Week 5: Designing a Hello World PCB Board (Echo Board)

Idea

This week, I am going to design my first custom PCB board from scratch. The functionality of this board will be rather simple – but it's also the first board I am ever designing. The board will be based on the ARM SMD D11C chip and has an LED connected to one and a tactile switch (button) connected to another pin – giving us a simple Hello World (Echo Board) that should light up the LED when pressing the button. I used Eagle (fully integrated into Fusion 360) to design the board, a 1/64 in endmill to mill the traces, a 1/32 in endmill to cut the board, and the Arduino IDE to write the code for the board. The design is based on Neil Gershenfeld's echo board and code, which can be found here (scroll down to "ATSAMD11C").

Designing the Echo Board

This was the first time I designed an electric circuit. I decided to use Eagle, which was bought by Autodesk and is fully integrated into Fusion 360, since I am a Fusion user already. Moreover, the Fusion integration makes it easy to switch between a view of the CAD model that houses the board, and the board itself with correct dimensions. I started by following along Jake's great tutorial "Eagle in 10 Commands" to get familiar with the most common commands. Although it took me about twice as long to get through the tutorial, it greatly sped up my progress afterwards. A great starter!

Before starting my design, I imported the design rules (.dru) from the class and the library (.lbr) for the footprints of the components I wanted to use, which can be found in this repository. First, I added all components I wanted to use in my board, and roughly organized them in the way I wanted to assemble them on my board later. I also added value labels to my resistors and capacitors to later remember which one to use during soldering. Next, I used nets to wire up all components accordingly. The name command was especially helpful here, since it allows you to connect two pins without drawing a wire all the way throughout the design. Basically, I compartementalized all parts of my board into the individual "pockets of components" on the board.

Adding all components to my board schematic

Using Google to find the pinout of the microprocessor...

...and then finding the right pin in my Eagle schematic

Last, I used Eagle's design rule check (DRC) to see if I had any errors with my board. Now I think is when I realized that Eagle is much more powerful for designing circuits than just a graphics program. I quickly fixed up the errors I had and then moved putting the PCB board together.

Finding errors in my schematic using Eagle's DRC feature

The final schematic of the board

Designing the PCB board

After finishing the schematics, I switched to Fusion's PCB tab. I began by placing the components onto the plane and playing around with the arrangement of them until I found positions and orientations that gave me the least overlap (and headache) with respect to the wiring.

The components without any arrangement

The rough arrangement that leaves the least overlap I could achieve

I continued by wiring the components according to their airwires. Unfortunately, some of the wires I wanted to fit under components did not go under the components I wanted to run them through (the mill clearance did not allow for these traces to fit under my components). As I could not find any way to fully eliminate this problem, I had to add a second 0-Ohm resistor as a bridge over a trace. Furthermore, I routed some wires very close to each other under the processor without any necessity to do so. I moved these wires further apart, not because it was required by the design rules, but because I felt it would just make it easier to manufacture. Last, I drew the board outline around the board.

I wanted to run this connection under the 2x5 connector, but it did not fit

The USB data lines fit nicely under the microprocessor

The arrangement of the LED, a resistor, and a tactile switch (button)

The finished PCB board

Before moving on to exporting my board design, I ran a design rule check on the PCB board to make sure everything was right. Again, I love this feature, since it allowed me to quickly fix up some minor issues and know that my design is printable by the endmill. Luckily, just before exporting, I realized that I should have added a pull-up resistor to the button to prevent its input pin from reading a floating input (if you want to know more about this problem, this YouTube video by AddOhms helped me understand it greatly). So I went back into the schematics, added a resistor from the button input pin to the 3.3V line, and tried to fit it onto my PCB board in the PCB tab. To save space and avoid rewiring the PCB, I fit the resistor under the for now button and hope it will fit under it.

The pull-up resistor I added later to my schematics

Checking and fixing my final board with a final design rule check (DRC)

The finished traces to be milled with the 1/64 in mill

The finished board outline to be cut with the 1/32 in mill

Milling the board and soldering the components

I was confident in my design and made sure both exported PNGs with the traces had just the areas to mill in white (there were some issues around the border of my exported image, which I fixed manually in Photoshop). I exported both designs to the endmill machine as PNGs. After mounting a fresh copper board and fixing it with double-sided tape at the bottom, I started milling the traces. Unfortunately, I discovered that although I adhered to the design rules and did not override the spacing, the wires came out very thin and fragile. Moreover, one section of the 3.3V and the ground wires on the bottom were not cleanly separated. At this point, I realized I could have given all wires a wider with from the start, as space was not too limited for the board I designed. I selected all wires as a group and used the change tool (wrench icon) to change the wire width of all wires from 0.15 to 0.39. I also manually redrew some wires to give them more space.

The milled traces came out quite thin, and there was a short in the bottom

The wires with thin wiring

The wires with extended wire width and extra spacing where possible

Both interventions fixed the problem with milling. When re-milling the traces, they came out cleanly separated and much stronger than before. Cutting the board with the 1/32 in mill also went clean. After scratching off any excess copper with a metal ruler and washing the board with dish soap under water, I proceeded to soldering. For soldering, I again stuck the board under the microscope with double-sided tape and worked entirely under the microscope. Again, I used a very thin lead wire at a temperature around 750F and needed no extra flux. The learnings I had written down two weeks ago when soldering my first board under the microscope (here) were extremely helpful here and I could solder all components on within about an hour.

The fixed board with wider trace width and clearances

Soldering the components under the microscope

The finished and soldered board

Programming the Echo Board

To finish this week's project, I tested and programmed the Echo board with a script that sets the pin of the button to an input pin and uses the LED as an output pin. I used the programmer I built in week 3 , in its programmed state, to write the bootloader and then Neil Gershenfeld's Echo Board script to my Echo board, by connecting the 5x2 output (TARGET) side of my programmer to the 5x2 input (SWD) side on my Echo board. I used OpenOCD to write both to my board which went without issues. After writing the script to my board I tested it's functionality – it worked just as planned and the LED lights when the button is pressed. Way to go to design more complicated circuits, but this is a great success for me for this week.

The programmer and the Echo board are both connected through a 5x2 connection

Writing the script to my Echo board

The finished and functioning Echo board in action