Homebrew PCBs

Hello,

I am making a board using the toner transfer method, using the Staples 
471861 glossy paper.  I am laying this out with the EagleCAD 
application. 

What I am wondering is what kind of trace width I can reasonably 
expect to be able to render with acceptable results.  I'm probably 
looking at about 25 or 50 mils right now, and would appreciate any 
input into what widths you are able to go down to using the 
paper/toner transfer method.

Thanks,

Mike

On Thu, 30 Nov 2006 05:39:47 +0100, Eugene Michael <eugenemic@...>  
wrote:

> Hello,
> I am making a board using the toner transfer method, using the Staples
> 471861 glossy paper.  I am laying this out with the EagleCAD
> application.
> What I am wondering is what kind of trace width I can reasonably
> expect to be able to render with acceptable results.  I'm probably
> looking at about 25 or 50 mils right now, and would appreciate any
> input into what widths you are able to go down to using the
> paper/toner transfer method.
> Thanks,
> Mike


The limit is about 6.66mil (odd number comes from DPI), more becasue of  
undercutting than the TT itself.

<http://trethan.at.tf/pub/img1/PCB2.JPG>

of course it also depends on printer resolution and probably about a  
zillion other factors, but that's what i use.


ST

"Stefan Trethan" <stefan_trethan@...> writes:
> The limit is about 6.66mil (odd number comes from DPI), more becasue of  
> undercutting than the TT itself.

I see that too.  I can sometimes do thinner traces if there's more
space between them, or smaller spaces if there's more copper around
them, but 7/7 is about the limit for me.

On Thu, 30 Nov 2006 17:01:47 +0100, DJ Delorie <dj@...> wrote:

>
> I see that too.  I can sometimes do thinner traces if there's more
> space between them, or smaller spaces if there's more copper around
> them, but 7/7 is about the limit for me.


I think spray etching could improve performance, but 7mil is good enough  
for me, usually i don't use anything under 10mil.

ST

"Stefan Trethan" wrote:
> 
> The limit is about 6.66mil (odd number comes from DPI), more becasue 
of  
> undercutting than the TT itself.
> 
> <http://trethan.at.tf/pub/img1/PCB2.JPG>
> 
> of course it also depends on printer resolution and probably about 
a  
> zillion other factors, but that's what i use.

Thanks, Stefan, that's great to have the picture to refer to for this; 
looks like I'm within the ballpark.  

BTW, is 300 DPI a good enough resolution to use?

Thanks,

Mike

"Eugene Michael" <eugenemic@...> writes:
> Thanks, Stefan, that's great to have the picture to refer to for this; 
> looks like I'm within the ballpark.  

For an additional picture, see http://www.delorie.com/pcb/first.html

The IC on the right is 7.87mil traces on 7.87mil spaces (0.4mm pitch)
and the 01005 cap is over a 6 mil gap.  So, I'm getting about 2 mil of
variability along the edges.

> BTW, is 300 DPI a good enough resolution to use?

Keep in mind that your "pixels" are 3.33 mil across, so lines below
about 7 mil risk being rendered with only one pixel.  You'll have to
try and see, but I'd think 10mil would be a reasonable limit for
300dpi printers (~3 pixels across).

Note that PCB, the software I use, has a "bloat" setting that can
systematically shrink or expand the copper to try to compensate for
the way your printing/etching system works.  If you're going to push
the limits, you might want to make sure your eda software can make
such adjustments.

I have a test board I use to "calibrate" my etching, which tries
different pitch/bloat settings: http://www.delorie.com/pcb/bloat.png
There's a bloat.gbr and bloat.pdf too, if you want to try it.  The
"bloat" setting on that board is 0.01 mil units.

On Fri, 01 Dec 2006 03:16:38 +0100, Eugene Michael <eugenemic@...>  
wrote:

>
> Thanks, Stefan, that's great to have the picture to refer to for this;
> looks like I'm within the ballpark.
> BTW, is 300 DPI a good enough resolution to use?
> Thanks,
> Mike


I think 10 mil is about the limit for 300DPI, i could not do 0.5mm spaced  
qfn with my old 300DPI HP IIID because the lines were all squiggly.

ST

is this THE dj delorie?  Holy crap, i was an avid djgpp user back in
highschool!


--- In Homebrew_PCBs@yahoogroups.com, DJ Delorie <dj@...> wrote:
>
> 
> "Eugene Michael" <eugenemic@...> writes:
> > Thanks, Stefan, that's great to have the picture to refer to for
this; 

> > looks like I'm within the ballpark.  
> 
> For an additional picture, see http://www.delorie.com/pcb/first.html
> 
> The IC on the right is 7.87mil traces on 7.87mil spaces (0.4mm pitch)
> and the 01005 cap is over a 6 mil gap.  So, I'm getting about 2 mil of
> variability along the edges.
> 
> > BTW, is 300 DPI a good enough resolution to use?
> 
> Keep in mind that your "pixels" are 3.33 mil across, so lines below
> about 7 mil risk being rendered with only one pixel.  You'll have to
> try and see, but I'd think 10mil would be a reasonable limit for
> 300dpi printers (~3 pixels across).
> 
> Note that PCB, the software I use, has a "bloat" setting that can
> systematically shrink or expand the copper to try to compensate for
> the way your printing/etching system works.  If you're going to push
> the limits, you might want to make sure your eda software can make
> such adjustments.
> 
> I have a test board I use to "calibrate" my etching, which tries
> different pitch/bloat settings: http://www.delorie.com/pcb/bloat.png
> There's a bloat.gbr and bloat.pdf too, if you want to try it.  The
> "bloat" setting on that board is 0.01 mil units.
>

"joedaddyshagg" <joedaddyshagg@...> writes:
> is this THE dj delorie?  Holy crap, i was an avid djgpp user back in
> highschool!

Yup.  I wrote DJGPP when I worked at Data General, back in the late
80's.  At the time, my job was designing PC motherboards (DG tools on
Eclipses, then Cadstar on PCs).  I designed a 12MHz 80286 one, and a
16MHz 80386 one (the new 80386 was *why* I created DJGPP).  The '386
one got fabbed and I manually built the first board myself :-) (yes,
it worked 100%)

Ah, those were the days.

I can reproducably make 10 mil. Never tried thinner, but sometimes i 
find that the Board outline from eagle also beeps the circuit tester 
after etching. Must be 3 mil or such....
using:
HP2100, original toner, modified laminator, magazine papers