§ 01The four things that earn their place on day one

A multimeter comes first — full stop. A decent mid-range unit with continuity beep, DC voltage and resistance ranges handles ninety percent of first-day faults: missing power rail, cold joint, backward diode. Spend enough to get auto-ranging and a fast continuity beeper; the ten-dollar units are slow enough to be genuinely annoying.

Next: a way to get firmware into the chip. A USB-to-serial adapter covers a huge range of microcontrollers that ship with a UART bootloader — ESP32, STM32 with BOOT0 pulled high, many AVRs. For anything ARM Cortex-M without a built-in bootloader, a low-cost SWD probe (an ST-LINK clone or a Raspberry Pi Debug Probe) costs very little and unlocks both programming and debugging. Buy one of these before you buy an oscilloscope.

A USB-to-serial adapter covers a huge range of microcontrollers that ship with a UART bootloader — ESP32, STM32 with BOOT0 pulled high, many AVRs.

Third: a regulated bench power supply, or at minimum a quality USB power meter inline with whatever you're powering. Knowing your board draws 340 mA at startup instead of the expected 40 mA tells you something has gone wrong before you've written a line of code. A cheap variable supply with a current-limit knob has saved more boards than most people admit.

Finally: a handful of jumper wires, a small breadboard and a bag of known-good resistors. These aren't instruments, but they are the physical grammar of debugging — pull a pin high, add a series resistor, isolate a sub-circuit.

What can wait? A logic analyser is genuinely useful but not day one. A soldering station matters, but a temperature-controlled iron without the full station is fine to start. An oscilloscope is worth owning eventually; it is not worth buying before you have a fault that actually needs one.

Start lean. Instruments earn their place by solving real problems, not by sitting on a shelf looking credible.[1]

Soldering iron resting on a workbench Enlarge ⤢

Notes

  1. Buy the meter you will still want in five years. Everything else can start cheap. ↩