Etching PCBs

[dominicbeesley]

Hello all,

Just a quick tip for anyone who like me has been put off from making their own printed circuit boards by the price of UV light boxes. Last week I thought I'd have a go but instead of a fancy light box I built my own.

I got a 5W UV LED Star from Farnell (167-8962). And stuck this to half the carcass of an old computer PSU to use as a heatsink and stand, this holds the LED a few inches above the workbench and gets warm but not too warm after about 10 mins. The pool of light is big enough to cover a 4"x4" (10cm x 10cm) board a couple of small blocks of wood would lift it high enough to do a bigger board.

Adjusted for 1200mA current through a 1R resistor the exposure time at 6 minutes was possibly a tad too high (4 would probably do it) for my rather leaky inkjet negatives but a good board was made (after a couple of false starts due to getting the negatives the wrong way round).

As usual I dived in and designed my own double sided board and this came out ok, a few adjustments next time (bigger pads for less than accurate drilling!) and it would be spot on.

Whatever you do watch your eyes though. These things are very very bright and will probably do you some serious damage. I managed to briefly flash it whilst looking directly at it and had a dose of welders eye the next day! so a nice light-tight box over the top would be an good idea. I can just close the door to the workshop and switch on/off from outside!

Dom

[Steve Anderson]

Dom,

Good thinking! Presumably this could double up as an UV EPROM eraser too if you're still using those devices. No idea what the 'dosage' would be, trial and error I guess. At some twenty quid plus each I would insist it did double-duty!

Steve A.

[dominicbeesley]

You know I'd not thought of that! I'll have to have a rummage I'm sure I've got some old UVPROMs somewhere then I could muck about with the Flash programmer I made to program one - I keep meaning to make a test card generator!

I'm sure with a bit of shopping around you could find one for less than £20 and it's worked far better than I'd hoped so a 1W one would probably do the trick!

Anyway far better than the cheapest Farnell eraser £90 (bulbs £47!) or £140 (bulb £18) for their smallest PCB exposure kit its a bargain! Doesn't look quite so professional but the boards do though!

Dom

[Steve Anderson]

Doesn't look quite so professional but the boards do though! Dom

The tasting is in the eating.

Steve A.

[Klaas Robers]

I fear the LED will not work as an EPROM eraser. To erase EPROMs you need a UV-C source. I have a small lamp that does the trick. It is a gas discharge lamp with transparent glass (quartz?) bulb. It runs directly on 230V and I mounted it inside a cake form. EPROMs on the table, cake form over them and the lamp on a mechanical timer clock. Erasing takes about 15 to 20 minutes. I never tried it for PCBs.

[DrZarkov]

A friend of mine used an old facial tanner from a bootmarket, he took all the electronics, and built it into the empty case of an old flatbed scanner. It works perfectly, and costed together about 5 EUR.

[Steve Anderson]

To erase EPROMs you need a UV-C source.

Yes, correct. But quoting an ST datasheet for a very typical 27C256, "The erasure characteristics of the M27C256B is such that erasure begins when the cells are exposed to light with a wavelength shorter than approximately 4000 A (Angstrom)."

Further on.."The recommended erasure procedure for the M27C256B is to expose it to short wavelength light which has a wavelength of 2537 A."

4000A=400nm, 2537A=253nm. It goes on to say sunlight or even fluorescent lighting will (in time) also erase the device, hence the use of opaque labels over the quartz window for long-term installation.

The UV wavelength of these LEDs peaks at 400nm so it is quite probable that erasure would happen. Depending on the intensity it may take a bit of time though. As I suggested it's worth a bit of experimentation.

But with EEPROMs around why use EPROMs? No awkward programming voltage required which varies from device to device and no sitting around waiting for the things to cook.

Steve A.

[dominicbeesley]

"But with EEPROMs around why use EPROMs?" - For the sheer hell of it!

As Steve says sunlight will erase them...I remember we used to line them up on the window sill in my mates house and leave for about a week - the sun's not too bright up here in Yorkshire.

Anyway I've got about 20 Flash EEPROMS I got from ebay a while back - trouble is they've all got their "boot block" write protected and I can't work out how to unprotect them! (I think it's easier with high voltage but that will mean making another programmer!)

Ho hum....

Dom

[Steve Anderson]

Anyway I've got about 20 Flash EEPROMS I got from ebay a while back - trouble is they've all got their "boot block" write protected and I can't work out how to unprotect them! (I think it's easier with high voltage but that will mean making another programmer!) Ho hum.... Dom

I've not used Flash devices myself but reading a few random datasheets for them it appears that once a 'boot block' has been protected that's it. You cannot even do a chip erase, the command will simply be ignored. The only reference to a voltage over 5V I've found so far is an Identification Voltage. But there may be devices that can be programmed/erased once the 'boot block' has been protected, but not those in the sample I looked at.

The only disadvantage I've found with EEPROMs is that generally they're not available in such large sizes. About the largest EEPROM that is 'home construction' friendly is 4MB (512k x eight), whereas EPROMs and Flash devices come in much larger capacities.

But in the context of NBTV there's not much need for them.

As so often happens I've taken this thread well and truly off-topic!

Steve A.

[dominicbeesley]

Some others EEPROMS I have there is a "secret" invocation that unlocks the lock I think they were AMD ones. Trouble is I can't find the secret for these other ones

On these ones Pm29F002T-12PC there is a 12V programming system and a 5V. I think the main reason for the protection is that they're used as the BIOS in PCs. The boot block cannot be reprogrammed by accident when its in the PC as that only does 5V programming....

Anyway I'm on revision 3 of my PCB now, the first one I messed up, the second one was ok except for a couple of mistakes. The third one I've reworked to be a single-sided board but with larger solder pads and no crossings between pins - should be a bit easier to solder and work with. Registering the front and back on the double-sided board was pretty tricky and with tiny solder pads drilling the holes was a bit critical!

Dom

[Steve Anderson]

Some others EEPROMS I have there is a "secret" invocation that unlocks the lock I think they were AMD ones. Trouble is I can't find the secret for these other ones. Dom

I attach the datasheet for the AMD version of the 29F002. There might be a 'secret' way of unlocking these devices but it is of course not in the datasheet that I can see.

Often devices that require 12V or thereabouts for programming actually don't need it, it's to ensure once the device is in the target system with only 5V available it cannot be accidentally erased or altered. PIC microcontrollers are an example but they can be altered with only 5V available, but it's a long(ish) sequence of events to get there.

Well done on the PCB front, something I've never attempted myself before.

Steve A.

[dominicbeesley]

Thanks Steve,

I can't remember where I found the "secret" out - it's a bit like the sector/chip-erase sequence but it unlocks the chip. I'm not 100% sure it was AMD chips though.

As to 12V programming as far as I can see it's just a way of unlocking certain features without needing an extra pin. On the PIC the programming is similar for low-voltage but an i/o pin is lost.

Anyway here's a few pictures. As can be seen the pool of UV is nice and uniform, the boards aren't bad - though they look a bit scruffy after tinning! This could be improved if I'd not run out of flux so having to just use normal solder and lots of it.

The final thing to get everything perfect would be to improve the acetate printing. Currently the black areas are patterned - I've tried new ink cartridges and messing with the various paper settings but not found the best setting yet....Anyway it's "good enough" now if not perfect

Dom

[Steve Anderson]

Well, it all looks pretty good to me, agreed the tinning needs a bit more practice but all in all a good result. Something I've not done myself and don't intend to. Even with a Dremmel in a pillar stand drilling cannot be easy, ideally a small CNC machine would be the answer.

I found this one at Farnell UK, might be worth investigating at around 300 quid inc. VAT. (Stock No. 4140369). It appears to be a kit of parts and you have to supply the drill, so it's not quite 'plug & play'. It might be worth contacting Milford directly, maybe cheaper?

http://www.milinst.co.uk/

Steve A.

[dominicbeesley]

I'm far too tight - I'm doing all this on a (near zero) budget. I drilled them all out last night and out of all the holes I only got 2 more than 1/2mm out. If I make a decent amount of money this year I might get something like that - I was tempted by the milling machine kit in Elektor as well....

With holes, there seem to be two main problems:

1) Getting the drill to bite in the right place. I use a drill press at its highest speed. It's not too bad and there isn't too much float or rattle in the head. But I tinned the board before drilling! The main problem then is that the drill won't bite dot in the middle but bends to one side of the solder "blob". Better tinning or tinning after drilling would fix that. Not sure what order you're meant to do it.

2) The other problem is drilling a load of holes and keeping them in a nice steady line. For IC sockets, headers, etc. When I get the time I'm going to try and make up a little board holder and a slide so that I can just set the angle and slide onto the next hole. Notches every 1/10th" should then give me nice neat lines of holes.

Thanks for the encouragement though!

Dom

[AncientBrit]

Very neat work Dom.

"Dossytronics"?.
There must be a story there!

Cheers,

Graham