Homebrew_PCBs

Short version:

10W ~380nm LED exposes Riston in about 60 seconds at 30cm.
A 3W one would be less expensive and require about 3 minutes.

Previously I used a 500W quartz halogen linear incandescent lamp (with
aluminium reflector) as my source of UV for exposing Riston. However,
it radiates a lot of visible and infra-red and so heats up the PCB and
the phototool at the distances I was using it at: about 40cm, with a 4
minute exposure. This could cause problems with the phototool expanding
at a different rate to the PCB and so messing up the image. I described
this on 2012-09-03:

http://tech.groups.yahoo.com/group/Homebrew_PCBs/message/30595?var=0&l=1

including the ~7:1 contrast ratio phototools which I made with a Brother
laser printer onto a particular laser transfer film.

Riston, such as the MM540 I get from the Czech Republic:

http://www.ebay.com/usr/gaminn
http://www.tech-place.com/en/photosensitive-materials/23-photosensitive-film.html

is exposed with near visible UV light. According to the datasheet:


http://www2.dupont.com/Imaging_Materials/en_US/assets/downloads/datasheets/mm500series.pdf

the "peak response" of the material is 350nm (nanometre) to 380nm. This
is the same as in the other Riston datasheets I looked at, but I did not
look at them all. The shorter the wavelength of light, the more energy
per electron is coupled, so 350nm light is more energetic than 380nm.

UV LEDs are now obtainable in high power versions. A 1 watt LED is a
single LED chip and can be bought for a few dollars. 3W LEDs may have
three chips in parallel. 10W LEDs have three in series, in parallel
with two other sets of three in series.

I found a wholesaler in China which evidently is source of many high
power LEDs sold individually by eBay merchants. This wholesaler has
minimum order quantities such as 10.

In their 10W High Power LED section they list the following UV wavelengths:

365nm
380nm
390nm
400nm
410nm

The shorter the wavelength, the less light they put out, the more they
cost and the harder they are to find.

This is page for a 10W 380nm LED is evidently for the same LED I bought
from an eBay retailer:


http://www.leds-global.com/ultra-violet-380nm-high-power-led-modules-p-7.html


This is item number G-P10UC140A1-XT.

It is specified to run at 1050mA at which it will have 10 to 12 volts
across it. It will then produce 400 to 500mW of light in the 375 to
385nm range.

The nine chips are arranged in a 3x3 array with outside edges of 5mm, so
this is a nice small light source which should give sharp edge shadows
on the photoresist even if the phototool is not pressed into direct
contact with it.

I bought the LED for USD$50 including shipping to anywhere:

http://www.ebay.com/itm/120896174810

It arrived here in Melbourne Australia 9 days later with signed for
delivery and a hand-written "Made in China" note. This is a Hong Kong
retailer:

http://www.ebay.com/usr/lucky_guy2010
http://stores.ebay.com/lucky4u2bid/

It is titled "10W UV Ultra Violet 380~385nm High Power LED Light for
Recognize Banknote". There's no mention there of a part number.

Running a few mA through it I see a rather broad spectrum with violet -
like a pastel violet. I assume I can hardly see the real peak of
emission, so I won't be looking at it when it is running properly.

I attached it a heatsink with small fan from a PC video display card. I
found an 18 to 19 volt regulated switch mode power supply and made up 9
ohms of power resistors to wire in series with the LED. This is one way
of making a reasonably good constant current source. I got about 1 amp
going to the LED, including a few tens of milliamps for the small 12V
fan motor. This takes the LED voltage nearly to 10 volts. So it is
running pretty close to its recommended current and voltage.

I found the ideal exposure time with 30cm distance, as before through
6mm of glass and the laser-printed phototool, was around 60 seconds.

20 seconds resulted in inadequate exposure - some photoresist remained
but not enough. 40 seconds produced a good image. So did 60 seconds
and 100 seconds.

This is at least a 2:1 exposure latitude. Many people think that laser
printed phototools are inadequate for exposing photoresist. Perhaps
with low-contrast photoresist (maybe some positive or non-Riston
negative photoresists) this may be the case. However, I find this
Riston MM540 is high contrast. This contrast enables the approximately
7:1 contrast ratio laser-printed phototool to work well, without being
excessively fussy about exposure time. "7:1" means that when I use a
photodiode with visible light to see how much light gets through the
dark part of the phototool, I find 1/7th the light than comes through
the clear parts. The dark parts are not entirely even, so the contrast
between the clear sections and the lighter parts of the dark parts is
probably 4:1, 5:1 or 6:1. Anyway, it is still high enough to get good
results with Riston.

- Robin http://www.firstpr.com.au/pcb-diy/

A correction: the exposure times I tested were 20, 40, 60 and 80
seconds, not 100 seconds. I got good results at 40, 60 and 80 seconds -
a 2:1 exposure range.

- Robin

At least some usable data about led's.
So 10W led at 30cm need 60 seconds, That's seems little long for
photoploting. But maybe isn't all lost as I thinking to put led way
closer. Maybe just few milimeters (to have rom for aperture whell) or
few centimeters if I need to put some lenses in betwen..
And 6mm thick glass should block a lot of light in that wavelength. I
think it's worth to try.

On 09/17/2013 06:08 PM, Robin Whittle wrote:
> Short version:
>
> 10W ~380nm LED exposes Riston in about 60 seconds at 30cm.
> A 3W one would be less expensive and require about 3 minutes.
>
> Previously I used a 500W quartz halogen linear incandescent lamp (with
> aluminium reflector) as my source of UV for exposing Riston. However,
> it radiates a lot of visible and infra-red and so heats up the PCB and
> the phototool at the distances I was using it at: about 40cm, with a 4
> minute exposure. This could cause problems with the phototool expanding
> at a different rate to the PCB and so messing up the image. I described
> this on 2012-09-03:
>
> http://tech.groups.yahoo.com/group/Homebrew_PCBs/message/30595?var=0&l=1
>
> including the ~7:1 contrast ratio phototools which I made with a Brother
> laser printer onto a particular laser transfer film.
>
> Riston, such as the MM540 I get from the Czech Republic:
>
> http://www.ebay.com/usr/gaminn
> http://www.tech-place.com/en/photosensitive-materials/23-photosensitive-film.html
>
> is exposed with near visible UV light. According to the datasheet:
>
>
> http://www2.dupont.com/Imaging_Materials/en_US/assets/downloads/datasheets/mm500series.pdf
>
> the "peak response" of the material is 350nm (nanometre) to 380nm. This
> is the same as in the other Riston datasheets I looked at, but I did not
> look at them all. The shorter the wavelength of light, the more energy
> per electron is coupled, so 350nm light is more energetic than 380nm.
>
> UV LEDs are now obtainable in high power versions. A 1 watt LED is a
> single LED chip and can be bought for a few dollars. 3W LEDs may have
> three chips in parallel. 10W LEDs have three in series, in parallel
> with two other sets of three in series.
>
> I found a wholesaler in China which evidently is source of many high
> power LEDs sold individually by eBay merchants. This wholesaler has
> minimum order quantities such as 10.
>
> In their 10W High Power LED section they list the following UV wavelengths:
>
> 365nm
> 380nm
> 390nm
> 400nm
> 410nm
>
> The shorter the wavelength, the less light they put out, the more they
> cost and the harder they are to find.
>
> This is page for a 10W 380nm LED is evidently for the same LED I bought
> from an eBay retailer:
>
>
> http://www.leds-global.com/ultra-violet-380nm-high-power-led-modules-p-7.html
>
>
> This is item number G-P10UC140A1-XT.
>
> It is specified to run at 1050mA at which it will have 10 to 12 volts
> across it. It will then produce 400 to 500mW of light in the 375 to
> 385nm range.
>
> The nine chips are arranged in a 3x3 array with outside edges of 5mm, so
> this is a nice small light source which should give sharp edge shadows
> on the photoresist even if the phototool is not pressed into direct
> contact with it.
>
> I bought the LED for USD$50 including shipping to anywhere:
>
> http://www.ebay.com/itm/120896174810
>
> It arrived here in Melbourne Australia 9 days later with signed for
> delivery and a hand-written "Made in China" note. This is a Hong Kong
> retailer:
>
> http://www.ebay.com/usr/lucky_guy2010
> http://stores.ebay.com/lucky4u2bid/
>
> It is titled "10W UV Ultra Violet 380~385nm High Power LED Light for
> Recognize Banknote". There's no mention there of a part number.
>
> Running a few mA through it I see a rather broad spectrum with violet -
> like a pastel violet. I assume I can hardly see the real peak of
> emission, so I won't be looking at it when it is running properly.
>
> I attached it a heatsink with small fan from a PC video display card. I
> found an 18 to 19 volt regulated switch mode power supply and made up 9
> ohms of power resistors to wire in series with the LED. This is one way
> of making a reasonably good constant current source. I got about 1 amp
> going to the LED, including a few tens of milliamps for the small 12V
> fan motor. This takes the LED voltage nearly to 10 volts. So it is
> running pretty close to its recommended current and voltage.
>
> I found the ideal exposure time with 30cm distance, as before through
> 6mm of glass and the laser-printed phototool, was around 60 seconds.
>
> 20 seconds resulted in inadequate exposure - some photoresist remained
> but not enough. 40 seconds produced a good image. So did 60 seconds
> and 100 seconds.
>
> This is at least a 2:1 exposure latitude. Many people think that laser
> printed phototools are inadequate for exposing photoresist. Perhaps
> with low-contrast photoresist (maybe some positive or non-Riston
> negative photoresists) this may be the case. However, I find this
> Riston MM540 is high contrast. This contrast enables the approximately
> 7:1 contrast ratio laser-printed phototool to work well, without being
> excessively fussy about exposure time. "7:1" means that when I use a
> photodiode with visible light to see how much light gets through the
> dark part of the phototool, I find 1/7th the light than comes through
> the clear parts. The dark parts are not entirely even, so the contrast
> between the clear sections and the lighter parts of the dark parts is
> probably 4:1, 5:1 or 6:1. Anyway, it is still high enough to get good
> results with Riston.
>
> - Robin http://www.firstpr.com.au/pcb-diy/
>
>
>
> ------------------------------------
>
> Be sure to visit the group home and check for new Links, Files, and Photos:
> http://groups.yahoo.com/group/Homebrew_PCBsYahoo! Groups Links
>
>
>
>

UV transmission of glass and acrylic.
UV LEDs for photoplotting direct to Riston.

Hi Slavko,

I think this exposure time is short. Looking at the way various items
in the room lit up with flurorescence, I guess this LED (really 9 LED
chips in a single package) is putting out 50% at least of the light
produced by an old (they have been around for decades) 40W linear
fluorescent "black light", though I haven't fired mine up for years.

"Ultra violet" covers many wavelengths. The UV at the central emission
band of these LEDs is about 380nm, which is very close to the short
wavelength limit of human vision, variously quoted as 390, 400nm or
whatever.

The transmission curves at:

http://www.rayotek.com/techincal_info_glass_sapphire.htm

show that soda-lime glass (the basic mass produced glass I am using - it
was the protective cover on the 500W floodlight) does not reduce
transmission until the wavelength gets to 300nm and below.

The transmission curve for acrylic (PMMA) falls of much closer to the
visible edge. This chart:

http://www.goodfellow.com/larger-quantities/polymers/tpx-characteristics/

shows acrylic (PMMA) falling off around 400 to 405nm.

I think you you could use a microscope objective lens and a 1W single
chip ~380nm LED to expose Riston. However, I guess the exposure time
would be something like a second or a fraction of a second, which is
probably too long considering that you need to expose tens of thousands
of ~10um to ~30um pixels one after another.

If you were using an aperture wheel then you would be exposing more of
the PCB at a time, so such exposure times might be OK.

For small pixel exposure of the Riston, I think a UV laser of the
correct wavelength - I guess 360 to 390nm - when focused by a microscope
objective lens would be much better, since the laser delivers all its
light in a beam, some or most of which can be used by the objective
lens. Then, I guess, depending on the power, you might get the exposure
time down to a millisecond or so. This would make it suitable for
direct photoplotting onto the Riston photoresist. However, you would
have some challenges keeping the focal point of the lens at exactly the
right height. If your lens had a very wide angle, the depth of field
would be too small to get good resolution unless your PCB was flat down
to 20um or so (which it surely won't be, unless you vacuum-held it to a
genuinely flat base) and if the geometry of your plotter was totally
even and aligned to this plane with similar accuracy.

I am perfectly happy with the resolution I am getting with laser printed
phototools and with the ease of getting sharp images in the Riston from
these. Since I am only doing small hand-drilled PCBs - single sided so
far - I don't need to worry about slight scale errors which are no-doubt
present in the laser printed phototools.

If I did want to make larger PCBs for NC drilling, then absolute
accuracy would be important and a laser and lens based exposure system
would be required.

- Robin

On 2013-09-18 3:00 AM, Slavko Kocjancic wrote:
> At least some usable data about led's.
> So 10W led at 30cm need 60 seconds, That's seems little long for
> photoploting. But maybe isn't all lost as I thinking to put led way
> closer. Maybe just few milimeters (to have rom for aperture whell) or
> few centimeters if I need to put some lenses in betwen..
> And 6mm thick glass should block a lot of light in that wavelength. I
> think it's worth to try.

A further correction . . . The LED I bought was advertised

http://www.ebay.com/itm/120896174810

as:

380 to 385nm
1000mA
9 to 11V
900 to 1100mW output

actually has a peak output at 394nm. I measured it with a monochromator
from an old spectrometer. The monochromator reads correctly within 1nm
for both a 532nm green laser and for the 404.7nm mercury emission line,
so I think my reading is reasonably accurate. Half power was +/- 6nm.

When I wrote my first message I thought this was probably the same as an
identical-looking device for which there is a little more data:


http://www.leds-global.com/ultra-violet-380nm-high-power-led-modules-p-7.html

375 to 385nm
1050mA
10 to 12V
400 to 500mW output

However, there is a serious mismatch in the specified output.

This LED runs at almost 10 volts with 1 amp, which matches the eBay
specification exactly, but would be at the low end of the range for the
p-7 device above. The shorter the wavelength the higher the voltage, so
this is consistent with the eBay specification being for a LED with a
longer wavelength than "380 to 385nm".

I think the eBay LED is a closer match for this one:

http://www.leds-global.com/uv-10w-390395nm-high-power-led-p-146.html

390 +/- 5nm
1050mA
9 to 10V
800 to 1000mW

I don't have a chart for the wavelength sensitivity of Riston - I was
referring to:


http://www2.dupont.com/Imaging_Materials/en_US/assets/downloads/datasheets/mm500series.pdf

which states that the "peak response" of the material is 350nm
(nanometre) to 380nm.

I was keen to get a high powered LED with as short a wavelength as possible.

While I may have paid more than I needed to for a 395nm LED, the good
news is that Riston seems to respond just fine to 394nm.

Searching eBay for UV LED 10W 395nm I find various cheaper LEDs, but
395nm is not their center wavelength - they are typically 395 to 405.

There are likely to be uncertainties regarding the actual wavelength of
LEDs vs. how eBay seller advertise them. However, if a 3 minute
exposure is OK at 30cm, which is fine for small PCBs, then I think a 3W
390 to 395nm LED would be a good, inexpensive, choice.

For instance, this supposedly 3W 390 to 395nm LED:


http://www.ledfedy.com/products/1-500w-led/3w-led/3w-uv-led-390-395nm-double-chips-with-star-pcb-1492.html

costs USD$4.69 - not counting postage.

A LED like this would be easy to run - just get a 12 regulated power
supply and choose a resistor which drops (12 - 3.3) volts at 0.7A = 12
ohms. The LED would need a heatsink and the resistor would dissipate
about 6 watts.

- Robin

I had ordered
http://www.ebay.com/itm/221248115201?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1497.l2649

and this should match my photoresist. They are marked as peak
sensitivity at 400nm. So this led should bje just right (and blueray
laser at 405 too)

Will check...

On 09/18/2013 01:06 PM, Robin Whittle wrote:
> A further correction . . . The LED I bought was advertised
>
> http://www.ebay.com/itm/120896174810
>
> as:
>
> 380 to 385nm
> 1000mA
> 9 to 11V
> 900 to 1100mW output
>
> actually has a peak output at 394nm. I measured it with a monochromator
> from an old spectrometer. The monochromator reads correctly within 1nm
> for both a 532nm green laser and for the 404.7nm mercury emission line,
> so I think my reading is reasonably accurate. Half power was +/- 6nm.
>
> When I wrote my first message I thought this was probably the same as an
> identical-looking device for which there is a little more data:
>
>
> http://www.leds-global.com/ultra-violet-380nm-high-power-led-modules-p-7.html
>
> 375 to 385nm
> 1050mA
> 10 to 12V
> 400 to 500mW output
>
> However, there is a serious mismatch in the specified output.
>
> This LED runs at almost 10 volts with 1 amp, which matches the eBay
> specification exactly, but would be at the low end of the range for the
> p-7 device above. The shorter the wavelength the higher the voltage, so
> this is consistent with the eBay specification being for a LED with a
> longer wavelength than "380 to 385nm".
>
> I think the eBay LED is a closer match for this one:
>
> http://www.leds-global.com/uv-10w-390395nm-high-power-led-p-146.html
>
> 390 +/- 5nm
> 1050mA
> 9 to 10V
> 800 to 1000mW
>
> I don't have a chart for the wavelength sensitivity of Riston - I was
> referring to:
>
>
> http://www2.dupont.com/Imaging_Materials/en_US/assets/downloads/datasheets/mm500series.pdf
>
> which states that the "peak response" of the material is 350nm
> (nanometre) to 380nm.
>
> I was keen to get a high powered LED with as short a wavelength as possible.
>
> While I may have paid more than I needed to for a 395nm LED, the good
> news is that Riston seems to respond just fine to 394nm.
>
> Searching eBay for UV LED 10W 395nm I find various cheaper LEDs, but
> 395nm is not their center wavelength - they are typically 395 to 405.
>
> There are likely to be uncertainties regarding the actual wavelength of
> LEDs vs. how eBay seller advertise them. However, if a 3 minute
> exposure is OK at 30cm, which is fine for small PCBs, then I think a 3W
> 390 to 395nm LED would be a good, inexpensive, choice.
>
> For instance, this supposedly 3W 390 to 395nm LED:
>
>
> http://www.ledfedy.com/products/1-500w-led/3w-led/3w-uv-led-390-395nm-double-chips-with-star-pcb-1492.html
>
> costs USD$4.69 - not counting postage.
>
> A LED like this would be easy to run - just get a 12 regulated power
> supply and choose a resistor which drops (12 - 3.3) volts at 0.7A = 12
> ohms. The LED would need a heatsink and the resistor would dissipate
> about 6 watts.
>
> - Robin
>
>
>
> ------------------------------------
>
> Be sure to visit the group home and check for new Links, Files, and Photos:
> http://groups.yahoo.com/group/Homebrew_PCBsYahoo! Groups Links
>
>
>
>

Hello....

After long time I got time and stuff to check thing about this topic.
I had some liquid photoresist (similar to positiv 20) produced localy. I smear it to test board and dry it.

After that I put 50mW/405nm laser on stand to get same height all the time and focused to have spot under 0.1mm. The shape of laser is not round as if I move in one direction the width is 0.06 and perpendicticular is 0.98mm. So I just moved by hand test sample under the beam with aprox 2cm/s and 5cm/s. After developing I got clear trace. So system works and speed can be higher too.

If I got laser out of focus to have 0.5mm spot the photoresist is not developed.


In same sample I do 3W UV led test too. I just put LED few mm over the board (had some plastic with 8mm hole under) and does "flash test" with 0.5, 1,2,3,4,5,10,20,30 second turned led on. Only 0.5s sample is not developed! the 1s has blury edge all other are perfect!!! So with little focusing should be nice too.

So both things works.
for laser the width is under 0.1mm (need litle better lenses that my module have) and to correct astigmatism. At my opinion this system is best suited for raster scanning of pcb.

Led can develop 8mm thick line with no problem but need lower speed. With vector scanning this method should work faster. As only "gaps betwen traces" should be developed it should be quite fast. (at least my boards have all unused space connected to ground polygon).

Need to got some lenses with wery short focal point to try to "colimate" LED.

And I do yet another test. I buy UV solder resist (green) to and do same test. Both Laser and UV led does polimerize the photoresist!! So same system can do both. (and same timing seem to be needed!)

Does someone know how to colimate led?

At 07:59 PM 17-11-13, you wrote:

>After that I put 50mW/405nm laser on stand to get same height all
>the time and focused to have spot under 0.1mm. The shape of laser is
>not round as if I move in one direction the width is 0.06 and
>perpendicticular is 0.98mm.

To get the beam round you have to expand it with a lens, to pass the
expanded beam through a round aperture and to re-focus it.
Of course, you will lose a lot of power: 0.98 divided by 0.06 = 16.3
times is the amount of power you'll lose.
BTW: you have to choose a laser with better x/y ratio or to try
rounding a LED beam.
Cristian

On 11/18/2013 07:25 AM, Cristian wrote:

> At 07:59 PM 17-11-13, you wrote:
>
>> After that I put 50mW/405nm laser on stand to get same height all
>> the time and focused to have spot under 0.1mm. The shape of laser is
>> not round as if I move in one direction the width is 0.06 and
>> perpendicticular is 0.98mm.
> To get the beam round you have to expand it with a lens, to pass the
> expanded beam through a round aperture and to re-focus it.
> Of course, you will lose a lot of power: 0.98 divided by 0.06 = 16.3
> times is the amount of power you'll lose.
> BTW: you have to choose a laser with better x/y ratio or to try
> rounding a LED beam.
> Cristian
>
>

Actualy I don't know what aproach is better. Both seems to work, and
both have specific problems.
I know that I can make laser beam round with aperture in widened beam.
But maybe there is option to use some cylindrical lense too. If that
works then there is no power loss. But finding the proper lense can be
problematic.

And LED is little problematic too as light is not collimated. And I dont
have any lenses on stock to try to make beam more paralel. In the led
there is one chip of aprox 2x2mm and covered with transparent dome. I
think to get most of them I need lense with short focal length to
capture most of led. Maybe the gless ball is the right one (I don't have
it to test).
Just need to find the source of lenses and make few tests...

At 12:11 PM 18-11-13, you wrote:

>
>
>I do not think you want to use an aperture to get a round beam since
>it will waste power.

Expander-round aperture-focusing is the easy way for an amateur.

> As you suggested, cylindrical lenses will allow you to selectively
> focus X & Y independently.
>
>Mirrors curved in one plane will also work.
>
>Bertho

Curved mirror will do the job, but I think is hard to find.
If you know a cheap 'one piece' producer, let me know, please.
I'd like to round-up a red laser.
Cristian