Ravi
633 posts

Ravi
@_raviguy
corporate hate tweeting / stream of consciousness / broadcasting my creations into the void


for the @i2cjak 48V dc/dc converter challenge: okay so covering 10mA to 10A current output is going to be more difficult than I thought now that I'm looking at behavior in simulation... approach: - 2 phases, fw will support phase shedding depending on iout. - iout >2A, both phases enabled, CCM - 2A > iout > 500mA, shed 1 phase, CCM - iout < 500mA, 1 phase with some kind of burst/skip scheme still need to figure out how to measure current across the whole range using some combination of sense Rs or DCR of the inductor or some combination.




@willreil Inquire at your local PCB assembly house to use their V-score cutter

I FORMALLY CHALLENGE THEE (@i2cjak) TO AN EE DESIGN CHALLENGE (rendered in the TONE of an i2cjak wannabEE ARTICLE): ANON… I HAVE A PROPOSAL. a DESIGN CHALLENGE. a TEST of SKILL. a TEST of GRIT. a TEST of whether you are a REAL ELECTRONICS ENGINEER or merely someone who COPY PASTAS application notes written by some TI intern named KYLE. let me set the stage... imagine a world where you need to convert 12v into 48v. this is not a crazy number. this is just BOOSTING a voltage. people have been doing this since before your GREAT GREAT grandparents knew what a transistor was (i2cjack just turned 14). and yet… what does an i2cjak do? he goes to MOUSER (he was banned from digikey) and he types: "BOOST CONVERTER" he grabs some $6.37 switching regulator IC that hardly has any stock, has 87 internal op amps, slope compensation, cmc, undervoltage lockout, soft start, thermal shutdown, brownout detect, a coffee maker, and a SMALL GOBLIN that tunes the loop compensation FOR HIM. he then looks himself in the mirror and calls himself A HARDTECH ENGINEER. he slaps down one inductor, a diode, some caps, and a prayer. NO. not today anon. THE CHALLENGE: • design a 12v to 48v dc/dc converter • highest efficiency @ 10A/480W wins but there are RULES. and they are CRUEL. RULES: • NO SWITCHING REGULATOR ICS • no boost controllers • no flyback controllers • no "smart power stages" • no little black boxes that say "PWM controller" on the datasheet • if the datasheet contains the phrase "switching regulator" you are DISQUALIFIED ALLOWED COMPONENTS: • a microcontroller • gate drivers (maybe) • power transistors • diodes • discrete analog • discrete logic • passives THE CONTROL LOOP: • must be written entirely BY YOU • no OTS analog compensation network that MAGICALLY stabilizes everything YOU WILL: • fab the board at a fab house of your choice • solder it YOURSELF • measure voltages YOURSELF • measure currents YOURSELF • perform step response YOURSELF • perform load sweeps YOURSELF • measure EFFICIENCY YOURSELF • you know, REAL EE SHIT YOURSELF like the founding fvcking fathers intended. ANY TOPOLOGY IS FAIR GAME: • boost • flyback • forward • pushpull • interleaved • some cursed topology you invented at 3am i do not care. but it must convert 12v to 48v. and it must fvck. SCORING: highest measured efficiency @ 10A wins (adjusted for BOM cost). WHY THIS IS FUN: because suddenly all the stuff those nice little regulator ICs do for you becomes YOUR PROBLEM. you must engineer. ELECTRICALLY. things like: • current limiting • duty cycle saturation • loop stability • slope compensation • dead time • gate drive timing • startup behavior • transient response EXPECTED OUTCOMES: • REAL EE is harder than it looks, i might fail. you might fail. BUT WE ALL LEARN. • writing a control loop from scratch is CURSED. • switch node ringing is DEMONIC. and guess what i2cjak, the inductor you'll pick will be WRONG!!! your efficiency will mysteriously drop by 15% because physics HATES YOU!!! but, you are a MANIAC. and you will eventually produce something BEAUTIFUL even IF it is covered in your BLOOD, SWEAT, AND TEARS. so anon. do you ACCEPT the challenge? or will you retreat back to your TPS5430 application notes like a frightened little B*TCH? MAY THE SUPREME WIZARD OF ELECTRONS WIN!!!





















