Homebrew PCBs

Richard wrote:
>  
>
>
> Excellent and very powerful PCB CAD software already
> exists(KICAD,Eagle,etc.)for FREE! Do you really want to start your 
> project with trying to first write your own routing software?? I don't 
> think the "homebrewer" has on staff a seasoned team of code writers at 
> their disposal : (
>
No, I don't want to reinvent and reimplement existing router apps. I am 
thinking of using the scan to e.g  correct component placement in the 
input file of the Router, using maybe a router produced Gerber and 
slightly shifting components, modifying/extending Kicad  etc. etc. There 
are many approaches thinkable and they don't seem insurmountable to me.




> Well, yes it does! The CAD software un-forgivenly demands this. It 
> expects what you place in its "virtual design board" to be absolutely 
> representative of the real world "outside physical placement" of the 
> actual components location. When I have you print the component layout 
> screen to the substrate, the CAD software later makes reference to 
> this fine placement when you flip the module upside down and now needs 
> to print micro thin conductive traces. Think about a 128 pin QFN IC 
> with lets say 50 0404 SMDs. Your SMDs better be right on the money!
>
That problem is exactly what I tried to solve with my suggestion. Did 
you ever try to manually place something onto something adhesive? When I 
do, it's usually always a tiny little bit off. And then there is no 
pushing around anymore and cursing won't help either, as soon as  one 
corner touches the adhesive surface, that's it. Well, I guess some 
mechanical and optical (think parallax) help can be constructed. I'd 
rather write some code, trying to take human failure out of the equation.

> In re of "probing" .... your wrong! All the SMDs remain 
> 100%accessible, even the connections "inside" of the constructed 
> "micro-hybrids". Yes, they are covered by the insulating mask, simply 
> run your finished module through the printer again, this time printing 
> the "inverted version" of the component screen. With a micro-sized 
> drill or sharp needle point probe, slightly tag the surface to pass 
> through to the conductive layer underneat
>
I like your idea of  'running your finished module through the printer 
again, this time printing the "inverted version" of the component 
screen'. Good thought. But I have difficulties to think of how that 
would work in the real world. What is if there is a conductive layer in 
between, e.g. a  ground plane, which I can imagine you want considering 
all this tight spacing and possibilities of capacitive coupling etc? Or, 
if no layer, what if a trace crosses just there and you are poking your 
needle through possibly more then one of them? It doesn't seem practical.

>
> That's very true! But think about it. Is the "homebrewer" equipped 
> with special and very expensive "solder re-work stations" that allows 
> changing out a 128 pin QFN or 0404 SMDs??? Your talking about ICs that 
> cost on average $3 to $5 and descretes around 20 cents a piece.
> Have you ever tried to "solder wick" out a defective IC before? 
> Chances are you'll lift every "pad" right of the board!
> My point is .... for the cost and time involved,if you have a 
> defective module...PRINT OUT A NEW ONE!! In about 4 hours, your back 
> up and running.
>

Yes, I hand solder surface mount components quite frequently. And yes, I 
have tried and succeeded to "solder wick" out a defective IC before, 
more then once. O.k. not a 128 pin QFN, but regular SOPs etc.. The 
majority of SMT components ain't 128 pin QFN's anyway. Regular solder 
station and a fine tip is all you need. I want to be able to do that 
with my prototypes where an IC  (your quote) 'cost on average $3 to $5'. 
PRINT OUT A NEW ONE (your quote) and throwing away maybe 10 of the other 
chips in the circuit worth $3 to $5 apiece does not sound like a great 
option to me. 


> Manufacturers today must commit their prototypes to the fabracation of 
> "real" sized PCBs to even develop the circuitry through many ite
> ations of the "true sized" PCBs.
>
Very true. But that's not what you are doing with your technique. Your 
'hybrid' is as far away from a a 'real' sized PCB as it gets.

The advantage of real prototype PCBs are also that I can put my finger 
onto components or traces, introducing capacitance and changing circuit 
behaviour, take a soldering iron and quickly add a cap, change an R etc 
etc (with my cheap soldering iron), until i am ready for the next 
iteration of the prototype pcb.That would possibly save substantially on 
the number of iterations.  Each iteration with your technique, even the 
smallest, requires a complete redo. I can change an R in a minute while 
you have a 4 hr turn around.

Derk