A joint comes back from the field open after a month in service, and the failure report says cold solder joint. On the bench the fillet looks dull, rounded, and slightly grainy, with the solder sitting on the pad instead of flowing into it. Nothing on the board says where it came from, and by then the process that made it has moved on.
That is the problem with this defect: it forms in a second, at a temperature the profile says was reached, and it is quiet until the product is in someone's hands.
What makes a joint cold
Solder that does not reach its liquidus and alloy with the pad leaves a mechanical contact instead of a metallurgical one. The joint conducts when it is new and opens when thermal cycling works the contact loose. Three process conditions produce it: heat removed from the joint faster than the solder can absorb it, movement while the joint is between liquid and solid, and a surface the solder cannot wet at all.
The first is the common one on modern boards. A through-hole lead sitting in a ground plane has a plane of copper pulling heat away, and a wave or an iron that was set for a 1 oz pad gives a joint that never fully melted. The second is a handling problem, not a thermal one, and it shows as a fillet with a crack across it rather than a dull surface. The third is contamination or oxidation, and it shows as a joint where the solder balled up and refused the pad.
Cause, appearance and the check that confirms it
| Cause | What the joint looks like | What confirms it |
|---|---|---|
| Heat pulled out by a plane or a heavy pad | Dull, grainy fillet, rounded instead of concave, solder sitting on the pad | Thermocouple on the joint during the run, compared against the paste or wire specification |
| Movement before the joint solidified | Fillet with a visible crack or a step in the surface | Handling and conveyor review, plus a pull test on the sample |
| Oxidised pad or lead | Solder dewetted, balled up, pad visible under the solder | Cross section, and a solderability check on the incoming lot |
| Too little flux, or paste past its working life | Patchy wetting, solder balls beside the joint | Print record, paste lot and open time, reflow profile |
| Iron too cool, or contact too short | Rounded joint with no fillet rise on the lead | Iron tip temperature log, joint time on the work instruction |
| Contaminated barrel from a previous process | Joint forms on the lead but not in the barrel | Cross section of the barrel, which the optical check cannot see |
The table is a starting point for the investigation, not a verdict. Two of the rows look identical on the bench, and only the cross section or the thermocouple tells them apart.
What inspection can and cannot prove
Optical inspection reads the fillet: angle, surface, and whether the solder has wet the lead and the pad. That catches a cold joint on a visible lead and misses everything inside a barrel or under a body. X-ray shows the fill as a density column, so a barrel that is short of solder is visible even when the top fillet looks formed, but it cannot tell a cold joint from a good one if the volume is right.
Where a joint is critical, the check is destructive. A cross section shows the grain structure and the intermetallic layer, and a pull test gives the force the joint carries. On a build where the customer has a service history of opens, a first article that includes both is cheaper than a field return.
Repair or scrap
A visible cold joint on a through-hole lead is usually repaired: flux, heat the joint until the solder flows, let it cool without moving it, then re-inspect. The rule is that the repair has to add solder only when the barrel is short, because a joint that was cold and is now simply bigger is still cold inside.
A cold joint under a BGA or inside a stack is a different decision. The part comes off, the site is dressed, and the part goes back on, which is a rework with its own risk. Where the part is a fine pitch BGA on a thick board, the cost of the rework often exceeds the board, and scrap is the honest answer.
How SUPERB PCBA keeps it out of the build
The profile is verified with a profiler on the product, not copied from a similar board, and the thermocouple goes on the heaviest joint on the panel. On the SMT side the print is checked on 3D SPI before placement, and reflow runs on a twelve zone nitrogen oven at under 500 ppm oxygen with a SAC305 profile, which keeps the soak long enough for a heavy pad to catch up.
Through-hole work runs on JT-450 and JT-350 wave machines with the preheat set against the pallet and the board, and selective soldering where a joint cannot take the wave. Hand soldering follows a work instruction with tip temperature and contact time per joint type.
Inspection is optical to IPC-A-610 Class 2 or Class 3, X-ray where a barrel or a hidden row is in question, and electrical test on in circuit or flying probe. The chain that catches a cold joint before it ships is the one on solder joint quality and inspection standard, and the profile side is on reflow profile verification. Where a part has to come off, the path is on BGA rework and reballing.
Board features that make it worse
Two board features make cold joints far more likely, and both are visible before the first panel. A through-hole pad connected to a plane without a thermal relief pulls heat out of the joint faster than any wave can fill it. A pad with a large copper area on both sides does the same. Where the customer's drawing has either, the note goes back with the quote, and the build runs with a longer preheat and a hotter contact rather than a hope.
Questions buyers ask
Is a dull joint always cold?
No. Some alloys and some finishes give a dull surface on a joint that is metallurgically sound. The test is the fillet angle and the wetting, not the shine.
Can X-ray find a cold joint?
It finds a short fill. A joint with the right volume and a poor bond looks correct on X-ray, which is why the destructive check exists.
Does lead-free solder make it worse?
It narrows the process window. SAC305 melts higher than eutectic tin lead and wets more slowly, so the same iron setting that worked on a leaded board gives a cold joint on a lead-free one.
Next step
Send the BOM, the Gerber set and the assembly drawing with the heavy copper areas marked. The reply comes back with the profile points that will be verified on the first article, the joints that get a cross section, and the repair rules the build will run under.