Hi, I'm trying to power my boost converter from a microcontroller and I've encountered stability problems. I tried to make pseudo-PI controller, a PID controller and few other tricks, but the unloaded output just doesn't want to be even close to stable.
Of course, I'm using a buffer between my MCU and the power mosfet (MCP1407). I run it from 72MHz clock on 100kHz, so PWM has limited resolution (only 720 steps). Of course, having the output voltage match one value of PWM is pretty much impossible, so I accept oscillation of PWM value by 1 as stable output. However, when the load is very light and the output capacitor is smaller than 10uF, the output is hard to control - it tends to oscillate by more than 100 steps, generating over 50Vpp AC wave on the output! This is unacceptable.
The only semi-working method is to reduce my PID controller to only I part with very small gain, which results with settling times of about 0,5s - which for a nixie clock isn't really a big deal. This is a compromise that I was able to achieve. I fear that a converter with so slow reaction will be unable to properly power a multiplexed clock.
I'd like to hear from you - have you tried to power your clocks with MCU driven boost converters? Did you encounter similar problems? Was there any other solution than placing a big cap (increasing "inertia" of the circuit) or pre-loading the output? Or is my PWM resolution just too low and lowering frequency to standard 20-30kHz would do the job?
I really like the idea of using MCU to drive the boost converter, in theory it is perfectly capable of doing the job, I just can't seem to get it to work properly.