PCB, stands for printed circuit board, and if you're here, you hopefully know what one is. They are used in almost every piece of electronic equipment known, and their purpose is to route electrical signals between components, using copper tracks attached to an insulator such as fiberglass.

Fig 1. How etching works.....

The process of making one involves getting the blank PCB (fiberglass type board or paper phenolic with a thin layer of copper) and removing some of the copper by chemical or mechanical means.

By chemical, we normally mean etching. The board is coated (using a variety of methods) with a resist, the area without the resist is removed by the chemical etchant. There are a few methods of getting the resist on the board, you can buy one of those dalo pens (the lumocolor pens available at stationary shops also work) and draw the pattern straight on. The disadvantage here is that it is far from precise, it's not easily repeatable, and can be a major pain for anything but really simple designs.

If you're lucky enough to have access to a laser printer, you can print the artwork straight onto one of those iron on sheets, and stick it onto the PCB using a clothes iron. The toner from the printer contains polymers which can act as a resist. Some people still think this method is kinda iffy, and can take a few tries to get right (which is a rather expensive exercise).

The best way, and the most time consuming, is using the photographic method.

Now, first you need a few ingredients....

1. Etchant. The stuff of choice seems to be ferric chloride. I hate it, use ammonium persulfate, it comes in crystal form. It's odorless, not very toxic, but it needs to be heated, and can't be reused like ferric chloride.

   Fig 2. Ammonium Persulfate.

2. Presensitised PCB. For this tutorial we will be using the MG chemicals positive photoresist PCB. Inexpensive, and they work great.
3. Developer. This stuff develops your exposed PCB, and removes the resist where the light has gotten through. Apparently, depending on the type you get, it can be quite toxic, though I've never dipped my hand in the stuff...
4. Light source. You need something to expose the PCB under, a fluorescent light, or even the sun will work, just make sure the light is even.
5. 2x Glass plates. You sandwich the presensitised boards and artwork in-between them, keeps everything still and flat.

   Fig 3. MG Chemicals Developer (NaOH)

6. Artwork. You can use stick-on tracks and pads, but it's much better to use computer generated artwork, printed onto a transparency with a laser printer. Or print it with an inkjet and get it photocopied onto a transparency.
7. Ice-cream containers. Eat the ice-cream first. These, I've found, make great dishes to develop, rinse, and etch your PCB, but can be a little small. Just make sure your container is made from glass or plastic, etchants eat metal, that's the whole point.
8. Stirring and scooping implement. McDonald's plastic fork does the trick.
9. Measuring device of some kind that can do 10ml - 250ml or so.
10. Gloves, goggles, apron. To protect you from all the nasty chemicals that can really destroy brand-name clothing, oh and your eyes and vital organs too.
11. Means of drilling the holes. 1mm and 0.8mm bits and drill press or dremel that can accommodate them. You can get PCB drill bits that have a 1/8 or so diameter for securing into the drill with a smaller diameter near the tip.

Having seperate implements, measuring devices and containers is a good idea in the case of ferric chloride as it leaves a residue in glass and plastic containers, and the developer and etchant neutralise each other if they mix.

Bruce Robinson has kindly made a few additional comments, directly quoted ones look like this, thanks Bruce!

A note on using Ferric Chloride (etchant) and Sodium Hydroxide (developer) from Bruce....

Step 1

First you need a PCB design. I use the freeware version of Eagle, it works great, excellent libraries and simple to use. I've also used Protel 99SE, it's the big expensive one, and for the price, doesn't work terribly well, that said, it's very powerful once you've fugured it out. You may want to use the design from a magazine, download and print it off the Internet, or photocopy from the magazine.

For small simple stuff, I use MS paint at five times normal size. That's 600 pixels per inch. I keep a file of standard pads, traces, etc. To print it, I convert to GIF format (perhaps not necessary) and insert the figure into MS Word. Then I scale the inserted figure to 20%. This gives a much smoother finish to curves and angled lines than I would get by scaling the figure before inserting it.

A few tips on design....

• Make everything big, pads tracks, spacing, etc. It takes practice to get these things to work well, so you'll have the best chance of success if you start simple and work your way up.

• Avoid having power tracks go all over the place (especially in loops), creates noise problems.

• Programs sometimes assume that pads are to be thru hole plated (have metal surrounding the inside of the drilled hole), thus the pads are much too small, and once drilled, can have very bad connections. I've had up to 50% bad connections when this happened on my hexapod robot PCB and I didn't notice until it was too late.

• Keep parts inline, no weird angles, makes things much easier to troubleshoot

• Ensure there are center holes in the pads, it's near impossible to drill on target otherwise.

• If you are good at soldering, it may be a good idea to use SMD parts, saves hours of drilling holes.

• Proof it. Print a big copy on plain paper, compare it to the schematic, and make sure that every trace is there, ticking them off as you go. Check to make sure that there is adequate spacing between all pads/traces, if not, redraw them.

Step 2

Now we need to get the artwork onto a reasonable transparent material. Where there is toner/ink, that's where the copper will be (if we were using negative photofabrication, it would be the other way around). If you have a laser printer with good resolution and opacity, go ahead and print the artwork onto a transparency, tracing paper may also work.

Print quality is probably the single most important factor in making a good PCB. A good dark image with no smudging on the "white" areas will allow a lot of leeway in exposure, developing, and etching times.

Fig 4. Artwork not totally opaque...but good enough.

If you have an inkjet printer, print it at maximum resolution, and get it copied onto a transparency (costs about AUS 63 cents at the local copy center around here). If the resolution is inadequate, print the artwork at double size, and have it reduced 50% when you get it photocopied.

The artwork should be printed so that when it comes time to expose, the toner side will be against the PCB, you also have to take into account that since the traces will be on the opposite side of the components (usually), the artwork will need to be flipped over. The best way to test all this, is to just print a draft on plain paper, and try and visualize how it will all come together, using IC's and other parts to test it if you want. Using text on the PCB is also good (it will appear backwards if the transparency is flipped).

Fig 5. Copied onto a transparency.

Finally, weird things can happen with the scaling. I've used the photocopier trick and although normal parts went in fine, a 30pin header wouldn't as the artwork was slightly too large. Laser printers are reported to mess with the scaling too by up to a few percent. Make sure you measure the printouts.

Your observations on scaling are bang on. A Xerox repair man told the secret once upon a time. People often complain about the shadow line along the edge of a copy (happens when the original is not PRECISELY in the right place). So since most copied documents have a blank margin, the copier actually scales up the image by 1 to 2%, on purpose. This pushes the shadow line off the page. Unfortunately, with a big IC, that 1% works out to have the width of a pad.

Step 3

Get ready... The best way to do this is to get everything set up beforehand. It also allows a sort of production line setup in case you wish to produce a few PCB's at a time.

First set up the developer. This stuff is nasty apparently. Don't drink it, though I almost did in chemistry, who the hell invented the pipette anyway? The mixing ratio varies, but with the MG Chemicals developer (NaOH Sodium Hydroxide), it is 1 part developer to 10 parts water. I usually do 20ml developer to 200ml water, which seems to cover the bottom of the ice-cream container nicely...

Pour the concentrated developer slowly into the measured quantity of water. Not the other way around. Just like diluting an acid. If you do it wrong way around, the solution can spatter at you.

Fig 6. Measuring cylinder and beaker stolen from school science dept. Works for solids when measuring ratios by volume, but not terribly accurate.

Set up the etchant tank. Ammonium persulfate is supposed to be one part to five. I usually use a bit more, as it can take a long time, especially if the board isn't developed fully. It comes in crystal form, use your plastic fast food implement to scoop some into a container, which holds the hot water, and give it a stir. The water needs to be about 75 degrees Celsius.

You need a water bath to rinse the board in-between stages, saves running to the sink (and the chemicals are corrosive to some metals anyway). So, I have three ice-cream containers inline, first is the developer, next is the water bath, and finally, the etchant. It always helps to have the chemicals arranged in the order you use them, same idea as photography processing.

Fig 7. Ice-cream containers minus ice-cream.

Finally, set up the light source. I use a fishtank fluorescent tube (grow-light), as it is mobile, and provides reasonably even light coverage. It's mounted about 20cm above where the board goes. Play with the set up until you get something that works well. It's important that you eliminate stray light (you can work under a 40w incandescent lamp though), such as sunlight shining directly onto your work area. You can put it out in the sun to expose if you want, saves having to find a fluorescent light, but is somewhat more variable.

Step 4

Now for the exposure. Get your presensitised board, peel off the white protective plastic, lay it, copper (shiny green side) side up, and place your artwork, toner side down, over the board. Line it up, and place the second glass plate over the top. Turn on the fluorescent light. Start timing. I sure hope you read the instructions that came with your stuff. The MG chemicals board is supposed to be exposed for 5 minutes with the light source 5 inches away. You won't notice much of a change to the exposed PCB until you develop it.

Fig 8. Peel....	Fig 9. Place....	Fig 10. Expose...

Step 5

When nicely tanned (aka. 5 minutes or so), take the board out and drop into your already mixed developer. You have to gently agitate the developer, or use the little sponge-on-a-stick that comes with the MG Chemicals kit. There is no set time here, but you have to etch until the area around the tracks has no trace of that green and is a bright shiny copper color. I've found that even leaving a thin layer of that resist makes it impossible to etch, at least with ammonium persulfate. Tricky because sometimes a part of the board will be overdeveloped while the other underdeveloped (you didn't ensure your light source was even...). Once all the green resist is gone from the area around the artwork, remove the board (while wearing gloves) and place it in the water bath, rinse it, and inspect.

Step 6

Hopefully you'll have a nice bright copper colour around the green artwork, if it's like that where it shouldn't be (ie. over a track), you'll have to repair it. Let it dry fully, and use a resist/dalo pen (lumocolor pen works well) to draw in the fixes.

Fig 11. Get the idea?

Step 7

Now you're ready to etch. There's a little trick I use here to ensure that the water stays hot (ammonium persulfate only works while warm) - get another container that the etchant container can sit in, and put some almost boiling water into it. Float the etchant tray in this hot water. This way, the hot water keeps the etchant warm, and if it cools off, you can always top up the water, whereas it's a pain to make up more etchant.

You should agitate the etchant as well, either by gently shaking the container, or using a fishtank aerator to bubble agitate.

Fig 12. Double double, toil and trouble....

There is no set time, so you have to keep a close eye on the progress of the etchant. First the copper dulls and goes a dull pink color, and then the exposed copper starts to disappear, letting us see through to the substrate (phenolic, fiberglass, etc.).

Make sure all the copper not part of the artwork is removed, if there's a pesky area that just won't etch, you may have to leave it and cut it with a sharp knife to prevent shorts later.

Step 8

Inspection time, ensure there are no shorts, or gaps in the tracks of the circuit board. You can fix gaps by soldering some wire in, and get rid of shorts by cutting through the copper with a sharp knife.

Now we can drill. Use a 1mm for standard thru-hole parts, and 0.8mm for IC's. You can use a handheld dremel, but I highly recommend you use a drill press. Your drill press probably doesn't take drill bits that small, but you can always get a small keyless chuck that acts as an adapter.

The 1/8" shaft PCB drills are great. If you can't get them, be sure to etch small holes in the copper where you want to drill. That acts as a pilot hole for starting the drill and can save many, many broken bits.

You need to drill from the copper side (of course), with plenty of light to see what you're doing.

Leave the green resist on, it protects the board from oxidization, looks cool, and can easily be soldered through.

Step 9

All done! It may seem a little complex, but I've just tried to cover every little issue that could affect the end product, it's really not all that difficult.

Fig 13. Sample board, although not fully drilled yet.

This tutorial is © David Perry 2001.

Portions are the property of Bruce Robinson, also © 2001.