CURRENT ISSUE
Contests
FEATURE ARTICLE
|
|
Issue #216 July 2008
Second Place Microchip 2007 Design Contest
’Net-Enabled Alarm Clock
by DJ Delorie
Start | System Overview | Network | Display | MP3 | ADC | Memory | Power | Software | Time | Alarms | GUI | Remote Protocols | Construction & Packaging | Smart Combinations | Sources & PDF
CONSTRUCTION & PACKAGING
The schematics for this project were created using the gschem program from the gEDA suite (S. Brorson, “Optical Proximity Sensor for Robots Part 1: Simple PCB Design with the gEDA Suite,” Circuit Cellar 188, 2006). The board was designed using the pcb program (S. Brorson, “Optical Proximity Sensor for Robots Part 2: Open-Source PCB Layout Editor,” Circuit Cellar 189, 2006).
While the board is a four-layer board, there are a number of PCB fabrication companies that offer affordable four-layer boards, such as PCB-Pool, Advanced Circuits, and Sierra Proto Express, some of which advertise in Circuit Cellar.
I chose to use a four-layer board for this project because I wanted solid ground and power planes for reliability and simplicity. With prototype fab prices dropping regularly, there’s little excuse for avoiding four-layer boards when the project calls for them. Having a solid ground plane under the majority of the signals helps reduce crosstalk and provide cleaner power to the chips. However, there are some exceptions to this. There is no ground plane under the switching power supply because you don’t want the ground plane to pick up the spikes from it. The three crystal circuits have no ground plane either, both to avoid stray capacitance and to keep noise on the ground plane from getting into the clock circuits.
The power plane is divided into the 3.3-V region and the raw “15-V” region. The raw power region passes under each of the regulators to provide full power to those, as well as to each amplifier. The 3.3-V region passes under everything else, except of course the crystals and switcher.
The traces used for the switcher are large and close together. This reduces their inductance and the “loop area.” The circuitry forms a crude antenna, so reducing its area reduces the amount of noise it transmits.
To prepare for debugging and potentially modifying the circuit, I made sure that all of the vias were untented. This means that they’re not covered with solder resist, so they can be soldered to later, and the hole size is just right for wire wrap wire. I also laid out a number of bare grids on the back where SOIC and 0603-sized components can be soldered in case more parts needed to be added later.
Contrary to popular belief, soldering SMT parts is not hard. Once you get used to it, it’s easier than through-hole because you don’t have to keep flipping the board over. My technique is to use a stencil to apply solder paste to the right parts of the circuit board, manually place all of the SMT parts, and use an electric hotplate to reflow the solder. I got this idea off of the Internet, but it’s proven to be easy and reliable. If there are any solder bridges, it’s easy to use copper desoldering braid to remove them. After all of the SMT parts on the top are soldered this way, the SMT parts on the bottom can be added using solder paste as before, but using your iron to reflow the paste. To do this, apply the iron to the board and/or part and let the heat melt the paste under the part. Lastly, the through-hole parts are added.
Previous | Next *| Subscription Deals |*
|
