https://muffinbytelabs.com/blog Practical notes on KiCad PCB design: layout, power, manufacturing packages, signal integrity, and board bring-up. en-us https://muffinbytelabs.com/blog/firmware-constraints-behind-the-layout https://muffinbytelabs.com/blog/firmware-constraints-behind-the-layout Mon, 10 Aug 2026 00:00:00 GMT Firmware Bulk caps sized for Wi-Fi current spikes, strapping pins left free, and a flashing path you can still reach: layout choices that only make sense if you have read the firmware. https://muffinbytelabs.com/blog/opto-isolation-and-creepage https://muffinbytelabs.com/blog/opto-isolation-and-creepage Thu, 30 Jul 2026 00:00:00 GMT Signals An optocoupler only isolates if the layout does too: no copper on any layer crossing the barrier, a slot to extend creepage, and an LED resistor sized for worst-case CTR. https://muffinbytelabs.com/blog/rev-a-notes https://muffinbytelabs.com/blog/rev-a-notes Sun, 19 Jul 2026 00:00:00 GMT Bring-up The friction you notice during assembly disappears from memory within a week. A REV_A notes file and a CHANGES log keep it, and handing both to a client builds trust rather than doubt. https://muffinbytelabs.com/blog/protecting-the-input https://muffinbytelabs.com/blog/protecting-the-input Wed, 08 Jul 2026 00:00:00 GMT Power Polyfuse, TVS, and reverse-polarity protection each cover a different failure, and the order matters. How I size and sequence all three on a 24V input. https://muffinbytelabs.com/blog/decoupling-placement-beats-value https://muffinbytelabs.com/blog/decoupling-placement-beats-value Sat, 27 Jun 2026 00:00:00 GMT Layout Loop inductance, not capacitance, decides what a decoupling cap can do. Why a well-placed 100nF beats a bigger one two millimeters away, and what DC bias really costs you. https://muffinbytelabs.com/blog/bom-discipline https://muffinbytelabs.com/blog/bom-discipline Mon, 15 Jun 2026 00:00:00 GMT Manufacturing Every BOM line needs a real MPN and a distributor code, checked for stock and lifecycle before you commit a footprint. The discipline that keeps an assembly job from stalling. https://muffinbytelabs.com/blog/level-shifting-for-ws2812b https://muffinbytelabs.com/blog/level-shifting-for-ws2812b Thu, 04 Jun 2026 00:00:00 GMT Signals A 3.3V GPIO sits 200 mV below what a 5V WS2812B needs. Why that passes on the bench and fails in the field, and why a 74AHCT125 — not a FET shifter — is the fix. https://muffinbytelabs.com/blog/bring-up-order-smoke-rails-buses-firmware https://muffinbytelabs.com/blog/bring-up-order-smoke-rails-buses-firmware Sun, 24 May 2026 00:00:00 GMT Bring-up Unpowered checks, a current-limited first power-up, rails one at a time, then buses, then firmware. The bring-up order that finds a mistake before it costs you the board. https://muffinbytelabs.com/blog/buck-converter-layout-the-hot-loop https://muffinbytelabs.com/blog/buck-converter-layout-the-hot-loop Wed, 13 May 2026 00:00:00 GMT Power The hot loop — input cap, high-side switch, return path — is the one thing in a buck layout that has to be physically tiny. How I lay out the 24V to 5V stage on PCB 2. https://muffinbytelabs.com/blog/footprint-libraries-that-dont-rot https://muffinbytelabs.com/blog/footprint-libraries-that-dont-rot Sat, 02 May 2026 00:00:00 GMT KiCad Ship a KiCad project a client can still edit in five years: project-local libraries, land patterns verified against the datasheet, and a checklist for third-party footprints. https://muffinbytelabs.com/blog/trace-width-is-a-thermal-problem https://muffinbytelabs.com/blog/trace-width-is-a-thermal-problem Tue, 21 Apr 2026 00:00:00 GMT Layout An 8x8 WS2812B matrix pulls close to 4A at full white. A worked IPC-2221 example: how wide that 5V trace must be, why 2 oz copper beats it, and where copper necks down. https://muffinbytelabs.com/blog/usb-c-wont-enumerate https://muffinbytelabs.com/blog/usb-c-wont-enumerate Fri, 10 Apr 2026 00:00:00 GMT Power A Type-C source never turns on VBUS until it sees 5.1k on CC. Why the pull-down goes on both CC1 and CC2, and why one shared resistor fails only with good cables. https://muffinbytelabs.com/blog/reading-your-fabs-capability-sheet https://muffinbytelabs.com/blog/reading-your-fabs-capability-sheet Mon, 30 Mar 2026 00:00:00 GMT Manufacturing Minimum trace, drill, annular ring, mask dam, silkscreen height: pull them off your fab's sheet into KiCad's Board Setup before you route, not after the gerber review. https://muffinbytelabs.com/blog/one-pull-up-pair-per-i2c-bus https://muffinbytelabs.com/blog/one-pull-up-pair-per-i2c-bus Wed, 18 Mar 2026 00:00:00 GMT Signals Four breakouts with 4.7k pull-ups give you a 1175 ohm bus. Here is the math, the 967 ohm floor at 3.3 V, the rise-time ceiling, and how to fix a bus that already has too many. https://muffinbytelabs.com/blog/ground-planes-and-return-paths https://muffinbytelabs.com/blog/ground-planes-and-return-paths Sat, 07 Mar 2026 00:00:00 GMT Layout Return current flows directly under the trace, not wherever it likes. What a slot in the pour really does, and how PCB 2 keeps logic ground and field ground apart. https://muffinbytelabs.com/blog/ap2112k-vs-ams1117 https://muffinbytelabs.com/blog/ap2112k-vs-ams1117 Tue, 24 Feb 2026 00:00:00 GMT Power The AMS1117 needs 1.1 V of dropout and burns 5 mA doing nothing. The AP2112K-3.3 needs 250 mV and 55 µA. Worked numbers for USB and LiPo boards, plus when to use a buck instead. https://muffinbytelabs.com/blog/why-i-standardized-on-kicad https://muffinbytelabs.com/blog/why-i-standardized-on-kicad Fri, 13 Feb 2026 00:00:00 GMT KiCad Why I deliver native .kicad_pro, .kicad_sch and .kicad_pcb files and refuse Altium conversions: what a conversion silently destroys, and where KiCad is honestly still weaker. https://muffinbytelabs.com/blog/what-actually-ships-to-the-fab https://muffinbytelabs.com/blog/what-actually-ships-to-the-fab Mon, 02 Feb 2026 00:00:00 GMT Manufacturing A fab never opens your .kicad_pcb. Here is every file that actually ships (Gerbers, Excellon, BOM, CPL, docs, DRC proof) and what each machine on the other end does with it.