How I found work that still scratched the itch after RTL writing stopped being the job

For me the answer ended up being firmware on microcontrollers, wich I stumbled into almost by accident. Our group got downsized and I took a support role writing drivers for an STM32 part, and it turned out I liked it more than I expected. You still get that feeling of poking at registers and watching something respond, except the feedback loop is faster and nobody hands you a netlist to argue with.

The thing I miss is the math. Timing closure on a design pushing 300 MHz felt like real engineering in a way that bit-banging an I2C bus never quite does. Some weeks that bugs me. Most weeks I am fine with it, because I go home on time and my weekends are mine.

One thing I whould add to the advice above: try before you leap. I spent two months doing evening projects on a cheap dev board before I applied anywhere, and that told me the itch was about building things, not specifically about FPGAs. Saved me from chasing the wrong job.

I found that same itch but it led me toward systems integration using field bus protocols. I started working with industrial sensors and the documentation for the RS485 protocol really clicked for me. It was less about optimizing a complex clock tree and more about physically wiring up and verifying the data flow across multiple devices. The fun was making sure the bits made it from point A to point B. I think the desire for that immediate feedback might point more toward physical instrumentation than pure digital logic design.

For me the physical layer was the thing. I found that working in hardware validation for power systems was a really solid fit. I wasn't writing any logic, just tracing signals with an oscilloscope and making sure the power delivery kept the whole setup stable. It was pure troubleshooting, chasing a voltage drop or a noisy clock signal. That felt like the same kind of detective work as FPGA debugging, just on a totally different scale. If you enjoy that process of chasing a bad trace on the scope, you might look into system diagnostics or physical testing platforms. It keeps your fingers on the breadboard without the huge HDL codebases.

Most folks today don't appreciate the grind of hand-optimizing layout. When I was running these chips back in the day, if you wanted a decent clock speed, you were spending half your week tweaking trace lengths and signal integrity. It wasn't just about the logic; it was about the physics of the metal. I remember spending hours just fixing a noisy power plane because the sim didn't catch it. I ended up moving into physical design verification. The focus shifted from what the logic did to how the silicon behaved under stress. If you liked the feeling of seeing a circuit actually behave how it was designed, but hated the endless HDL tweaking, that might be the move. You're still dealing with the whole hardware stack, just at a deeper, less algorithmic level.