Homebrew PCBs

I found out today that the Linux printing software (Gutenprint) now
supports my R280 printer (the R280 came with a camera, I've mentioned
this before).  So now I can control the dots more accurately.
However, it seems that the printer just can't position a dot with,
say, 1 mil of accuracy.  I haven't done much experimenting with the
driver to try to reduce the variables that are causing this variation
(like weaving method).  Note that this is with unidirectional
printing; bidirectional makes it much worse.  The printer is 5760 x
2880 DPI, I've been using 2880x2880 for my tests.

Even so, if I can solve the bleeding problem (see below), it looks
like 4 mil can be done, maybe 5-6 mil reliably.  Assuming the
photoresist doesn't make things worse.  I haven't tried doing anything
with these plots other than plotting them and looking at them under a
microscope.

Fortunately, Gutenprint allows you to adjust the ink density.
I've experimented with values from 0.8 (the default) up to 2.0.  I've
found that 2.0 gives a really dark print - pretty much solid - but the
ink quickly bleeds on my vellum, corrupting the print.  Any tips on
that?  Heat gun? (tried - it takes long enough to print that the
bleeding happens before I can get the page out) Less ink leaves spotty
results, so for this to work with these inks, I'd need to be able to
print at 2.0 without ink bleeding.

Also - I've figure out my Vellum problem (paper "jam").  Putting a
small piece of masking tape in the upper right corner of the vellum
lets the printer see the leading edge, so it won't reject the paper.
The tape can go on the front or back of the vellum.

The R280 has a cd-rom tray, but I haven't tried printing directly to
copper yet (resist) or to copper with photofilm on it.  I suspect the
results will be no better than on the vellum.

http://www.delorie.com/pcb/r280/

DJ

You may care to investigate the possibility of 'reducing' the head to
paper distance. This may gather up the dots. It appears the ink
droplets  are launched randomly in a cone pattern from each nozzle.
Reducing the flight height should reduce the spread. 

Regards

DJ Delorie.> wrote:

>> 
> I found out today that the Linux printing software (Gutenprint) now
> supports my R280 printer (the R280 came with a camera, I've mentioned
> this before).  So now I can control the dots more accurately.
> However, it seems that the printer just can't position a dot with,
> say, 1 mil of accuracy.  I haven't done much experimenting with the
> driver to try to reduce the variables that are causing this variation

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

From: "DJ Delorie" <dj@...>
To: <Homebrew_PCBs@yahoogroups.com>
Sent: Saturday, May 10, 2008 4:32 AM
Subject: [Homebrew_PCBs] Epson R280 inket notes


>
> Fortunately, Gutenprint allows you to adjust the ink density.
> I've experimented with values from 0.8 (the default) up to 2.0.  I've
> found that 2.0 gives a really dark print - pretty much solid - but the
> ink quickly bleeds on my vellum, corrupting the print.  Any tips on
> that?  Heat gun? (tried - it takes long enough to print that the
> bleeding happens before I can get the page out) Less ink leaves spotty
> results, so for this to work with these inks, I'd need to be able to
> print at 2.0 without ink bleeding.

Try using Mega Electronics JetStar or JetStar Premium film. It's especially 
made for PCB artwork. I use the Premium film with my HP printer with 
excellent results, and the ordinary stuff worked very well with the Epson 
printer I used to use. You can buy it from Farnell in the USA.

Leon

DJ Delorie wrote:
> I found out today that the Linux printing software (Gutenprint) now
> supports my R280 printer (the R280 came with a camera, I've mentioned
> this before).  So now I can control the dots more accurately.
> However, it seems that the printer just can't position a dot with,
> say, 1 mil of accuracy.  I haven't done much experimenting with the
> driver to try to reduce the variables that are causing this variation
> (like weaving method).  Note that this is with unidirectional
> printing; bidirectional makes it much worse.  The printer is 5760 x
> 2880 DPI, I've been using 2880x2880 for my tests.
> 

..

Inkjet transparency films vary widely between brands. the Epson film 
(now obsolete ) had a unique dye ink absorption coating that would 
always produce %100 solid fills. The ink did not sit on the surface 
unlike other brands I investigated at the time. You can buy an 
equivalent film from
here http://www.des-pl.com.au/products/products.asp?pd=PD00068
sorry, I don't know of any other suppliers.
You really have to test other brands of media.

As for dot placement fluctuations. Let us know when you've solved it as 
this is the major limitation of all inkjet printers.

Some follow-up notes from today's experiments...

It occurred to me that I've got negative acting film, so I need ink
where there will be NO copper.  This changes the rules a bit :-)

Specifically - the printer can print dark fine lines, even if I say "1
mil" it can print dark enough (I hope) to properly expose the
photofilm.  The problem with the positive print was that bleeding
would cause shorts, but I needed to make the pads as big as they
needed to be, which put lots of ink there.

So what I tried tonight was something different.  I printed a
negative, which is a lot less ink because I normally flood copper onto
the board to reduce etchant usage Thus, rather than print a big pad,
I'm printing a thin outline around it.

I also told PCB to bloat all the copper by 1.5 mil to compensate for
the blurring caused by dot misalignment.  With a 5 mil space, this
leaves 2 mil of ink, plenty for the printer to print with.  It also
moves the ink away by more than the misalignment amount.  If this
works out, I'll just change my design rules to use larger traces and
smaller spaces, like 8/4 instead of 6/6, or 7/3 instead of 5/5.  That
will handle the copper floods properly.

I also tried printing on an old laser transparency film.  Note that
inkjet ink pretty much doesn't dry on this, as it is not absorbed at
all.  I haven't experimented with a heat gun on this; I assume that
something like JetStar will resolve the ink issues.  Anyway...

I'll have to experiment with the ink density.  What happens is that in
large ink areas, there's so much ink that the surface tension pulls
the ink away from the edges.  So for this to be useful, either a more
absorbant transparency will be needed (as noted :), or software to
find these areas and checkerboard their interiors (much like we have
to break up big solder paste areas under thermal pads).  If ink lines
are kept to a few mil, they seem to hold their shape and density
nicely - they're pretty much solid black.  Since they don't dry
though, they will eventually bleed enough to corrupt the plot.

I tried printing on heavy cardstock (8 mil thick vs 3 for paper) with
no change in ink drop misalignment.  Of course, I have no idea how far
the print head is from the paper, or how to change it with my
software.  If I get too close, it smears the ink all over the paper.

If the ink is forming globules of any description then that 
automatically means the media you are using is not compatible with the 
ink. There should be no density difference between large areas and 
narrow line features. I suggest you give up on the media your 
experimenting with. I found a number of inkjet transparency sellers for 
the screen printing industry were happy to provide free samples of film. 
The JetStar brand was mentioned before.

For example, I bought a 100 sheets of "HP everyday semi-gloss gloss 
photo paper" that would simply not absorb ink correctly in either my 
Epson or Canon printers. Some inkjet clear films I came across were 
similarly ineffective.

Adam

DJ Delorie wrote:

> Some follow-up notes from today's experiments...
> 
> It occurred to me that I've got negative acting film, so I need ink
> where there will be NO copper.  This changes the rules a bit :-)
> 
> Specifically - the printer can print dark fine lines, even if I say "1
> mil" it can print dark enough (I hope) to properly expose the
> photofilm.  The problem with the positive print was that bleeding
> would cause shorts, but I needed to make the pads as big as they
> needed to be, which put lots of ink there.
> 
> So what I tried tonight was something different.  I printed a
> negative, which is a lot less ink because I normally flood copper onto
> the board to reduce etchant usage Thus, rather than print a big pad,
> I'm printing a thin outline around it.
> 
> I also told PCB to bloat all the copper by 1.5 mil to compensate for
> the blurring caused by dot misalignment.  With a 5 mil space, this
> leaves 2 mil of ink, plenty for the printer to print with.  It also
> moves the ink away by more than the misalignment amount.  If this
> works out, I'll just change my design rules to use larger traces and
> smaller spaces, like 8/4 instead of 6/6, or 7/3 instead of 5/5.  That
> will handle the copper floods properly.
> 
> I also tried printing on an old laser transparency film.  Note that
> inkjet ink pretty much doesn't dry on this, as it is not absorbed at
> all.  I haven't experimented with a heat gun on this; I assume that
> something like JetStar will resolve the ink issues.  Anyway...
> 
> I'll have to experiment with the ink density.  What happens is that in
> large ink areas, there's so much ink that the surface tension pulls
> the ink away from the edges.  So for this to be useful, either a more
> absorbant transparency will be needed (as noted :), or software to
> find these areas and checkerboard their interiors (much like we have
> to break up big solder paste areas under thermal pads).  If ink lines
> are kept to a few mil, they seem to hold their shape and density
> nicely - they're pretty much solid black.  Since they don't dry
> though, they will eventually bleed enough to corrupt the plot.
> 
> I tried printing on heavy cardstock (8 mil thick vs 3 for paper) with
> no change in ink drop misalignment.  Of course, I have no idea how far
> the print head is from the paper, or how to change it with my
> software.  If I get too close, it smears the ink all over the paper.
>

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

From: "Adam Seychell" <a_seychell@...>
To: <Homebrew_PCBs@yahoogroups.com>
Sent: Sunday, May 11, 2008 8:43 AM
Subject: Re: [Homebrew_PCBs] Epson R280 inket notes


> If the ink is forming globules of any description then that
> automatically means the media you are using is not compatible with the
> ink. There should be no density difference between large areas and
> narrow line features. I suggest you give up on the media your
> experimenting with. I found a number of inkjet transparency sellers for
> the screen printing industry were happy to provide free samples of film.
> The JetStar brand was mentioned before.

Mega Electronics sent me samples of the JetStar Premium film when I had 
problems with the standard stuff in my new HP printer. Being in the UK where 
Mega is based helped, of course.

Leon

Assuming you are testing for the photoresist you got from me, I got
execellent results using Pictorico Premium OHP Transparency Film. 

http://www.pictorico.com/index.asp?PageAction=VIEWCATS&Category=51

Part of the reason that I used this is that people who do alternative
photography that involves etching, use this transparency with
excellent results.  According to the wrapper my transparency is
enclosed in it is good for pigment and dye based inkjet inks.

Adam Seychell <a_seychell@...> writes:
> If the ink is forming globules of any description then that
> automatically means the media you are using is not compatible with
> the ink.

Yeah, I figured that.  I have some old HP glossy inkjet paper that
does the same thing, but the new Epson glossy works just fine.

However, I'm just playing with whatever I've got sitting around here.
The transparencies are for laser printers anyway, I expected them to
reject pretty much any ink.  They don't bleed though, and the puddling
seems to help.  It does show that the printer supports thinner
features than I expected, though.  As of this morning, the
transparency prints from last night are a mess - they bled all over
the place and the ink is still wet.  So yeah, I can't *use* those
sheets for anything, but it was worth trying them anyway.

Staples has inkjet compatible transparencies, but they're expensive.

"javaguy11111" <javaguy11111@...> writes:
> Assuming you are testing for the photoresist you got from me,

Yup.  I'm looking into a UV LED based exposure box too, I haven't
tried the sunlight exposure yet.

Hey - anyone tried printing on the photoresist film itself?  I have
some 8 mil clad I could try it...

--- In Homebrew_PCBs@yahoogroups.com, DJ Delorie <dj@...> wrote:
>
> 
> "javaguy11111" <javaguy11111@...> writes:
> > Assuming you are testing for the photoresist you got from me,
> 
> Yup.  I'm looking into a UV LED based exposure box too, I haven't
> tried the sunlight exposure yet.
> 

With classes done for the semester I am finally moving forward with my
UV LED work. I have started back working on my driver board using the
MAX16809. If I do not spend too much time goofing off on various
forums I might even get most of the design work done today :)

It is very likely that the driver software is dithering and that is
why you see the spread on a line.

DJ, are you printing that as a raster graphic or a vector file?

Steve Greenfield

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

>
> You may care to investigate the possibility of 'reducing' the head to
> paper distance. This may gather up the dots. It appears the ink
> droplets  are launched randomly in a cone pattern from each nozzle.
> Reducing the flight height should reduce the spread. 
> 
> Regards
> 
> DJ Delorie.> wrote:
> >> 
> > I found out today that the Linux printing software (Gutenprint) now
> > supports my R280 printer (the R280 came with a camera, I've mentioned
> > this before).  So now I can control the dots more accurately.
> > However, it seems that the printer just can't position a dot with,
> > say, 1 mil of accuracy.  I haven't done much experimenting with the
> > driver to try to reduce the variables that are causing this variation
>

"Steve" <alienrelics@...> writes:
> It is very likely that the driver software is dithering and that is
> why you see the spread on a line.
> 
> DJ, are you printing that as a raster graphic or a vector file?

It's a raster, printed with the Gimp under Linux.  The image
resolution matches the print resolution dot-for-dot (2880 dpi), and
the image is a 1bpp black and white.  I've tried single-dot-rows and
single-dot-columns, and pretty much all the settings available in the
print settings block.

Note that enabling bi-directional printing makes it a LOT worse, as
expected - pretty much two copies of everything.

The only thing I haven't tried is producing my own escp2 file for the
printer that uses a single jet in a single pass to see how much of the
problem is physics-related.