Homebrew_PCBs

Alright, call me crazy, but an idea just occurred to me, and I can't see why
it can't work. To me, the most difficult (expensive, complicated,
unreliable, etc) part of creating a PCB is getting the pattern laid out on
the board so it can be exposed, etc. The rest is science ... standard
procedure, standards times, standard results. Yes, laying out the pattern
can be science, but at a higher price, and still takes time to line up the
image transparencies, etc.

So this idea just randomly came to me ... why not use a projector to project
the image on the board. Use a positive-coated board and an XGA LCD
projector, with a UV bulb instead of the usual bulb. I get quite good
results at 300dpi, so if we go with a projector resolution of 1024x768 for
now, that's about 3.4" x 2.5". There may need to be some optical
manipulation done to get it to focus clearly at that distance, and a bunch of
other little things sorted out (such as calibration for
non-linear/inconsistent pixel spread), but I can't see why it can't work.
No, it's not cheap, but I believe it would be really fast, going from
computer straight to PCB.

Maybe this has been thought of or done already? Or maybe I'm going off the
deep end?

Cheers,
-Neil.

>From: Neil <cobra_neil@...>
>Reply-To: Homebrew_PCBs@yahoogroups.com
>To: pcblist <Homebrew_PCBs@yahoogroups.com>
>Subject: [Homebrew_PCBs] I have a dream...
>Date: Fri, 23 May 2003 03:36:00 -0500
>

Even better would be something like a home-made laser plotter with a UV
laser plotting directly onto the resist-coated PCB. Send it the Gerber file
and take out the exposed PCB a few minutes later, into the developer and
etchant, and one has an etched PCB. That's what I dream of!

Leon
--
Leon Heller, G1HSM Tel: +44 1424 423947
Email:leon_heller@...
My web page: http://www.geocities.com/leon_heller

_________________________________________________________________
Get Hotmail on your mobile phone http://www.msn.co.uk/msnmobile

>Alright, call me crazy, but an idea just occurred to me, and I can't see why
>it can't work. To me, the most difficult (expensive, complicated,
>unreliable, etc) part of creating a PCB is getting the pattern laid out on
>the board so it can be exposed, etc. The rest is science ... standard
>procedure, standards times, standard results. Yes, laying out the pattern
>can be science, but at a higher price, and still takes time to line up the
>image transparencies, etc.
>
>So this idea just randomly came to me ... why not use a projector to project
>the image on the board. Use a positive-coated board and an XGA LCD
>projector, with a UV bulb instead of the usual bulb. I get quite good
>results at 300dpi, so if we go with a projector resolution of 1024x768 for
>now, that's about 3.4" x 2.5". There may need to be some optical
>manipulation done to get it to focus clearly at that distance, and a bunch of
>other little things sorted out (such as calibration for
>non-linear/inconsistent pixel spread), but I can't see why it can't work.
>No, it's not cheap, but I believe it would be really fast, going from
>computer straight to PCB.
>
>Maybe this has been thought of or done already? Or maybe I'm going off the
>deep end?

There was a thread about this in a german usenet group not so long
ago. Anyways, while this might be possible, let me tell you that I do
not agree with you on getting the pattern laid out on the board being
difficult. Actually this is a very straight foreward process which can
also be standard procedure, standards times, standard results.

I simply print the artwork using the CAD software onto cheap
transparency paper. I do agree that aligning - with no experience -
can be difficult. However let me tell you how I do it and you will see
that it can be very simple in fact.

I align the top and bottom layers by puting the two transparencies
directly ontop of each other. Since they are transparent, aligning is
very simple by moving them around until the vias and holes macht
perfectly. I cut the top layer transparency so as it's a bit smaller
than the other one Once I have them aligned, I simply tape some tesa
film on one side to fix them. I then take two cut off's of FR4
material that I aling along the OPOSITE side of the just added tesa
film and one in the right angle of it. If you do it once this way, you
quickly see that it's actualy a lot simpler than it may sounds. With
this angle, you can slide the FR4 material into this "pocket" and
hence have it propperly aligned no matter if you have to move it
around or turn it over. Exposing this unit of PCB and transperncies
then becomes really simple. Building an exposing box (or buing a ready
made one) is also much easier and cheaper then using your sugested
prjector setup.

Again, I do agree that you may can do it the way you proposed, but the
described method is magnitudes cheaper and simpler.

Markus

Hello Neil,

23. May 2003, 10:36:00, you wrote:


N> Maybe this has been thought of or done already? Or maybe I'm going off the
N> deep end?

N> Cheers,
N> -Neil.

You forgot only one thing: CRT/LCD screen resolution is 72 dpi! Best results
are obtained if you can find a prepress bureau and give them your postscript
files, and for very little money tey will produce the perfect artwork on
graphical film. Resolution of photoplotters is at least 2000 dpi!

I always have my artwork made this way. 1m of artwork (width 355mm) costs me
7 EUR. My largest board that I've made was 620x250mm, but all the rest are
mostly 160x100 mm.


--
Best Regards,
Zoran
mailto:zasto@...

At home I have a Photographic 'Wet' Darkroom (as opposed to a 'Digital'
Darkroom) and I have often wanted to project UV light to expose some of
the older forms of emulsion (Cyanotype, for instance). Conventional
wisdom has it that this cannot be done with glass lenses as they absorb
far too much of the short wavelength light. I have tried a jury-rig
and it didn't work at all. I have also observed that if I use a
plastic negative, of only a few thou' thickness, the exposure times
double (this is when I use a direct light source).
Now, if someone out there is able to push short wavelength UV through a
glass lens, or perhaps they have a source of lenses made from quartz or
some other material that passes UV, I would most definitely like to
know.
There are sources of special lenses but at a high price only. The URL
below is such a company.

http://www.edmundoptics.com/index.cfm

Just a thought,
Simon

On Friday, May 23, 2003, at 09:36 Europe/London, Neil wrote:

> Alright, call me crazy, but an idea just occurred to me, and I can't
> see why
> it can't work. To me, the most difficult (expensive, complicated,
> unreliable, etc) part of creating a PCB is getting the pattern laid
> out on
> the board so it can be exposed, etc. The rest is science ... standard
> procedure, standards times, standard results. Yes, laying out the
> pattern
> can be science, but at a higher price, and still takes time to line up
> the
> image transparencies, etc.
>
> So this idea just randomly came to me ... why not use a projector to
> project
> the image on the board. Use a positive-coated board and an XGA LCD
> projector, with a UV bulb instead of the usual bulb. I get quite good
> results at 300dpi, so if we go with a projector resolution of 1024x768
> for
> now, that's about 3.4" x 2.5". There may need to be some optical
> manipulation done to get it to focus clearly at that distance, and a
> bunch of
> other little things sorted out (such as calibration for
> non-linear/inconsistent pixel spread), but I can't see why it can't
> work.
> No, it's not cheap, but I believe it would be really fast, going from
> computer straight to PCB.
>
> Maybe this has been thought of or done already? Or maybe I'm going
> off the
> deep end?
>
> Cheers,
> -Neil.
>
>
> ------------------------ Yahoo! Groups Sponsor
> ---------------------~-->
> Get A Free Psychic Reading! Your Online Answer To Life's Important
> Questions.
> http://us.click.yahoo.com/Lj3uPC/Me7FAA/uetFAA/bGYolB/TM
> ---------------------------------------------------------------------
> ~->
>
> Be sure to visit the group home and check for new Bookmarks and files:
> http://groups.yahoo.com/group/Homebrew_PCBs
>
> To unsubscribe from this group, send an email to:
> Homebrew_PCBs-unsubscribe@yahoogroups.com
>
>
>
> Your use of Yahoo! Groups is subject to
> http://docs.yahoo.com/info/terms/
>
>
>
>
>
Simon Whitehead
s.whitehead@...

how about just a uv led with a lens to focus to a
diameter you want i bet it would fit in the pen holder
without modifications you might just have to
repeat the number of plots till it is dark enough
--- Leon Heller <leon_heller@...> wrote:

>
>
>
> >From: Neil <cobra_neil@...>
> >Reply-To: Homebrew_PCBs@yahoogroups.com
> >To: pcblist <Homebrew_PCBs@yahoogroups.com>
> >Subject: [Homebrew_PCBs] I have a dream...
> >Date: Fri, 23 May 2003 03:36:00 -0500
> >
>
> Even better would be something like a home-made
> laser plotter with a UV
> laser plotting directly onto the resist-coated PCB.
> Send it the Gerber file
> and take out the exposed PCB a few minutes later,
> into the developer and
> etchant, and one has an etched PCB. That's what I
> dream of!
>
> Leon
> --
> Leon Heller, G1HSM Tel: +44 1424 423947
> Email:leon_heller@...
> My web page: http://www.geocities.com/leon_heller
>
>

_________________________________________________________________

> Get Hotmail on your mobile phone
> http://www.msn.co.uk/msnmobile
>
>

__________________________________
Do you Yahoo!?
The New Yahoo! Search - Faster. Easier. Bingo.
http://search.yahoo.com

That is a great "Hi-Tek" idea.

BUT, I subscribe to the K.I.S.S. method... keep it simple

I used to work in a professional PCB manufacturing lab and they also
did things the simplest and cheapest ways.

For double sided PCBs you take both films (top/bottom) and register
them together. Tape a scrap piece of PCB material to one edge. Now
you have an exposure "set". All you need to do is slip in the PCB
into the "set" and expose. We had a top/bottom exposure frame, lights
underneath and in the lid.

We found that certain acrylics transmit the UV better than others.
Quartz is the best, but extremely expensive. Lexan BLOCKS UV totally.
Glass works, but not as well as the acrylics. The only problem with
acrylics was that it needed to be replaced occasionally since it
would become "crazed" and brittle after a while.

John M...
=================

--- In Homebrew_PCBs@yahoogroups.com, Neil <cobra_neil@a...> wrote:
> Alright, call me crazy, but an idea just occurred to me, and I
can't see why
> it can't work. To me, the most difficult (expensive, complicated,
> unreliable, etc) part of creating a PCB is getting the pattern laid
out on
> the board so it can be exposed, etc. The rest is science ...
standard
> procedure, standards times, standard results. Yes, laying out the
pattern
> can be science, but at a higher price, and still takes time to line
up the
> image transparencies, etc.
>
> So this idea just randomly came to me ... why not use a projector
to project
> the image on the board. Use a positive-coated board and an XGA LCD
> projector, with a UV bulb instead of the usual bulb. I get quite
good
> results at 300dpi, so if we go with a projector resolution of
1024x768 for
> now, that's about 3.4" x 2.5". There may need to be some optical
> manipulation done to get it to focus clearly at that distance, and
a bunch of
> other little things sorted out (such as calibration for
> non-linear/inconsistent pixel spread), but I can't see why it can't
work.
> No, it's not cheap, but I believe it would be really fast, going
from
> computer straight to PCB.
>
> Maybe this has been thought of or done already? Or maybe I'm going
off the
> deep end?
>
> Cheers,
> -Neil.

nice dream...

but i don't think any plotter attempt will work.
it is not the problem of getting some source (a laser with fiberoptics etc) on the plotter but much more
the problem of making a good file for the plotter. i couldn't get it to work with hpgl to be really
good. i have a 0,18mm pen, perfectly working. but the software doesn't really what i want it.
so now i try it with the inkjet....

regards
stefan

23.05.2003 17:46:00, "John Myszkowski" <myszka_us2000@...> wrote:

>That is a great "Hi-Tek" idea.
>
>BUT, I subscribe to the K.I.S.S. method... keep it simple
>
>I used to work in a professional PCB manufacturing lab and they also
>did things the simplest and cheapest ways.
>
>For double sided PCBs you take both films (top/bottom) and register
>them together. Tape a scrap piece of PCB material to one edge. Now
>you have an exposure "set". All you need to do is slip in the PCB
>into the "set" and expose. We had a top/bottom exposure frame, lights
>underneath and in the lid.
>
>We found that certain acrylics transmit the UV better than others.
>Quartz is the best, but extremely expensive. Lexan BLOCKS UV totally.
>Glass works, but not as well as the acrylics. The only problem with
>acrylics was that it needed to be replaced occasionally since it
>would become "crazed" and brittle after a while.
>
>John M...
>=================
>
>

--- In Homebrew_PCBs@yahoogroups.com, "Leon Heller" <leon_heller@h...>
wrote:

>
>
>
> >From: Neil <cobra_neil@a...>
> >Reply-To: Homebrew_PCBs@yahoogroups.com
> >To: pcblist <Homebrew_PCBs@yahoogroups.com>
> >Subject: [Homebrew_PCBs] I have a dream...
> >Date: Fri, 23 May 2003 03:36:00 -0500
> >
>
> Even better would be something like a home-made laser plotter with a UV
> laser plotting directly onto the resist-coated PCB. Send it the

Gerber file

> and take out the exposed PCB a few minutes later, into the developer

and

> etchant, and one has an etched PCB. That's what I dream of!
>
> Leon

It would appear that several people are thinking along the same lines.
The LCD screen idea occurred to me (either with changing the
backlighting to a UV source or using one of the older overhead
projector LCD screens), as did using a TI Digital Mirror Device
(something that looks like it might already be in the works at some
companies for some photoprocessing equipment and PCB manufacturing).
Given the resolution needed, obviously you'd probably want an image
that could be stepped across the board -- no single DMD has sufficient
pixels to cover a good-sized board with thousandths-of-an-inch resolution.

As for the laser idea, well, that's something I've been thinking about
a lot lately, too. I recently dismantled a malfunctioning Laserjet
III and obtained the laser diode scanning assembly; replacing the
laser diode should be easy enough once Nichia and others start
mass-producing them, although I'll need to ask HP whether the optics
will pass UV light and focus it sufficiently.

The challenges would be installing the scanner on a precision-motion
shuttle with some sort of stepper motor for moving the laser down the
PCB as each line is "drawn" (maybe using the original LJIII stepper),
correctly spacing the optical assembly for accurate to-scale imaging
for several thicknesses of copper-clad board, and controlling the
whole deal based upon image or gerber files. It would also be nice to
know whether the laser scanning assembly is actually capable of
1/600th of an inch (or better) spot size focusing.

It would be necessary to slow down the laser scanning from that used
in the original LJ III; with a 2 mW laser and reasonably sensitive
(e.g. 50 mJoule/cm^2) photoresist, covering an entire board could take
more than an hour (25 seconds per square centimeter, give-or-take).
Of course, if one used higher-sensitivity resist, one could go faster.
It would be nice to be able to adjust speed to compensate.

I guess if I make it work, I'll let you guys know :)

Jonathan

24.05.2003 06:13:44, "Jonathan W" <n0ym@...> wrote:

>--- In Homebrew_PCBs@yahoogroups.com, "Leon Heller" <leon_heller@h...>
>wrote:
>>
>>
>>
>> >From: Neil <cobra_neil@a...>
>> >Reply-To: Homebrew_PCBs@yahoogroups.com
>> >To: pcblist <Homebrew_PCBs@yahoogroups.com>
>> >Subject: [Homebrew_PCBs] I have a dream...
>> >Date: Fri, 23 May 2003 03:36:00 -0500
>> >
>>
>> Even better would be something like a home-made laser plotter with a UV
>> laser plotting directly onto the resist-coated PCB. Send it the
>Gerber file
>> and take out the exposed PCB a few minutes later, into the developer
>and
>> etchant, and one has an etched PCB. That's what I dream of!
>>
>> Leon
>
>It would appear that several people are thinking along the same lines.
> The LCD screen idea occurred to me (either with changing the
>backlighting to a UV source or using one of the older overhead
>projector LCD screens), as did using a TI Digital Mirror Device
>(something that looks like it might already be in the works at some
>companies for some photoprocessing equipment and PCB manufacturing).
>Given the resolution needed, obviously you'd probably want an image
>that could be stepped across the board -- no single DMD has sufficient
>pixels to cover a good-sized board with thousandths-of-an-inch resolution.
>
>As for the laser idea, well, that's something I've been thinking about
>a lot lately, too. I recently dismantled a malfunctioning Laserjet
>III and obtained the laser diode scanning assembly; replacing the
>laser diode should be easy enough once Nichia and others start
>mass-producing them, although I'll need to ask HP whether the optics
>will pass UV light and focus it sufficiently.
>
>The challenges would be installing the scanner on a precision-motion
>shuttle with some sort of stepper motor for moving the laser down the
>PCB as each line is "drawn" (maybe using the original LJIII stepper),
>correctly spacing the optical assembly for accurate to-scale imaging
>for several thicknesses of copper-clad board, and controlling the
>whole deal based upon image or gerber files. It would also be nice to
>know whether the laser scanning assembly is actually capable of
>1/600th of an inch (or better) spot size focusing.
>
>It would be necessary to slow down the laser scanning from that used
>in the original LJ III; with a 2 mW laser and reasonably sensitive
>(e.g. 50 mJoule/cm^2) photoresist, covering an entire board could take
>more than an hour (25 seconds per square centimeter, give-or-take).
>Of course, if one used higher-sensitivity resist, one could go faster.
> It would be nice to be able to adjust speed to compensate.
>
>I guess if I make it work, I'll let you guys know :)
>
>Jonathan
>

the hp3d has 300x300 dpi.
i don't think this would work.
but it would really be nice to simply print to a normal lj.

(in a laser printer, how is the laser deflected in the horizontal?)

you would also have to adopt all the paper sensors for x/y exposing.

regards
stefan

>
>
>Be sure to visit the group home and check for new Bookmarks and files:
>http://groups.yahoo.com/group/Homebrew_PCBs
>
>To unsubscribe from this group, send an email to:
>Homebrew_PCBs-unsubscribe@yahoogroups.com
>
>
>
>Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>
>

Using an LCD instead of a transparency saves you only the very small step of
printing the transparency. All the optics and alignment issues remain pretty
much the same, you simply replace a printed transparency with a LCD image.

See the following URLs for discussion of home-built LCD projectors:
<http://www.hommie.net/>
<http://www.louisville.edu/~wchall01/diyprojector/>
<http://www.diyprojection.com/>
<http://diyaudio.com/forums/forumdisplay.php?forumid=20>

----
Doug

On Fri, May 23, 2003 at 03:36:00AM -0500, Neil wrote:
> Alright, call me crazy, but an idea just occurred to me, and I can't see why
> it can't work. To me, the most difficult (expensive, complicated,
> unreliable, etc) part of creating a PCB is getting the pattern laid out on
> the board so it can be exposed, etc. The rest is science ... standard
> procedure, standards times, standard results. Yes, laying out the pattern
> can be science, but at a higher price, and still takes time to line up the
> image transparencies, etc.
>
> So this idea just randomly came to me ... why not use a projector to project
> the image on the board. Use a positive-coated board and an XGA LCD
> projector, with a UV bulb instead of the usual bulb. I get quite good
> results at 300dpi, so if we go with a projector resolution of 1024x768 for
> now, that's about 3.4" x 2.5". There may need to be some optical
> manipulation done to get it to focus clearly at that distance, and a bunch of
> other little things sorted out (such as calibration for
> non-linear/inconsistent pixel spread), but I can't see why it can't work.
> No, it's not cheap, but I believe it would be really fast, going from
> computer straight to PCB.
>
> Maybe this has been thought of or done already? Or maybe I'm going off the
> deep end?
>
> Cheers,
> -Neil.
>
>
>

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

>
> the hp3d has 300x300 dpi.
> i don't think this would work.
> but it would really be nice to simply print to a normal lj.
>
> (in a laser printer, how is the laser deflected in the horizontal?)
>
> you would also have to adopt all the paper sensors for x/y exposing.
>
> regards
> stefan
>

Actually, the resolution depends upon a lot of things. It might have
been limited by the optics, or it might have been limited by the
control circuitry (e.g. memory for the bitmap, processing speed, etc.).

The laser is deflected horizontally by a servo-motor-driven spinning
mirror (I think it's a hexagonal mirror in the LJIII). If I recall
correctly, the speed is set by a clock crystal, and the laser is
sychronized by sensing when it reaches a certain point of deflection
(at that point, it's reflected with a small mirror into a fiber-optic
cable that leads to a photosensor on the control board).

At first, I was thinking that the original LJIII control circuitry
could be fooled into doing all of my work for me by adapting the
paper-path mechanism to move the laser scanner down the length of the
PCB board, rather than its normal activity of moving the paper through
a printer. I haven't given up on this notion, but it might be easier
and more robust to simply develop my own control interface for
managing the scanning mechanism and possibly do the image-to-bitmap
conversion in the attached computer.

Incidently, I understand that you can buy machines to do stuff like
this, along with ultra-sensitive photoresist to speed up the exposure
process. The biggest problem, of course, is throughput: it's simply
faster for mass production right now to do everything with a mask.

Jonathan

thank you for that information.
i did know there is a laser but had no idea how it is deflected.
i thought the fiberoptic cable is for adjusting the intensity of it, but
possibly it is for both - sync and intensity.

i think it would be possible to convert it, but i would try to move the pcb
under (or over, which might be easier but then you can't mount small pcbs)
the laser. assembly.

the biggest problem with it i think is th uv laser, its flimsy power needs
high sensitivity resist (which requires darkroom?). also it will not work with
the lenses in the unit, and not with the fiberoptic sync.


is it true there are pretty standard flatbed plotters on the market which
have a fiberoptic cable to the
pen holder with a lcd shutter in it?
and a more powerful laser on the side?
if you are good in programming, and can get your hand on a powerful laser
you may also manage to use a real lcd screen (small) with some resolution to
directly use the gerber apertures and make flashes on the board, not
substituting every pad with lines.

i believe i have seen that somwhere but am not sure.

regards
stefan

> --- In Homebrew_PCBs@yahoogroups.com, Stefan Trethan
> <stefan_trethan@g...> wrote:
>
> >
> > the hp3d has 300x300 dpi.
> > i don't think this would work.
> > but it would really be nice to simply print to a normal lj.
> >
> > (in a laser printer, how is the laser deflected in the horizontal?)
> >
> > you would also have to adopt all the paper sensors for x/y exposing.
> >
> > regards
> > stefan
> >
>
> Actually, the resolution depends upon a lot of things. It might have
> been limited by the optics, or it might have been limited by the
> control circuitry (e.g. memory for the bitmap, processing speed, etc.).
>
> The laser is deflected horizontally by a servo-motor-driven spinning
> mirror (I think it's a hexagonal mirror in the LJIII). If I recall
> correctly, the speed is set by a clock crystal, and the laser is
> sychronized by sensing when it reaches a certain point of deflection
> (at that point, it's reflected with a small mirror into a fiber-optic
> cable that leads to a photosensor on the control board).
>
> At first, I was thinking that the original LJIII control circuitry
> could be fooled into doing all of my work for me by adapting the
> paper-path mechanism to move the laser scanner down the length of the
> PCB board, rather than its normal activity of moving the paper through
> a printer. I haven't given up on this notion, but it might be easier
> and more robust to simply develop my own control interface for
> managing the scanning mechanism and possibly do the image-to-bitmap
> conversion in the attached computer.
>
> Incidently, I understand that you can buy machines to do stuff like
> this, along with ultra-sensitive photoresist to speed up the exposure
> process. The biggest problem, of course, is throughput: it's simply
> faster for mass production right now to do everything with a mask.
>
> Jonathan
>
>
>
> Be sure to visit the group home and check for new Bookmarks and files:
> http://groups.yahoo.com/group/Homebrew_PCBs
>
> To unsubscribe from this group, send an email to:
> Homebrew_PCBs-unsubscribe@yahoogroups.com
>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>

--
+++ GMX - Mail, Messaging & more http://www.gmx.net +++
Bitte lächeln! Fotogalerie online mit GMX ohne eigene Homepage!

--- In Homebrew_PCBs@yahoogroups.com, stefan_trethan@g... wrote:

> thank you for that information.
> i did know there is a laser but had no idea how it is deflected.
> i thought the fiberoptic cable is for adjusting the intensity of it, but
> possibly it is for both - sync and intensity.
>
> i think it would be possible to convert it, but i would try to move

the pcb

> under (or over, which might be easier but then you can't mount small

pcbs)

> the laser. assembly.
>
> the biggest problem with it i think is th uv laser, its flimsy power

needs

> high sensitivity resist (which requires darkroom?). also it will not

work with

> the lenses in the unit, and not with the fiberoptic sync.

The Nichia UV laser diode has a rated output of 2 mW. Now, that's not
a whole lot of power, but when none of it is lost to the sides or to
other frequencies (like it is with standard UV lamps), it can go a
long ways. Standard photoresist, from what I've seen, requires a
minimum exposure of around 50 millijoules of UV energy per square
centimeter (give or take, depending upon the photoresist), which
translates into 25 seconds of 2 mW-power UV light per square
centimeter. This is certainly manageable for reasonable-sized PCB
boards -- not quick, but not horrendously slow, either. If one uses a
higher-sensitivity photoresist (e.g. 5-10 mJ/cm^2 sensitivity, which
may have been designed specifically for Laser Direct Imaging), the
time can be decreased dramatically.

I'm more and more intrigued by this idea. It may not end up being the
fastest or cheapest way to homebrew a PCB, but it sure would be
interesting :)

Jonathan