Homebrew PCBs

Hi guys,

From this ground I have learned about toner transfer and PCB making 
and have been doing it for a couple of years now.  Recently, I have 
been starting to work more and more with surface mount components.  
As they are smaller and you can fit more stuff in the same space, I 
am finding that I want to use traces that are more narrow than I have 
been doing.

I am about 85~90% reliable with my current process doing traces that 
are .015" (380 microns).  I would like to get down to .008"~.010" 
(200 to 254 microns). 

Most of my current failures with narrow traces are in getting the 
toner to stick and I think I can improve that part of the process (I 
just started using a laminator instead of a clothes iron).  

I etch with Ferric Chloride and bubbles.  I am wondering wherein the 
limit is there with regard to narrow traces.  I recall a long while 
back that there seemed to be a certain barrier where the trace width 
was so narrow, "under etching" would cause breaks in the traces. 

Anyone one care to share some experiences with narrow traces and DIY 
processes?

Also, I am not opposed to using PCBs pre-coated with etch resist if 
that would improve results.

Thanks

Chris

On Wed, 1 Oct 2008, lcdpublishing wrote:

> I am about 85~90% reliable with my current process doing traces that
> are .015" (380 microns).  I would like to get down to .008"~.010"
> (200 to 254 microns).

With toner transfer, 10/10 is quite easy. I've made my own board for a
0.4mm pitch LQFP-80 package, which will be something like 8/8. The biggest
problem I found was not broken traces, but shorts due to paper fragments
getting stuck between the fine gaps between traces and preventing full
etching. The trouble is you can't scrub the paper remnants off too harshly
because the thin etch resist with 0.2mm traces and 0.2mm spacing is very
easy to break (and it means it takes quite a long time to prepare the
board for etching). I use a clothes iron for the transfer process. 10/10
seems about an order of magnitude easier to do than 8/8.

So I've decided for when I'm doing stuff with a 0.4mm pitch components,
I'll send it off to a board house :-) (Which offers some other advantages
such as smaller vias and gives me more flexibility about where I put vias,
for the tradeoff of extra cost and time waiting for the boards to come
back).

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

From: "lcdpublishing" <lcdpublishing@...>
To: <Homebrew_PCBs@yahoogroups.com>
Sent: Wednesday, October 01, 2008 10:54 AM
Subject: [Homebrew_PCBs] How fine can you etch with toner transfer?



> Anyone one care to share some experiences with narrow traces and DIY
> processes?
>
> Also, I am not opposed to using PCBs pre-coated with etch resist if
> that would improve results.

I get 8 mil without any problems with photo-etch, although I normall use 10 
mil. I have got 5 mil on test boards but haven't actually tried it on a real 
design.

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

I have found that a sponge, or piece of foam rubber, will take those off
easily.
The foam rubber is soft enough not to damage the traces, and you can rub
quite firmly with it.

ST

On Wed, Oct 1, 2008 at 12:01 PM, Dylan Smith <dyls@...> wrote:

> The biggest
> problem I found was not broken traces, but shorts due to paper fragments
> getting stuck between the fine gaps between traces and preventing full
> etching.


[Non-text portions of this message have been removed]

I've gotten down to 8.33 mil with TT with a fair degree of
reliability.  6.66 I can do if I'm careful and put extra copies of the
board on the print - there's usually one short or open somewhere,
often I can fix it with a knife or solder.  Note the off-mil sizes; my
laser printer is 600dpi or 1.33 mil per pixel, so I use multiples of
that to get consistent results.  The limitation with my printer is its
ability to place the toner particles precisely.

With photomask I've been able to do 6 mil with high reliability, and 5
if I'm careful.  In theory I can do 4, but I haven't needed to try to
get that to work since I don't use parts that small.  The limitation
here, with my printer, seems to be its ability to feed the paper
accurately.  Quarter sheets are too small for a good grip, half sheets
work much better.

Thanks guys,  that gives me a bit of confidence to give it a go with 
toner transfer.

I also checked into various forms of photoresist and it turns out that 
it isn't all that expensive to go that way.  Using dry film resists and 
the developer isn't as expensive as I thought.  But still more 
expensive than toner transfer.

chris

--- In Homebrew_PCBs@yahoogroups.com, DJ Delorie <dj@...> wrote:
 
> I've gotten down to 8.33 mil with TT with a fair degree of
> reliability.  6.66 I can do if I'm careful and put extra copies of 
the
> board on the print - there's usually one short or open somewhere,
> often I can fix it with a knife or solder.  Note the off-mil sizes; 
my
> laser printer is 600dpi or 1.33 mil per pixel, so I use multiples of
> that to get consistent results.  The limitation with my printer is 
its
> ability to place the toner particles precisely.

8.33mil? is that a multiple of 1.33? ;-)

--Calvin

"calvingrier" <cgrier@...> writes:
> > laser printer is 600dpi or 1.33 mil per pixel, so I use multiples
> > of that to get consistent results.  The limitation with my printer
> > is its ability to place the toner particles precisely.
> 
> 8.33mil? is that a multiple of 1.33? ;-)

Sigh, wrong math, right results.  1/600 is 1.666 mil, not 1.333 mil,
resulting in possible trace widths of:

Pixels   Width (mils)
------   -----
0        0.000
1        1.666
2        3.333
3        5.000
4        6.667
5        8.333
6       10.000
7       11.667
8       13.333

Traces of other widths will be printed as either the next smaller or
next larger size, depending on your rasterizer.  Same with the space
between traces.  So you could be off by as much as 0.833 mil (half a
pixel).

Switching to a 2880 DPI inkjet lets me choose any size and get within
0.2 mil of the right size.

> Sigh, wrong math, right results.  1/600 is 1.666 mil, not 1.333 mil,
...
> Switching to a 2880 DPI inkjet lets me choose any size and get 
within
> 0.2 mil of the right size.


No this only means, hat you can adress this resolution. May be the 
stepsize comes in this region, but there is nothing said about the 
size of one pixel. Especialy inkjets are designed to produce 
overlapping pixels.

Martin

"martin_schoenegg" <Martin.Schoenegg@...> writes:
> No this only means, hat you can adress this resolution. May be the 
> stepsize comes in this region, but there is nothing said about the 
> size of one pixel. Especialy inkjets are designed to produce 

Well, yeah.  Relatively speaking though, I can compare 0.833 mil to
0.174 mil and say "it's much more accurate".  What I really mean is
that the line sizes - as drawn - will be much more consistent, because
a variation of one pixel is a much smaller percentage of the line
size.

This means that, if I know my process can handle 6 mil lines, I won't
have to worry about the 5 mil lines, or the spaces between 6.666 mil
lines - because the films will more accurately reflect my desired
sizes, compared to a laser printer.

My inkjet is good to about 0.5 mil as far as drop placement accuracy
goes.  However, I know that "fuzzy edge" will be in the same spot
every time.  So, if I know I can do 6/6 rules with a 0.5mil fuzzy
edge, at least the results will be consistent - a one pixel difference
is negligible.  One pixel on a laser printer is BIG compared to what
I'm trying to achieve.

On a laser printer, for 6/6 rules (er, 6.66 rules), off by one pixel
means the line is 25% smaller.  On an inkjet, off by one pixel would
only be 5% smaller.

Hello,
I think that this URL might be useful for those who want understand the chemistry behind an etching process.
Bye
 
http://members.optusnet.com.au/~eseychell/PCB/etching_CuCl/index.html
 


      

[Non-text portions of this message have been removed]