This summer I took a class on power electronics through MIT, focused on voltage management and dealing with higher-power systems. It taught me a ton about PCB design as well as schematic design. It helped me conceptualize a board that I just finished designing for a secret new venture I’m
working on – my most ambitious project yet!
In fact, I’m excited to share that I recently finished the schematic and layout for my first 4-layer PCB design.
Why a 4 layer board?
The first thing that I occurred to me – I need a very complex board. I quickly realized that having four layers to route signals was going to be necessary on my new board. Two wires can’t cross on any given layer of a board, and a 4 layer board offered me versatility with routing that a 2 layer board couldn’t.
4-layer PCBs vs 2-layer: What’s the difference?
Many of my readers may know that 2-layer boards are simply two layers of copper separated by an insulator. 4-layer boards are more complex, with four layers of copper (2 on the outside, 2 on the inside), separated by insulators. However, working with four-layer boards isn’t just “add two extra layers for routing and call it a day”. When you consider that 2 layers are sandwiched inside a 4-layer board, you have to take into account thermal dissipation and the fact that the copper is typically thinner for these inside layers. Traces have to be bigger on the internal layers in order to counteract this.
2-layer PCBs are cheaper to get manufactured and work for simple applications. In fact, my model rocket control boards STARLIGHT and STARLIGHT MINI are both 2 layer board designs. And, 2-layer boards are easier to understand and debug. All layers are exposed to the air, so if you have an issue with one of the traces you can see it. With 4 layer boards if the issue is in an inside layer, it is more difficult to identify the issue. This means 4-layer boards require a lot of trust with your manufacturer.
So… what kind of board did I design?
After all of this research and hours of labor, I was able to complete my PCB design. The PCB is designed to be a 350w peak power brushless DC motor controller. It’s run by an on-board DSC and tons of support circuitry in order to allow the board to take a hybrid drive approach. The board uses both square wave control and sinusoidal control. Square wave control with Hall sensors is much better at low speeds for getting the motor up and running. However, square wave control at higher speeds leads to choppy acceleration, loud running noise, and an overall worse feel. This board is designed to have both Hall sensors AND sinusoidal back-EMF control. Hall-sensored square waves are used to get the board up to speed, then the controller automatically switches to sinusoidal control for maximum efficiency and reduced noise.
I designed the board with four different voltages – 36V, 12V, 5V, and 3.3V. Unfortunately, the parts I’m using don’t all play well with the same voltage. For this reason, the board takes 36V as input and to drive the motor, but uses three switching voltage regulators for 12, 5, and 3.3 volts (for logic, hall sensors, and power MOSFETS).
What was the biggest challenge with all of this?
Honestly, the largest amount of work for me was the parts research. I had to spend a good deal of time figuring out what parts I needed and researching semiconductor manufacturers’ component options. The challenge was in selecting components that are not only in stock but new enough that they won’t be phased out in the near future leaving me out of luck. The actual board design was not too difficult as I have experience with PCB design, but it was tedious and took several hours.
I’m excited about the next steps for this project, and I think it will be well worth the time and effort. And I learned a ton. Stay tuned for updates on my secret application in the coming months!
