Automated Circuitry Making
Release status: unknown
This is the process of making electronics for the RepRap board. Primarily it is aimed at using a RepRap, combined with modified software to take an electronic representation of a circuit and create it as a physical object. The traditional process for making a printed circuit board assembly (PCBA) involves many manufacturing steps, including forming the copper traces, drilling, component placement, soldering. Automating any one of these steps using RepRap technology would be useful. A complete cycle electronics assembly machine would be an amazingly awesome and marvelous thing.
Obviously, all of these ideas require an actual RepRap machine as a prerequisite. (Ideas that do not require an actual RepRap are collected at alternative electronics).
All of these processes start with a blank copper clad board and the desired end result of a circuit board with the traces required for the board. There are a variety of ways we are examining for this approach.
Milling Head Route
This method would use a router attached to the RepRap machine and would selectively mill away the copper to separate the copper tracks from each other. The things to be investigated are:
- Milling and Drilling Head -- how to make PCBs the next-most-common way, starting with copper-clad PCB and milling off the unwanted bits of copper.
- how accurate can the milling be done.
- how small the area milled can be... at least between pins
- if a reprap machine could withstand the stress (if not, can you re-inforce it / reprap new parts?)
- Does the Replath software (Builders/ReplathHowto) handle milling a PCB correctly?
Apparently the Fabio board is already being made by a milling process, and one RepRap opto endstop circuit has been successfully milled (see example circuits below).
Existing and proven toolpaths are discussed on the PCB Milling page.
Circular saw route
Is it possible to cut PCBs with a relatively large-diameter (but thin) circular saw on a horizontal axis? (i.e., the same setup khiraly used to mill his bed flat: 1mm (0.040") thick 24 mm (15/16") diameter: Dremel #420 cutting disk.   or perhaps thinner wheels to allow finer-pitch SMT parts: 0.6 mm (0.023") thick: Dremel #545 Diamond Wheel. 0.6 mm (0.025") thick: Dremel #409 cutting wheels. ... is it possible to make or buy 0.4 mm (0.016") thick cutting wheels? ) Perhaps:
- mount the wheel North-South to cut the North-South isolation gaps (and gaps running at a small incline from N-S), and then
- re-mount the wheel East-West to cut the East-West isolation gaps (and gaps running at a small incline from E-W). (Or automate this with yet another motor).
Compared to most PCB routers (which use a small-diameter milling bit on a vertical axis), a cutting wheel:
- doesn't need the CNC machine frame to be as rigid: when cutting a N-S slot, the wheel pulls the head N or S, so very narrow slots with very narrow traces between them can be placed very accurately in the East-West position
- A (hypothetical) 0.40 mm thick cutting wheel has a larger abrasive surface and seems (theoretically) stronger than a 0.40 mm diameter milling bit, and so perhaps could cut faster and last longer before needing to be replaced.
Electrochemical Machining Route
With the first variant of this method, you move the tool head over the isolation gaps near (but not touching) the copper clad board, immersed in a shallow pan of saltwater. A positive voltage wired to the copper clad board and a negative voltage wired to the tool head drives a current through the saltwater that dissolves the workpiece near the tool head.
This was tested and found to be not really working, see Electrochemical Machining.
The second variant of this method doesn't move the toolhead, but covers the copper tracks to be produced with some paint, similarly to the etch resist method. Instead of submerging the board in an acid, it's also submerged in salt water and a voltage is applied. The copper in the uncovered areas dissolves.
The advantage over etching is, there are no special or hazardous chemicals involved. The problematic point is, the copper stops to dissolve as soon as the electrical contact to the positive voltage is dissolved away. You have to prepare for dissolving in a predefined direction.
Etching / Resist Route
With this method you would be applying some sort of etch resist to the copper clad board by means of the extruder head. There are a couple ideas that we have been tossing around and are interested in pursuing. We basically only need one way to work. The etching itself is a simple process that is basically foolproof if one is using the correct precautions. Simple objects needed to do the lab work could even be printed with a RepRap machine (etch bath tray / bubbler / heater / beaker, stirring rods, etc.) This would make the etching process much simpler.
- MakePCBInstructions -- how to make PCBs the traditional way, starting with copper-clad PCB and using an acid that can dissolve the unwanted bits of copper.
- these would be fairly easy to do and pose the fewest questions of if it will work.
- etching and semi-nasty chemicals are involved. not super fun.
- Does the Replath software (Builders/ReplathHowto) handle placing thermoplast resist or ink resist correctly?
This is the ideal route. You are using something that already exists in the form of an extruder. You simply need to find the right combination of plastic, extrusion rate, etc. It would be a very elegant solution.
- you already have an extruder, and practically no modifications required.
- you could print a holder for the board to get perfectly lined up 2 sided boards.
- the traces should be protected really well and thusly survive the etch process nicely.
- afterwards, the plastic could also serve as an excellent insulator.
- downside is that it may be tough to get the plastic off the board to solder.
ABS may end up being a material of choice in this process. It is temperature resistant (hot traces/soldering process), it has excellent insulation properties, and one of the most crucial: it dissolves in acetone. it remains to be determined just how easily this process occurs, however it is highly likely that the abs could be made thin enough for this to be simple. Also, if the ABS was left on prior to the drilling process, it might possibly serve as a nice guide for the bit before it hits PCB. It could also prevent/reduce burrs from forming during the drilling process. Afterwards, it could simply be cleaned off and the pads would be ready to solder.
also see Scratch n' Etch PCBs
Cover all the metal with a layer of etch-resist (perhaps dunk in melted wax, or paint with machinist's "layout blue" paint). Use a metal scribe tool to scratch off the etch-resist at "the right places". Drop the board in the same acid used in all the other etch/resist techniques. (This technique moves the toolhead over all the isolation gaps, like the milling techniques, the opposite of most other etch/resist techniques).
This technique has the simplest possible toolhead -- a pointed piece of metal.
laser diode as a UV source
Needless to say Please, safety first!
Lasers are dangerous, not only when attached on sharks.
With presensitized pcb's and a laser (<450nm-laser diode) such as a a140/m140-diode one could expose pcb's. Note: "DuPont Riston" type dry films are most sensitive to a wavelength of 365nm for which several high power diodes are available (June 2013) from eBay China. For a proof of concept/working example, see: LaserExposer (video).
There are "support for this" in the most common firmwares aka; "Fan-control through pwm" that is used with most RepRap electronics. Used with a "image -> gcode-software" and some "electronics-so-far-above-your-narrators-head", it ought to work™.
Cover all the metal with a layer of etch-resist (perhaps black spray-paint). Use a Laser Cutter to vaporize the etch-resist at "the right places". Drop the board in the same acid used in all the other etch/resist techniques. (This technique moves the toolhead over all the isolation gaps, like the milling techniques, the opposite of most other etch/resist techniques). Arcol has some very promising progress making a 2-sided PCB on a commercial laser cutter.
- Main article: plotting.
PCBs have been successfully made with the ink resist process.
Instead of using an extruder print head, the Mendel moves plotting head holding a marker that draws the circuit traces onto the copper-clad board. Then the operator drops the board in the etch tank, where the copper dissolves where it is not covered by a protective ink coating.
The holder itself could be printed out of RepRap parts. The pen holder toolhead does not need a controller or any electronic parts.
- Sharpie pens are cheap and universal.
- you would also then have the makings of a drawing bot, useful for making art.
- What pen and ink have good combinations of tip size, flow rate, and etch-resistance? testing needed.
Drilling the PCB
- Main page: MakePCBInstructions#Drilling the PCB
The comfortable way, of course, is to mount a Dremel-like tool on your RepRap. Forces required for drilling small holes are pretty low, so even wobbly Mendels can do this. PCB editors like gEDA/PCB can export G-code drill files directly. A useful feedrate is about 80 to 100 mm/min.
Doing this by hand is another task that is pretty mundane and repetitive. It also requires either extremely steady hands, or a drill press. Don't forget a fair share of patience, typical RepRap electronics boards have some 300 to 400 holes to drill.
For copper-clad boards, it's a good idea to use Liquid Tin to deposit a layer of tin over the copper. This will prevent oxidation and corrosion over time, and it provides a surface that is easier to solder. Liquid Tin is quite dangerous though, so observe all precautions (goggles, thick rubber gloves, etc.)
solder paste dispenser
Surface mounted parts are most easily attached by using solder paste. Fortunately for us, we theoretically have one, so before the SMT parts are placed, the syringe goes over each pad and deposits the required amount of solder paste on each contact.
This task appears to be a pretty daunting one to automate. It deals with components, except some of them have multiple contacts and a variety of form factors. Things could be simplified with moving to a reduced component part, and a technology more suited to automated production like SMT (surface mount technology)
SMT component placement Method
The pick/place head (Pick and Place ToolHead) deposits a component directly onto the solder paste which will ideally secure it in place until its ready to be soldered.
Through-hole Components placement
This is much more tricky, and it may be simpler to assemble by hand. However, through-hole ICs could be converted to surface mount by simply bending their pins outwards so they lie flat against the PCB surface. This would be easier than true SMT parts because less accuracy and fine detail would be required. However, the same technique would not be as easy to apply to other types of through-hole components, although it may be feasible ("surface mounting DIP/DIL package?") (FIXME: link to photos of NASA AGC and satellite PCB that use this technique)
After the surface mount parts have been placed on the solder paste, the board needs to be heated up. This can be accomplished with the help of a very common, simple household item: A toaster oven! This has actually been done before, and its a pretty elegant process. The solder will actually center the SMT parts, and will bead up nicely around each component leg. Here are more links on the topic from people who have actually done it:
- Have you seen my new soldering iron? (tutorial)
- another tutorial
- more technical and automated process
- some people claim skillet reflow is better than toaster oven reflow -- see HotplateReflowTechnique.
- do we want a Heated Bed hot enough to melt solder?
alternatives to standard copper-clad FR4 PCBs
- The first reprapped circuit was a RepRap opto endstop circuit (see examples below), a relatively solid chunk of plastic with some channels that were later filled with a special extruder that prints out molten solder.
- Or was this the first reprapped circuit?: Construction of PCBs using Silver Paint and EVA Hot-melt Glue. by VikOlliver 2005-May-05.
- "Copper Plating Wire Glue"
- Simple PWM Output Design - Spoondriver ???
- Instead of using copper-clad FR4 to connect components, somehow use copper wire placed by a SpoolHead and somehow stapled/glued/soldered to an appropriate substrate?
- Wikipedia: Occam Process
Fields Metal Deposition
This technique is definitely a departure from the norm of circuit manufacture, but its potential is still great nonetheless. The basic idea is that you use a heated extruder to melt and extrude Fields Metal, an alloy with an especially low melting point. Its lower than the plastic used to print with. Anyway, as you're constructing the part, you could lay down tracks of Fields Metal within the part itself, which leads to some very interesting options such as truly 3-dimensional circuits, parts with completely integrated circuitry, and a nicely integrated manufacturing process.
However, there are still some issues and unknowns involved. It is not known how well the circuits will stand up to regular usage (what if components heat up past the melting point of the metal) and it also doesn't lend itself well to automated pick and place processes. That being said, it is still promising and theres a good chance one of these techniques will pan out and become the dominant way of creating circuits with RepRap.
Field's metal liquefies at ~ 62 °C (144 °F) while most CPU's safely operate until 110 °F. melting the fields metel is unlikely. If we were create flow paths in the material itself we may be able to use the toaster oven method or HotplateReflowTechnique.
Further details: Automatic deposition of molten alloy into a casting channel to create a (very) simple electro-mechanical component by EdSells, May 2005.
Carbon black mixed plastic
This article from the University of Warwick describes how to mix PCL and carbon black to make a ~0.1Ω⋅m conductor, possibly a variant of multi head or mixing could print a conductor that way. They demonstrated capacitive and piezoresistive use with it.
How much precision does a automated circuitry making machine require?
Practically all through-hole components come with pins at 0.1" (2.54 mm) pitch. ... hole-to-trace alignment ... ... hole-to-hole alignment ...
DIP chips -- such as the DIP ATmega328 used in many Arduino boards -- all have 2.54 mm pitch (10 pins/inch). As mentioned above, it's possible to surface-mount such a DIP chip by "flattening" the pins. A PCB designed for this surface mounting technique may be easier to fabricate than a PCB designed for through-hole mounting the same chip, since surface mounting doesn't require precisely-aligned drill holes. It may be possible to design such a "flattened" PCB with 1.27 mm trace/space (0.050 inch trace/space): the pins rest on 1.27 mm wide parallel metal traces, separated by 1.27 mm isolation gaps. If some places cut away a bit too much ( 0.2 mm trace / 2.3 mm space), or other places cut away a bit too little (2.3 mm trace / 0.2 mm space), it will work just as well.
Many popular SMT microcontrollers have pins at 0.80 mm pitch (~32 pins/inch) -- such as the 32TQFP ATMega168 in the Fabio board or the 44TQFP ATmega644PA used in some other Arduino-compatible boards. Ideally the PCB has 0.4 mm trace/space (0.015 inch trace/space): the pins rest on 0.4 mm wide parallel metal traces, separated by 0.4 mm isolation gaps. If some places cut away a bit to much (0.2 mm trace / 0.6 mm space), or other places cut away a bit too little (0.6 mm trace / 0.2 mm space), it often will work just as well.
(Some SMT microcontrollers have pins at 1.27 mm pitch (20 pins/inch), giving 0.63 mm trace/space (0.025 inch trace/space) -- such as the Atmel AVR AT90PWM3B and ATtiny85. Alas, they don't appear to be Arduino-compatible).
If your cutter or etch mask has enough resolution to "fan out" metal traces from those 32 (or 44) pins of the microcontroller to a wider pitch to the rest of the board, it has plenty of resolution for the traces and all other parts on the RepRap controller boards.
The pins of a 0.80 mm pitch part must be less than 0.40 mm from their corresponding pads, or it will "jump" during reflow -- any particular pin will look properly soldered to a pad, but they all will be off-by-one soldered to the *wrong* pad.
- The first reprapped circuit was a RepRap opto endstop circuit; it was also used in a PCB milling experiment: "We are constantly pushing to increase the number of parts the RepRap can make for itself."
- single-sided, lots of holes: some Pololu Electronics adapter boards
- single-sided, lots of holes: the Arduino Single-Sided Serial Board
- single-sided, no holes: the Fabio Arduino compatible microcontroller board
- single-sided, no holes: "An ARM dev board you can make at home" by BarsMonster. I'm guessing this is an ARM STM32F100Cx which has 48 pins at 0.5 mm pin pitch. (Not the ARM STM32F100xC linked to from the article, which has at least 64 pins at 0.5 mm pin pitch).
- Open Source Ecology: routed circuit board
- video of Hydra doing PCB milling; the Hydra-MMM Prototype uses standard RepRap electronics on a highly customized multi-head Extruded Aluminum moving gantry RepStrap.
- "Adventures with ... CNC PCB engraving." ???
- "Printed Circuit Board Isolation Routing" tutorial
- Wikibooks: PCB layout
- The MillPCBS.com has information on how to mill PCBs using any CNC mill, and has useful utilities -- for example, PCB-GCODE converts Eagle .brd files to G-code. Presumably this would also be useful on a RepRap-based CNC mill cutting a PCB.
- "Mini CNC case study" shows it milling a TinyPOV PCB and a (much easier to mill) blinkie board.
- PCB manufacturing on a mill controlled by LinuxCNC has some tips on bit selection and speed.