Moto GP.

Honda didn't put them there just to look cool, and ease of maintenance certainly wasn't the primary design factor.

Yup. Previous incarnation of a VW Golf had an engine compartment that fit together, and you could replace the headlight bulb in about 20 minutes with common tools. Current one has an engine compartment that fits together, and you can't replace the very same bulb unless you take half of the damn thing apart, and then you hit a snag, because the very last screw is unaccessible unless you're a nanobot, has a special nut, and some sort of seal guarding it. I suppose it does make sense if you have the whole workshop at your disposal and can disassemble and reassemble entire vehicles at will.

CAD allows much more optimization. In general the more you optimize, the better defined and all-encompassing your metrics have to be. Otherwise you will end up optimizing away qualities you took for granted (in this case, serviceability. or more specifically being able to easily remove inexpensive wear parts).

An old colleague worked for a mechanical engineering consultancy that helped companies optimize things like engines, etc.

They built measuring equipment and software that allowed the companies to squeeze the biggest engine possible into the smallest compartment.

Costs have big impacts on manufacturing. Saving $0.50 on some little widget may allow for much bigger savings in the build process. There’s also a general goal to make minute cost changes just to eliminate inflation without retooling.

I wonder if something like this was at work when the 96 Eclipse was designed. My then-gf in college had one; I changed her timing belt and water pump. It was miserable:there was only 1/2"of clearance between the engine and bay everywhere. I joked that they must have used a giant Play-Doh pasta maker to install the engine by squishing it in to conform to the engine bay.

Reply on news.ycombinator.com