Thermal Profile: Reflow Thermal Profiling of Thick Copper PCB Assemblies
A board with heavy copper planes does not heat like a thin signal board, and a profile that works on one will not transfer to the other. The copper conducts heat away from the joint faster than the laminate does, so the solder joint lags the air temperature and the assembly reaches its peak later than the oven controller suggests. This article explains why thick copper behaves differently, how to attach a thermocouple so that the reading means something, and how to develop a profile that is repeatable rather than lucky.
Why Thick Copper Changes the Profile
Copper has a thermal conductivity several hundred times that of a typical laminate and a high volumetric heat capacity. A plane of copper therefore spreads heat laterally and stores it, which makes the board act as a heat sink for every joint attached to it. The more copper there is, the more heat the assembly needs before the joint reaches liquidus.
The effect is not uniform across the board. A joint on a small isolated pad sees almost the same temperature as the surrounding air, while a joint connected to a large plane through several thermal vias sees a much lower temperature at the same moment. A single profile has to satisfy both.
Heat Capacity and In-Plane Conduction
In-plane conduction is what makes thick copper useful for thermal management and awkward for soldering. Heat applied to one part of the plane travels along it rather than into the joint, so the plane equalises the temperature of the whole assembly and delays the local peak. The result is a slower rise and a flatter peak than the oven set point.
The out-of-plane direction behaves differently, because the laminate between copper layers is a poor conductor. A board can be hot on the surface copper and still cool at an inner layer, which is one reason a profile developed on a bare board can fail on a populated one.
Thermocouple Selection and Attachment
A fine wire thermocouple responds quickly and disturbs the joint least, which makes it the usual choice for profiling a small solder joint. A heavier wire is more robust but slow, and a slow sensor under-reports the peak of a fast event. The junction itself should be as small as possible for the same reason.
Attachment is where profiles go wrong. The junction must be in contact with the joint or the copper immediately next to it, and the attachment has to survive the oven without adding thermal mass. High temperature tape and a bead of adhesive are common, and the tape has to be small enough not to insulate the joint from the air it is meant to be measuring.
Preparing a Profiling Board
The profiling board should be a real production panel, populated with the actual components and the actual paste volume, because an unpopulated board has a different thermal mass. Using a scrap board from the same design is far better than using a test vehicle, and where a scrap board is not available, one should be sacrificed rather than profiled on an approximation.
The attachment points should be chosen to represent the extremes: the largest thermal mass on the board, the smallest component, and a location on a heavy plane in the middle of the panel. Leading and trailing positions matter too, because a board in a conveyor oven sees a different history at its front and back edges.
Peak Temperature and Time Above Liquidus
Peak temperature is the highest temperature the joint reaches, and it has to be high enough for good wetting and low enough to avoid damage. Time above liquidus is how long the alloy stays molten, and it controls the formation of the intermetallic layer between the solder and the pad. Both work together: a short time at a high peak can deliver the same wetting as a longer time at a lower one.
The measurement has to come from the joint, not from the oven. An oven that reports a peak of a certain value is reporting the air or the heater, and the assembly will be lower by an amount that depends on the copper content.
Soak and Flux Activation
The soak zone gives the assembly time to equalise before the ramp to peak. For a thick copper board this stage is essential, because the difference between the hottest and coolest point on the panel is at its largest during the rise. A soak that is long enough lets the plane, the joints and the components approach a common temperature.
The soak also activates the flux and drives off the volatiles, and a soak that is too short leaves flux unactivated and volatiles trapped. Both appear later as voids or as poor wetting on the joints with the largest thermal mass.
Ramp Rates and Component Limits
The ramp rate is limited from above by the components, which have a maximum rate of change they will tolerate, and from below by the paste, which needs a profile within its published window. Thick copper assemblies usually need a slower ramp than thin ones because the internal gradients are larger and the risk of thermal shock is higher.
The cooling rate matters as much as the heating rate and is often overlooked. A fast quench produces fine grains and a shiny joint, while a slow cool produces a coarser structure; both can be acceptable, but the rate should be a deliberate part of the profile rather than whatever the oven happens to do.
Reading a Profile from a Real Board
A profile is a set of traces, and the useful work is the comparison between them. The difference between the hottest and coldest thermocouple shows the thermal gradient across the assembly, and that figure should be small enough that every joint falls inside the paste window.
Our thermal management notes describe how copper distribution is planned on the board, and our current capacity guide explains why heavy copper is used in the first place. Our laminate properties guide covers how the base material conducts and stores heat.
Repeatability and Verification
A profile is only valid for the conditions under which it was taken, so the oven zones, the conveyor speed, the board design, the paste and the component loading all have to be recorded with it. When any of them changes, the profile should be re-measured rather than assumed.
Periodic verification with the same profiling board detects drift in the oven before it produces defects. At gopcb that record sits with the fabrication notes for the design, so that the thermal history is known when a soldering question comes back from the field.
FAQ
How many thermocouples should a profile use? At least three for a thick copper board: one on the largest thermal mass, one on a small component and one on a heavy plane. More points give a better picture of the gradient, which is the number that actually matters.
Can a profile be transferred between ovens? Only if the zone layout, the airflow and the conveyor speed are equivalent. The profile describes the board history, not the oven settings, so it should be re-measured on any oven it is moved to.
Does thick copper need a higher peak temperature? It usually needs a longer soak rather than a higher peak, because the problem is getting the mass to temperature evenly. Raising the peak to compensate for a short soak risks damaging the components without improving the joint.
