PCB fabrication.

Warning - this is LONG but bear with it.

Printed Circuit Boards, are a necessary and somewhat evil requirement to realising high performance when it comes to a lot of audio circuits. Without them, you are severely limited when it comes to a lot of the newer ICs, and as the electronics world moves further and further towards surface mount, options are ever decreasing when it comes to through-hole.

The goal of this guide is to hopefully remove any and all of the evils of PCB fabrication. The end resulting in a process that you're confident in and that will yield predictable and repeatable results.

This guide is going to focus on the process of light exposure using Fotoboard. There are other methods, but this is the only one I have used and as it's never let me down. As a result I've never seen a reason to try anything else.

Fotoboard works on the same principles as the traditional production of photographs, hence the name. In its simplest form a light-sensitive film is exposed to a flash of light, whereby the chemicals within the film are altered depending on how much light has landed on them. The film is then be submerged within another chemical, known as developer. Introducing the developer would set a chemical reaction in motion where the film would change colour based on how much light it had originally absorbed, creating a photo.

Throw an unexposed piece of Fotoboard into developer and the light-sensitive coating will change colour, green in the case of the stuff I use. However if you first exposed the board to some light, then threw it into the developer, the light sensitive coating would turn black and peel away from the board leaving bare copper beneath it. This obviously sets up a system that will allow you to create patterns of green and copper on the board. These patterns would be the tracks and pads of copper that form the basis for a PCB.

So how do you go about creating this pattern?

Fortunately the processes of PCB manufacture have spawned the creation of several CAD products aimed directly in assisting you with the task.

http://www.olimex.com/pcb/dtools.html

Has a good list of programs. Eagle tends to be popular, the free-ware version gives you enough functionality to make decent PCBs. The only real limitation is that the board size can't be bigger then 8x10cm. Worry not though as this is large enough for almost everything I've made.

I won't go into PCB design as that's an entirely different topic, needless to say, there is a wealth of information available on DIY Audio. Datasheets sometimes come with recommendations on how to design a PCB too.

Once you have created your design you now need to go about turning it into a real PCB.

You're going to need a few basic tools and some chemicals.

First, you need the raw PCB material itself. This you can most likely purchase from a number of the common electronic component suppliers. I know Farnell sells everything but they are a little pricey. I use http://www.megauk.com/ for all my PCB creation needs. The prices are very reasonable and they have great customer relations. As you can see there are lots of things you can buy to ease your life when it comes to making PCBs. You will also notice, as you browse their website that these things are rather expensive! Don't worry though, you don't need to purchase any of them for first-time success.

http://www.megauk.com/pcb_laminates.php

Now you've found a place to buy materials from, what exactly is it you need? As you can see there are a lot of options. What you're after here is the Fotoboard (2) Pre-sensitised 'FR4' Copper Board

The most obvious requirement is the size of the board. Don't buy too little, this is important as cutting/preparing the boards can use up more then you'd think. If this is your first time, you're also going to need some extra to experiment with.

The next option is single or double-sided? This may start alarm bells ringing for increased complexity and surely making double-sided is a lot harder and risk-prone than a single-sided? Well not really, once you know what you're doing. The only thing single-sided has going for it is cost. Double-sided, on the other hand, gives you tremendous flexibility when it comes to board design and will usually result in a much better end product, if only for the fact that you can use one side as a ground plane. This is crucial if you want to achieve the specifications that a lot of the digital surface mount components carry.

Next is the thickness of the copper. Lower resistance is always better. One point along a PCB trace is not electrically the same as 2cm further along that same trace. This is simply because the trace carries its own internal resistance. Thicker copper reduces the difference between two points and obviously reduces the overall resistance of the entire trace. It increases the maximum current that the trace can carry and it also offers better cooling for surface mount components.

The final property is the thickness of the board itself. This isn't nearly as crucial as the others but can bring about some benefits. First of all the board is stronger and less flexible. The second benefit is inherent capacitance. In it's most basic form a capacitor is two conductors, (usually metal plates), with a nonconductor, (usually called the dielectric), between the metal plates. Sound familiar? Well, it should do. The entire double-sided PCB makes up a simple capacitor, the two layers of copper foil with the epoxy resin in the middle. Stray capacitance is bad with high-speed electronics and increasing the thickness of the middle reduces any capacitance set up by tracks on the top coupling with the ground plane, or traces on the bottom.

Personally I buy the 2oz 1.6mm double sided 18x12” boards. This may seem like a lot, and it is, and I don't recommend you buy too much. The sensitivity of the photo-sensitive layer does reduce over time. I've still had success with FR4 that's a few years old, but you do notice the extra time it takes to develop.

After having completed your PCB design, using whatever CAD program you've chosen, you need to turn it into a format that's useful. This simply means printing it out onto a transparency. If your printer has an economy mode now's the time to turn it off. You want the black part of your design to be as dark and impenetrable to light as you can make it. If your printer has exceptional paper handling you might be able to put the printed design through again to thicken up the ink layer. My printer does not, so I don't, it doesn't affect the final PCBs too badly, but it does allow more margin for error.

When printing you often have the option to print a 'mirror image' this is important. For the best results, you want the ink layer, printed on the transparency, pressing against the light-sensitive film on the raw PCB material. What you don't want is there to be a layer of transparency between the ink and the film. Why? Because light can get in. The plastic the transparency is made of has a thickness and it might not be much, but light can find its way in. So what you don't want is Ink > Transparency > Film. What you do want is Transparency > Ink > Film. So print a mirror image if you have to. Usually, this means printing a mirror image for the copper top layer and leaving the copper bottom layer as it is.

An example of the copper top and copper bottom printed onto transparencies using a laser printer.



Now you've printed your design out and have two transparencies, one for the copper top and one for the copper bottom. Now what? Simple!

First, you might want to lay the two transparencies on top of one another to check that they line up well. There's no reason that they shouldn't but it's worth checking.

Remember to flip the copper top over. You should end up with Copper Top transparency>Copper Top ink>Copper Bottom ink>Copper Bottom transparency. This way both sides of the PCB will press directly up against the ink layers.



Now what? Take a pair of scissors and chop off a strip of the transparency on the copper top. This should leave the copper top say 5-10cm shorter than the copper bottom.

Now lay the two transparencies back on top of one another, line them up really well and tape the top in place as in the next photograph.



You should now be able to fold the copper top back and put it back down in place, the tape keeping them both aligned. Make sure you can do this so that they keep alignment if they don't try taping them again. Once you're satisfied the next stage is how to go about securing the PCB in place.

The next picture shows how.



You want to fold back the copper top, then place two strips of double-sided sticky tape as the photo shows. Obviously you leave the backing on the top side

At this stage, you're ready to start chopping up PCB material.

The next picture shows roughly how large the PCB needs to be in comparison to the image on the transparencies. So get out a ruler and a pen and get marking up that PCB material ready to chop it up. A good hacksaw should have no problems easily cutting the raw sheets. A band-saw also makes light work of it but will make an awful noise so use earplugs or ear defenders.



If you're going to use a hack saw you're going to want to clamp the raw sheet in place. This is best accomplished like the next picture.



You want to clamp the raw sheet between two pieces of MDF or wood, or anything else that's suitable. The protective sheet that covers the light-sensitive coating on the PCB material must not be perforated in any way, otherwise light will get through and ruin that which remains.

After you've chopped everything to size the ends of the boards will look something like...



This is obviously unacceptable, the protruding edges will push up the transparency, stopping it from lying flat against the PCB. So you must file down the edges.

When doing so you want to move the file in the direction the next picture indicates – press down and move the file to the right. Do not pull the file to the left otherwise, you can cause the protective coating to pull back a little, or worse, cause the copper layer to bend back a little.



After you've filed, the edges should look much better. Something like this.



Now your board has been cut to shape and prepared, you're almost ready to expose the PCB.

You are going to need something flat to place the PCB on. I find this works well if the supporting thing beneath it has a little bit of play, ie it can move a little. I use a phone directory for this. The next picture shows exactly how you want things to look when exposing.



Here you can see the phone directory underneath everything. On top of this, you've got the transparencies with the raw PCB material inside.

The next important thing is a sheet of glass. This can be anything of an appropriate size. This sheet was from a cheap picture frame, anything will really do here. I'd just make sure that the edges of the glass aren't sharp as you don't want to cut yourself. The sheet of glass is merely there to keep everything pressed tightly together.

If you happened to have a rough protruding edge of copper left on the raw PCB material, the glass wouldn't lie flat against the PCB and the transparency would/could lift off a little. This really does create a shadow beneath and causes you to have blurred tracks, widening them in the process. If you're trying to make a PCB capable of working with SSOP surface-mount packages, this will ruin any chance of success. So check that everything lies perfectly flat. This is also why you want the ink layer pressing directly against the PCB and not with the transparency material in-between.

(Obviously, I've left the backing on the PCB material and not stuck it to the transparencies, this is just to illustrate how everything should look).

The exposure stage is the most vital part of the whole process. Whereas everything else also needs to be done right, if the exposure isn't good enough or goes wrong in any way, the rest of your efforts will count for nothing.

First and foremost you need to do this in a dimly lit room, so when you pull the backing off the raw PCB material, the light-sensitive coating will remain shielded from any high levels of light. If it is exposed to something you've already shot yourself in the foot. So close the curtains and turn any nearby lights off. Or do this at the dimming of the day.

Next is the light source. Ideally, this needs to be bright and provide an even covering of light over the area that it will be illuminating. I have found that energy saving/florescent bulbs work far better than the standard incandescent fair.

I suffer somewhat from SAD syndrome during the winter and also have trouble sleeping as a direct result of not getting enough light during the day. To this end, I've got a lightbox. This thing is very bright and provides a nice even light.

In the next picture, you can see how I have it positioned relative to the phone directory. Yes, I had it turned on for this and it doesn't look that impressive. But bare in mind this was taken at something like 1/500th of a second and as a result the rest of the surroundings are very dark, showing really how bright this is. Only half of the tubes were on also.



The first thing you will notice is how close it is to the phone directory. The lightbox is supported by 4 soup cans (yes, hi-tech I know!), so there are only a few centimetres between the lightbox and the top side of the phone directory. Having the lightbox close to the exposed PCB keeps the light intensity very high and reduces the exposure time dramatically. Also you will notice that the light is very even right across from one side of the phone directory to the other.

If I were using a standard bulb, rather then tubes, it would be bright directly beneath the bulb, but the intensity would drop off towards the edges of the phone directory. If you're only making a small PCB this won't be important, but if the board size is significant the differences in light levels at the centre to the outer edges would alter the optimum exposure time across the span of the PCB. This isn't ideal and moving the bulb further away from the PCB will reduce the effect somewhat. It will obviously result in a greater exposure time.

Reading all that probably sounds like you're doomed to succeed, but that's far from the truth. Before I had the lightbox I made successful PCBs using two 100 watt equivalent energy saving bulbs as the light source.

After assembling or choosing a light source what you need to do is have a play around. Cut up some small squares of PCB material and experiment. Try exposing the small squares for different amounts of time and remember to use a proper PCB design whilst doing this. When I used the energy-saving light bulbs exposure took around 10 minutes, but this is largely dependant on how far away you have them, anywhere between 5-20 minutes should work fine. The lightbox I've got only needs 1 minute 40 for an exposure, which is a great time saver I can tell you that.

The other crucial part to exposing is the developing process. It is better to have the developer too weak then too strong. If it's too strong it will go way too fast and you stand the risk of removing even the parts you want to keep.

To start out mix up the developer to the strength directed by the instructions that come with the chemicals.

(Once upon a time there were two types of board. One was far more sensitive to light and I think was directed at fine quality work, ie for surface mount. This required you use half-strength developer. In the last couple of years, I think the older type of PCB has been largely abandoned. The developer stopped referring to the two different types of board and the more sensitive boards stopped carrying stickers warning you to use half-strength developer. Either way you've been warned should you happen to get some of the older stuff.)

I use order code 600-007 @ http://www.megauk.com/pcb_chemicals.php

Let the mixture cool to room temperature. The hotter it is the faster the reaction will go, if you get impatient and use it when it's hot you could lose the entire PCB and that's if your exposure time is perfect. Up until now you're still trying to determine your best exposure time so we don't want the chemicals to be a cause for concern.

Pour the developer into a suitable plastic or glass container, an old ice cream container will do nicely.

Now take a small square of the PCB material, peel off the protective layer and throw it straight into the developer. Without having been subject to any light at all, this would represent what would be a 'track' of copper, something you want to keep. As time passes the board should start to turn colour, my stuff turns green. After a certain amount of time, the colour change should stop and everything should settle down. Leaving the PCB in the developer after this point shouldn't cause anything else to immediately occur. If the developer is too strong the green layer will start to fade and will disappear entirely.

The last board I made, with almost brand new developer at room temperature, took around 1-2 minutes to turn fully green. After this point, it didn't change colour. With the previous board I had made up some new developer and I got impatient, the mixture was still warm. As a result, I almost lost the board as the green stuff I wanted to keep started to disappear. BAD! MATT! So let it cool. The room temperature developer works perfectly at the concentration the instructions tell you to use.

If you're happy with the strength of the developer take a small square and do the exact opposite to before. Pull the backing off and shine a bright light at it, keeping the light source very close. Expose it for say 20 minutes, to make sure, then place the small square into the developer.

This represents the other end of the spectrum. Light has hit the light-sensitive coating on the PCB material and has changed its structure. When you place this in the developer it should turn black and if agitated (that is you move the PCB so that the developer flows against it.) it should move away, almost like a cloud of black. As time goes on more of the coating turns black and peels away from the copper. Eventually, it should stop. At this point, all the light-sensitive coating should have been removed from the small square of PCB material leaving bare copper exposed. This represents what you don't want, hence the copper is exposed and will be removed by the etchant, leaving the green covered copper traces behind.

If that was successful now you're ready to start exposing. That is in a dimly lit room, peel off the backing of another small square. Then place a transparency with part of a design over the small square of PCB, place a sheet of glass on top, then turn on your light source.

I recommend you start with a time of something like 5 minutes.

What goes wrong if the time is incorrect?

1)The ink layer isn't entirely opaque, in other words, it will let some light through. If the exposure time is far too long this will cause the green layer, that you want to keep, to disappear. If this is happening, either the exposure time was too long or you need to thicken up the ink layer. If the exposure time was too long the areas that you don't want to keep should still turn black leaving the copper beneath it.
2)Everything still turns green. Your exposure time isn't long enough, try again at a longer exposure time.
3)The parts you don't want to keep are turning black, but it's taking a very long time and it isn't really working properly. Once again your exposure time isn't long enough, try again at a longer exposure time.

When you land on the correct time the PCB should turn green in the parts you want to keep and the areas you don't should turn black. After a certain amount of time the green will stop getting darker and remain stable and dark. At the same time the black stuff will continue to peel away, eventually leaving the bare copper beneath it.

Once you're at this stage take the PCB out of the developer and give it a wash in cold water.

It should look something like this.



This is the time you feel good about yourself in that you've almost succeeded, albeit with a trial version, at getting this stage right, but! All might not be well. What can happen is the copper you want to etch away isn't perfectly exposed. A very fine film of the light-sensitive coating can remain. Now, this can be difficult to see, but under good lighting (you should have a good light source to inspect the board under by this point though you should be able to notice that the texture of the copper isn't quite right.

The best way to test this is to put your test square into some etchant.

There are two types of etchant that I am aware of. Ferric Chloride and Ammonium Persulphate. If you're to buy Ferric Chloride, buy it in the dry form as it's a lot more expensive ready-made up. It's easy enough to make up, simply add tap hot water and shake. You do need a decent plastic container for this as you don't want ANY of it to leak, it's rather horrible stuff and it will stain. Ammonium Persulphate is nicer to work with, but does require a higher temperature to work well at. This makes Ferric Chloride easier to use in my opinion if this is your first adventure into making PCBs. It will work at room temperature, but works better a little warmer.


Here is a photo of my ultra hi-tech lab of devious development. Note the darkened developer in the ice cream container on the far left. The ice cream container in the foreground contains Ferric Chloride, as you can see, its lovely stuff and great for the skin! (Although this is written as a joke I won't be held responsible if someone takes this seriously. Please observe the box of rubber gloves. These are a required necessity if you want to go dipping your fingers into the chemicals, otherwise don't even think about it.) As written before Ferric Chloride works best at an elevated temperature. To this end, the tub containing the etchant is placed in a large tub that I fill with hot water from the kettle.



After you've put your test square into some etchant, leave it for say 10 seconds then remove it from the solution. If successful the areas of bare copper should turn a nice pink colour, as indicated in the following picture.



It will be obvious if this doesn't happen and in that case, you either need to rinse the board and place it back into the developer for a bit more time, or you really should use a slightly longer exposure time.

If you've successfully got this far then you're ready to try making a proper PCB. Place your prepared transparencies on a flat surface and peel away the backing left on the double-sided sticky tape. Darken the room. Peel away both sides of the protective backing on the PCB material and place it in an appropriate position over the sticky tape. Press down on the PCB to ensure its secured properly.

Now place the PCB in the location you're going to expose it at, place your sheet of glass over the top (I start with the copper top) and turn on your light source. Expose the PCB for the predetermined time and then turn off the light source. Gently take away the sheet of glass and flip over the PCB, re-place the glass sheet and turn the light source on again. Once again expose the board for the pre determined time and turn off the light.

Pull the PCB away from the transparencies, this can require a decent amount of force and you may pull the tape away from the transparencies themselves. Don't worry about this just make sure to remove the tape from the PCB if this happens. Now, still in a dark room put on your rubber gloves and then place the PCB in your solution of developer. You can now turn on the lights.

Watch as your design unfolds, you may wish to agitate the PCB in the solution. I recommend you do so, moving the board up and down to create small waves that ripple through the container.

Once done and the board is washed and dried you will have something that looks like, the previously shown.



You're almost ready to etch the board but not quite. Notice how there is a lot of copper exposed around the edges of the boards. Whilst this isn't directly a problem, it does mean that your etchant is going to be used up removing copper that isn't important to the design. To this end you can take a permanent ink marker and touch up the board till it looks something like this.



Let the ink dry, depending on the pen this can take a bit of time. If your pen only puts down a thin layer of ink, you may wish to double up the thickness as sometimes a thin layer isn't enough. After the ink dries you can now start your board etching. Depending on the strength of the etchant, the temperature of the etchant and the thickness of the copper this can take some time. I do recommend you agitate the board again during this stage.

Although not necessary this is one area where buying the appropriate equipment would be a great help. Bubble etch tanks, however, are rather expensive. If you want you could make some contraption out of a fish tank pump + plastic tubing that would blow bubbles over and around your PCB, helping to speed up the process. This would allow you to do other things, rather than create small waves in your plastic tub for what can take 20-30 minutes.

After the required time your boards should end up looking something like this.



As you can see these boards were a roaring success. Note here also that the finest trace I used in this design was 15 mil. When using SSOP devices I use between 15 and 20 mil, so if you can make a board like this, you can make boards that will work with chips such as the PCM1794 or the TAS5630. This process can yield rather amazing results and I have managed to etch traces as fine as the printer is capable of printing. This is a 600x600 dpi laserjet from HP.

We are almost done, but one other thing that requires a little bit of description is PCB drilling. This isn't trivial, no matter how simple a task it sounds.

The tiny micro drills used to drill PCBs are very hard and as a result, are extremely brittle. Adding to this their delicate proportions, it's not unusual to snap them even if you know what you're doing.



The larger the diameter the drill the more resistant it will be to breaking, but at the same time you also need a drill off sufficiently small diameter so that you're not forced into making huge pads, or worse yet, find it hard to work with 8 pin DIL size components.

I make no excuses for the fact that I love using surface mount components. With a good technique, these are a breeze to work with (1206 style cases are really not that tiny) and as a result, require ZERO drilling. This saves you the time of having to drill holes, place components through the holes and then have to snip off the excess legs. It also reduces the likely hood that you will destroy a drill. Sadly not everything comes in an SMD package and there are other times where the design necessitates a through-hole component.

To this end, I find a 1.1mm diameter drill a good compromise. This is wide enough that it will accept most components (it will accept trim pots and T0220/TO126 style leads), but also leave behind enough copper after drilling a 75-80 mil pad in a DIL device.

You will need a drill stand/pillar to do this effectively. However, if you've only got a small number of holes to drill a hand-held, hand-operated, or small powered drill would be acceptable. You have to be very careful.

When drilling, make sure to hold the PCB firmly, clamp it if you're using a handheld drill, or hold it down extremely firmly if you're doing this with a pillar drill/stand. Sometimes once you've drilled through the PCB the change in resistance as you penetrate the material causes the board to 'jump' or move a fraction, this is all it takes to sometimes snap a drill. At the same time, you want to place something like a piece of MDF underneath the PCB whilst your drilling. This helps to maintain pad integrity on the copper bottom, as they can sometimes rip right off if the drill is slightly blunt. The MDF also helps to prevent drills from snapping.

Micro drills of this kind are expensive, you don't want them to snap. I buy mine through golstertooling on eBay, where a pack of 10 costs ~£13 including shipping.

I've hit the word limit

This page has been seen 5,301 times.

Wiki