Homebrew PCBs

I'm wondering what other people are doing out there for PCB that use 
SMD. I finding over the years my PCBs need smaller and smaller tracks to 
suit the ever shrinking SMD packages. Well, we know BGA is out, because 
that needs a multilayer board. Luckily, only the really big chips are 
BGA. These days 0603 chip resistors/caps are the norm. Often I need to 
run a trace under a 0603, or my latest project has two 0.5mm pitch TQFP 
(leaded) and QFN (leadless) packages. I'm wondering how other people 
deal with PCBs for SMD.

I use photomasks from a inkjet printer. Recently I bought a Canon MP810, 
and noticed the resolution and line edge sharpness was a noticeable 
improvement over my old printers (Epson Stylus 660, Canon i550). Its 
definitely true 1200x1200 dpi resolution. Using "3M Universal 
Transparency" film, and printing with the driver set to "Photo Paper 
Pro", "Quality 1" and Cannon OEM inks, I can just get away with 
0.008"(0.20mm) lines/spacings. 18um foil instead 36um makes etching fine 
traces a lot better.

Adam

On Tue, 10 Apr 2007 00:34:34 +0200, Adam Seychell  
<a_seychell@...> wrote:

> I'm wondering what other people are doing out there for PCB that use
> SMD. I finding over the years my PCBs need smaller and smaller tracks to
> suit the ever shrinking SMD packages. Well, we know BGA is out, because
> that needs a multilayer board. Luckily, only the really big chips are
> BGA. These days 0603 chip resistors/caps are the norm. Often I need to
> run a trace under a 0603, or my latest project has two 0.5mm pitch TQFP
> (leaded) and QFN (leadless) packages. I'm wondering how other people
> deal with PCBs for SMD.
> I use photomasks from a inkjet printer. Recently I bought a Canon MP810,
> and noticed the resolution and line edge sharpness was a noticeable
> improvement over my old printers (Epson Stylus 660, Canon i550). Its
> definitely true 1200x1200 dpi resolution. Using "3M Universal
> Transparency" film, and printing with the driver set to "Photo Paper
> Pro", "Quality 1" and Cannon OEM inks, I can just get away with
> 0.008"(0.20mm) lines/spacings. 18um foil instead 36um makes etching fine
> traces a lot better.
> Adam


8 mil is no problem with toner transfer, but my printer won't go much  
below that (6 mil maybe). I usually use 10mil smallest size unless i  
really need less.
0.5mm qfn and stuff works just fine, only trouble are thermal pads (lack  
of VIAs).
I avoid using passives smaller than 0805 (why should i?).

I don't think pitch will go much lower, but we will see more and more  
trouble with thermal impossibilities. A case the size of a pinhead can not  
cool the same as a to220 case, no matter what the manufacturers want to  
make us believe.....

ST

Adam Seychell <a_seychell@...> writes:
> I'm wondering what other people are doing out there for PCB that use 
> SMD.

I normally use 0603 and SSOP/TSSOP sizes, with either SOT-23 or
SOT-323 depending on power need.

I've done down to 01005 (8x16 mil) and VSSOP (0.4mm pitch) with toner
transfer.

http://www.delorie.com/pcb/first.html
http://www.delorie.com/pcb/smd-challenge/
http://www.delorie.com/pcb/smd-challenge/old/proto-boards.html

> Luckily, only the really big chips are BGA.

The really small chips are CSP, which have similar type of issues as
BGA.

> I use photomasks from a inkjet printer.

I can do down to 6.7 mil (four dots wide) reliably with toner
transfer.

>I can just get away with 0.008"(0.20mm) lines/spacings.

That should be walk in the park. At least with a $100 Epson and Epson's 
transparencies it is. Here's 6/6 rules in a worst case scenario;

http://www.pbase.com/eldata/6_6_rules

Spacing irregularities are due to the printer and not the process.

--- In Homebrew_PCBs@yahoogroups.com, Adam Seychell <a_seychell@...> 
wrote:
>

>8 mil is no problem with toner transfer

It shouldn't be with photo either. And best case photo will always look 
better than best case TT.

--- In Homebrew_PCBs@yahoogroups.com, "Stefan Trethan" 
<stefan_trethan@...> wrote:
>

> Adam Seychell wrote:
>
> I'm wondering what other people are doing
> out there for PCB that use SMD.
> <SNIP>
> I use photomasks from a inkjet printer.
> <SNIP>
> I can just get away with 0.008"(0.20mm)
> lines/spacings. 18um foil instead 36um
> makes etching fine traces a lot better.

I do pretty much the same as you do
except I use a large colour laser printer
instead of the inkjet.

Routinely get 8/8 thou across a large
board (150x300mm).  Will do 6/6 but will
try limit PCB size to 50x75mm so I can
fit 4 boards on small blank to make sure
I get one of the 4 that does not look too
thin.  That is mainly due to uneven etch
I think.

We don't all have spray etchers set up
in the back yard yet :D

I do find the cupperic chloride is not as
good for fine resolution as the old
ferric chloride used to be.  That might
be my haphazzard monitoring of the acid/
oxygen/density though.

>That is mainly due to uneven etch I think.

Still guessing after all these years? (Apologies to Paul Simon).

--- In Homebrew_PCBs@yahoogroups.com, "Andrew" <andrewm1973@...> wrote:
>

"derekhawkins" <eldata@...> writes:
> Still guessing after all these years? (Apologies to Paul Simon).

In my case, variations in etch are due to two primary factors:

1. Mapping the pattern onto discrete toner dots sometimes results in
   slightly wider traces, and sometimes slightly narrower, depending
   on how it falls on dot boundaries.  A 6 mil trace is sometimes 5
   mil (three dots) and sometimes 6.67 mil (four dots).

2. My toner and drum are old, so the edges of the toner aren't crisp.
   I'm getting new ones tomorrow, maybe I'll do some test prints.

I suspect that variations in the iron transfer may cause
discrepancies, but I use a laminator because of that.  Also, the etch
itself causes differences.  I use the sponge wipe method, so I can
manually control how fast each part of the board etches.  However, the
sideways motion causes more etch creep under the toner.

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

From: "Adam Seychell" <a_seychell@...>
To: <Homebrew_PCBs@yahoogroups.com>
Sent: Monday, April 09, 2007 11:34 PM
Subject: [Homebrew_PCBs] who does SMD ?


> I'm wondering what other people are doing out there for PCB that use
> SMD. I finding over the years my PCBs need smaller and smaller tracks to
> suit the ever shrinking SMD packages. Well, we know BGA is out, because
> that needs a multilayer board. Luckily, only the really big chips are
> BGA. These days 0603 chip resistors/caps are the norm. Often I need to
> run a trace under a 0603, or my latest project has two 0.5mm pitch TQFP
> (leaded) and QFN (leadless) packages. I'm wondering how other people
> deal with PCBs for SMD.
>
> I use photomasks from a inkjet printer. Recently I bought a Canon MP810,
> and noticed the resolution and line edge sharpness was a noticeable
> improvement over my old printers (Epson Stylus 660, Canon i550). Its
> definitely true 1200x1200 dpi resolution. Using "3M Universal
> Transparency" film, and printing with the driver set to "Photo Paper
> Pro", "Quality 1" and Cannon OEM inks, I can just get away with
> 0.008"(0.20mm) lines/spacings. 18um foil instead 36um makes etching fine
> traces a lot better.

I get similar results, using an HP 5940 printer and Premium JetStar film for 
the transparencies.

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

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

From: "Andrew" <andrewm1973@...>
To: <Homebrew_PCBs@yahoogroups.com>
Sent: Tuesday, April 10, 2007 2:27 AM
Subject: [Homebrew_PCBs] Re: who does SMD ?


>> Adam Seychell wrote:
>>
>> I'm wondering what other people are doing
>> out there for PCB that use SMD.
>> <SNIP>
>> I use photomasks from a inkjet printer.
>> <SNIP>
>> I can just get away with 0.008"(0.20mm)
>> lines/spacings. 18um foil instead 36um
>> makes etching fine traces a lot better.
>
> I do pretty much the same as you do
> except I use a large colour laser printer
> instead of the inkjet.
>
> Routinely get 8/8 thou across a large
> board (150x300mm).  Will do 6/6 but will
> try limit PCB size to 50x75mm so I can
> fit 4 boards on small blank to make sure
> I get one of the 4 that does not look too
> thin.  That is mainly due to uneven etch
> I think.

It might be due to the board being bowed, so that the transparency isn't in 
contact with the resist over the whole board. I had that problem recently, 
with 12 mil tracks and a largish board - some of the tracks disappeared. I 
thought it was uneven etching and made three boards before I realised the 
cause.

Leon

Leon wrote:
> 
> 
> 
>  > I use photomasks from a inkjet printer. Recently I bought a Canon MP810,
>  > and noticed the resolution and line edge sharpness was a noticeable
>  > improvement over my old printers (Epson Stylus 660, Canon i550). Its
>  > definitely true 1200x1200 dpi resolution. Using "3M Universal
>  > Transparency" film, and printing with the driver set to "Photo Paper
>  > Pro", "Quality 1" and Cannon OEM inks, I can just get away with
>  > 0.008"(0.20mm) lines/spacings. 18um foil instead 36um makes etching fine
>  > traces a lot better.
> 
> I get similar results, using an HP 5940 printer and Premium JetStar film 
> for
> the transparencies.
> 
> Leon

Its good to know people are not finding it a problem making PCBs for SMD.
When I said 8mils, I meant to say thats size while still being very 
reliable. I should make up a test panel one day to see how fine I can 
go. I find the real limitation is etching. You need better etch 
uniformity the finer the lines you go. I'm currently building my spray 
etcher, which should solve the uniformity problem.

Adam

>I find the real limitation is etching. You need better etch
>uniformity the finer the lines you go. 

8/8 rules on a reliable basis should be easy with the crudest of 
etching setups provided you can see the process and not just depend on 
a timer. Etching has always been the no-brainer step for me.

It's the printer and/or transparent media that's the limitation at 
finer rules (below 6/6) IMO. A combination of irregular spacing and 
rough edges in the case of inkjet or toner fuzz in the case of laser 
seems to be the show stopper. Magnification of the artwork is all 
that's required to confirm this.

--- In Homebrew_PCBs@yahoogroups.com, Adam Seychell <a_seychell@...> 
wrote:
>

On Tue, 10 Apr 2007 16:07:01 +0200, derekhawkins <eldata@...> wrote:

> 8/8 rules on a reliable basis should be easy with the crudest of
> etching setups provided you can see the process and not just depend on
> a timer. Etching has always been the no-brainer step for me.
> It's the printer and/or transparent media that's the limitation at
> finer rules (below 6/6) IMO. A combination of irregular spacing and
> rough edges in the case of inkjet or toner fuzz in the case of laser
> seems to be the show stopper. Magnification of the artwork is all
> that's required to confirm this.


I'm not actually sure about that. I've transferred finer lines with TT  
that looked ok, but they disappeared completely during etching.
I do not know what the under-cutting rate is, but let's keep in mind 35u  
is 1.4mil, and it looks always like areas close to resist etch faster than  
large copper areas. So i would not be surprised if there is quite some  
undercut.


Anyway, even the pro board houses have 5/5 as their "standard" product.  
Some even 8/8...


ST

DJ Delorie <dj@...> writes:
> 2. My toner and drum are old, so the edges of the toner aren't crisp.
>    I'm getting new ones tomorrow, maybe I'll do some test prints.

Ok, I did a test print and compared it to the same print a few days
ago.  Frankly, I can't tell the difference, even under a 200x scope.

http://www.delorie.com/pcb/toner.html

>I've transferred finer lines with TT that looked ok, but they 
>disappeared completely during etching.

No surprises there, transferred toner is actually a poor resist in 
comparison to resist that was developed properly. And when you 
say "looked ok", what magnification are you referring to? 

I did that 6/6 rules torture test to debunk a collimation myth. The 
only reason why it wasn't 5/5 was because the artwork didn't make the 
grade. I can do less tortuous 5/5 or 4/4 rules (to the limit of the 
printer) to debunk the etching myth as well. How about .003" or .002" 
lines with .005" or .004" spacing? What will it take to convince you 
that etching should be the least of your worries provided you know what 
you're doing? Do I have to walk on water?

--- In Homebrew_PCBs@yahoogroups.com, "Stefan Trethan" 
<stefan_trethan@...> wrote:
>

> Adam Seychell wrote:
> <SNIP>
> I'm currently building my spray etcher,
> which should solve the uniformity
> problem.

Adam,

I thought you already had 2 versions of
it up and runnig.  Are you doing a 3rd
itteration ?

> derekhawkins wrote:
>
> I can do less tortuous 5/5 or 4/4 rules
> (to the limit of the printer) to debunk
> the etching myth as well. How about .003"
> or .002" lines with .005" or .004"
> spacing? What will it take to convince
> you that etching should be the least of
> your worries provided you know what 
> you're doing? Do I have to walk on water?

Derek,

I can print and etch much smaller than 8/8
in a small area of PCB but uneven etching
kills me.  The laser print-out and the
resist on the copper board are all perfect
before the etch.

I have a Kinsten bubble etcher similar to
the ET-10 bubble etcher mentioned recently.

When it is running it creates a swirl in
the tank.  This swirl makes about 40% of
the left side and 20% of the right side of
the tank etch faster than the center.

If I etch till the center area is good the
sides are over etched.  The tracks are
generally still intact but I don't call it
a success becuase there is no uniformity
and some tracks are getting very thin.

I won't say "I can do 4/4 no problems"
unless I can get 4/4 tracks on any part of
a 300mm x 150mm board looking good while
still having differing levels of copper
density across the board.

I am interested in Adams spray etcher for
this reason.  I did build myself a crude
rotary spray etcher but was too scared to
use it because of the amount of aerosol
it created during the water tests.

> derek wrote:
> <SNIP>
> It's the printer and/or transparent media
> that's the limitation at finer rules
> (below 6/6) IMO. A combination of irregular
> spacing and rough edges in the case of
> inkjet or toner fuzz in the case of laser 
> seems to be the show stopper. Magnification
> of the artwork is all that's required to
> confirm this.

These are not up to Dereks photographic
standards but here is a close up of a board
recently made.

http://www.thehacktory.com/88Wide.jpg
http://www.thehacktory.com/88Wide.jpg

The traces there are 8 thou.

Traces are going underneath 0402 components.

The photo was taken for another purpose than
showing of track widths so it is not of
the best part of the board for that.

When I get home tonight I will take photos
of the laser print-out on the transperancy
to show the edges of it.  Will see if the
printer I use has the same fuzzy edge
problems as DJDelorie.

derekhawkins wrote:
> 
> 
>  >I've transferred finer lines with TT that looked ok, but they
>  >disappeared completely during etching.
> 
> No surprises there, transferred toner is actually a poor resist in
> comparison to resist that was developed properly. And when you
> say "looked ok", what magnification are you referring to?
> 
> I did that 6/6 rules torture test to debunk a collimation myth. The
> only reason why it wasn't 5/5 was because the artwork didn't make the
> grade. I can do less tortuous 5/5 or 4/4 rules (to the limit of the
> printer) to debunk the etching myth as well. How about .003" or .002"
> lines with .005" or .004" spacing? What will it take to convince you
> that etching should be the least of your worries provided you know what
> you're doing? Do I have to walk on water?

What is your etching technique ?
What is your foil thickness ?

Ok, you got me curious. Now you've made me waste the past 3 hours of my 
time  determining if etching undercut does indeed limiting line widths.
My test showed for 18um foil  PCB the limited factor is the printer, and 
not the etch process. I have compile a quick web page showing the 
results of this mornings experiment, enjoy.

http://members.optusnet.com.au/~eseychell/pcbtest1/line_width_tests.html

Even thought my results demonstrated a printer limitation, I still think 
etch undercut can be an issue, especially for bubble etchers or 70um 
("2oz") foils.

Adam

> Adam wrote:
>
> <SNIP>
>
> My test showed for 18um foil  PCB the
> limited factor is the printer, and 
> not the etch process. I have compile a
> quick web page showing the results of
> this mornings experiment, enjoy.

WOW - the dot scattering in the
perpendicular direction is dramatic.

Are there options in the printer driver
to slow down the speed the head traverses
the paper or maybe do 12 passes (1 pass
for each of the 12 multiplex rows) to
limit that?  Probably unlikely as
marketing types only care that it is 2400
dpi for printing photos and accurate dot
placement is not needed.


> 
> <SNIP>
>
> Even thought my results demonstrated a
> printer limitation, I still think 
> etch undercut can be an issue,
> especially for bubble etchers or 70um 
> ("2oz") foils.


I am going to print out your test circles
on my laser here and take some photos of
it when I get home.

I dont think I will bother exposing or
etchign it as I am pretty certain no
matter how good a resist mask I had my
bubble etcher would limit me to 8/8
across a large board.

>I still think etch undercut can be an issue

No more so than over-exposure or over-development IMO. I actually use 
undercut to get rid of the inkjet jaggies;

Artwork with jaggies;

http://www.pbase.com/eldata/image/45947532

Etched board using above (warning: very large image);

http://www.pbase.com/eldata/image/46673206/original

The etched board looks like it was from laser artwork. The undercut 
is no greater than the width of the jaggies. Why would the undercut 
be any greater with a 5 mil track?

6/6 rules artwork snippet;

http://www.pbase.com/eldata/image/62430331/original

Etched 6/6 snippet from the above;

http://www.pbase.com/eldata/image/62430370/original

All I'm losing to undercut is the jaggies. Why on earth should I get 
rid of this bubble etcher in favor of a spray etcher?

Was this etched or milled (warning: large image);

http://www.pbase.com/eldata/image/57044297/original


Bottom line, you shouldn't have to try 6/6 or 5/5 rules in order to 
determine whether it can be done. You should be able to determine 
that by inspecting your 10/10 rules results. If the printer is the 
limitation then the artwork is going to tell you that before having 
to go any further.

--- In Homebrew_PCBs@yahoogroups.com, Adam Seychell <a_seychell@...> 
wrote:
>

> derek wrote:
>
> <SNIP>
>
> The etched board looks like it was from
> laser artwork. The undercut is no greater
> than the width of the jaggies. Why would
> the undercut be any greater with a 5 mil
> track?

I don't think yor undercut has to be any
wider than Adams photo shows his "jaggies"
for it to be marginal at 6 thou does it ?

Those jaggies in the perpendicular photos
look like they are 25% of the 5 thou track
width.

I would consider a track that came out 25%
undersized compared to another track on
the same board was a failure.

Does the Epson peizo print head have less
scatter of dots than the Canon Adam uses ?
Your photos of the 6/6 artwork certainly
look less jaggy than Adams.

> <SNIP>
>
> Why on earth should I get rid of this
> bubble etcher in favor of a spray etcher?

Not saying you should.  I am saying my
bubble etcher is bad and I need to get a
new etcher.  You obviously have a better
bubble etcher than I do.

In my case the bubble etcher will fully
etch a stripe down one side of a board in
1/2 the time it takes the middle of the
board to etch.

derekhawkins wrote:

> The etched board looks like it was from laser artwork. The undercut
> is no greater than the width of the jaggies. Why would the undercut
> be any greater with a 5 mil track?

Because from past experience with bubble etching there is anywhere up to 
a 2:1 ratio between fastest and slowest etch rates across the board. 
This _does_ effect reliability of fine traces. A 6 mil line is 150um. If 
you etch 35um down, and 35um laterally, then you already reduced your 
line to 50% of its original width (150um - 35*2 = 80 um). If your bubble 
etcher has uniformity problems then thats even more undercut. Even if 
lateral etching is slower than downward, its still is a problem and its 
becomes critical you don't over etch, period.

> 6/6 rules artwork snippet;
> 
  > Etched 6/6 snippet from the above;
> 

If thats 6/6 mils, then those laser print lines edges look a lot cleaner 
than what I've ever seen from any inkjet at its perpendicular edge. I've 
looked in detail at many inkjet printer for photomasks, and all of them 
suffer this same problem. There is no way of slowing down the print head 
necessary for elimination of perpendicular jaggedness. Its also 
impossible to see any of this this from the naked eye.

The fuzziness of the edges are not as bad as the seem. The photoresist 
has ability to average out the edges because its polymerization reaction 
is quite abrupt. When exposed to sufficient energy it suddenly 
polymerizes, meaning the tapered light intensity due to the dispersed 
inkjet droplets at the vicinity of line edges will in effect produce a 
straighter polymerized resist pattern.


> 
> Bottom line, you shouldn't have to try 6/6 or 5/5 rules in order to
> determine whether it can be done. You should be able to determine
> that by inspecting your 10/10 rules results. If the printer is the
> limitation then the artwork is going to tell you that before having
> to go any further.

True. But experimental 5/5 lines are good for testing reliability of 
10/10 lines. I've never needed to go below 8/8 lines to date.

Adam

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

From: "Adam Seychell" <a_seychell@...>
To: <Homebrew_PCBs@yahoogroups.com>
Sent: Wednesday, April 11, 2007 2:02 AM
Subject: [Homebrew_PCBs] printer limitations or etch undercut


>
> derekhawkins wrote:
>>
>>
>>  >I've transferred finer lines with TT that looked ok, but they
>>  >disappeared completely during etching.
>>
>> No surprises there, transferred toner is actually a poor resist in
>> comparison to resist that was developed properly. And when you
>> say "looked ok", what magnification are you referring to?
>>
>> I did that 6/6 rules torture test to debunk a collimation myth. The
>> only reason why it wasn't 5/5 was because the artwork didn't make the
>> grade. I can do less tortuous 5/5 or 4/4 rules (to the limit of the
>> printer) to debunk the etching myth as well. How about .003" or .002"
>> lines with .005" or .004" spacing? What will it take to convince you
>> that etching should be the least of your worries provided you know what
>> you're doing? Do I have to walk on water?
>
> What is your etching technique ?
> What is your foil thickness ?
>
> Ok, you got me curious. Now you've made me waste the past 3 hours of my
> time  determining if etching undercut does indeed limiting line widths.
> My test showed for 18um foil  PCB the limited factor is the printer, and
> not the etch process. I have compile a quick web page showing the
> results of this mornings experiment, enjoy.
>
> http://members.optusnet.com.au/~eseychell/pcbtest1/line_width_tests.html
>
> Even thought my results demonstrated a printer limitation, I still think
> etch undercut can be an issue, especially for bubble etchers or 70um
> ("2oz") foils.

Manual agitation by rocking a container might be best for very narrow lines. 
It's also easier to see what is happening, and it could be combined with 
rubbing the surface with a swab.

Leon

Andrew wrote:
> 
> 
> Does the Epson peizo print head have less
> scatter of dots than the Canon Adam uses ?
> Your photos of the 6/6 artwork certainly
> look less jaggy than Adams.

Unfortunately all inkjet printers seem to suffer the same problem. Since 
its invisible to the naked eye , printer manufactures are unlikely to 
bother improving it. All that is needed is to slow the print head by 10 
times. You may wait 5 minutes for a printout, but who cares ? I can't 
think of any applications other than for hobby PCBs. Professional screen 
printers who use inkjet printer transparencies for photomasks don't even 
need that accuracy and any PCB manufacture will use a laser photoplotter 
(3000+ dpi). So the marketers of inkjet printer manufactures aren't 
going to get any calls from customers demanding reduced line edge 
jaggedness.

If your interested, I can send you similar images produced from an Epson 
Stylus 660 printer.

> In my case the bubble etcher will fully
> etch a stripe down one side of a board in
> 1/2 the time it takes the middle of the
> board to etch.
> 

This agrees with my experience with bubble etching too. For really fine 
lines, such large etch non-uniformity is unacceptable. Another thing to 
note is Derek's test pieces had 6 mil spacing between 6 mil lines. I bet 
if one of those 6 mil lines were far away from surrounding copper then 
it will get etched faster, and therefore more undercut.

Adam

Adam Seychell wrote:
> Andrew wrote:
>>
>> Does the Epson peizo print head have less
>> scatter of dots than the Canon Adam uses ?
>> Your photos of the 6/6 artwork certainly
>> look less jaggy than Adams.
> 
> Unfortunately all inkjet printers seem to suffer the same problem. Since 
> its invisible to the naked eye , printer manufactures are unlikely to 
> bother improving it. All that is needed is to slow the print head by 10 
> times. You may wait 5 minutes for a printout, but who cares ? I can't 
> think of any applications other than for hobby PCBs. Professional screen 
> printers who use inkjet printer transparencies for photomasks don't even 
> need that accuracy and any PCB manufacture will use a laser photoplotter 
> (3000+ dpi). So the marketers of inkjet printer manufactures aren't 
> going to get any calls from customers demanding reduced line edge 
> jaggedness.
> 
> If your interested, I can send you similar images produced from an Epson 
> Stylus 660 printer.
> 
>> In my case the bubble etcher will fully
>> etch a stripe down one side of a board in
>> 1/2 the time it takes the middle of the
>> board to etch.
>>
> 
> This agrees with my experience with bubble etching too. For really fine 
> lines, such large etch non-uniformity is unacceptable. Another thing to 
> note is Derek's test pieces had 6 mil spacing between 6 mil lines. I bet 
> if one of those 6 mil lines were far away from surrounding copper then 
> it will get etched faster, and therefore more undercut.

Using a Stylus 460 printer driver on linux (CUPS printing), i've got it on
a mode where a 50mm long pcb will take atleast 10 mins to print. The resolution
is very high.

>> Adam Seychell wrote:
>> <SNIP>
>> Unfortunately all inkjet printers seem to
>> suffer the same problem. Since its invisible
>> to the naked eye , printer manufactures are
>> unlikely to bother improving it. All that is
>> needed is to slow the print head by 10 times.
>> You may wait 5 minutes for a printout, but
>> who cares ?
>> <SNIP>

> Russell Shaw wrote:
>
> Using a Stylus 460 printer driver on linux
> (CUPS printing), i've got it on a mode where a
> 50mm long pcb will take atleast 10 mins to
> print. The resolution is very high.

Russell,

Can you make the printer print from the same
bank of nozzels every pass?  If it doesn't does
it alter the offset on where it fires a bank so
they all have the same position?

I imagine that if you can control the printheads
like that under linux the direct inkjet resist
people would love it.

> Adam Seychell wrote:

> > Andrew wrote:
> > 
> > Does the Epson peizo print head have less
> > scatter of dots than the Canon Adam uses ?
> > Your photos of the 6/6 artwork certainly
> > look less jaggy than Adams.
>
> Unfortunately all inkjet printers seem to
> suffer the same problem. Since its invisible
> to the naked eye , printer manufactures are
> unlikely to bother improving it. All that is
> needed is to slow the print head by 10 times.
> You may wait 5 minutes for a printout, but
> who cares ?
> <SNIP>

Slow the print head down or make it do 12
passes (one for each bank).

Probably going to require major firmware
hacking for either method I think.

> AndrewM wrote:
> 
> I am going to print out your test circles
> on my laser here and take some photos of
> it when I get home.

Again sub par compared to Mr Hawkins photos
but the best I was able to get in several
attempts.

I must say you don't have a lot of depth of
feild this close up.

http://www.thehacktory.com/LaserWide.jpg
http://www.thehacktory.com/LaserClose.jpg

They are the 5.3 thou lines at the center
of Adams circles.  I didn't bother doing
horizontal and verticle lines in the photo
as they looked identical on the laser
printer.

Also I must say the lines have more fuzz/
random dot noise around them than they did
6 months ago.  Might be a sympton of an
ageing transfer roller or something.

Andrew wrote:
>>> Adam Seychell wrote:
>>> <SNIP>
>>> Unfortunately all inkjet printers seem to
>>> suffer the same problem. Since its invisible
>>> to the naked eye , printer manufactures are
>>> unlikely to bother improving it. All that is
>>> needed is to slow the print head by 10 times.
>>> You may wait 5 minutes for a printout, but
>>> who cares ?
>>> <SNIP>
> 
>> Russell Shaw wrote:
>>
>> Using a Stylus 460 printer driver on linux
>> (CUPS printing), i've got it on a mode where a
>> 50mm long pcb will take atleast 10 mins to
>> print. The resolution is very high.
> 
> Russell,
> 
> Can you make the printer print from the same
> bank of nozzels every pass?  If it doesn't does
> it alter the offset on where it fires a bank so
> they all have the same position?
> 
> I imagine that if you can control the printheads
> like that under linux the direct inkjet resist
> people would love it.

I had a quick look in the gutenprint source code and
can see that every possible parameter you'd want to
change could be done (for epson, look in escp2-driver.c)

The whole thing is too big to figure out immediately.
There might be an external config file setting you could
change.

http://gutenprint.sourceforge.net/p_Download.php3

>I bet if one of those 6 mil lines were far away from surrounding 
>copper then it will get etched faster, and therefore more undercut.

Look at the outside edge of the outermost track in the link below. The 
only real issue here is the pronounced lower printer resolution on 
tracks perpendicular to printhead movement. This contributes to the 
printer limitations I've been referring to since it's just as evident 
in the artwork.

http://upload.pbase.com/eldata/image/77010930/original


--- In Homebrew_PCBs@yahoogroups.com, Adam Seychell <a_seychell@...> 
wrote:
>

>Not saying you should. I am saying my
>bubble etcher is bad and I need to get a
>new etcher. You obviously have a better
>bubble etcher than I do.

Check out messages 14958 and 14969. I've always favored the smallest BE 
(ET-10) mainly because 6"X4" is my standard board size and you need 
half as much etchant to handle that in an ET-10 than the larger BEs. 
But as message 14969 indicates, all ET-20s are not the same so it's 
quite possible that all ET-10s are not the same either. Whatever, I've 
never had any of the issues usually attributed to BEs.

--- In Homebrew_PCBs@yahoogroups.com, "Andrew" <andrewm1973@...> wrote:
>

>This contributes to the printer limitations I've been referring to 
>since it's just as evident in the artwork.

I found the artwork since the last post.

6/6 artwork snippet;

http://upload.pbase.com/image/77013545/original

6/6 etched board snippet from above;

http://www.pbase.com/eldata/image/77010930/original

Yes, the etched board looks better that the artwork. And there you have 
it folks "Undercut can be our friend" or "Jameson is fidelity" 
or "Jameson is gravity". Whatever...
 









--- In Homebrew_PCBs@yahoogroups.com, "derekhawkins" <eldata@...> wrote:
>
> >I bet if one of those 6 mil lines were far away from surrounding 
> >copper then it will get etched faster, and therefore more undercut.
> 
> Look at the outside edge of the outermost track in the link below. 
The 

> only real issue here is the pronounced lower printer resolution on 
> tracks perpendicular to printhead movement. This contributes to the 
> printer limitations I've been referring to since it's just as evident 
> in the artwork.
> 
> http://upload.pbase.com/eldata/image/77010930/original
> 
> 
> --- In Homebrew_PCBs@yahoogroups.com, Adam Seychell <a_seychell@> 
> wrote:
> >
>

On Tue, 10 Apr 2007 23:30:22 +0200, derekhawkins <eldata@...> wrote:

> How about .003" or .002"
> lines with .005" or .004" spacing? What will it take to convince you
> that etching should be the least of your worries provided you know what
> you're doing? Do I have to walk on water?


I would be impressed with 2 mil lines allright.
At least with an air agitated vertical etcher using CuCl i don't think  
it'll work.

I don't doubt it is possible with the right etcher, they do crazy stuff  
etching ICs, but in my etcher i don't think so.

You are welcome to try, and i'll be appropriately impressed if you  
succeed, but it doesn't really matter since nobody needs such fine lines  
anyway.

ST

> >AndrewM wrote:
> >Not saying you should. I am saying my
> >bubble etcher is bad and I need to get a
> >new etcher. You obviously have a better
> >bubble etcher than I do.

> derekhawkins wrote:
> 
> Check out messages 14958 and 14969. I've
> always favored the smallest BE (ET-10)
> mainly because 6"X4" is my standard board
> size and you need half as much etchant to
> handle that in an ET-10 than the larger
> BEs. But as message 14969 indicates, all
> ET-20s are not the same so it's quite
> possible that all ET-10s are not the same
> either. Whatever, I've never had any of
> the issues usually attributed to BEs.

Had a look at the supplier.  Mine is an 
ET-20

http://www.computronics.com.au/kinsten/tools/

I actually have 2 of them and I checked them
both out last night.  Both have the big
swirl happening.

After you posted you pictures of your 6/6
afterwork last night and I compared them
to the test print I did of Adams pattern
I would say with pretty high certainty my
problem is the etching.  The printout from
the laser here looks sharper than the ink
jet printouts.  And as I have said before
this laser has very high density and no
pinholes.