You’ve got the ECU lid off. You’ve desoldered the flash chip, you’re feeling good about the day, and then you see it — a thin copper line running off from the pad like a scratch on a vinyl record. Broken trace. Or worse, the pad itself came up with the chip leg and is now sitting on the tip of your iron like a tiny gold flake.

I’ve been there. Everyone who has ever put an iron to a Bosch MED17, a Siemens SID803, a Delphi DCM, or a boring old Denso body module has been there. The good news is that a broken track is one of the most fixable faults in all of electronics. The bad news is that most people fix it badly, and it comes back three months later when the car is 400 km from home.

So let’s do it properly.

What actually breaks on a PCB

Copper traces on a car module are usually 35 µm thick (1 oz copper), sometimes 70 µm on power layers. That’s thinner than kitchen foil. Four things kill them:

1. Heat and mechanical stress during rework. Too much time on the pad, too much tip pressure, or lifting a chip before all the solder is molten. The adhesive bonding copper to the fibreglass gives up around 200–250 °C, so if you’re sitting at 400 °C for twenty seconds, the pad is going to peel.

2. Corrosion. Water in the loom, a leaking bulkhead grommet, a coolant weep onto the connector. Green crust eats copper fast. You’ll often find the trace is fine under the solder mask but completely gone where a via or a component pad was exposed.

3. Physical damage. A slipped screwdriver, a dropped board, over-tightened case screws flexing the substrate, or an ambitious drill during a “quick” bench flash socket install.

4. Overcurrent. An injector driver shorts, and the track leading to it acts like a fuse. Very satisfying to look at, very annoying to repair, because the copper is often gone over 10–15 mm.

Knowing which of these caused your break matters, because a corroded trace means there’s likely more corrosion hiding under the mask nearby. Fix one, and the neighbour fails next winter.

Step 1: Find both ends of the break

Don’t guess. Grab your multimeter, set it to continuity, and buzz from a known good point to the component pad. If it beeps, the copper is intact and your problem lives elsewhere. If it doesn’t, walk along the trace pad by pad until you find where continuity stops.

A cheap USB microscope is worth its weight in gold here. Under 20× magnification, a hairline crack that’s invisible to your eyes looks like the Grand Canyon.

Pro move: use the ECU’s own service manual pinout or a boardview file if you can get one. Tracing a track that dives under a BGA and reappears on the other side of the board will burn an hour of your life if you’re doing it by eye.

Step 2: Expose clean copper

You need shiny, solderable copper on both sides of the break. Scrape the green solder mask away gently with a fibreglass scratch pen or the very tip of a scalpel held almost flat. Work in small strokes, always away from the component. Never dig — the copper underneath is 35 µm and your blade doesn’t care.

Once you see bare copper, clean it with isopropyl alcohol and tin it immediately with a small amount of leaded solder. Sn63/Pb37 melts at 183 °C and flows beautifully; if you’re forced to use lead-free on an automotive board for compliance reasons, bump your tip to about 340 °C and use plenty of no-clean flux.

Copper oxidises within minutes once exposed. Tin it or lose it.

Step 3: The jumper wire — do it like a professional

Here’s where most repairs go wrong. People grab whatever wire is on the bench and drape it across the board like a washing line.

Wire choice matters:

• Signal traces (CAN, sensor inputs, logic): 0.1–0.2 mm enamelled copper wire (magnet wire, sometimes sold as “kynar wire wrap”). Thin, stiff enough to stay put, and the enamel coating means it can’t short against anything it touches.
• Medium current (5 V rails, relay drives): 0.3–0.5 mm enamelled or PTFE-insulated wire.
• Power and ground (injectors, coils, main supply): stranded PTFE-insulated wire, sized to the original trace width. A 2 mm wide, 1 oz trace carries roughly 2 A comfortably — match or beat that. Underspec the jumper and you’ve built yourself a fuse with a delayed action.

Routing matters even more. Follow the original trace path as closely as you can. Don’t take shortcuts across the board — a long jumper flying over a switching regulator will pick up noise, and on a CAN or knock-sensor line that noise becomes a fault code. Keep jumpers short, flat against the board, and away from heatsinks and inductors.

Soldering technique: tin the wire end, place it on the tinned pad, touch with the iron for one second, remove. That’s it. A good joint is shiny and concave. If it’s a grey blob, reflow it with fresh flux.

For enamelled wire, don’t bother stripping mechanically — just hold the end in a blob of molten solder for two seconds and the enamel burns off cleanly.

Step 4: Rescuing a lifted pad

A lifted pad is a different animal, because the pad is no longer anchored to anything. Solder it as-is and the first vibration event will tear it off completely, usually taking a bit more trace with it.

If the pad is still attached but loose: Clean it, press it flat, and put a small amount of thin CA glue or low-viscosity epoxy under the edge. Let it cure fully before you go anywhere near it with an iron.

If the pad is gone entirely: You have two options. Use a pad repair kit (Circuitmedic and similar sell adhesive-backed replacement pads in a range of sizes) — bond the new pad down, cure it, then jumper it back to the trace. Or, for through-hole and larger SMD footprints, build a pad from a small piece of tinned copper foil or a flattened bit of component leg, glued down with epoxy and jumpered.

The critical rule: the mechanical load must never sit on the solder joint. Solder is glue-strength, not structural. The component body needs to be anchored to the board with adhesive, and the jumper needs a strain relief point so any flex is absorbed by the epoxy, not by your beautiful joint.

Step 5: Mechanical protection — the part everyone skips

This is the difference between a repair and a time bomb.

An engine bay ECU lives through -30 °C winter starts and +100 °C summer heat soak, constant vibration, oil vapour, and humidity cycling. A bare jumper wire tacked to two pads will survive that for a while. “A while” is not what you want to hand back to a customer.

Here’s the protection stack I use, in order:

1. Clean the flux off. Isopropyl alcohol, a stiff brush, and a lint-free wipe. Flux residue attracts moisture and, over years, becomes mildly conductive. On high-impedance sensor inputs that’s enough to cause drift.

2. Anchor the wire. Small dots of UV-cure solder mask, epoxy, or even hot melt at two or three points along the jumper run. The wire should not be able to vibrate freely anywhere. Free-standing wire work-hardens and snaps at the solder joint — always at the joint, never in the middle.

3. Reinforce the joints. A drop of two-part epoxy or UV solder mask over each solder connection, spread just past the pad edge onto adjacent solder mask. This spreads mechanical load across a wider area and stops moisture ingress. UV-cure solder mask is my favourite — it goes on green, cures in 30 seconds with a cheap UV lamp, sands smooth and looks factory.

4. Conformal coat the area. Acrylic conformal coating (like MG Chemicals 419D or equivalent) sprayed or brushed over the repair zone. Mask connectors and any programming headers first. This is your barrier against humidity and corrosive vapour, and it’s what the factory did to that board in the first place.

5. Photograph it. Genuinely. Before you close the lid, take a clear photo of the repair. If that unit ever comes back, you or the next tech will know exactly what was done. It’s also proof of workmanship — which matters more than people admit when a customer questions a job six months later.

Common mistakes I still see every week

• Using solder wick as a jumper. It’s braided copper with no insulation and it frays. Please don’t.
• Bridging a break with a blob of solder. Fibreglass and solder have wildly different thermal expansion rates. The blob cracks. Always use wire.
• Too much heat on an already-damaged area. If a pad lifted once, the surrounding adhesive is already compromised. Drop your temperature, add flux, work fast.
• No strain relief where the jumper crosses a board edge or a case screw boss. That’s a guaranteed break point.
• Skipping the continuity check afterwards. Buzz the repair before you reassemble. Then buzz it again after you’ve screwed the case down — sometimes reassembly is what breaks it.

Tools worth owning if you do this regularly

A temperature-controlled iron with a fine conical and a small chisel tip. Good tweezers (ESD-safe, curved). A fibreglass scratch pen. No-clean flux in a pen or syringe. Kynar/magnet wire in a couple of gauges. UV-cure solder mask and a UV torch. Two-part epoxy. Conformal coating. A USB microscope or a proper stereo scope if the budget allows. Isopropyl alcohol at 99%.

None of that is expensive. All of it will pay for itself the first time you rescue a module that would otherwise cost four figures to replace and code.

When not to repair

Be honest with yourself. If the corrosion has spread under a BGA, if multiple inner-layer traces are gone, if the substrate is delaminated or charred through, or if the board is a safety-critical airbag or ABS module — walk away. Source a replacement. A questionable repair on a braking or restraint system isn’t a technical decision, it’s a liability one.