[sdiy] making PCB's with PNP blue

[ChristianH]

On Wed, 2 Nov 2011 09:08:35 +0100 Louis van Dompselaar <louis at dompselaar.org> wrote:

> That's why I opted for the photographic method, far less error prone and almost entirely dependend on fixed parameters only.
> 
> Plus you can reuse the sheets to do mulitple pcbs.
> 
> 
> >> They are both right, and here is why: lots of variables.


I can only second that.

This is my way of doing PCBs:
- design with EAGLE
- print on inkjet overhead projection transparency sheets,
    setting print settings to transparency (important!),
    black&white printing, max quality. No over-printing, no layered
    transparencies needed - that can only reduce precision.
    If possible, I let it dry for a day to prevent smearing.
    Then I sandwich the PCB, the transparency, and a glass plate for
    close contact of the transparency to the PCB, held together by two 
    strong clamps similar to alligator clips to avoid misalignment.
    (using a laser printer didn't appear to work significantly better)
- using only pre-coated photo PCBs (using blanks with spray-on photo
    resist is far too unpredictable)
- I got me a used portable face tanning lamp, uses the same flourescent
    lamps as commercial exposure boxes - but much cheaper, some 20 Euros
- I made a simple mounting rack from wood boards for it, facing down
    upon the PCB, so exposure distance is fixed and repeatable
- let it warm up for 1-2 minutes to reach full UV intensity, then
    insert the prepared PCB sandwich below, exposing for 2-3 minutes
    (I once did a test board with several strips, covered for different
    periods of time to find out the required time. Turned out that it's
    hard to over-expose, even doubling the time doesn't really hurt
    if the print is good)
- develop using 1.5 small spoons of NaOH in 1 liter of barely luke
    warm water. Too cold will delay solution of the grainy NaOH, if
    there are still visible grains, they will destroy the exposed resist
    pretty soon. Too warm will make it too agressive to the resist. 
    It appears to be best to let it simply stand for a couple of hours
    to reach room temperature, and at the same time it helps to completely
    resolve all grains.
    As with exposure, it's better to develop a bit longer than it needs
    for a visible image to appear. If the developer isn't made overly
    strong, it won't remove the dark patterns too soon.
- etch in a flat, vertical plexiglass tank, using sodium persulfate.
    Temperature is about 45 degress celsius. There's a perforated
    plastic tube at the bottom, connected to an aquarium pump. The air
    bubbles help the etching process. Earlier models also had a second
    pump for moving the etchant, but it turned out the rubber pump
    membranes got destroyed over time (bad news fo those with fish
    asking for a sodium persulfate bath ;-)
    Takes about 10 minutes to etch.

This gives repeatably good results. 10 mil tracks and even 0.8 mm
pitched TQFPs can be done this way easily. (I tried 0.5 mm as well, but
the printout doesn't look good enough, I didn't even try that on a PCB)

The only thing that caused bad results were pre-coated PCBs that were
too old. After more than 3-4 years in storage, the photo resist gets
lazy, and the results can be disappointing.

However, for real complex boards, I now have them made professionally.
For years I considered that to be too expensive, but with the very last
complex board I made earlier this year (3 ATmega32s and lots of other
stuff), I came to the conclusion that all the additional work after
etching (drilling, oodles of wire bridges, and wire vias for double
sided boards) are taking so much time that the expense for a real board
is justified.
It's still different for e.g. simple panel boards just carrying a couple
of pots, I keep making those at home.

Christian