I’ve been making double-sided printed circuit boards in my home workshop for quite some time now. This page is an attempt to document the processes I usually follow and experiences I’ve obtained making PCBs.
The common methods of making PCBs at home (e.g. using toner transfer) have been quite well documented throughout the net. My intention here is not to provide a tutorial covering the basics. Instead, I briefly cover my own setup and processes from start to finish and share some of my own experiences as well as elaborate on some usually overlooked topics such as vias and soldermasks. Since I am continuously experimenting and finetuning the processes, I expect to make frequent updates to this page to reflect any progress I may achieve over time.
Due to my almost non-existant photography skills and poor setup, most of the photos on this page do not do justice to the actual results. Especially the soldermasked PCB below looks much better than the photo here.
Laser Printer (Kyocera FS-1100, for toner transfer), CNC Mill/drill (converted Proxxon MF70, for drilling copperclad), Office Laminator (Generic brand, for toner transfer), Light-box (self-made, for pattern aligment), Via press (self-made, for making vias), Airbrush (Proxxon MK240 compressor and additional airbrush, for soldermask application), Toaster Oven (Generic brand, for soldermask curing).
Copperclad board (Phenolic or epoxy fibreglass substrate), Toner transper paper (Glossy/magazine or Press’n Peel Blue), Etchant (Ferric Chloride), Drill bits (Tungsten Carbide), Cleaning supplies (Scotch-brite, Acetone, Paint Thinner), Copper wire (for vias), Soldermask (Pebeo Vitrea 160 Glass Paint).
Layout: Apart from the usual guidelines of PCB design, there are a few basic rules that I generally follow to make sure that I can actually manufacture it. For example, since I am not able to make plated through-holes and vias, I ensure that through-holes can be soldered on both sides if they are to complete an electrical connection. If this is not possible, I either route the board in a different way or I place a mechanical via nearby. So far, I have been able to achieve 8 mil wide traces with 6 mil clearance, but I use wider traces/clearances wherever I can. I also maximize the use of copper pours to save on etchant and to speed up etching.
Drilling: Since I use a CNC to drill the board, drilling comes before the toner transfer, as it is easier to align artwork to drilled holes than the other way around. I fix the copperclad to a piece of MDF using strong double-sided adhesive tape to make sure that the copperclad stays flat during drilling. I’ve written a little Python script that converts .nc drill files to G-code for use on my MF70 CNC.
It is crucial to use sharp tungsten carbide bits to have clean holes with no burrs. I usually use 0.5mm bits for vias and 0.8mm for through-hole components.
Toner transfer: I print the artwork to the transfer paper using my laser printer. Then, I align the artwork to drilled holes on the copperclad using the light-box I’ve made. Finally, I pass the combination through the laminator several times (usually about 8 to 10 times) to perform the actual transfer. A clothes iron may be used for really small boards, but I find that a laminator is a must for accurate, high density transfers for larger boards.
I have found out that, since laser printers usually have poor dimensional accuracy, it is usually necessary to scale the output slightly to have it match properly to the holes. I export gerber files from the design software and I use an excellent little utility called gerber2pdf.py to obtain the scaled artwork. The scaling errors for both axes are usually less than 1% but even such seemingly small errors are enough to cause misalignment. As long as the errors are linear, they are not difficult to compensate for.
I’ve tried two different kinds of paper for transfers: The Press’n Peel Blue and glossy (magazine) paper. Both work well for relatively high density artwork. Here is a comparison of the two, based on my experiences:
- Stability: Compared to magazine paper, deformation of PnP Blue under heat is more irregular. This can make alignment of PnP Blue prinouts to drilled holes is more difficult and less accurate.
- Cleanliness: PnP Blue tends to pull small dust particles which means touch-ups are invariably required after the transfer. This is less of a problem with glossy paper.
- Removal: PnP Blue can be removed right away from the copperclad after the transfer. Glossy paper requires the board to be soaked in water for a while and very careful cleaning afterwards to remove residue.
- Transfer Quality: PnP Blue creates much better copper fills/pours compared to the glossy paper. With PnP Blue, copper fills look completely solid, but in the case of glossy paper, copper fills usually have tiny little pits in them. This can be fixed to a degree by strengthening those areas with resist pen. Thin tracks are fine with both kinds of paper.
- Resolution: I’ve successfully made boards with 8 mil tracks and 6 mil clearance with toner transfer using both kinds of paper.
- Price: Glossy paper wins hands down as there is simply no comparison.
Currently, my primary choice of toner transfer paper is glossy paper, with price and stability being the determining factors.
Etching: My etchant of choice is Ferric Chloride. I warm the etchant beforehand and I keep agitating the board until etching is complete. It usually takes around 6-7 minutes. I would like to experiment with other chemistries such as Cupric Chloride, but I have not had the chance yet.
For producing double-sided boards, I usually do toner-transfer and etch one side at a time. I mask the other side using wide clear adhesive tape.
Vias: The “right way” of making connections between the top and bottom layers of a double-sided board is by “plated through-holes” (PTH). It is not impossible to do PTH at a home workshop such as mine but it is not easy to obtain the required chemistry and equipment (i.e. electroplating tank). I also doubt that it would be economical in very low volumes.
The simplest way of making a via is by putting a wire in the via hole and soldering both sides. This can work very well, except when you have an SMD board and the need to put vias under an IC arises (this happens almost always). Mechanical vias are a solution that can work well in that case. Special rivets can be used for this purpose, but they are not necessarily cheap.
I built a device that I call a “via press” that can be used to make mechanical vias. I place a small piece of copper wire in the via hole. It needs to be a tight fit so that it won’t slip out of the hole. I trim the ends of the wire so that very little (perhaps less than a millimeter) of it sticks out of the hole in both sides. Then, I place the board in the via press and applying pressure to the board crushes both ends of the copper via and produces a reliable connection between the two sides. More details on the via press can be found in this post.
Soldermask: If one is assembling PCBs by hand, a soldermask is probably unnecessary, especially for a prototype. Nevertheless, a soldermask is beneficial for protection of the board, especially from corrosion. Not to mention that it looks much nicer.
The method I use (I first encountered it here) is based on a commonly-available, inexpensive glass paint called Pebeo Vitrea 160. This is a water-based, transparent glass paint that works quite well as a soldermask once it is cured in a domestic oven. It comes in several different colours and in bottle, tube or felt tip marker containers. After application, it requires a drying period of 24 hours before oven curing.
To be able to use this as a soldermask, the solderable areas of the PCB need to be masked from the glass paint. Fortunately, toner transfer also works quite well for this purpose.
The procedure is as follows:
- Toner transfer is applied on the etched board, using the “solder layer” as the pattern. This covers all solderable areas (e.g. IC pads) with toner.
- The whole board is painted with Pebeo Vitrea 160. A 0.5mm nozzle airbrush works best for me. I thin the paint slightly (add 1 part water to 4 parts paint). I haven’t tried paste or markers, those may also work well.
- The board is left to dry. The whole 24 hours may not be required, but I have not experimented enough to specify a minimum.
- The board is cured in the oven. I use an inexpensive toaster oven and an oven thermometer to monitor the oven temperature. I bake the board for 40 minutes and I do not let the temperature to go above 130 C. The paint manufacturer specifies 160 C for curing but I find that sometimes the copperclad stock I use are not able to withstand those temperatures for extended periods of time.
- The areas masked with toner are removed using a combination of paint thinner and acetone, by light rubbing with a sponge. Depending on the thickness of paint, it may take a while for the solvents to take effect.
- After cleaning the solvent remains, the board is ready for soldering.
I do not fully understand the mechanism behind the toner masking of glass paint, but I am guessing that it somehow interferes with the adhesion of the paint on the board surface and this reduces the resistance of the paint against solvents.
Due to the my laser printer’s poor coverage of black areas, copper fills come out with tiny little pits in them. I am looking for a solution to this problem other than strengthening all affected areas manually with a resist pen.
I have not yet exactly determined the scale correction factors to compensate for poor laser printer dimensional accuracy, so there are still minor alignment errors, as is clearly seen from the board photos.
The method I use for making the mechanical vias works quite well, but it is very labor intensive. I am open to exploring less time-consuming alternatives.
The soldermask method needs more testing and experimentation. It is not as strong as a real soldermask. Curing it at higher temperatures (as specified by the manufacturer) may make it stronger, but it could also make mask removal more difficult (if not impossible) and there is a risk of burning the paint and/or the board if the temperature is not controlled well enough. It may also be beneficial to divide the curing step into two: First, a quick cure step before mask removal and a later, longer cure step to really harden the paint. It is also not clear how resistant the glass-paint is against certain fluxes (and solvents used for cleaning those fluxes).