Week 5 Assignment: Electronics Design

Home Final Project Week 1 Week 2a Week 2b Week 3 Week 4 Week 5 Week 6 Week 7 Week 8 Week 9 Week 10 Week 11 Week 12 Week 13 Week 5 Assignment: Electronics Design

Week 5: Electronics Design

Based on an echo hello-world board, design and fabricate a PCB incorporating a button and LED with at current-limiting resistor.
Demonstrate the programmability and operation of the PCB.

As with most facets taught during this class, PCB design was a completely new topic for me. Having used Fusion 360 previously, I decided to use its Eagle package for designing my PCB. For milling I used the same Roland SRM-20 end mill described in Week 3. In the design of my PCB, I wanted to incorporate capabilities that could be used for testing my final project prototype. Programming this board would be beyond my current capabilities, but developing a functional layout with MOSFETs, motor diode protection, and pull up resistors were the learning objectives for me this week.

Autodesk Eagle Application

My PCB design is based on Neil’s hello.D11C.serial.5V.1 board which utilizes a ATSAMD11C microcontroller. Adding to the base design, I incorporated a circuit containing a button and LED that can be controlled via the MCU per the week's assignment (outlined in blue below). I also added an area to incorporate an additional power source (e.g., 12V) and set of N-type MOSFETs to power electromagnet coils to drive an induction machine (outlined in orange below). To provide opportunities to expand use of the PCB, I added resistors and a connector to jump out unused pins of the MCUs (outlined in green below). To simplify trace routing I opted to create a board with the bottom layer set to ground with connections to ground made through vias. As will be discussed in detail below, the incorporation of vias proved difficult with a number of lessons learned. I used 20 mil width traces throughout the 12 V supply region of my board and used 10 mil traces everywhere else. The lower right image shows a 2D depiction of my board with additional component information and is hyperlinked to the CAD file. The bottom image is a 3D rendering of my board.

Note, 3D rendering shows a 12V barrel jack that is not available through the fab lab inventory. This component was used to provide an adequate footprint for soldering a wired barrel jack to the board.

After altering the display of my schematic document in Eagle, I used the export image function to create monochrome .PNG trace and outline files with a resolution of 1000 dpi. Images of these documents are shown below and both are individually hyperlinked.

Milling the traces of my board with a 1/64 in end mill progressed well, except for one 10 mil ground trace from the programming header that lifted (circled in red on the image to the right). Based on the end mill design rules completed during week 3, I did not expect this to occur. I failed to inspect the end mill prior to milling and suspect copper debris build up on the end mill may have caused with discrepancy. However, to help prevent this in the future I would recommend increasing trace width and/or .PNG file resolution to limit 'staircasing' along angled trace runs. While milling the outline with a 1/32 in end mill, I immediately noticed a discrepancy while milling the board's vias; they were significantly oversized (circled in yellow on the image to the right). After recognizing this I stopping further milling of the outline.

With several students waiting to use the micro end mills in the EECS lab, I decided to remove my board rather than attempting to correct via sizing on the .PNG file. I used a drill press to finish drilling via holes and a band saw to cut the board’s outline. From here I proceeded to weeding and deburring the board with tweezers, needle nose pliers, and a razor blade.

Prior to soldering on components, I focused on tackling my oversized via hole quandary by attempting to create a solder bridge between the top and bottom layers. As one would expect this was unsuccessful, because the solder only flowed towards the heated side. I then made the poor decision of applying additional heat with a heat gun to simultaneously heat both sides. This resulted in overheating my board and caused traces and pads to delaminate from the backing material. In hindsight I should have used short wire segments that I could have routed through the oversized via holes to create a path to ground between the layers. Looks like it is time to head back to the end mill, but first I need to figure out my via sizing issue.

After reviewing my schematic in Eagle, I found I used vias with an outside diameter of 32 mils, which is nearly equivalent to the diameter of the 1/32 in (31.25 mil) end mill being used to create them; this checks out. In looking closer at the via component symbol, which is shown to the right, I realized while adjusting color of my .PNG outline file in Paint, I used diameter 2 instead of diameter 1. With this correction I was more confident with via sizing going into round 2 of milling my board.

My added confidence was quickly subdued while milling the traces of my 2nd board. After an audible change in radiant noise coming from the mill, I noticed the end mill was no longer cutting through the copper. I stopped the mill, verified the origin, and reset the z-axis height. I determined that I had not tightened the end mill's set screw enough and it shift up slightly while milling. After this correction there were no further issues with milling the traces. I will point out that I increased most trace widths to 20mils and used the same 1000 dpi file resolution. After milling I pressed 0.6 mm vias into the via holes and soldered into place on both sides.

After soldering vias into place, I verified continuity of all ground traces to the bottom layer with a multimeter. My general soldering skills are gradually improving, but I still struggle with heat control and patience. I then discovered another discrepancy in my design, the USB port was slightly oversized. I removed excess material from the USB port with sand paper, but pulled off a jumper connection in the process (pads and all); luckily I had immediate plans for using this.

While attempting to program my board a fatal fault was discovered; my programmer pins did not align with the correct legs of the microcontroller. Looking back, I must have swapped between two base design images while developing my board in Eagle. The traces between my programmer header and MCU were designed for a ATSAMD11C microcontroller, but I used a ATSAMD11D footprint during my design. Lesson learned in reading the fine print. Well, the saga will continue as a return to the mill for round 3 (see later weeks).

D11C Base Design

D11D Base Design

I will not bore with the details of Round 3, but have provided images and file hyperlinks below. I made several alterations to the design, such as MOSFET heat sinking and ground trace routing, but general functions remained the same. This board was successfully programmed and used for coding assignments during future weeks.

The group assignment for this week involved using test equipment to observe the operation of a microcontroller circuit board. The EECS lab is outfitted with a number of multimeters which are helpful in checking continuity, measuring resistance, and identifying shorts. In addition to multimeters, the EECS lab also has oscilloscopes, which provide a time domain plot of voltage. An oscilloscope can be used to observe the operation of a microcontroller circuit. Using an oscilloscope, Anthony demonstrated a phenomenon known as button bounce, which involves a button only creating an intermittent connection when it is initially depressed. This bounce is not noticeable to the human eye because its duration is measured in milliseconds, but can be identified using an oscilloscope.

The oscilloscope window to the right shows the button bounce I was able to recreate while coding my board in a later week. Note, a bounce is visible on both the button depress and release. I used a Keysight DSOX1204G Digital Storage Oscilloscope to capture this image.

In your design program, set design rules for trace and pad spacing based on end mill sizing (e.g., if milling traces with a 1/64 in end mill, set spacing to 16 mils).
Know your MOSFIT (P-type verse N-type). I used N-type MOSFETs which are placed downstream of a load with the drain wired to the load, source wired to ground, and gate wired to a MCU pin in my case.
For motor diode protection, the diode must be placed in parallel with the load and orientated in the opposite direction of current flow.
Recommend exporting .PNG files with resolution greater than 1000 dpi if milling diagonal traces smaller than 12 mils.
Avoid placing board outline along the edge of your export file, because mods will reject milling edge lines due to clearance limitations. This can be circumvented by lying to Mods and entering a tool diameter of zero.
Remember the Roland SRM-20 End Mill, similar to the Roland Vinyl Cutter in the EECS lab, mills off set from black to white. Pending display options, your outline file may need to be inverted in Mods to produce solid white vias surrounded by a black board.
Instead of plug soldering a board hole, use a wire segment to form a bridge through the hole.