What's Actually in a Hardware Engineer's Debug Kit (PCB Bench Edition)

May 10, 2026

What's Actually in a Hardware Engineer's Debug Kit (PCB Bench Edition)

You can tell a working bench from a staged one in about four seconds. The staged one has a fresh roll of solder, a clean mat, and the multimeter probes coiled like they came out of the box yesterday. The working one has a coffee ring on the ESD mat, three logic analyzer clips with the insulation chewed off, and a populated board sitting in a vise with one of the inductors hand-soldered at forty-five degrees because rev A footprint was wrong. If you've spent any real time chasing a bug down through four layers of copper, you already know none of the manufacturer kit lists describe what's actually on the bench. So here's the actual hardware engineer debug kit, as observed in the wild.

The Instruments You Actually Touch

Every bench has the marquee items: scope, bench supply, soldering station, multimeter. What's interesting is which ones get used. The scope is on every day. The bench supply is on every day. The function generator is on for maybe six hours a year and the rest of the time it's a monitor stand. The spectrum analyzer is treated like a religious relic until the day EMC pre-compliance lands on your desk, then it doesn't get turned off for two weeks.

Real list, in order of times-touched-per-day:

**Multimeter.** Probably a Fluke 87V if the company has money, a Brymen if the engineer has taste, or a generic yellow thing from a Pacific Rim factory if the company doesn't and the engineer is tired. Continuity beeper is the single most used feature on the bench. Diode mode is second.
**Oscilloscope.** Four channels is the floor. If you have two channels you are buying a four-channel scope within eighteen months whether you mean to or not. Bandwidth matters less than people pretend until it suddenly matters more than anything.
**Bench supply.** Two channels, current limiting that actually works, and a knob you can grab without looking. The smart-supply with the touchscreen interface is a war crime and you know it.
**Soldering iron.** Hakko FX-888D, JBC if someone got a budget approved, Pinecil if the bench is at home. The chisel tip lives on the iron. The conical tip lives in the drawer with the audit cert.
**Hot air rework.** For pulling QFNs, drying flux, and one specific coworker uses it to reheat ramen, which is a fireable offense everyone pretends they didn't see.
**Logic analyzer.** The Saleae is on the desk. The HP 1670G is on the floor under the desk because nobody can throw it out.

The pattern: the more sophisticated the instrument, the less often it gets used, and the more reverently it gets unboxed when it does.

Probes, Clips, and the Tyranny of the Ground Lead

Probes are where benches diverge from idealized lab photos. Nobody has matching probes. You have the two that came with the scope, one passive probe you bought yourself because the lab inventory probes have been borrowed by someone who left the company in 2019, one differential probe that lives in a Pelican case, and a current probe you're scared to drop.

The ground lead situation is its own discipline. The long alligator-clip ground lead is fine for poking around at 1 MHz. The moment you try to look at a switching edge with one, you get a 200 MHz ring that isn't actually there. So you keep:

A spring tip for the probe (lost, then found, then lost again)
A ground spring that you bent yourself from a cut resistor lead
A short flying lead with a clip you crimped in 2021
Twelve of those tiny hook clips that come ten to a bag, of which three still work

There is also, somewhere in the drawer, a set of micro-grabbers that someone ordered for an 0402 debug session and which have since become the only way to probe anything on the QFP-100 because the actual test points were value-engineered out of rev C.

The Soldering Drawer

The drawer beneath the iron contains more accumulated wealth than most people realize. Inventory, roughly:

**Solder.** Sn63/Pb37 for actual work, lead-free for boards going to customers, a spool of Kester 245 that's been on the bench since the previous engineer retired.
**Flux.** A pen, a syringe of tacky flux, and a jar of rosin paste with the lid welded on.
**Solder wick.** Three widths, two of which are unusable because someone touched the bare copper with their fingers.
**Tweezers.** The good pair that you hide. The shared pair that's bent. The reverse-action pair that nobody asked for.
**Kapton tape.** A roll that will outlive you.
**Isopropyl.** 99%, in a squeeze bottle, with a label that's been worn off by 99% isopropyl.
**Brass wool.** The sponge is a lie. The brass wool is the truth.
**Microscope or loupe.** If you're under thirty-five, a loupe. If you're over forty-five, a stereo microscope on a boom arm. Between thirty-five and forty-five, denial.

Hand Tools That Live Within Arm's Reach

The unspoken rule is that if a tool isn't within rolling-chair distance, it doesn't exist during a debug session. You will use the wrong tool sitting next to you before you stand up to get the right tool ten feet away. Therefore the bench accumulates:

Flush cutters (one decent pair, one pair for cutting steel wire that you're not telling anyone about)
Needle-nose pliers
A pick set, originally for dental work, now bent into shapes their manufacturer would not recognize
Hex keys, metric and imperial, half of them missing
A small Phillips, a small flat-head, and one Torx that fits the one screw on the one product
A scalpel for cutting traces, with a blade that should have been changed two years ago
A pair of ESD-safe scissors, used for everything except ESD-safe work

The Drawer of Shame

Every bench has it. It's the bottom drawer, or it's a cardboard box behind the monitor, or it's a static bag that says DO NOT OPEN. Contents include: dead-bug prototypes from previous projects, a board with magnet wire soldered to a BGA ball that someone managed once and never repeated, a USB cable with a known-good D+ line and a hand-labeled note, a 3D-printed jig for a product that shipped four years ago, and a small Ziploc of 0402 resistors that fell on the floor and were collected with tape.

You will never use these things. You will also never throw them away. When you switch companies, you will leave a similar drawer for the next person, and you will inherit theirs.

Software, Because Half of Hardware Debug Is Software

The bench computer is its own archaeological site. Installed: KiCad or Altium, the scope vendor's terrible Java-based capture utility, PuTTY, a serial terminal that isn't PuTTY because PuTTY did something annoying once, Saleae Logic, the JTAG vendor's IDE, Python with pyserial and pyvisa, a spreadsheet with the pinout, and a text file called notes.txt that contains the only documentation of what GPIO 17 actually does on rev B.

There are also seventeen browser tabs open: three datasheets, two errata documents, the schematic PDF, a Stack Exchange question from 2014 that almost answers your question, the JIRA ticket, two unrelated tabs from lunch, and a YouTube video of someone reflowing a PS5 that you swear you'll watch later.

Q&A from the Bench

Q: Why do I have three multimeters on my bench?

A: One is reading voltage on the rail you're debugging. One is on continuity from the previous session and you forgot. The third is the good one and you don't want anyone to know where it is.

Q: Is the no-clean flux actually no-clean?

A: No.

Q: How long should a debug session take?

A: It is exactly twice as long as you told your manager in the morning, regardless of what number you said.

Q: When do I escalate to a colleague?

A: After you've explained the problem out loud to nobody for the third time and realized halfway through that the issue is your ground reference. You will not escalate. You will fix it and not tell anyone.

Q: What's the right way to label test points?

A: Sharpie on the silkscreen. Anything else is theater.

The Bench Tells the Truth

There is a version of this job that exists in datasheets, app notes, and LinkedIn posts. In that version, the engineer follows the recommended reference design, the layout review catches everything, the bring-up takes a single afternoon, and the spectrum analyzer never sees an EMI failure. That job is not real. The real job is the one where the regulator is oscillating, the I2C bus is held low by something you can't identify, and the only person on the floor with a working differential probe is on PTO.

The debug kit isn't a list of parts. It's the residue of every previous problem, kept around because next time the problem will be slightly different but the residue will still apply. The bench accumulates faster than it gets cleaned, and that's mostly fine, because the engineer who keeps a perfectly tidy bench is either between projects or lying about what they do all day.