Not quite so long ago, we would be using techniques that, to some degree remain quite appropriate for circuit construction—I refer to a variety of bread-board, point-to-point, wire-wrap, or dead-bug construction. Printed circuit boards (PCB) are increasingly required for some circuits, have become relatively inexpensive, and can accommodate many of the prototyping techniques. Surface-mount components inevitably require PCBs. The small, light, reliable, and easy-to-copy attributes of the PCB are offset to some degree by the difficulty of modifying and the greater effort of planning and execution that the PCB requires.

The Community PCB Mill Project lies towards the end of a chain of tools that are demanded by the greater effort that the printed circuit board requires. The purpose of this article is to describe the tools that lie between the circuit concept and its rendering by the PCB Mill, whether the Mill drills a chemically etched board, or the Mill both etches and drills the board.

More recently, we use the computer to enforce the rigor of the process of capturing the conceptual design and rendering it (eventually) onto paper while simultaneously allowing us to revise the concept relatively easily. The design capture tool, Computer-Aided Design (CAD) software, is essentially a tool that allows us to draw pictures on a computer, save that picture, and alter that picture, and (if the need presents itself) actually print that picture on paper. Depending on the focus of the CAD tool, the design that is captured can be as varied as the manual processes allowed—for this discussion, we will be limiting ourselves to the CAD tools that permit us to capture pictures of a PCB:

• the logical electrical connections of the circuit (Schematic Capture);
• the arrangement of the electronic components of the circuit (Placement);
• and the arrangement of the wires that actually connect the devices to each other (Routing).

A large body of knowledge specific to electrical circuits and devices is embedded in the CAD tools mentioned. This allows the CAD tool to act as an auditor for the designer of the circuit. To some degree, the CAD tool has sufficient understanding of the specific problem domain that circuits can be automatically generated after the tool has asked a few questions of the user (for a simple example, take a look at this 555 timer designer; and a more complex example, check out this power supply design tool).

We will use, by way of example, information that Parallax publishes about the Basic Stamp 1 OEM.

More recently, we have applied machinery to allow us to faithfully produce many copies of the design.

Even more recently, we have developed means to have the computer transfer the design articulated to the CAD tool to the machines so that the machines can quickly and inexpensively mass-produce identical copies of the design. This process of having the machines render a physical instance of a design is called Computer-Aided Machining (CAM). We will limit the focus of Computer-Aided Machining for this article to the production of printed circuits boards. The completed circuit board has the components connected and arranged according to the CAD design of the circuit. We will further limit, for this article, the focus of CAM to the production of the circuit board itself.

In the same way as for CAD, a vast body of machining knowledge can be embedded into the CAM tool that will influence the tool's output. Sometimes, this machining knowledge can be very machine-specific. Once a particular machine has reached a degree of popularity, the CAM tools from various vendors will support that machine. This has the effect, then, that when another machine-tool manufacturer wants to introduce a new machine to the marketplace, the CAM interface of the popular tool is built into the new machine.

Thus, it is that certain methods of communicating the design to the machine have become standards that the machinist and tool-builders cannot generally ignore with impunity.

By way of analogy, a programmer uses a CAD tool (whether a simple ASCII editor to a full-blown UML graphical tool) to capture the design of a software design, and uses a CAM tool (the compiler/assembler and the downloader) to communicate that software design to the microcontroller to execute that design. The machine code that is squirted into the microcontroller is the analog of the CNC language.

In essence, the CNC language is a script of motions that the machine is to follow to make an instance of the design. The CAM tool processes the CAD information about the concept into this motion script, sends it to the machine; assuming that the machine is a PCB mill, that Milling Machine then follows the CNC script to move its cutting tool so as to cut the outline of the PCB traces, and then to drill the pad-holes through the board.

For a variety of reasons, the defacto standard (and one of the most wide-spread) of CNC language is called G-Codes. G-Codes are also called by the name of the ANSI standard number RS274. I hasten to point out that there are other CNC languages, many of which tend to be isolated to rather vertical applications.

The manufacturing of printed circuit boards is a rather vertical industry. While general-purpose CNC machines are present, specific purposed machines quickly were developed. As the machines specialized, subsets of the G-Codes language were developed. Two of these subsets are known as Gerber format and Excellon format.

Let's assume the schematic capture CAD tool we have only outputs Gerber files. The commercial board house we have selected requires Excellon files for drilling our PCBs. Instead of getting a new CAD tool, or a new board house, we can elect to just do a Post Process conversion of the available Gerber file into the required Excellon file.

Back in the days when punched paper tape was all the rage for CNC program storage, it was also realized that any reduction of the amount of data to be stored would yield a significant savings, if not money directly, certainly in filing cabinet volume. The G-Code motion scripting language was purposefully invented as a terse, but rather flexible, language.

The basics of the G-Code language syntax can be summarized as follows. The command line is composed of a series of numbers interspersed by letters. These all have quite specific meaning. Nearly all of the letters of the English alphabet are used—these are little more than names of specific registers in the CNC machine. The number that follows any letter is the value to store in that register. As the CNC machine parses the G-Code input, it stores the stated values into the named registers. Typically, when the parser reaches the end of the line, the machine executes the motion command based upon the current content of all of the machine registers. The scripting language allows that the G-Word (or many of the other Words) need not be repeatedly specified for sequential similar action—one only needs to identify the desired change in axis position to re-trigger that command.

The G-Word (the combination of the letter 'G' and a number) is called a preparatory function. All this means is that the G-Word is the command that tells the machine what it is to do; other words are essentially parameters for this action, and the action is triggered when, typically then end of the command sentence is reached. Approximately 100 preparatory functions are defined in the RS274D G-Code standard; machine tool manufacturers sometimes extend this list.

Some of the G-Codes deal with commanding motion in a certain way; others toggle between Inch/Metric measurement systems or act as switches to influence the machine's motion planning logic. Still others deal with calling subroutines, or determine the compensation to motion planning that must occur due to geometry of the cutting tool.

Of all of these, I'll discuss in detail the ones that intimately involved in command linear motion. These are the G-Codes that would be the basis of PCB milling or drilling. Do understand that all commanded motion starts with the current position of the axes, and axis words specify the end-point of the motion.

As already mentioned, there are about a hundred variations on the G-Word, depending on the particular machine's implementation. You can look up the meaning of the additional preparatory functions in the references provided below.

Other of the miscellaneous functions are enumerated in a number of the links below. Just like the G-Codes, depending on the particular machine's implementation, upwards of 100 or more M-Codes may be defined.

The T-Word is used to identify a particular tool (and sometimes tool geometry information) that the machine is to select for the following operations. The D-Word is often used to specify either an additional angular motion, a feed function, or a function secondary to the current tool—D-Codes specific to PCB manufacturing are discussed below.

The F-Word establishes the effective feed-rate of the tip of the tool (how fast it moves); for interpolated motions in multiple axis planes, this is the velocity of the resultant movement vector. The S-Word sets the rotational velocity of subsequent spindle commands.

The N-Word is kind of label within the G-Code file—it is similar to the BASIC line number/label. The O-Word identifies particular subroutines—again comparing to BASIC, this would be the label that the GOSUB would call. Differing levels of branching and program structure capabilities are present in the various G-Code processors, depending on the toolbuilders requirements; some provide rather complete scripting capabilities.

The machine's linear axes of motion are labeled, typically as the X-Axis, Y-Axis, and Z-Axis. U, V, W, and P, Q, R are additional labels for axes, parallel to the X, Y, Z axes, respectively. A, B, C are labels of axes that rotate around the X, Y, Z axes. Each of these axes has a corresponding register. The meaning of the value stored in these axis registers is nothing more than where the machine should have positioned the tool/work-piece on that axis by the end of the movement.

The I, J, K words are mostly used as interpolation parameters. These will be often found in the circular interpolation commands G02 & G03. Think of them as just specifying a secondary coordinate on the X, Y, Z axes for some motion sequence.

So, that pretty well uses up the alphabet. With this brief review of the G-Code commands, let's move on to how these are used in PCB fabrication.

The Gerber data file format is an adaptation of the G-Code standard. Much of what has been said about G-Codes applies directly to Gerber Files. The tools that are used by the photoplotter are the apertures; these are selected by the G54 G-Code. The D-Codes are used to draw the image on the film.

Recently, there has been a push to advance the Gerber file standard towards what is known as RS274X, an extended Gerber format. RS274X only really differs from RS274D specification by the definition of some new parameters (for more information, see the "Gerber RS-274X Format" in the Selected Links. Remove these parameters; you have a standard G-Code file. It seems that one big advantage to the extended Gerber format is that what used to be several files that described a circuit board layer is now combined into one file. Fewer separate files mean that there is less to go wrong in the communication between a customer and the board fabrication facility.

Let's take a look at the Gerber files that details how to make a Basic Stamp 1 OEM circuit board (for this tutorial, the APT, GDG, & GDD files will be ignored). Let's pick, say, the Top Pad Master (GPT).

bs1Gerber.zip

GTO Top Overlay (Top Silkscreen) GBO Bottom Overlay (Bottom Silkscreen) GTL Top Copper Layer GBL Bottom Copper Layer GPT Top Pad Master GPB Bottom Pad Master GTS Top Solder Mask GBS Bottom Solder Mask APT Aperture List GDG Gerber Drill Guide GDD Gerber Drill Drawing

This pattern of selecting an aperture, and flashing that image on the film at a series of positions is repeated until all of the pads for the top layer have been imaged on the film.

Having looked a bit into the innards of the Top Pad Master (GPT) file, we can now peek into the Top Copper Layer (GTL) file. We can see the same extended Gerber parameters included in this file.

The most significant difference between Excellon Files and Gerber files is that the Excellon format files script out a specific drill bit selection, and each X,Y coordinate in the drill file specifies the location of a hole to be drilled. In other words, the Excellon drill machine selects the bit commanded, and then moves the drill to the specified coordinate, plunges the drill to create the hole, and retracts the drill. This process is repeated for all holes that use that particular bit. Then the next size of bit is selected, and all the holes for that bit drilled.

See, just a machine process simplification, or subset, of what is can be now viewed as a standard G-Code motion scripting language. It makes sense for something like PCB drilling, but the Excellon script might be inadequate for a number of other types of machining processes. Incidentally, most anyone that makes PCB drilling machines enables their machine to understand the Excellon motion scripting language.

Going back to the Basic Stamp 1 OEM example, let's now examine the Excellon Drill Files (in the same ZIP file as the Gerber Files).

bs1Gerber.zip

DRR Drill File Report DRL Excellon Drill File (binary) [not present in this ZIP] TXT Excellon Drill File (ASCII)

The ASCII Excellon Drill File (TXT) starts out with this preamble.

Then it moves right on to the business of drilling the holes: use the T-Word to select the drill bit, and then enumerating all the holes to be drilled with that size of bit (one hole per line). The instructions for all of the first size of hole are shown.

The binary Excellon Drill File (DRL) has not been included by Parallax in this collection of the Gerber files that detail the fabrication of the circuit board. In this case, it is likely that the board house they have selected for production runs of these PCB use the ASCII format. The binary Excellon file has essentially the same information as the ASCII file, but a format compatible with paper tape readers from the time when paper tape reigned as the NC storage media of choice. Don't you think that the ASCII G-Codes are much nicer, anyway?

There is (typically) no difference in the Gerber/Excellon files that define positioning of the holes to be drilled in the PCB.

And since this Community PCB Mill project is oriented for the hobbiest/prototyper, we are largely not very interested in the solder mask and silk-screen layers. In any case, these would be applied by a process other than the PCB Mill.

The biggest difference is in the G-Code files that specify the circuit paths and pads. The PCBs destined to be chemically etched have the traces drawn on the board's chemical resist—that stuff that prevents the copper layer from being dissolved by the etching chemical. When the board etching is completed, the chemical is flushed off, and the resist layer washed away. What is left is a network of flat wires in the PCB. Chemical etching is the process assumed by the Basic Stamp 1 OEM circuit board Gerber Files.

For PCBs that are going to be fabricated by a milling process, the G-Code files that are required, instead of the tracing path, actually an outline of the trace's path. This difference is subtle but important. Both methods involve removing copper metal from the PCB substrate to isolate one trace from another. The PCB mill must do this by drawing around the trace. Perhaps a picture will help—take a look at the first page of this PCB prototyping tutorial.

Some schematic capture tools emit the set of G-Codes required for PCB Milling operations. Lacking this, there are converters that will read a set of Gerber Files for chemical etching and output a set of G-Code files for milling PCBs; it is probably better, however, to get a software package that will process the required PCB milling G-Codes direct from a CAD file, such as DXF, or direct from the schematic capture tool's data file..

But specific software is outside the scope of this article; which ends with this list of reference links.