Need help design UV Curing heads

[vizion]

Hi

I am working on a project which means I need to build another tool for my cnc. It is an led UV curing head and have no idea how to design the circuit or what components would be best.

The project is in two parts - part 1 being complete

.Part 1.
A UV adhesive dispensing tool mounted on my cnc & laying a 0.15 diameter bead on the upward facing cut edge of a number of 2mm thick sheets of glass.

A sheet of 2mm thick clear glass is then rested on those edges and now I need a UV lamp tool to mount on the cnc and shine through the clear glass to set the adhesive below it. hence PART 2

Part 2.
The UV curing light needs to be 365nm
The recomended power requirement to set the adhesive is 100mw/cm^2
The cnc can reliably keep a UV head 2mm above the top glass sheet
The speed of travel of the lamp controlled can be using the cnc electronics so that the adhesive gets the appropriate exposure time. A short cure time is prefered so it is a question of p[acking as much power as practical into a small space bearing in mind there are changes of direction!
Microswitches to turn on/off the curing light
I want to keep the head as small as possible preferably using no more than four tightly packed LEDS.
Can anyone recomend components and an appropriate circuit and power supply?
I am used to building these thing but am pretty dumb when it comes to circuit design and I know the UV leds are v. expensive so do not want to wreck them!
#

Thanks in advance for any help

david

 

[Arouse1973]

To work out the mW/sr of the UV device we first need to know the distance from the LED to the work piece. Any calculations you do will depend greatly on this distance so you must give this some thought first.
Adam

 

[vizion]

The distances are given but not in one piece.

In part 1:
"A UV adhesive dispensing tool mounted on my cnc & laying a 0.15 diameter bead on the upward facing cut edge of a number of 2mm thick sheets of glass."

A sheet of 2mm thick clear glass is then rested on those edges

(this means the adhesive is pressed between the upward facing cut edge and the 2mm sheet)
In part 2
"The cnc can reliably keep a UV head 2mm above the top glass sheet"

So in one statement it means:
The minimal distance from the head to the adhesive is 4mm with the light travelling through 2mm of air and 2mm of clear glass.
The head could be held further away but I thought as this is spot curing the closer the better.

David

 

[Arouse1973]

Ok thanks. Yes closer is better obviously. I need to have a look at this because glass is reflective and will not let all of the radiant intensity of the light through it. The attenuation needs to be calculated.
Thanks
Adam

 

[vizion]

One further issue is how to meet the need for cooling

 

[Arouse1973]

Ok so I have had a go and as you would expect it's a bit more complex than you might think. I have given an example of a unit distance of 1 to indicate 1 unit of solid angle (Steradian) This can make approx. validation easier as without any losses.

Example:

A light source at a distance of 1m producing a irradiated area of 1m^2 with a output of 40 mW/sr will impart a radiance of 4uW/cm^2. This is because the area of 1m^2 is 10*10^3 cm^2 (10,000) So you then divide the original power by this to give you 4uW/cm^2.

I have done a similar example but with losses. So if you take these calculations and put them into excel you will be able to fiddle with them to get the desired irradiance. You may find it easier when you scale down to still keep the irradiance as 1 sr. This is where the Area is equal to the distance (or radius of your circle) squared. But as the distance from the object gets smaller things are no so accurate.

You will have to suck it and see I am afraid and I recommend getting an LED a bit more powerful than you need so you can adjust the current if needed.

The losses I have include are for Infrared through 0.25cm tinted glass. I couldn't find anything for UV, only that UV-A passes through glass well. So if you are able to find them then just change the constants AF, Nair and Nglass. I will see if I can refine this over the next few weeks but this is the best I can come up with at the moment. I hope it helps.

Adam

 

[vizion]

This is brilliant impressive work. I cannot thank you enough.

My next query is what led(s) Awould be suitable and a circuit. The I can experiment.
Can anyone identify a suitable LED? Also cooling is going to be important.
Heat sink enough?david

 

[Arouse1973]

Pacer
http://www.analog.com/en/switchesmultiplexers/analog-switches/products/index.html
Adam

 

[Arouse1973]

This is brilliant impressive work. I cannot thank you enough.

My next query is what led(s) Awould be suitable and a circuit. The I can experiment.
Can anyone identify a suitable LED? Also cooling is going to be important.
Heat sink enough?david

I liked your post so much I have asked one of my distributers to get involved and confirm my findings. They are one of the leading experts in optos, Osram. Lets see what they come back with.
Adam

 

[Arouse1973]

I must admit 100mW/cm seems quite small. I was thinking more of 4 to 8 Watts.
Adam

 

[Fish4Fun]

Hey David,

There are a bunch of choices in 365nm LEDs ranging from a few mW right on up to 100W....I don't have any experience with UV LEDs, but I have a fair amount of experience with LEDs in general....Working on the assumption that UV LEDs are similar to other LEDs, first thing I would suggest is that you "de-rate" the published output, especially if you go with China-Cheap LEDs....Most LEDs produce more than 50% of the light they are capable of producing with somewhat less than 50% of their rated current....in the case of the "white" LEDs I have worked with I typically find that I get considerably more light from three 3W LEDs in parallel driven @ 3W (1/3 of the rated current per LED) than I do from running two @ full output....and the three 3W LEDs in parallel create far less heat than even a single 3W LED run @ full power.....The LED life is also considerably longer if run at lower current....

The next thing that needs to be addressed is cost of materials vs cost of machine time AND number of units to be processed.....I am assuming this is a fairly large production run since you are going to the trouble to use a CNC machine, and I am also assuming that you would like to run the machine as fast as is practical....if both of these assumptions are correct then maximizing the amount of UV should minimize overall costs....If the glue is run in straight lines then you might consider one of two things 1) Making a rotating head with a "line" of UV LEDs spaced so that the output is overlapping and fairly consistent, OR 2) If the glue lines are pure X-Axis runs followed by pure Y-Axis runs, you might create a "Cross Pattern" so that there are two "lines" one would be on for Y-Axis Travel, the other would be on for X-Axis travel....this might help with cooling the LEDs.

You might also consider simply having a line of LEDs the entire width of one Axis, Let's say the X-Axis....in this case once the glue has been laid and the tops pressed on, a simple run from Y=0 to Y=Max at a particular feed rate would cure all of the glue in all of the parts in a single pass....This might prove faster and more reliable than attempting to "chase the glue lines'....You might even be able to use Fluorescent tubes instead of LEDs, UV "Grow lights" are available in various lengths and wattages and might be more cost effective than LEDs...

Fish

 

[vizion]

OK what you say makes sense but looking at the practical requirements I am suggesting the following approach - if it can be made to work (big IF?)

With the setup I have got and the variety of shapes I want to cure I am suggesting a spot system with up to 4 LEDs at the tip.

A. The outside diameter of the led holder would ideally be no more than 12mm for 4mm
B. The next section(moving away from the tip) reduce to 10mm diameter for about 80mm length to conveniently slot into an existing vertical motion holder on the cnc carriage.
C. The final part of the head could be enlarged to as much as 15mm.
Some detail of my thinking :

A , primary function - secure the leds, with the part turned from aluminium (up to 12mm diamand 4mm long) (male thread to female B) or pressed onto B
B. primarily function - a heat sink (again may be turned from aluminium) with the power feed to the leds running through it. Made of a cylinder 10mm diameter 80mm long. B again threaded or pressed onto C
C. Primary function - holds additional circuitry and connect the head assembly to coiled flexible power feed cables bring the that convey power from a transformer. The coiled flexible cables need to be about 1200mm long with the transformer receiving its input from 220-240v supplies.

I have no idea how to design the led holder so heat is efficiently transferred to the heatsinks. The heatsinks will hopefully be of sufficent area to dissipate the heat into the air.

So now what LEDS to choose and what circuit is required? It would be great to get away with one or at most three LEDs. That would make assembly easier and more compact.

 

[vizion]

I should have mentioned that tubes are not as good at delivering their power at a precise wavelength (in this case 365nm) and create much more heat. so they are not really suitable for spot cure as LEDs. They also mean more weight and directing the light from tubes to generate a spot source is very complex involving design of light tubes.

A spot system will also reduce the amount of stray light and facilitates hooding which, in combination, may be important safety considerations.

Does anyone have any experience of heatsink design for LEDs - incidentally it would be easy to have additional heat sink clamp onto part C!

 

[vizion]

Would this LED be suitable?
http://docs-europe.electrocomponents.com/webdocs/0fdf/0900766b80fdfb20.pdf

 

[Arouse1973]

I think it's worth a try. You will probably need a small PCB made and the device assembled correctly to the PCB so any heat can get away as soon as possible. You might be able to find one on a PCB already that you just connect power to. Manufacturers have been combining heat sinks into PCBs for quite some time now. The substrate is aluminium instead of FR4.
Adam

 

[vizion]

Hi
It would be great if there is something available off the shelf but failing that and bearing in mind my incompetence in that direction would anyone be willing to design/build and perhaps mount the chip?

Also necessary to make sure heat sink works!

The circuit will be relatively simple - apart from maybe a simple overheating protection and current protection (against shorting or incorrect polarity) all that would be needed is some on/off circuitry triggered by microswitches (need to think of being able to use multiple pairs) and a specification for the power supplies.

Willing to pay if someone is willing to design & build!

BTW I should mention I am in Bristol

 

[Arouse1973]

And I am in Cornwall I know a few people who will place the part down for you. It might be more cost effective to have more than one made. Let me look into this for you.
Adam

 

[Arouse1973]

This the sort of thing you need.

http://www.ebay.com/itm/3W-365nm-UV...t-High-Power-bead-with-20mm-pcb-/221264871940

Adam

 

[Arouse1973]

Some more
http://www.ebay.co.uk/bhp/led-pcb

 

[vizion]

They look interesting but everyone in china is on holiday until 28th!

Maybe I will talk to the guys in the states tomorrow and see whether they have an off the shelf solution. The prices for the US LEDs are a lot higher but I may be able to finish up more quickly.

It seems RS components have their chip at:
http://docs-europe.electrocomponents.com/webdocs/0fdf/0900766b80fdfb20.pdf
in stock so I could order one and pick it up on tuesday.

It looks as though that should work well but I must admit I am very nervous about assembling it correctly.

Can anyone tell me how I should go about it or, if it is atall complex, would offer to do the assembly for me?.

David