Homebrew PCBs

On Wednesday 28 April 2004 01:38 am, Phil wrote:
> actually, I think that this could be done with what is emerging as a
> some what standard CNC PCB machine (3 axis basis plus "special"
> axises (axees?)) with a wrap gun attachment.  I see two problems that
> need solution:
>
> a) routing the wires. probably done by hand but you need to handle
> the issue of binding and wire build up.

I'm not real sure about how you'd handle this part.

> b) ww socket tails (i.e. the wire posts you wrap on) are not terribly
> accurate in their position.  I've used em and its inevitible that
> they get bent a little.  Finding the post to slide the wire spinner
> onto would be tricky.  maybe just have a funnel on the wrap tool to
> guide the sleeve to the post.

It couldn't be very big,  as close as those pins are to each other.  I guess 
that's why the ends of the pins are pointed...

> Of course, this is kind of a moot point as WW appears to be
> dissapearing.  Guess those pesky SMDs dont wrap very well...

Is it?  I can't say that I'm tracking things well enough to have seen that one 
way or the other.

> But this does bring up a kind of wild idea I've thought about during
> episodes of low blood sugar.  Why not just have a direct wire
> machine?  Stuff the components (TH, of course) into a predrilled
> board. Invert the board (securing the components somehow) 

Bending the wires will usually do that.

> and then a machine strips a wire, solders it to a lead, moves (er, routes
> the wire) to the next lead, cuts the wire (if terminal run), solders it
> to the lead and moves to the next lead.  There was a company in the
> 70s (could still be around) called multiwire or some such that did
> this for fast turn prototypes. It was quite expensive but it produced
> some very complex boards fast.  If I remember correctly, the first
> intel 386 logic simulator (made out of random logic gates) was built
> with this technology.  I think fast turn PCB houses pretty much
> killed their business.

I can't imagine anybody prototyping a 386 chip with random logic gates...!

Sounds like such a machine would be possible,  but you'd need to have the 
feeding of wire,  stripping,  soldering,  and so forth all working right.  I 
guess we could lump most of this stuff under the heading of "automated 
assembly",  which seems like a pretty good topic to me.  Even if we don't go 
with your suggestion or with wire-wrap,  something useful may come out of 
this...