Thermocouple Attachment: Getting an Honest Reflow Profile
A thermal profile is only as good as the joint between the thermocouple and the board. Attach the bead badly and the curve still looks plausible, still peaks in range, and still misleads: it reports the temperature of the adhesive rather than of the copper, and every decision made from it is built on a measurement artifact.
Why Attachment Decides the Result
The bead has to reach the temperature of the material it is measuring, and it can only do that through contact. A bead sitting on top of a lump of adhesive is insulated from the pad and lags the board by tens of degrees during ramp. The lag is largest where the profile changes fastest, which is exactly where peak and time above liquidus are read.
Because the error is systematic rather than random, it does not average out. Two boards profiled with the same bad method will agree with each other and both be wrong. The check is not repeatability between runs but agreement with an independent measurement of the board temperature.
Bead Size and Wire Gauge
A fine bead responds quickly and disturbs the board little, while a large bead responds slowly and adds thermal mass. A bead of roughly 0.1 to 0.25 mm with 30 AWG wire is the usual compromise for board profiling, because it settles in a fraction of a second and does not act as a heat sink on a small pad.
Wire gauge matters beyond flexibility. Thicker wire conducts heat away from the bead along its length, lowering the reading at the joint and creating a cool tail on the curve. Twisting the pair and taping it down along the board surface reduces both conduction and the antenna effect that picks up oven noise.
High-Temperature Solder Versus Adhesive
Attaching the bead with a small amount of high-temperature solder gives the best thermal contact on a metallic surface, because the metal wets and the bead becomes part of the joint. The solder must have a melting point well above the profile peak, and the amount used must be small enough not to change the mass of the pad being measured.
Adhesive is used where the surface cannot be soldered, such as a plastic connector body or a coated pad. High-temperature epoxy rated above 260 degrees Celsius is the common choice, but it has to be applied thin: a blob of adhesive insulates the bead, and the resulting curve understates the peak and stretches the apparent soak. Where an adhesive is used, its cure must be complete before profiling.
Placement: Which Locations Represent the Board
A profile is taken at the locations that are hardest and easiest to heat, and the pair defines the process window. The hot spot is usually a small component on a thin section with little copper, while the cold spot is a large body, a heavy ground plane or a board edge near a conveyor rail. Profiling only the middle of the board gives a curve that describes neither.
Place at least three thermocouples on a mixed assembly and label each one. Attach one to a large thermal mass, one to a small component adjacent to that mass, and one to a representative area. The temperature difference between them is the number the profile has to accommodate, and it is often 15 to 30 degrees Celsius.
Attaching to the Mass, Not the Surface
The most common profiling error is measuring the surface of a component body instead of the joint beneath it. For a large package, the joint under the corner ball is 10 to 20 degrees cooler than the top of the package at the same instant, so a bead on the lid reports a comfortable peak while the solder is still solid. Attach to the pad, the joint or the copper it is connected to.
Where the joint cannot be reached, attach beside it on the same copper and record that the reading is an approximation. A profile is a map of the board, and its usefulness depends on knowing which point on the map each curve represents.
Routing the Wires Without Disturbing the Profile
The wires leave the board and travel out of the oven, and they must not lift a component, hold the board off the conveyor, or drag on the rail. Route them along the board surface with high-temperature tape, exit over the trailing edge, and give enough slack that the board is not pulled as it moves. A board that is lifted off the rail heats differently at that edge.
Wire routing also affects the reading through conduction. A wire that is taped flat to the board conducts less heat away than one that stands up in the air, and long unsupported runs act as a radiator that cools the bead. Keep the run short and supported, and secure the exit so the bundle does not swing onto a heater.
Verifying the Instrument Chain
Before the profile is trusted, verify the chain: a calibrated logger, cold-junction compensation, and the correct thermocouple type selected in the software. A K-type wire read as a J-type produces an offset that looks like a real process condition, and it is far more common than a genuinely drifting oven.
Check the logger against a known reference at two points in the range, and record the result with the profile file. Verifying only at room temperature does not validate the range that matters, since the errors that affect a reflow decision appear at 200 degrees Celsius and above.
Reproducing a Profile Across Lines
A profile belongs to a board, an oven and a conveyor speed together. Moving the same product to a different oven requires a new profile, because zone lengths, airflow and heat transfer differ even when the set points are copied. Assuming that a profile transfers between machines is one of the most reliable ways to produce a batch of cold joints.
Where several lines run the same product, build one reference profile on each and compare the resulting curves rather than the set points. If the curves agree within a few degrees at peak and a few seconds above liquidus, the process can be treated as equivalent; if they do not, each line carries its own numbers.
Common Artifacts and What They Mean
A curve that shows an impossibly fast rise, a flat plateau where a component should be absorbing heat, or a peak that matches the oven set point rather than the board, all point to an attachment or instrument problem rather than a process one. The same applies to a curve with no soak, which usually means the bead is insulated from the board and simply follows the air.
Read artifacts as diagnostics rather than noise. They are the fastest indication that the profile being reviewed does not describe the board, and correcting the attachment is cheaper than acting on a curve that was never true.
Related reading: our fabrication notes, board quality and design release notes cover the same ground.
FAQ
What size thermocouple bead should be used for profiling? Roughly 0.1 to 0.25 mm with 30 AWG wire, which responds quickly and adds little thermal mass to the pad being measured.
Is solder or adhesive better for attaching a thermocouple? High-temperature solder gives the best thermal contact on metal. Adhesive is for surfaces that cannot be soldered, and it must be applied thin or it insulates the bead.
Why does the profile peak look lower than the oven setting? Usually because the bead is insulated from the board by adhesive or is measuring a component body instead of the joint beneath it. The board, not the air, sets the solder temperature.
