To answer the second phrase of the question first, making sure that every joint has the intended temperature experience is the only way of ensuring that all the joints have the same high quality and low defect rate. Too low a temperature, and the solder may not wet board and components; too hot, and the assembly may be damaged; too long above liquidus, and the growth of intermetallics may impair joint reliability; too fast a ramp rate and chip ceramic capacitors may crack, which will reduce their long-term reliability.
The source of difference in temperature experience is that components differ in thermal mass and area, so that both the amount of heat input and the resulting rate of temperature increase will be different. The art of profiling lies in ensuring that all the components lie within an acceptable envelope.
Ovens are controlled by setting the conveyor speed, zone temperature settings (top and bottom) and flow rates in each zone, although the ability to control the last of these varies considerably between different oven types. Because of temperature lags and the comparatively fast movement of the belt, the correspondence between the oven setting and the impact on a component has to be measured, and cannot be calculated from first principles.
The measurement technique is to fix thermocouples (using adhesive or high melting point solder) to a number of points on the assembly that are as representative as possible of the whole range of thermal masses and locations. Typically between three and eight thermocouples will be used for this. These thermocouples are the inputs for a profile unit, which logs temperature against time. This travels through the oven behind the board being profiled, at a sufficient distance from the board not to affect the outcome, and protected from the oven environment by a thermally-insulating sleeve.
Analysis of the information collected by the profiler is analysed with the help of suitable software, which allows the result from different thermocouples to be overlaid and compared, both against each other and against the target profile. The information from each test run will be analysed and appropriate corrections made to the oven settings, and the measurement sequence iterated until all the joints on the sample board can go through the reflow process and be within the profile limits.
Measuring such a sample profile board is of course a proxy for making measurements on each joint during reflow, the latter being impractical. In order to improve the likelihood that the profile experiment is representative of real production, the assembler uses a combination of feedback from the oven (controlling temperature, conveyor speed, and sometimes convection flow rate) with repeated profiling of the sample board. Depending on the company, profiling may be carried out each shift or once a week. The aim is to alert the operator to major changes in the intended profile, and in particular to pick up long-term drifts. These are often due to changes in oven characteristics caused by the accumulation of flux residues.