I saw your post on this lately and found it quite intersting. Here, I'm
using a fairly "industrial" aproach in that I use an almost identical
process than they use.
Basically I first cut out the base material to form a stack. In case of
a four layer, the inner two layers consist of a double side claded base
material of one third of the usual board thikness. The outer layers are
similar in thikness, but only cladded on one side. I decided a while ago
to always work on a "mini pannel" and dump execss material at the end of
the process by cutting it off. This has helped a lot in serveral areas
to simplify the process. I "support" here two pannel sizes, but could do
other if the need really arises. The "norming" just helps to keep things
consistant, fast and simple. The most common "pannel" size I use is
160x100mm (euro card size), then there is a 180x100 variant in case the
final product does not have the mounting holes where I usually place
them (at 5/5mm and 155/95 mm, 3mm diameter). These holes are the
alignement holes for the shafts and in many cases at least the first one
is also at the same time a later on mounting hole for the PCB. I drill
the first hole through the whole stack, put a shaft in and then drill
the second hole. I meanwhile have a CNC drill, but I did this also back
when I did not had one (in other words the method also works with hand
drilling).
Once the stack is defined this way, I disasemble it, make the films for
the inner layers, laminate, expose develop and etch the inner layer. I
then glue the outer layers using five minute epoxy and a home made press
(which basically is a very simple wood construction with two vices to
generate the pressure). Needless to say that for the glueing process, I
insert the shafts to keep the layers aligned. One of the pressure
applying plates is haveing holes for the shafts for this purpose too, so
as the shafts do not get bent from the pressure and remain in a right
angle compared to the PCB.
After glueing (the nice thing compared to the industrial process is that
it's a lot faster, just 5 minutes), I drill the holes through the
complete stack. The CNC is also having positioning holes on a
permanently mounting plate, so this process is easy, fast, precise and
reporduceable. I simply stick the stack onto the shafts of the CNC
drill. No further mounting or positioning requiered. Without CNC I used
to laminate, expose and develop (but not etch) a pattern on one outer
side to serve as a drill gudance, then stripped it away after drilling.
The next step then is through plating the stack in my home made thourgh
plating machine. I used to present it here some years ago, and to
describe the hole process is a bit long for a single post. I meanwhile
have overworked the design of this machine to reflect the experiance
made and will document this in the near future (no time yet). Anywas, I
may should say that throughtout all those years I had to replace the
chemistry (which is in case of the palladium bath somewhat expensive,
but not THAT expensive) only ONCE. But maintaining the baths in a helthy
state is important. If you master this, making four layers is just plain
and simply said dirt cheap and fast (4 - 6 hours depending on your
experience and equipement).
After the throughplating, the outer layers are laminated, exposed, and
etched. In simple cases this completes the process. In cases where the
PCB has to be seen by external people, I also apply a chemical thin
imersion step and then laminate, expose, develop and harden a solder
stop mask I have available as a laminate. I also do this in cases where
I have to reflow solder components due to their nature (BGA etc.).
Ok, that was a bit short, but I hope you get the picture.
Markus
DJ Delorie schrieb:
>
>
> Markus Zingg <homebrew-pcb@...
> <mailto:homebrew-pcb%40shdesign.info>> writes:
> > As far as the aligning is concerned, I'm mostly doing 4-layers here,
>
> Woot! Another one! http://www.delorie.com/electronics/usb-gpio/
> <http://www.delorie.com/electronics/usb-gpio/>
>
> How are you managing vias and stackup details?
>
> > The printed black rings I use to position the hole are just about
> > 1mil bigger than the physical hole in the puncher.
>
> What I've been doing is adding a registration target, and using an awl
> to mark a dimple for a 13 mil drill in the PCB stock. I use that to
> line up the two sides. For four layer boards, the outer layers are
> see-through anyway, so I just line them up visually when I tape them
> on. Or, I drill all the 13 mil registration points and use spare
> drill bits or wire to line them up.
>
> > I have no personal experience with the toner transfer method but
> > since it relies on lasers I'm not sure if one can achive a
> > reproduceable match on multiple printouts as it would be requiered
> > for the purpose.
>
> The size is not 100% scale, but it is very predictable. I've added a
> correction factor in my output specs, and can usually get the two
> layers lined up within a few mil of each other.
>
> I'm guessing that photomask alignment would be easier; just draw
> registration marks outside the pcb area and line them up visually. At
> least, if you're exposing both sides at once.
>
>