> I wonder if they use something like the acid on "frosted" glass...

There are both mechanical methods and chemical etching methods to make the 
patterns.  Basically you are just making a pattern on the surface of the 
metal in the mould.  For one of our plastic products (which is an optical 
component so is transparent plastic) they make "fine adjustments" to the 
surface texture by some dude manually "roughening up" the surface in certain 
positions :-)

> Yeah, well, that's a function of far more than the plastic shell. Gotta 
> assemble the whole thing for that! ;-)

Usually you don't do any testing on individual parts, only on the assembled 
product.

> Apparently by dad spent some time working in a plast where they do "vacuum 
> forming", and each time you do that, you have to cut a bunch of wasted 
> plastic off the edges. He says they sent it back to the suppliers and they 
> got a discount. (But then, how hard can reprocessing a flat sheet of 
> plastic be?)

Well yeh, if from every single shot you have some waste then it makes sense 
to work out how to reuse that.  But working out how to reuse 5000 samples is 
a bit pointless if you are planning to make 2 million every month later.

> And that brings us back to my question about components which are supposed 
> to "snap together". I guess it depends on how deformable the plastic is? 
> (And that surely must depend on shape...)

Yep, or technically "stiffness", which is how much something deforms for a 
given force (units DISTANCE/FORCE).  You can of course design the shape so 
that things deform easily for clips, or make springs to hold things in place 
etc.

> It kinda makes you wonder how anything ever gets made in the first place!

Yeh I often wonder that too...

> Heh, and I bet that's even *more* damed expensive again... not to mention 
> the quality hardened steel you need for it to mill...

And then the actual milling bit, which presumably must be harder than the 
hardened steel :-)

> I wonder... do companies buy all this stuff themselves? Or just 
> subcontract it out?

Well you get quite long supply chains sometimes, like take for example car 
company 1.  They will have supplier 2 that makes some big chunk of the 
dashboard.  Supplier 2 will have a supplier 3 that provides one part (eg the 
radio).  Supplier 3 will then use supplier 4 to make the plastic part to go 
on the front of the radio.  Supplier 4 will own the injection moulding 
machine, but could well use a supplier 5 to actually manufacturer the tools 
to go in the injection molding machine.  And of course there will be another 
supplier 5 that supplies the raw plastic to supplier 4.

Companies like Nokia and Ford don't actually make much, they just design 
things and get other people to make them, then assemble bits together :-)  I 
once saw a Nokia factory, each production line was about 50 metres long, 
they simply screwed some components together, did some electrical and 
optical testing, clipped the plastic parts on and packaged it into several 
layers of cardboard boxes.  Very low tech and very simple.

> I also wonder... Today we have machines that make components of machines 
> that are used to make machines for making machines. Some of this stuff is 
> made to absurd tolerances - stuff no human could ever make by hand. So... 
> how the **** did we make it in the first place?!

Well just because you only have equipment to design and make stuff to 
tolerance X, it doesn't mean that you can't make a machine that can make 
stuff to a tolerance lower than X.  eg you could just build a motor and then 
gear it down a lot so that the movements are very small and accurate, or use 
optical lens to shrink your design, or use a lever so that a larger 
tolerance on one end corresponds to a smaller tolerance at the other end.

Post a reply to this message