Homebrew PCBs

I seem to be seeing a trend here, that regardless of the techniques
used, with care one can get to 6/6 rules.  How do we get past that?
It must be possible in theory, as the big fabs do it.

My laser printer does 1.3 mil "lines" (600dpi) but the quality isn't
good enough to do so reliably (fuzzy edges); I need at least 3-4
pixels across to get good definition with TT.

I think the various elements are:

* Getting a print with crisp edges at fine resolutions.

* Getting that print onto the board as etch resist.

* Consistent etching with little side-etch.

* Thinner copper?

Does anyone get 5/5 or better, reliably, with home etch?  If so, how?

Random thoughts:

Would a 1200 dpi laser printer give better edges, or just smoother
curves?  Is there a paper trick to keeping the toner were it's
supposed to be?

I read one home fabber printed their TT at a 45 degree angle to get
more consistent results.

Photo: Inkjets can do 1440 and sometimes 2880 dpi, what limits those
to 6/6?  Would having the UV source further away from the board result
in better edges?  What about the thickness of the photomask?

What about 1/4 oz copper?  I usually use 1/2 oz, and I haven't seen
1/4 oz in my usual places.

Injets: Anyone try writing a custom print driver that forces all the
colors to be used, overlapping, making a nice thick print?  I've tried
doing that with my color laser, but I haven't found the command to
turn off the "smarter than the user" mode.

Do you know what a GERBER machine is? It is used to
make a photo mask. The GERBER is a laser photo
plotter.

I worked on the original at Dickson Eletronics in
Scottsdale in 1969.

Bob Macklin
Seattle, Wa.

--- DJ Delorie <dj@...> wrote:

> 
> I seem to be seeing a trend here, that regardless of
> the techniques
> used, with care one can get to 6/6 rules.  How do we
> get past that?
> It must be possible in theory, as the big fabs do
> it.
> 
> My laser printer does 1.3 mil "lines" (600dpi) but
> the quality isn't
> good enough to do so reliably (fuzzy edges); I need
> at least 3-4
> pixels across to get good definition with TT.
> 
> I think the various elements are:
> 
> * Getting a print with crisp edges at fine
> resolutions.
> 
> * Getting that print onto the board as etch resist.
> 
> * Consistent etching with little side-etch.
> 
> * Thinner copper?
> 
> Does anyone get 5/5 or better, reliably, with home
> etch?  If so, how?
> 
> Random thoughts:
> 
> Would a 1200 dpi laser printer give better edges, or
> just smoother
> curves?  Is there a paper trick to keeping the toner
> were it's
> supposed to be?
> 
> I read one home fabber printed their TT at a 45
> degree angle to get
> more consistent results.
> 
> Photo: Inkjets can do 1440 and sometimes 2880 dpi,
> what limits those
> to 6/6?  Would having the UV source further away
> from the board result
> in better edges?  What about the thickness of the
> photomask?
> 
> What about 1/4 oz copper?  I usually use 1/2 oz, and
> I haven't seen
> 1/4 oz in my usual places.
> 
> Injets: Anyone try writing a custom print driver
> that forces all the
> colors to be used, overlapping, making a nice thick
> print?  I've tried
> doing that with my color laser, but I haven't found
> the command to
> turn off the "smarter than the user" mode.
> 
> 
> Be sure to visit the group home and check for new
> Links, Files, and Photos:
> http://groups.yahoo.com/group/Homebrew_PCBs 
> Yahoo! Groups Links
> 
> 
> mailto:Homebrew_PCBs-fullfeatured@yahoogroups.com
> 
> 
>

Answering from the perspective of "Home Brew PCBs"...

Bob Macklin <rottenrobbie0@...> writes:
> Do you know what a GERBER machine is? It is used to make a photo
> mask.

Yes, I know what a gerber file is, and what a commercial photoplotter
is.  I wrote a gerber viewer for Data General in 1988 (where I
designed PC motherboards), and the current gerber plugin for gEDA's
PCB program.

> The GERBER is a laser photo plotter.

"The gerber" these days is a text file with photoplotter commands,
with syntax conforming with the RS274D or (more recently) RS274X
standards.  It was originally designed for a mechanical photoplotter
with an aperture wheel, not a laser photoplotter.

> I worked on the original at Dickson Eletronics in Scottsdale in
> 1969.

And did you do better than 6/6 rules at home back then?  If so, how?
That's what I'm interested in.  As I said, I know commercial fabs can
do better, but how can we?

The plotter I worked on at Dickson in 1969 did use a
laser in addition to the aperature wheel. The laser
was fixed in position and the beam was positioned by
mirrors on the X/Y drive.

The high resoluion boards being made by commercial
fabs today are still done photographicly. The
photographic process has higher reseolution than can
be generatred by our common printers.

The semiconductor industry uses the same photo
techniques to make masks which are shrunk
photographicly. And the use an etching process similar
to the way PCBs are made but in far higher resolution.

My interest is not in making PCB as good as commercial
houses make. Just good enough for hobby level projects
and possibly some very low volume production. Foe any
real production these processes are just not cost
effective.

Bob Macklin
Seattle, Wa.
--- DJ Delorie <dj@...> wrote:

> 
> Answering from the perspective of "Home Brew
> PCBs"...
> 
> Bob Macklin <rottenrobbie0@...> writes:
> > Do you know what a GERBER machine is? It is used
> to make a photo
> > mask.
> 
> Yes, I know what a gerber file is, and what a
> commercial photoplotter
> is.  I wrote a gerber viewer for Data General in
> 1988 (where I
> designed PC motherboards), and the current gerber
> plugin for gEDA's
> PCB program.
> 
> > The GERBER is a laser photo plotter.
> 
> "The gerber" these days is a text file with
> photoplotter commands,
> with syntax conforming with the RS274D or (more
> recently) RS274X
> standards.  It was originally designed for a
> mechanical photoplotter
> with an aperture wheel, not a laser photoplotter.
> 
> > I worked on the original at Dickson Eletronics in
> Scottsdale in
> > 1969.
> 
> And did you do better than 6/6 rules at home back
> then?  If so, how?
> That's what I'm interested in.  As I said, I know
> commercial fabs can
> do better, but how can we?
> 
> 
> Be sure to visit the group home and check for new
> Links, Files, and Photos:
> http://groups.yahoo.com/group/Homebrew_PCBs 
> Yahoo! Groups Links
> 
> 
> mailto:Homebrew_PCBs-fullfeatured@yahoogroups.com
> 
> 
>

DJ,

My limiting factor on a small board is etching.

I can make a phototool with 4/4 on a 1200dpi
laser that has very little "Fuzz".

The photomask on the copper looks perfect.

Then after etching some traces magicaly
vanish or are so thin they may as well have.

If I am doing a tiny PCB I can put down 20
or so in a panel and I will get a few that
are usable at 4/4


I found I used to get less undercut whith
FeCl3 than I do with CuCl - but not enough
less for me to bother keeping FeCl3 on hand.

One of the posters here advocates "float
etching" with FeCl3 to get better line
def.  I have yet to try it myself though.

1 Oz copper should be OK for 4/4 (I see
comercial shops offering this) but for less
than 4/4 they drop down the copper thickness.

Maybe Adam Seychell will be able to say if
spray etching is the answer to this part of
the problem.



--- In Homebrew_PCBs@yahoogroups.com, DJ Delorie <dj@...> wrote:
>
> 
> I seem to be seeing a trend here, that regardless of the techniques
> used, with care one can get to 6/6 rules.  How do we get past that?
> It must be possible in theory, as the big fabs do it.
> 
> My laser printer does 1.3 mil "lines" (600dpi) but the quality isn't
> good enough to do so reliably (fuzzy edges); I need at least 3-4
> pixels across to get good definition with TT.
> 
> I think the various elements are:
> 
> * Getting a print with crisp edges at fine resolutions.
> 
> * Getting that print onto the board as etch resist.
> 
> * Consistent etching with little side-etch.
> 
> * Thinner copper?
> 
> Does anyone get 5/5 or better, reliably, with home etch?  If so, 
how?
> 
> Random thoughts:
> 
> Would a 1200 dpi laser printer give better edges, or just smoother
> curves?  Is there a paper trick to keeping the toner were it's
> supposed to be?
> 
> I read one home fabber printed their TT at a 45 degree angle to get
> more consistent results.
> 
> Photo: Inkjets can do 1440 and sometimes 2880 dpi, what limits those
> to 6/6?  Would having the UV source further away from the board 
result
> in better edges?  What about the thickness of the photomask?
> 
> What about 1/4 oz copper?  I usually use 1/2 oz, and I haven't seen
> 1/4 oz in my usual places.
> 
> Injets: Anyone try writing a custom print driver that forces all the
> colors to be used, overlapping, making a nice thick print?  I've 
tried

> doing that with my color laser, but I haven't found the command to
> turn off the "smarter than the user" mode.
>

----- Original Message ----- 

From: "DJ Delorie" <dj@...>
To: <Homebrew_PCBs@yahoogroups.com>
Sent: Monday, November 05, 2007 2:34 AM
Subject: [Homebrew_PCBs] Breaking the 6/6 barrier


>
> I seem to be seeing a trend here, that regardless of the techniques
> used, with care one can get to 6/6 rules.  How do we get past that?
> It must be possible in theory, as the big fabs do it.

They must find it quite tricky, though, as my favourite supplier, PCB-Pool, 
doesn't go below 6/6, and some don't even go below 8/8, like Olimex.

They use negative resist, which gives better definition, and thinner copper 
(I think they only offer < 6/6 on multi-layer boards). They also use much 
higher artwork resolution than we can achieve, with vacuum keeping it in 
contact with the resist, and highly collimated UV.

Leon
--
Leon Heller
Amateur radio call-sign G1HSM
Yaesu FT-817ND and FT-857D transceivers
Suzuki SV1000S motorcycle
leon355@...
http://www.geocities.com/leon_heller

On 05/11/2007, Andrew <andrewm1973@...> wrote:

>  1 Oz copper should be OK for 4/4 (I see
>  comercial shops offering this) but for less
>  than 4/4 they drop down the copper thickness.

They etch thin copper, then plate it back up afterwards. That's a
tricky process to do at home, I suspect.

However, by the time you need 4/4 or whatever, surely you're also
calling for tiny, plated-through vias and multi layers? There comes a
point when you just pay someone. I can't see me ever doing finer than
6/6 at home.

Steve

"Steve Wiseman" <sjwiseman@...> writes:
> However, by the time you need 4/4 or whatever, surely you're also
> calling for tiny, plated-through vias and multi layers?

My main reason for pondering this is to get fan-out on BGAs *without*
needing tiny vias.  Not that I'm considering using any at the moment,
but they seem to be getting more popular over time.

But I'm also hoping that I'll find a way to get cleaner results at 8/8
too, as a lot of the parts I *do* use these days are 0.5mm pitch.

--- In Homebrew_PCBs@yahoogroups.com, DJ Delorie <dj@...> wrote:
>
> I seem to be seeing a trend here, that regardless of the techniques
> used, with care one can get to 6/6 rules.  How do we get past that?
> It must be possible in theory, as the big fabs do it.
> 
> My laser printer does 1.3 mil "lines" (600dpi) but the quality isn't
> good enough to do so reliably (fuzzy edges); I need at least 3-4
> pixels across to get good definition with TT.
> 
> I think the various elements are:
> 
> * Getting a print with crisp edges at fine resolutions.

I don't think that's super critical for signal traces--as long as the
trace exists & survives etching without breaking; we don't care if
it's a little wavy.

> * Getting that print onto the board as etch resist.

That's what limits my toner transfer's resolution presently.  Fine
lines don't survive the transfer process intact.

A second, lesser problem is smearing of the toner during
transfer--lines that are too close together will be joined, producing
a short-circuit.

> * Consistent etching with little side-etch.

Floating the board face-down on FeCl works really well -- see post 14965.

> * Thinner copper?

Maybe, but why? (for high currents?)  Could electroless plate it up
afterwards--that's not hard.

> Does anyone get 5/5 or better, reliably, with home etch?  If so, how?

  8/8 has been my limit, with baked Sharpie markups needed to fix gaps
in traces that didn't transfer well.

  I've just done my first experiments with magazine paper at 0.010"
traces and spaces.  I found it transfers and releases *much* better /
more completely than the inkjet paper I had been using.

(Preliminary reason seems to be, in part, the magazine paper is
thinner, allowing better heating of the coppper foil underneath it
when passing through my laminator.  A hot air gun also helps keep
everything at temperature.)

Best,
James Arthur

"dagmargoodboat" <dagmargoodboat@...> writes:
> I don't think that's super critical for signal traces--as long as the
> trace exists & survives etching without breaking; we don't care if
> it's a little wavy.

I was thinking that less waviness in the print gives you more leniency
elsewhere.  At least, my prints don't have a definite "edge", it's
more of a 3 thou wide blur.  So I can't tell in advance where the edge
of the copper will be.

But if the mask is the issue, wouldn't inkjet+photomask be able to be
done at finer resolution?  Or do they still have issues with "fuzzy"
edges?

> > * Thinner copper?
> 
> Maybe, but why? (for high currents?)  Could electroless plate it up
> afterwards--that's not hard.

I'm thinking less side-etch, allowing you more time (relatively) to
over-etch everything else.  What I mean is, "in the time it takes you
to go 150% of the copper etch time, the side etch is still much less
than with thicker copper."

For example, when I etch 5 thou brass, I can't do a 5 mil space
between pads because the sides etch all the way across under the
resist.

>   I've just done my first experiments with magazine paper at 0.010"
> traces and spaces.  I found it transfers and releases *much* better /
> more completely than the inkjet paper I had been using.

I'm using Pulsar's paper, which transfers just fine, but the original
has a "fuzzy" edge, which results in variations in line widths.  The
2550N has a number of software knobs to fiddle with, but I haven't
been able to get any of them to make any noticable changes in the
actual print.

> (Preliminary reason seems to be, in part, the magazine paper is
> thinner, allowing better heating of the coppper foil underneath it
> when passing through my laminator.  A hot air gun also helps keep
> everything at temperature.)

I often pass the board alone through the laminator a few times to warm
it up.  I have some 8 and 16 thou SS on order for some
experimentation, I know I can't get a good transfer with 62 thou
phonelic as it sucks up too much heat from the laminator.

I've never tried magazine paper.  I suppose that's at least free to
try, but the TT paper releases so well...

----- Original Message ----- 

From: "DJ Delorie" <dj@...>
To: <Homebrew_PCBs@yahoogroups.com>
Sent: Tuesday, November 06, 2007 12:16 AM
Subject: Re: [Homebrew_PCBs] Re: Breaking the 6/6 barrier


>
> "dagmargoodboat" <dagmargoodboat@...> writes:
>> I don't think that's super critical for signal traces--as long as the
>> trace exists & survives etching without breaking; we don't care if
>> it's a little wavy.
>
> I was thinking that less waviness in the print gives you more leniency
> elsewhere.  At least, my prints don't have a definite "edge", it's
> more of a 3 thou wide blur.  So I can't tell in advance where the edge
> of the copper will be.
>
> But if the mask is the issue, wouldn't inkjet+photomask be able to be
> done at finer resolution?  Or do they still have issues with "fuzzy"
> edges?

Depends on the printer, of course. My cheap one gives 1200 dpi and the edges 
are fuzzy. One of the expensive printers should manage 5/5, but there are 
all the other problems I mentioned previously.

Leon

--- In Homebrew_PCBs@yahoogroups.com, DJ Delorie <dj@...> wrote:
>
> 
> James Arthur writes:
> > I don't think that's super critical for signal traces--as long as the
> > trace exists & survives etching without breaking; we don't care if
> > it's a little wavy.
> 
> I was thinking that less waviness in the print gives you more leniency
> elsewhere.  At least, my prints don't have a definite "edge", it's
> more of a 3 thou wide blur.  So I can't tell in advance where the edge
> of the copper will be.

We have to distinguish between sharply defined dots in staggered
positions (staggered due to aliasing, for example), from smudged dots.
 Angling the printouts 45 degrees might fill in gaps via improved
interpolation.

My 600dpi laser prints on magazine paper are razor-sharp to the eye,
but close-ups reveal a 1-to-1.5 mil-wide scalloping, sloping,
spattered transition at each edge/border.  (I'm not sure how to
quantify it better.)

There are also a surprising number of defects in the black areas,
something I hadn't noticed.

From the above estimates of the "sharpness" of my prints on paper, I
suppose that factor would limit me to no better than 6/6 with this
paper (giving traces that, with worst-case defects on both sides,
would be 3 mils, and in danger of etching through).

If the toner is smearing on your paper, it's also poosible the paper's
absorbing and spreading the toner a bit.  I've seen papers do that.

If the image is sharp on the paper, it could be smearing during
transfer to the PCB.  That's definitely a danger when using a clothes
iron, and could be a problem with multiple passes through a laminator.
 The artwork is bound to creep a little each time through.

Passing a gridded test pattern through the laminator multiple times,
keeping the same orientation each time, should reveal any creepage.


> But if the mask is the issue, wouldn't inkjet+photomask be able to be
> done at finer resolution?  Or do they still have issues with "fuzzy"
> edges?


Dunno.  I haven't used my inkjet since the first cartridge dried up,
virtually unused.  That was years ago, and I was too irritated to
replace it.

Whatever the inkjets' resolutions, one could always print the artwork
2x, then reduce it photographically or with lenses.

> > > * Thinner copper?
> > 
> > Maybe, but why? (for high currents?)  Could electroless plate it up
> > afterwards--that's not hard.
> 
> I'm thinking less side-etch, allowing you more time (relatively) to
> over-etch everything else.  What I mean is, "in the time it takes you
> to go 150% of the copper etch time, the side etch is still much less
> than with thicker copper."
> 
> For example, when I etch 5 thou brass, I can't do a 5 mil space
> between pads because the sides etch all the way across under the
> resist.


Yes, definitely--thinner copper will reduce undercutting.  I somehow
read that as "thicker" copper instead of "thinner."  Sorry.


> >   I've just done my first experiments with magazine paper at 0.010"
> > traces and spaces.  I found it transfers and releases *much* better /
> > more completely than the inkjet paper I had been using.
> 
> I'm using Pulsar's paper, which transfers just fine, but the original
> has a "fuzzy" edge, which results in variations in line widths.  The
> 2550N has a number of software knobs to fiddle with, but I haven't
> been able to get any of them to make any noticable changes in the
> actual print.
> 
> > (Preliminary reason seems to be, in part, the magazine paper is
> > thinner, allowing better heating of the coppper foil underneath it
> > when passing through my laminator.  A hot air gun also helps keep
> > everything at temperature.)
> 
> I often pass the board alone through the laminator a few times to warm
> it up.  I have some 8 and 16 thou SS on order for some
> experimentation, I know I can't get a good transfer with 62 thou
> phonelic as it sucks up too much heat from the laminator.

I preheat by putting the board part-way in.  There the laminator's
heaters clamp down and heat the board, but the board is short of the
feed rollers.  When the board's good and hot--judged by feel, and the
laminator's "Ready" light staying on--I give the board a push & let it
feed.  I then also pass the board through the laminator a few times,
simultaneously adding extra heat to the copper clad with a small heat gun.

More convenient and possibly better, I think, would be to preheat the
board with a heater of some sort (halogen torchiere lamp bulb?), and /
or to slow the laminator's feedrate.  One pass might then be enough
for a good transfer.

> I've never tried magazine paper.  I suppose that's at least free to
> try, but the TT paper releases so well...

I think the way to achieve finer lines is to go step-by-step in tuning
the process.  For toner-transfer, first get hi-res prints.  Next,
concentrate on hi-res transfers to the copper-clad.  I believe thin
copper and the "float" etch technique I referenced (post #14965) will
suffice for etching to 6/6, maybe better.

James Arthur

Not mentioned so far in this round, is the technique of allowing for 
the side etching of the trace in your pattern design.

Depending on the copper thickness and your etchant chemistry you can 
compensate for the undercutting of the etch resist.

So to achieve 5/5 reliably, depending on the under cutting at your 
place, you may have to set your design rules to 7/3 (1mil per edge 
undercutting)or 6/4 (0.5mil per edge undercutting)

Regards

"warrenbrayshaw" <warrenbrayshaw@...> writes:
> Depending on the copper thickness and your etchant chemistry you can
> compensate for the undercutting of the etch resist.

Actually, I already thought of this.  When I wrote the postscript
output plugin for PCB, I added a "bloat" option that lets you expand
all the toner by some amount (units of 1/100 thou, negative allowed)
to compensate for variations in the process, both printer and etch.

I put a test board here: http://www.delorie.com/pcb/ - scroll down to
"Goodies".  It has various line/space rules from 1/1 to 12/12, with
bloats in 0.25 thou increments from -1.5 thou to +1.5 thou.

Both gerber and pdf are available for download.