/diy/ - Do It Yourself

>>2403194
CD4046's VCO will do it, though it wouldn't be stable by itself. Could use a 555 or similar oscillator to generate a lower frequency clock (desired clock divided by a power of 2) and use digital logic ICs (counters probably, but shift registers or D-flip-flops also work) to make a phase-locked-loop of the desired frequency. Phase-locked-loops are really neat to learn about, so while using a 555+4046 isn't the most practical solution it's definitely an interesting one.

>a buck converter without a uC
Personally I used a comparator, they should be able to get up to a few hundred kHz but I'm not sure if common LM393s will be any better than 555s. If you're doing variable duty cycle then whatever you pick needs to output something akin to a triangle or sawtooth wave, which both the 555 and comparator options do. Not too sure about the 4046, but I think the pin that you put the timing cap on should have a triangle wave on it. Variable duty-cycle is quite an interesting challenge from the point of view of loop stability when it comes to selecting your error amplifier gain (and potential PID characteristics) but it's not the simplest of projects to start with.

If you're doing fixed duty-cycle on-off control (which I'd attempt once if you haven't done so already) then I'd challenge you to try and make the entire converter with a single dual-comparator IC. I've got such a design (constant-current fyi) ready to solder up on this board here, the red one at the top.
Keep in mind that dual comparator ICs usually have open-collector outputs.

>>2374082
Done. Hand painting is a bitch, but hopefully that will be a thing of the past since the red board is a boost converter designed to power an LCD lightbox's UVA backlight.

>>2374259
Then again, maybe I could actually make money selling custom 4-terminal MOSFETs, complimentary JFETs, those sorts of analog stuff. I'd buy aliexpress leadframes, stick the silicon chip to them somehow, do the wire-bonding, and blob JB weld over them.
>inb4 biodes