Addicted to Progress
Sunday, February 8, 2004
Note: much of this entry will sound like gibberish to most of you. Just smile and nod.
It's official: I love EE.
I spent my whole weekend on my EE 140 project (due Thursday)—and enjoyed every minute of it. We have to design an amplifier using only MOSFETs and IC resistors. Driving a 50-ohm load resistor, the amp has to have a voltage gain over 500. Our design goal is to minimize (power * area). Transistors and IC resistors have a specified minimum width and length.
On Saturday, I was reluctant to start. I spent the morning alternating between thinking about the project and playing Fire Emblem. The task seemed daunting, especially since I got schooled by a similar assignment last semester in EE 105. However, once I got started, I couldn't stop. Jumping between HSPICE and hand calculations, I worked from the afternoon until 2 AM. This morning, I was back at it. I didn't stop until I reached a finished product at 7 PM.
I worked like a fiend. I constantly scoured the HSPICE output and my scribbled notes, looking for ways to optimize my design. The data was the packaging; the optimizations were the drug. At first, it was easy to find optimizations. But as my designs got better and better, the longer I had to fumble through the data to yield the nuggets I desired. 40,000+ μm2 (don't ask) to 606 μm2, 23 to 12.8, 8.5 to 5; I couldn't get enough.
I ended up with a 6-stage amplifier. Five NMOS, common-source stages (without source degeneration), each with a diode-connected transistor (instead of a resistor), generate over 20k-worth of gain. The PMOS, common-drain stage at the end pisses away all but the necessary 500 gain—an intentional move to keep the size of the PMOS small.
3 μm2, 16.8 mW. I love EE. (I might be abusing the level-2 MOSFET model. I need to check with hand calculations...)
(excuse the look of the archive)