PCB UV exposure box

[AdeV]

Some time ago, I started to make a UV box to expose PCB photoresist (and possibly other UV-curing applications, e.g. diazo process screen printing, of PCBs and/or front panels, and the like).

Un-seen by the camera, I soldered dozens of UV LEDs I got cheep from fleaBay onto a couple of Eurocard sized strip-boards, spending many happy hours getting the damn things the right way around, in the right holes, and all connected up...

Later, I spend many hours designing and breadboarding a supremely over-complicated timer control unit, complete with rotary encoder for programming, and individual control of the top and/or bottom light panels. Then everything went into hiding for a few months while the office was plastered/painted, until a Round Tuit got unearthed the other day...

So, finally, a couple of reasonably productive days in the office resulted in a pile of precision-ish parts (see exhibit 1):

• A piece of glass sized to exactly somewhere around 110x160mm
• Aforementioned circuit board covered in UV LEDs
• four aluminium sides, the two long sides with grooves cut 0.8mm down x 2mm wide x 2mm deep (for the glass) and a bit lower x 1mm deep for the light board

These were then carefully-ish assembled, and after much fettling, more or less went together as expected per photos 2 & 3.

Finally, the assembled box was popped into the case I prepared for it much much earlier (by stripping most of the foam padding out - mistake!), lashed up to the breadboard circuit, and tested. Works well! See photo 4.

Now I just need to remember to bring my special PCB tracing paper down to the workshop tomorrow and I'll have a crack at actually developing a PCB!

So... things I have learned so far:

• Never, ever, try to work with glass! What an evil material! Copper rod is a joy by comparison... Note the cracks in the glass (getting bigger all the time), some acrylic has been ordered to replace it.
• My milling machine has a problem with its brake. Currently, it'll only work with the centre bearing cap removed  Another fixer-upper project in the offing.
• Did I mention not working with glass?
• Wickes cheapo tile cutter is actually reasonable on glass, but it's definitely best not to work with it. The glass OR the tile cutter.
• Always always always use the proper tapping head on the mill when power tapping. Don't ask me why (see photo 3...).

[awemawson]

Beware that acrylic sheet is available in UV transmitting and UV blocking versions as I understand it.

(mine uses glass sheet and UV fluorescent tubes)

[tom osselton]

I have a carbon arc platemaker that was used for printing would that work for that blue resist paper?

[AdeV]

Hi Tom,

Interesting device... it'd probably do the job, I expect the exposure time would be only a few seconds!

Aaanyway, PCB experimentation time was limited to this afternoon, as the water supply was frozen down here this morning. Once the pipes had thawed & no bursts were apparent, it was time to crack on. See attached photo...

#1: Clearly badly over-exposed (that was 4m 30s). Tracks are missing and the big pad has a bit missing too.
#2: 2nd attempt, this time with 2m 30s. Better, but I'm still losing tracks in places.
#3: Aaah, now we see what's going on. This one is under-exposed... except where the tops of the LEDs are.
#4: Another attempt, I've lifted the board up on a temporary structure (and managed to finish breaking the glass  ) This one is nearly usable, except for the side tracks are still virtually gone.
#5: Last go tonight on a little offcut I had left over. Much better - somewhat shorter exposure time (2m 30s again). A smidgeon of resist has resisted dissolving, so I may need to go to 2m 45s or even 3m to get the final effect. Ignore the slightly blurry looking tracks on the RHS, the board was slightly curled up due to the way I cut it, and I didn't dare weigh it down properly.

On the bright side, my home-made Sodium Meta-silicate (aka water-glass) developer worked just fine, if a tad slowly. For hobby stuff, I think slow is probably better than fast, though. Next job, once I've re-made the box & put an acrylic top on it (hopefully all arriving in the next couple of days), is to mix up some Cupric Chloride acid to etch the copper 

[seadog]

..."mix up some Cupric Chloride acid to etch the copper"

Don't you mean ferric chloride?

[AdeV]

So... thinking about this scientifically  - on the underexposed one, the overexposed white blobs are about 7.4mm diameter. The tops of the LEDs are about 4.8mm diameter, and the lens sits approximately 7.5mm above the PCB which is 1.6mm thick. The light board slot is 16.2mm below the glass guide slot, giving a total lens-to-glass clearance of (drumroll) [wait, I need the calculator for this....] 7.1mm. Ish.

So... cranking that all around the ole' grey matter (with a piece of paper and this post as a reference guide!), we can arrive at the conclusion  that the light has spread from the top of the LED by 2.3mm in total, i.e. 1.15mm on each side (if a circle had sides). So now it's a simple  geometry job to find the angle of spread to find the minimum height difference needed to give full coverage & no dark spots - although without a frosted lens, it'll still be a bit brighter in a grid pattern, unfortunately, unless I mount the LEDs in the other room....

So, the angle of the dangle times the mass of the ass equals the heat of the meat - as they say, but actually we need a bit of Pythagoras here, so we'll go with that instead. In this case, TOA: Tan = Opposite over Adjacent, or 1.15/7.1 = 0.162ish; arctan of that is 9.2 degrees. Since that's only half the angle, double it and add 2(ish) to get about 18 degrees. Which is actually slightly better than the 15 degrees I vaguely remember the data sheet said; but that could be measurement error.

Last but not least, we need to cover the whole area with light - and now, my maths basically runs out, so it's google o'clock to find out what the minimum gap I need between lens & workpiece to get full coverage. Then, once I've fixed the blasted spindle brake, I'll make some new slots in the box, machine up an acrylic top to replace the shonky glass, and try again!

For a laugh, see the attached calculation sheet...

[AdeV]

..."mix up some Cupric Chloride acid to etch the copper"

Don't you mean ferric chloride?

Nope - I haven't got any ferric chloride; and given my production rate, I'd need to buy more every time I wanted to etch a PCB. Cupric Chloride is a mixture of HCL + Copper (and a little peroxide to get things going initially), has an unlimited shelf life, and just occasionally needs a top-up of acid to keep it going. It's a bit slower than ferric, but a shedload less messy, and better still I've got all the components on the shelf to make a batch 

[Brass_Machine]

Hi Ade!

Nice looking solution so far.

I have been away slightly because I have been working on shiney stuff... IE Jewelry. I just made a UV bath for curing UV Epoxy. Not sure how it would work for you. It does have set timers... 120 seconds, 180 seconds and 30 minutes.

I took 2 inexpensive UV nail cure machines that are used for curing finger nails. They are U shaped. Connected them together to form a 360 degree UV bath.

Eric

[AdeV]

Hey Eric,

Looks interesting! I presume you drop a glass/plastic container holding your cleaning solution + dirty things in there to be nuked/teleported? 


Anyway, I did some back-of-the-envelope Googling/maths, and arrived at the conclusion that I need ~ 38.5mm between the top of the glass and the LEDs to get 100% definite coverage. This is fine in the bottom of the box, where I have nearly 70mm height to play with; my current design has the lid section only 40mm deep. So... I need to revise that slightly & split the difference between the lid & the base. But for now, I'll just try to get it working with the one side. Hopefully.... this weekend coming, as I'm back to work tomorrow.

[tom osselton]

Just thinking could you not oscillate the led rack to manage the uv coverage?

[Brass_Machine]

...
Looks interesting! I presume you drop a glass/plastic container holding your cleaning solution + dirty things in there to be nuked/teleported? 
...

Would be interesting... Nope, so a ring gets sealed with the UV epoxy. Get's dropped in the UV bath for 30 min to cure.

[John Swift]

 Hi Ade

in the past Elektor magazine did a euro size LED light box I found on the net a long time ago

    john

PS
it started with  page 64 of the  May 2006 issue of  elektor electronics

[philf]

Hi Ade,

You can get a double sided UV box on eBay for £20 which has a timer and will do 6ft x 2ft boards:



Happy New Year.

Phil.

[awemawson]

Does it come with a free Tanned Lovely as a bonus 

[AdeV]

LOL! I've got enough space problems as it is... but.... so big and shiny!

Enough, stop tempting me to buy stuff!

Anyway... the acrylic showed up the other day, so I spent  bit of time trimming the box to perfection(ish) so it screwed together square. Threw out the old glass, and expanded its slot to 4mm to take the acrylic, also cut a new slot much further down for the LEDs. Cut the acrylic to size, and put it all back together, voila! It looks much better now, and goes together much more securely. No more wobble! Good job I left that extra 15 thou on the length for adjustment 

[AdeV]

As Andrew said somewhere up the thread, one of the things you have to watch for with plastic is whether it cuts out UV or not. Fortunately, there's a nice simple test - common or garden tonic water fluoresces a lovely cyan colour in UV light. So, pop the light box back on the timer, fire it up, and check it out with a bottle of the (apparently delicious) Fever Tree tonic water.

Sure enough, a lovely light blue glow. I think I'll be OK with this.

Now to go cut up some more circuit board and do the test over again!

[PekkaNF]

That is good to know. I need to check that trick one day. I have affinity to Fever tree tonic and I got bitten by acryllic sheet that would block over 95% of UV - no good for PCB work.

Pekka

[AdeV]

That is good to know. I need to check that trick one day. I have affinity to Fever tree tonic and I got bitten by acryllic sheet that would block over 95% of UV - no good for PCB work.

Yep - I can't remember where I read it, but it's a great check.

NEWS: I just made my first PCB!! There were one or two problems.... actually, just one broken trace, although it looked like there should be more dead ones  so I'm having another go now.

Pictures soon..., because I know that 

[AdeV]

Sooo, after a few experiments, 3rd time lucky! (pic 1)

After etching... one track has undercut (pic 2). Oddly, the tracks that LOOK broken are fine, and one that looks fine isn't. I tried to fix it with some solder, but no joy.

Decided to drill it anyway (pic 3), silly me forgot to check the drill bit size and the holes are a little too large, but there looks to be enough pad left to solder to. Too late to start now, that's tomorrow's job. I'll bridge the broken track with a tiny bit of wire.

So, if the soldering goes well, then the jury will be back in! The provisional verdict so far is:

• The LEDs are now doing the job nicely
• The drilling machine works fine; it's a bit of a chore setting it up without any alignment marks, but I pretty much got there OK
• Cupric Chloride works just fine as an etchant. A bit slow, perhaps, but maybe I didn't make it strong enough, I'll try a new batch another day.
• Will it solder....

[AdeV]

So, update...

Nope! I think I drilled a bit too much of the pad away... that and the copper oxidised too quickly, or maybe there was a hint of photoresist left on it, or something. Basically, the solder refused to stick properly.

So... I'll make it all again next weekend; in the meantime, I'm off to buy some tinning solution!

[AdeV]

I'm still waiting on my tinning crystals, so with nothing else PCB-ish to do immediately, I decided to have another go at making the board. To better my chances, I thickened up the tracks to 0.6mm, and rather than trying to carefully stick two prints together to thicken up the black ink, I gave it a go with just a single sheet... I also made some fresh waterglass (plenty of it) and etchant.

So... after a number of foxtrot umbrellas, I have the exposure time down to 1m 15s. The time in the developer was, approximately, 2 seconds (!) - I think I may have made it a smidgeon too strong this time, after making it too weak last week - The fresh acid etched the board down in just a few minutes....

...Success!

A MUCH better board, and this time all the tracks are fully connected. I'm going to tin it before I drill it this time, so no further progress today; if I get too impatient I might pop down mid-week to have a go at it 

[AdeV]

Tonight's episode: Amateur Hour in the PCB Factory!

Earlier today the tinning chemicals showed up. Since I happened (coincidence, honest) to be out of work early, I popped down to the workshop to have a play, as one does. I've bought enough tinning powder to make 500mls of solution, but since it only lasts 6 months when made up, I chose to only make 50ml. Weighed exactly 9g of powder into 50ml of near-boiling water, per the instructions; let it cool somewhat, then dunk the board in & watch some YouTube videos while I waited. After about 20 mins I took it out; it was tinned, but quite dull in colour.

It did polish up a little bit with some cloths, but I had to take it too far, didn't I? I very VERY lightly ran it across a sanding sponge I have, which promptly removed most of the tin... So, back in the bath for another couple of Marty's Matchbox Restoration videos (recommended if you like watching ancient tiny metal cars being resurrected from near-destruction back to better-than-factory), I removed it again and gave it the obligatory cold & hot shower, per the instructions. This time, I just buffed it as best I could with paper towels - it looks OK. Not as shiny as I'd like, I might give it a touch of a polish with some polishing compound, I think I got some with my dremel. I'll do that later.

Next job - carefully measure the X/Y offsets of the first pad from the edge of the board. Stick board to bed. Mount 0.6mm drill (so I get to keep more of the pad this time). Using the milled slot, carefully set the drill height so it doesn't drill any more holes in the plastic baseboard.... (duuuh). Then, one careless Y-move and  SNAP! Bye-bye 0.6mm drill. 

Undeterred, I used the 0.5mm drill anyway, results as per photo. I do have another 0.6mm drill somewhere, I think it's at home. If I can't find it, I think I'll re-drill the holes at 0.7mm, as the 0.5 looks a bit thin to take any component legs. Also, I'm not 100% sure I've drilled all the way through, and I don't want to move the board until I've made absolutely sure.

In the pic, I've stuck the board to a piece of scrap (a previous failed effort), using my aluminium fences to align it. It's not too bad, although the holes at the top of the board are slightly off-centre, probably because the board's not exactly square cut. Mea culpa, but that's what you get for having to cut all four sides by hand... I've had good results using a paper guillotine (the lever type) to cut circuit board, except it likes to move as you start the cut. In future, I'll probably stick the board to the cutter using double-sided tape, that way I should get a really good and dead-square cut. Not sure if it'll tear off the photoresist protective layer, though... some experimentation is required...

That's it for this week, I'll pick this up again on Saturday, which will hopefully involve finishing the re-drilling to the correct size, soldering the board up, and giving it a test run.

[AdeV]

So... thanks to Merseyrail (for abandoning me at Hamilton Square, 50 minutes walk from home, no taxis to be seen...), and my mate Ted who sublets part of my workshop... I ended up back here 

Further, thanks to Amazon, who delivered a nice new box of 50x PCB drills... so I recommenced drilling with the 0.6mm bit 

A little later, upon discovering that even 0.6mm is a real squeeze for some of those component legs, I opened them even further to 0.7mm. Which I could have drilled with my original set. 

Aaaanyhow. A little later, with a little help from some 50+ year old Fry's rosin flux, it's complete. And, even better, IT WORKS!!! Photos show back,side,  front side, and front size powered up... which isn't terribly exciting I know, but hey!

Hmm, a wave soldering machine would have made it MUCH easier to solder 

The End......



... or is it?   

[philf]

So Ade,

Now you've built it, what does it do?

[WeldingRod]

It's an optocoupler, right?

Sent from my SAMSUNG-SM-G891A using Tapatalk