Fullscreen HTML5Med Res (??MB)HTML4

Making PCBs Most of our projects use printed circuit boards (PCBs) because they make assembly so much easier and dramatically reduce the possibility of making mistakes. But PCBs are no longer available for our older designs, which may still be valid. And besides, you might want to make your own PCB for something you’ve designed yourself, or a modified version of one of our designs. Here’s everything you need to know to go about doing that! H ow handy would it be to be able to design and make your own PCBs in a short time frame? It could be that you need something a bit tidier or more compact (and reliable) than a breadboard. Or maybe you’re even considering commercialising your design. There’s just something satisfying about seeing your design made real in fibreglass and copper. We reviewed Altium’s free CircuitMaker software in the January 2019 issue (siliconchip.com.au/Article/11378), which can be used to design PCBs. We’ll refer to such EDA (electronic design automation) software in this article, but our primary intention is to explain what happens after you have completed a PCB design. 32 Silicon Chip by Tim Blythman As well as covering the commercial manufacturing services and traditional etching methods, there are a growing number of alternative techniques being described to make PCBs, especially with the rise of consumer and hobbyist CNC systems such as 3D printers, laser cutters and mills. And if you decide to take the commercial option, you may be surprised how reasonable the prices are, and the quality of the end result. Why make a PCB? You might still be wondering why you need to have a PCB made. There may be applications that you may not have considered for a custom PCB. As an example, take our April 2019 Flip- Australia’s electronics magazine dot Display project (siliconchip.com. au/Article/11520), which uses small custom PCBs as mechanical elements. That project also uses a PCB to form fifteen separate air-cored inductors from PCB tracks. You can also use PCBs as shielding between circuitry running at significantly different potentials, as we did in our Versatile Trailing Edge Dimmer (February & March 2019; siliconchip. com.au/Series/332). And you can use PCBs as part or all of a case for a project, as we have done on many occasions; you can even use the copper layers for shielding. It’s also possible to get flexible PCBs made. The cost to get this done professionally is still daunting, but we’ll cover more affordable hobbyist techniques for making flexible PCBs below. siliconchip.com.au Many PCB manufacturers can also create PCBs with aluminium cores, rather than fibreglass, which is used in high-dissipation devices, like radar systems and LED arrays. That’s because aluminium conducts heat away from parts much better than fibreglass. Anatomy of a PCB We covered the anatomy of a twolayer PCB in our CircuitMaker article, but it’s also possible to get four-layer (or more) PCBs made at a reasonable price. Here, we’ll explain a bit more about how commercial operators make PCBs, and how this changes with the numbers of layers. Whether the design has two or more layers, the early stages are not too dissimilar to the home etching process you might have tried. It starts with a sheet of fibreglass (the most common type is called FR4) clad on both sides with copper. A resist layer is applied to match the desired copper pattern, and the board is ‘etched’ by removing the exposed copper with a chemical that dissolves copper not covered by resist. The board is then drilled (and any slots to be plated are routed), but this is about where the similarity ends. A process for plating copper into the holes is used to create vias (which connect to the copper on both sides) and other plated-through holes. Then, an insulating solder mask layer is printed onto both sides of the board, followed by the silkscreen layer, which may be on one or both sides. The exposed copper is then coated with a protective layer of solder, or possibly silver or gold plating. Finally, the boards are ‘depanelised’ (ie, cut apart). Typically, several different designs (or copies of the same design) are processed at the same time on a large panel for efficiency (24in x 24in [610mm x 610m] is a typical panel size), so they need to be separated. This is usually done by a CNC routing machine, which can also rout slots and other shapes within the individual boards too. For a four-layer PCB, the inner layers are etched as for a two-layer board, using a thinner core than the final product. The outer layers of copper are then laminated to the core using ‘pre-preg’, which is actually uncured fibreglass laminated with copper foil. The outer layers of the PCB are then etched. The later steps proceed as for a two-layer board. siliconchip.com.au The four-layer technique can be extended to more layers as necessary, and there are variations where two or more two-layer boards can be sandwiched to give a similar result. In any case, to make a board, especially one with many layers, we need information about what each layer will look like. For a typical two-layer board, this amounts to six layers worth of information: two copper layers, two solder mask layers and two silkscreen layers (each pair is for the top and bottom). There also needs to be information about the final board shape and the size and location of the drill holes and slots, making for a total of eight files. All this information is typically exported from your EDA program of choice. On top of this will be information such as how thick the finished board will be and what thickness of copper is used. Other features such as silkscreen and solder mask colour can often be specified too. These specifications are usually made in a separate step, though. File formats Practically all PCB manufacturers will accept so-called ‘Gerber’ files for the manufacture of PCB designs. It is also called RS-274X. Fig.1: the eight Gerber files typically required to manufacture a doublesided PCB. In order, they are: bottom layer copper, bottom overlay (silkscreen), bottom solder mask apertures, board outline/routing, top layer copper, top overlay (silkscreen), top solder mask apertures, drilled holes and slots, and the zip package which contains the above. Australia’s electronics magazine A single PCB design results in not one, but rather multiple Gerber files, usually packaged in a .zip archive. We’ve emphasised the importance of the layers because, in the standard Gerber format, each layer is described by a separate file. The file extension of each file dictates what role it has. Fig.1 shows a typical set of Gerber files describing a single PCB. From top to bottom, the layer names refer to the bottom copper, bottom overlay (silkscreen), bottom solder mask, mechanical (board outline) layer, top copper layer, top overlay, top solder mask and drill file. The .zip file describing the board simply contains these eight files. The drill file is in a slightly different format to the other files, generally known as “Excellon” format; it is similar to Gerber but not identical. This is because the drill file was traditionally used to control a CNC drilling/routing machine, while the Gerber files were originally intended to be used with optical plotters that ‘exposed’ a lightsensitive resist layer. The Excellon file instructs the machine to select a particular bit, then use that bit to drill at a series of locations, while the other files contain an assortment of shapes, such as rectangles and circles, which are combined to create the board pattern. These shapes are called ‘apertures’. They literally were used as apertures for the optical plotters, but these days, the resist is applied differently and the Gerber files have simply become a standard way to describe the required patterns. The Gerber files are now rendered by a computer, but the photochemical resist process survives, with the apertures replaced by a single computerprinted transparency. The overlay and copper layers are rendered positively. That means that the Gerber file indicates where there should be copper or “silkscreen” ink. The solder mask is rendered negatively, meaning the file dictates where there are holes in the solder mask. In other words, an empty copper file would result in no copper on the board, while an empty solder mask file would result in the board being covered in the solder mask. The board outline layer is treated differently again. It consists of a series of lines or arcs which dictate the outline of the board. There may also July 2019  33 that can be transferred to a resist mask for home etching. Both formats store and preserve dimensions, which is critical. Some commercial manufacturers may be able to make a PCB from such files, but since they only describe the copper layers, you need Gerber files to have a proper board made with a solder mask, holes drilled to the correct sizes and so on. These days, the PDFs we supply are mainly useful so that you clearly can see where tracks run on the board. Unless you really want to make boards yourself, the commercial boards are quite inexpensive considering the high quality Fig.2: opening the files shown in Fig.1 in ‘gerbv’ produces this display. Colours are assigned randomly to each layer, for example, bottom layer copper is purple and top layer copper is cyan. Transparent rendering allows you to see all the layers in full, even where they overlap. The actual PCB produced by this file is shown at right. be lines inside the board itself, which indicate the presence of slots (for example, for isolation) and other cutouts. These lines are traced by a routing machine to give the board its final shape. As this is done as the final step, any slots defined here will not be copper plated. If you need copper plated slots, eg, to solder flat component pins into, they are defined in the drill layer, using something known as a “G85” command. These are made before the through-plating process is applied. It may seem odd that the drill file has a .TXT file extension, as if it is a text file, but Gerber and Excellon files are text-based; you can open any of these files in a text editor program like Notepad. You will see a series of coordinates and commands, which will look familiar if you are used to working with CNC machinery. The above is only a brief overview, but should give you an idea of what to expect when creating PCBs for your own use. We won’t go into any detail about creating Gerber files; if you are using CircuitMaker, we explained how to generate Gerber files in the January 2019 article. Other EDA programs will have their own instructions on how to export Gerber files. Just make sure that you provide all the required layers. In many cases, exporting the drill (Excellon) file is a 34 Silicon Chip How to view Gerber files separate step to producing the other Gerber files, so don’t forget to do it! And it’s always a good idea to check the Gerber files before sending them off for manufacture, as it’s quite common for some elements to be missing or extra elements to be present. We’ll explain how to do that shortly. Exporting PCBs as PDFs You might have noticed that SILICON CHIP has historically published our PCBs as EPS or PDF files, a tradition that we continue to this day, although we now also offer commercially produced boards for virtually all of our published designs. The main reason for doing this is that it’s easy to print such files at home to produce a negative or positive mask We use the free open source program “gerbv” to check and validate our Gerber files. It’s available for Windows and Linux. The latest Windows version can be downloaded from https://sourceforge.net/projects/gerbv/files/ and it is available as a system-installed software package for many Linux distributions (eg, “sudo apt-get install gerbv” in Debian-based distributions like Ubuntu). As well as displaying Gerber files and allowing you to view and manipulate the layers, it also has the option of exporting to PDF, which is handy if you want to make PCBs using some of the more hobbyist oriented techniques. But note that most versions of gerbv do a poor job of exporting to PDF when the PCB contains copper pours (large areas of copper which are not to be removed); these tend to get pixelated. An up-and-coming version claims to solve this. Fig.3: the code on the white ‘silkscreen’ overlay of this board (ringed in red), was added by the PCB manufacturer. It allows them to figure out to whom to send this PCB after it has been cut out of the large panel that was manufactured (known as “depanelisation”). Note how clean the tracks and pads are, and how accurately the holes have been drilled on this low-cost board. Australia’s electronics magazine siliconchip.com.au If you have a ‘zipped’ set of Gerbers, you will need to extract the individual files before opening them in gerbv. Multiple layers can be opened from the File → Open Layer dialog box. You can change layer colours, rearrange and hide individual layers with the panel at left. When exporting to PDF (File → Export → PDF), you can select one layer at a time by clearing all but one of the checkboxes in the layer tab. Change the layer colour to black by pressing F6 and picking the colour from the popup menu, if you plan to print the PDF as an optical mask. You may need to set the background colour to make the layer visible; this can be done via View → Change background colour. Fig.2 shows gerbv displaying the Gerber files for our recent (April 2019) iCEstick VGA Adaptor PCB (siliconchip.com.au/Article/11525). The colours shown are assigned essentially randomly when you open up the layers and are designed to make each layer distinctive. You can change them to more realistic colours if you want (eg, green for copper, light grey for solder mask openings etc). Getting PCBs made from Gerber files The first technique for making PCBs is the one we use most at SILICON CHIP. It sounds really easy, too – we get someone else to do it! In spite of what you might think, it’s not expensive, and the results are very good. Of course, the proviso is that you won’t get the PCBs right away unless you pay a lot for “fast turnaround” and express delivery. It typically takes a week or two between ordering the PCBs and receiving them, sometimes longer. So if you need a prototype today, you should probably look at one of the other options. For smaller orders (eg, less than 100 units), the cost of manufacturing PCBs is normally kept reasonable by aggregating boards from many customers. One minor side effect of this is that a small tracking code may be added to the silkscreen of your board, so that the manufacturer knows which board goes to whom. An example of this is shown in Fig.3. It’s usually quite small and placed in an out-of-the-way location. Some manufacturers have webbased ordering while others accept files via e-mail and will send you a quote (usually within one business day). Ordering via e-mail can be convenient because this makes it easy for them to point out any problems they may find with your files so you can correct them before manufacturing begins. Minimum quantities are usually in the order of 5-10, with a decreasing per-board cost as you order more. For prototypes, you’ll generally want to order a small quantity, but it’s good to have a couple of spares in case you make a mistake during assembly, or find it necessary to modify the board. Design rules If you’ve just started out using an Photomicrograph of a section through a multi-layer PCB complete with an IC soldered to the top layer. The copper section at right is a via which connects two of the internal layers. siliconchip.com.au Australia’s electronics magazine Fig.4: PCBcart’s specifications and requirements. You will need to make sure your design adheres to these rules shown here, or they will complain when you send them your files. Luckily, all the rules can be programmed into the Design Rule Check settings of most ECAD software, and the software will then automatically inform you of any problems (or may not even allow you to create them in the first place!). EDA tool like CircuitMaker, Eagle, KiCad or DipTrace, you may not be familiar with design rules. They are an important part of PCB manufacturing since they aim to ensure that the design does not incorporate any elements which can not be easily and reliably made. Board manufacturers generally supply a set of design rules which, if adhered to, guarantee that your design can be manufactured using their processes and equipment, with a minimal chance of failures. You can add your own, stricter design rules to ensure the safety of your design (such as ensuring separation between highvoltage tracks). For example, you can see PCBCart’s rules at www.pcbcart.com/pcb-fab/ standard-pcb.html, partially reproduced in Fig.4. In some cases, you can violate some July 2019  35 Fig.5: here is where you can enter the manufacturers’ requirements in CircuitMaker so that it can check there are no violations. For example, the Clearance rule is set to 10mil in all cases, so it will ensure that there is a minimum of 0.01in (0.254mm) between adjacent conductors. Generally, you only need to make a few small changes to the default rules to suit typical manufacturers. of the manufacturers’ design rules slightly if you are willing to accept a higher percentage of faulty boards. Or they may charge you extra for the more involved processes required to manufacture your boards correctly. Before we look at actual specs, let’s get a “trap for young players” out of the way! Track thickness and track gaps are generally specified in “mils”. A mil is not an abbreviation for millimetre! 1mil equals one thousandth of an inch, so a track width specified as 12mil will be 12 thousandths of an inch wide – about 0.3mm. Many people have been caught over the years – now you shouldn’t be! Most EDA software will naturally work in mils, although some have the ability to work to other standards. A tip: stick with mils, because that’s what PCB manufacturers are expecting. A typical rule is that copper tracks should be no less than 6mil (six thousandths of an inch or around 0.15mm) wide and no less than 6mil apart. Another common rule is that the drilled holes should be no less than 12mil in diameter (0.3mm). If you were to place tracks 5mil (0.13mm) apart, they might still make your board, but you may find that some boards have short circuits between adjacent tracks. Or they may just reject it. You should ideally set up the design rules before starting to lay out your PCB, although, as a general guide, if your board is easily hand-solderable and you aren’t after any special board 36 Silicon Chip finishes or colours, virtually any manufacturer should be able to make your board. Since most manufacturers have similar rules, once you have set them up, you should be able to have your board made by many different companies, perhaps with some slight tweaking to suit the stricter ones. Most EDA programs offer automatic design rule checking, so it’s worth entering the manufacturers’ rules into your EDA program. It will then alert you to any violations, so you can fix them. Some PCB manufacturers offer downloadable design rule files that can be imported into your EDA program directly. CircuitMaker’s design rules can be accessed from the “View → Rules and Violations” menu, which opens the dialog shown in Fig.5. The minimum width and clearance constraints correspond to the trace width and separation noted above. Our choice of 10mil should be achievable by most board manufacturers (see panel!). Order process As mentioned above, some manufacturers take orders via e-mail. So, for example, if you want to order some boards from Sydney-based LD Electronics, e-mail your zipped Gerber Fig.6: like many PCB manufacturers, PCBCart gives you an instant quote once you have put in your PCB’s particulars. You can then log in, add the design to your cart, upload the Gerber files, fill in your details (eg, delivery address) and pay for the order. They’ll start manufacturing your PCBs once your order has been submitted and will normally send you updates, and eventually a courier tracking number, via e-mail. Australia’s electronics magazine siliconchip.com.au files to quote<at>LDElectronics.com.au along with any special requirements (board thickness, copper thickness, solder mask colour etc) and they will e-mail you back a quote. They will then guide you through the order process. On the other hand, as the name suggests, Guangdong, Chinabased PCBCart offers web ordering. You can get an instant quote by visiting www.pcbcart.com/quote and entering your requirements. Fig.6 shows this page. We have already filled in the details of one of our boards, and you can see that the price (in US dollars) is being displayed at upper right. We can then change the board quantity and other requirements and the price is updated. The only fields that you need to fill in are those shown with an orange asterisk: the Part Number, Board Type and Board Size. The other defaults are fine unless you know you need something different. You can try changing some of the parameters and see how much non-standard features add to the cost of the basic PCB. If you increase the quantity, you will see that the price doesn’t go up all that much. In our example, five boards cost US$32.65 ($6.53 each) while 10 boards cost US$42.20 ($4.22 each) and 100 boards cost $213 ($2.13 each). This is typical, as there is a fixed cost associated with every different PCB made; making more copies of the same board has a lower incremental cost. You can vary the board thickness between 0.4mm and 2.0mm; the cost varies slightly as you do this. 1.6mm is a typical thickness and a good default. 35µm copper is also known as ‘1oz’ and is the default for single or double-layer boards; 70µm copper is ‘2oz’ and costs a little more. A green solder mask is usually the cheapest. In this case, there are other colours available at the same price (eg, blue and red) while other options increase the cost slightly. So does opting for a lead-free or gold finish, or a shorter lead time. Note that a 30mil (0.75mm) wide track on a 35µm (1oz) copper board can handle 1A with only a 10°C tem- RCS Radio’s Ron Bell and his 31-thou limit Older SILICON CHIP readers would no doubt remember the name RCS Radio, if not its owner, Ron Bell. If not the first manufacturer, RCS Radio was certainly a pioneer in this country, manufacturing “Printed Wiring Boards” for the military, industry and for the hobbyist from a factory in Canterubury (boy, were there some arguments when people started calling them that American name: printed circuit boards!) But mostly we remember Ron “doing his nana” when patterns were sent to him with less than a 31mil track width or spacing. In fact, he’d get upset at anything under about 40-50mil! This was long, long before computer software to produce PCB files. There weren’t even computers in those days! PCB patterns were hand-drawn with pen and ink; later this was superseded by black crepe tapes and pads. Often, the patterns were produced at 200% scale, so that when reduced photographically, minor errors in drafting were also reduced. They didn’t eliminate errors in the trackwork itself, though! After Ron Bell’s passing, RCS Radio was run by Bob Barnes, until his passing about ten years ago. By then, many production houses around the world were turning out PCBs which Ron Bell would have dismissed as “impossible!” perature rise, so unless you have a specific high current application, thicker copper is usually unnecessary. With the higher cost of 2oz copper, it’s generally worth using wider tracks instead, if possible. Like most online PCB manufacturers, PCBCart accepts payment by PayPal, including Visa or Mastercard. They offer delivery via DHL, UPS or FedEx. Other companies may offer cheaper options such as registered post. If you order from a local manufacturer like LD Electronics, postage will probably be quite a bit faster too! Doing it yourself Fig.7: here’s how the photochemical etch-resist process is used to produce a PCB (eg, using “Press ‘n’ Peel” film). Both positive and negative processes are shown. siliconchip.com.au Australia’s electronics magazine Of course, if time is of the essence, then ordering boards from China will not be your first choice. The age-honoured technique of etching copper from a pre-laminated board is still widely used, although modern methods put some twists on how the etch resist is applied. There are also other techniques July 2019  37 available for removing copper, and it’s now even possible to print a PCB using conductive ink, allowing the wiring to be ‘drawn’ directly onto a substrate. That really is a printed circuit board! PCB etching You might not think that PCB etching has changed much over time; indeed, the basic chemical technology is very much established and is still the primary method of commercial PCB manufacturers. What has changed is the generation of the etch-resist layer, with some clever people using novel techniques. If you have etched your own boards, you will have heard of ammonium persulphate and ferric chloride. But many board manufacturers use cupric chloride (green in solution) to etch their boards instead. When cupric chloride (CuCl2) reacts with copper, it turns into cuprous chloride (CuCl). These two compounds both contain only copper and chlorine, the difference being the ‘oxidation state’ of the copper atoms. The beauty of this method is that the cuprous chloride (CuCl) can be turned back into cupric chloride (CuCl2) by an oxidising agent. This oxidising agent can be something as simple as oxygen from the air we breathe. Of course, the chemistry is not that simple, and there needs to be a supply of chlorine atoms to supplement the copper atoms that are being added, although this can come in the form of hydrochloric acid. The result is an etchant that not only doesn’t get used up; it becomes self-generating. There are downsides, of course. Cupric chloride is nasty stuff, and is worse for the environment than ferric chloride if released, which makes it difficult for hobbyists to use, particularly if the amount of cupric chloride keeps increasing. That said, the actual etching works similarly to that of ferric chloride, with agitation and heat accelerating the process. Ammonium persulphate is similar, but has the advantage that it doesn’t stain anywhere near as much as ferric chloride. It has been said that if you walk within five metres of a ferric chloride bath, it will jump the gap and stain your clothes. A slight exaggeration perhaps, but . . . 38 Silicon Chip Fig.8: a PCB which was produced from a bare copper laminate board using a milling machine. A conical milling bit is normally used, as the copper and fibreglass are fairly tough and you want to cut a V-shaped groove. The main difficulty in doing a job as good as this is ensuring that the PCB is perfectly flat, and perfectly aligned with the bed of the mill. Toner transfer etch-resist process If you have access to a laser printer, toner transfer is one of the best etchresist methods for a hobbyist. While some toner transfer kits can be expensive, cheaper versions are available online. They aren’t as good, but they can be made to work. A PCB design is printed onto the glossy side of the toner transfer paper using a laser printer. It must be mirrored, as the transfer process mirrors the design a second time, so it ends up the right way around. The toner itself becomes the resist layer. For this to work, the copper clad board must be spotlessly clean. Even fingerprints can impede the etching process. The toner transfer paper is pressed against the copper cladding, and heat is applied. This can come from a clothes iron or even a laminating machine, although it appears some laminators can’t reach the temperatures needed to transfer the toner. After the board cools, the toner transfer paper is carefully peeled back, leaving the toner attached to the copper clad board, which can then be etched. The copper under the toner will remain intact, as long as it isn’t left in the etchant too long. You can also use this method to produce a ‘silkscreen’ layer by applying the toner to a pre-etched board. Incidentally, we’ve used the “toner transfer” method to produce a PCB using ordinary bond paper (ie, from a photocopier or laser printer). It takes quite a few attempts to get it right and importantly, the track spacing and gaps cannot be very fine. But it does work fairly well and is a great method for the hobbyist to try. (See siliconchip.com.au/Article/6884). Photochemical resist processes This involves a chemical which reacts to light, where the areas exposed Fig.9: a screen grab of the FlatCam software which can convert Gerber files into G-code which can then be fed to a milling machine, laser cutter or other CNC equipment. Australia’s electronics magazine siliconchip.com.au to light change in chemical composition, allowing the unwanted parts of the layer to be chemically removed, leaving just the areas required to protect the copper underneath during the etching process. You usually print the copper pattern as a mask on transparency film, then place that mask on top of the photochemical layer, which is attached to the copper laminate. You then expose it to UV light, either using a light box or by exposing it to sunlight. The resist layer is then treated in a developing solution to remove the undesired parts of the resist mask, after which the board is etched as it would be for other resist types. This is fairly close to the method used in factories for PCB manufacturing. It is vital to ensure that the resist layer is not exposed to light unnecessarily, as this lessens the effectiveness of the process. Options for using a photochemical resist include both pre-sensitised boards, films that can be laminated to copper and even liquid photo-resist that can be painted onto copper-clad fibreglass. There are also options for negative and positive resists. A negative resist is one that hardens where exposed to light, so the remaining etch resist layer corresponds to clear spaces in the transparency; the final PCB result is the negative of what is printed onto the film. With a positive resist, the areas which are exposed to light are the ar- eas which are then removed, and the areas which were not exposed remain to resist the etchant. Both options are shown in Fig.7. Again, there are variations on this theme where a pattern printed onto plain (bond) paper is used to expose the PCB photoresist. It is important that the PCB pattern is in contact with the resist (ie, it is printed “wrong reading”) so light scatter within the paper is minimised. Fairly obviously, exposure times are rather significantly longer than when using transparency film. Etch resist pens Etch resist pens are typically used to touch up or repair the resist layer already applied to a board, where it has not transferred or printed correctly. They are also sometimes used to quickly sketch a very small PCB design by hand. But they can also be used as part a CAM (computer aided manufacture) process. This involves the use of FlatCam (http://flatcam.org/) and a 3D printer. Rather than using FlatCam to mill an isolation path, it can also be used to trace a resist path using a pen. The etch resist pen is attached to the head of the 3D printer, and it is commanded to lay down a resist path by the Gcode that FlatCam generates. It’s a marvellously simple method, as it doesn’t require any permanent changes to your 3D printer; the pen can be held in place with a rubber band. The difficulty is in converting the Gerber files to an appropriate set of commands to drive the 3D printer. The best option we found is to use gerbv to convert the Gerber file to .png graphics, followed by using the http://svg2stl.com/ website to convert these to an .stl file. The .stl file can then be converted by any 3D ‘slicer’ program to files that can be printed on a 3D printer. You need a custom ‘slicing’ profile for the pen, so it can be lifted when moving between points; many programs offer a ‘lift between extrudes’ option, which is suitable. By the way, most etch-resist pens work much better if the board is ‘baked’ before etching, to cure the resist layer. This is also true of many other methods, especially photo resist. Filament extrusion We’ve also seen a similar method but without even needing the pen; a 3D printer can be used to extrude plastic filament onto a blank copper PCB, with the filament forming the etch-resist layer. Flexible filament appears to be the best choice. This helps to prevent the plastic from lifting off the PCB during the etching process. Printing conductive material If you have access to a 3D printer, you can also consider directly printing wiring using a conductive filament. But note that conductive filaments are not as good conductors as copper, so this method is mainly for low-power applications. It’s also pretty much impossible to tin the conductive filament; you need to melt the component leads into the filament. We’re not sure how permanent the result is! A typical 3D printer nozzle width of 0.4mm corresponds to a minimum track width of 16mil, so this method isn’t capable of producing the fine details of other methods, and small SMD footprints will be impossible. But it appears that having a 3D printer can still be a useful tool for making PCBs. Voltera V-One PCB Printer Fig.10: the Voltera V-One can “print” a double-sided PCB up to 127 x 104mm. It’s an expensive way to produce a board but when time is money . . . 40 Silicon Chip Australia’s electronics magazine An extreme example of this is the Voltera V-One PCB Printer, which can not only produce double-sided PCBs up to 127 x 104mm using proprietary siliconchip.com.au conductive inks, but can also apply solder paste and perform reflow of populated boards. You can see a video of the Voltera V-One in action at: http://youtu.be/ PeW1nURJ5ww According to the Voltera website, a complete, unpopulated board can be ready in around 35 minutes. Compared to a manufactured board, the Voltera PCBs will lack a solder mask and silkscreen layer, and the conductive ink is not as durable as bonded copper traces. But the Voltera V-One is not limited to fibreglass substrates, and flexible substrates or even glass can be used. The current listed price is US$4199 for the machine itself, with the cost of producing each board at around US$5 each. If speed is of the essence and price is not a problem, the V-One is certainly worth checking out. Milling PCBs To form tracks on pre-laminated board, rather than etching, copper can be removed by mechanical means. Open-source and do-it-yourself CNC (computer numeric control) machines such as desktop mills, as well as simi- lar commercial devices, can be used for this purpose. A PCB mill routes insulating grooves in the copper layer to separate the copper into the tracks and islands required to form a circuit, as shown in Fig.8. The same machine may be able to drill holes for the insertion of vias and through-hole parts. While such a technique does not inherently provide the option for silkscreen labelling or solder masks, the grooves formed by the routing action makes it harder for the solder to form accidental bridges and production can be very fast, taking just minutes for smaller designs. Double-sided boards are possible with accurate enough registration, although plated holes and vias must be created manually. Small copper rivets are available specifically for creating vias in such boards (they can also be used to repair commercially manufactured boards). While it is possible to completely remove all unneeded areas of copper from a PCB using a mill, it is usually unnecessary, wasting time and wearing the milling bits. So PCB mills generally remove just enough copper to provide the isolation necessary for correct circuit operation, and no more. An extra step is also needed if the copper needs to be tinned, although this is generally not necessary for a prototype board; tinning prevents surface corrosion, but if the board is assembled right away, that’s less of a problem. Another consideration for this technique is the waste produced, ie, copper and fibreglass dust. These are health hazards, especially glass dust, so a vacuum system is needed to keep this under control. Suitable off-the-shelf PCB mills are available; the Bantam Tools Desktop PCB Milling Machine is an example of this. It is available from Core Electronics. See: https://core-electronics.com. au/bantam-tools-desktop-pcb-millingmachine.html Many people are also attempting to build their own PCB mills, some even using 3D printers with their extruder heads replaced by a rotary bit. The lateral forces caused by the milling bit moving through the material are much higher than would be experienced during 3D printing, so not all 3D printers are suitable for this conversion. If you really do want to make your own PCBs . . In this article, we’ve briefly mentioned methods of producing one-off PCBs yourself – perhaps from a magazine project or indeed a prototype for a new product. And while we usually take advantage of today’s low cost, speed and quality of commercial PCBs (which is why we’ve given up making them ourselves!) there may well be a time when you want a PCB right now! SILICON CHIP has published quite a few articles over the years detailing methods of making one-off PCBs, using a variety of production processes. We’ll briefly recap on the most recent articles so if you really want to make a PCB yourself, you should be able to do so. February 2001: Toner Transfer, by Heath Young. This article showed how you can “transfer” the toner from a pattern reproduced on bond paper from a standard laser printer to the blank board You then use that toner as a resist for etching. The difficult part is to carefully remove the paper, which you do by breaking it down, rubbing it under running water. We’ve tried it, with mixed results, although we’ve proved it can be done. Be prepared for a few misses before you get the system to work! siliconchip.com.au March 2001: Making photo-resist boards at home, by Ross Tester. We followed the last article with a more “traditional” approach using commercial resist-coated boards and exposing them to special UV lights (or the Sun, which is very high in UV!) through PCB patterns which had printed on a photo copier onto either transparent or semi-transparent film. This is a time-honored method and is capable of very good results with fine tracks and spacing. Incidentally, you don’t have to buy pre-coated board – you can still buy blank board and photo resist, in either a liquid or spray-on form, or even as film which you can apply to the board. It’s certainly not as common as it used to be but it is available (Google is your friend!). February, 2012: Homebrew PCB via Toner Transfer Film, by Alex Sum. This uses a special film called “Press’n’Peel” which still available from Jaycar (HG9980). You print your pattern onto this film via a laser printer and use a hot laminator (or even a hot iron) to transfer the pattern to the PCB then etch, drill and cut in the normal way. The author even used Press’n’Peel to create a component image on the top side of the board (similar to the silkscreen found on virtually all commercial boards). Australia’s electronics magazine July 2019  41 There are some challenges to milling PCBs. To get good results, the PCB must be very flat and level, as the milling depth will vary if the PCB is not entirely flat. Some mills can compensate for this. Software for milling PCBs Appropriate software is also required to convert Gerber files to a language that a 3D printer understands; typically G-code. G-code is a slightly different subset of RS-274 than that used in Gerber files and is commonly used in CNC applications. We found two programs which can do this, but since we don’t have a mill, we couldn’t test them fully. FlatCam, mentioned earlier, is a very flexible and powerful program, and it can do the Gerber to G-code conversion that is needed to create a PCB using a mill – see Fig.9. Another suitable program is pcb2gcode, found at: https://github.com/ pcb2gcode/pcb2gcode This has a much simpler commandline interface, although a graphical version is available. Making PCBs with a laser cutter There are a couple of different approaches to creating PCBs with a laser cutter. One uses the laser to react with a photochemical resist layer. Rather than using a mask, the resist is directly cured by a pass of the laser. It appears that the software to do this is straightforward. We used gerbv to export a PCB layer in Gerber format as a PNG image, then imported this file into our laser cutting software. We then cut a scrap of acrylic as a test. The results can be seen in the photo below. Because many CNC laser cutters are used to do engraving, the software is It’s not a PCB but a PCB pattern cut into a piece of acrylic which we produced with our CNC laser cutter – just to prove it could be done! 42 Silicon Chip almost always capable of importing image files like this. Despite how easy it is to do this, we would be dubious to recommend it without further research into the specific chemicals being used and how they might react to being hit by laser radiation. That’s why we tried it on a piece of perspex and not a PCB. For example, it’s well known that vinyl should not be cut in a laser cutter as it releases toxic, corrosive chlorine gas which will poison you and damage your laser cutter. Any compound that contains chlorine will have a similar result. Also, you will have to tune the speed and laser intensity to get a good result, and in doing so, if the laser power is too high or cutting speed too low, you could cut through the etch-resist layer, with unknown consequences. Another variation we’ve seen, which may be more practical, is to coat the copper clad board with black paint and using the laser to blast it away to match the negative of the PCB pattern. The remaining paint forms the etchresist mask, and the board is etched. In this case, the development step is not needed. If you have an industrial power laser cutter, it may even be possible to simply vaporise copper off the board, producing PCBs in a single step. Drilled holes could also be completed by having the laser linger a little longer! In brief, a laser cutter could make a great tool for producing PCBs, but we have our doubts as to the safety of the process, both for human and machine. Printing circuits on other substrates We mentioned that the Voltera VOne PCB Printer can print on glass or even flexible substrates. PCB manufacturers can also create aluminium-core or flexible PCBs at a price. We’ll mention some techniques we have seen which allow hobbyists to create their own PCBs with unusual substrates. Just as it is possible to buy copper clad fibreglass panels (blank PCBs), so too is it possible to buy copped clad polyimide (DuPont calls this “Kapton”) in sheets, ie, blank flexible PCBs. The copper clad polyimide sometimes goes by the name “Pyralux”. Polyimide is hardy stuff and can handle Australia’s electronics magazine the harsh conditions of an etch bath. The substrate lends itself well to the toner transfer resist method, but we have seen some people comment that the Pyralux tends to curl when exposed to heat; for example in a heated etch bath. The curling may cause the etch resist to lift. We suggest fixing the sheets to a rigid backing during the etch process to prevent this. This method could also be used to create custom flexible flat cables (FFCs). It’s also possible to buy sheets of copper foil, in which case there is no limitation on what substrates are possible, as long as there is a way to bond the two together. We have seen home-made kevlar PCBs, where the copper is bonded to the kevlar using fibreglass resin. It seems the secret here is clamping the two together rigidly to ensure that the surface to be etched remains flat. We’ve even seen PCBs made on glass using a similar technique, although soldering onto such a board would be quite fraught; you would have to do it carefully to avoid breaking the glass from differential heating – possibly by directly heating the glass itself. Conclusion In this article, we have presented an assortment of PCB manufacturing techniques that are accessible to the hobbyist, but we haven’t been able to mention every possible variation. The rise of home CNC type machines such as mills, laser cutters and 3D printers is making it possible to do many things that we would not have dreamed of previously. Some techniques are still being developed and improved, including the traditional ones. Having a laser cutter at our disposal tempts us to try some of the methods we have mentioned above. However, we will have to do further research to ensure we do not damage our machine or risk our health. If we needed to make a prototype board today, we would use the toner transfer or a pre-sensitised photochemical board, followed by a bath in ferric chloride or ammonium persulphate. And when we’re in less of a hurry, we order commercial prototype boards. That is, until someone lends us a Voltera V-One . . . SC siliconchip.com.au