Article is from SMT Magazine

In many situations, EMS providers cannot waste a PCB for thermal profiling. Some ovens are equipped with profiling tools to generate an accurate reflow recipe without thermal profiling. This saves time, labor, money, and materials, but there are limitations.

By Brian O’Leary, KIC

There is a right way and a wrong way to set up a reflow oven to manufacture a new PCB assembly. This article suggests using the wrong method, but for the right reason. If an electronics manufacturer is prevented from following the correct method for setting up the reflow oven for a new production run, does a fallback position exist where they can still expect good results? For example, contract manufacturers find themselves in a not-so-uncommon situation where the manufacturer receives 100 boards and is expected to give a 100 assembled boards back. Sacrificing a single PCB to the profiling process is not an option. In another example, a manufacturer has PCBs that run in the several thousands of dollars. A suitable scrap board is not available for profiling, due to the cost incurred for the lost PCB.

Advantages and Disadvantages of Traditional Reflow Oven Set-up

The traditional method for setting up a reflow oven to manufacture a new PCB assembly is to attach thermocouples (TCs) to the PCB and run a series of profiles. Multiple profiles are usually required for the technician to adjust the oven recipe until an in-spec or deep in-spec profile is found. The introduction of lead-free assemblies has made this task more difficult and time-consuming. However, automatic prediction software and process optimization software have significantly cut down on the number of profile iterations required to determine the oven recipe that provides an in-spec process.

The benefit of this conventional reflow profiling method is clear: It achieves a deep in-spec and therefore stable process that is fundamental to good end-product quality. It also provides documentation to the client that proper process development work was performed.

These procedures, however, tend to sacrifice one or more PCBs. One reason for this concerns the TC attachment method. There are several TC attachment solutions, some more destructive to the PCB assembly than others. The use of high-temperature solder wire is a reliable method, but tends to damage the PCB assembly. Aluminum tape is also a reliable and repeatable method with the added benefit that the tape can be removed after the profile without damaging components.

A second cause of PCB damage is the fact that each subsequent thermal cycle through the reflow oven raises the risk of latent or real defects as solder joints are re-reflowed, components are exposed to multiple reflow cycles, and the properties of the substrate changing. The PCB gets lighter, discolored, and more brittle with multiple profiles. Therefore, even with non-destructive TC attachment methods, such as aluminum tape, the PCB may need to be discarded when several profiles are run.

A final risk is that the technician selects, often guessing, a wrong initial oven recipe prior to the first profile. The initial recipe could damage the PCB. This could happen when the peak temperature is too high, the slope too steep, the soak prematurely dries out the volatiles in the paste, etc.

Profiling the Reflow Oven, Not the PCB

Modern reflow ovens are a far cry from their legacy siblings. Each oven model produced in volume tends to have very tight and similar thermal characteristics to each other. Equally important, these properties do not change over time as rapidly as in the past due to better flux management, improved oven control systems, more precise mechanics, etc. This enables new thermal process tools that “learn” the behavior of each oven model. To capture the thermal properties of a specific oven model, numerous profiles are run on a variety of PCB assemblies under differing process windows. This database will cover all but the most unusual applications encountered in SMT production. Once this work has been done, it is a simple matter of copying the information onto all the similar oven models. At that point, the operator could simply enter the basic information of the application, such as the length, width, and weight of the PCB assembly as well as the appropriate process window, and the oven will find its suitable recipe (zone temperatures and conveyor speed). This recipe will yield an in-spec profile in the vast majority of the cases without the need to run a profile or attach TCs. Experience with such technologies also suggests that when the recipe generated by the new thermal process tools does not yield an in-spec profile, it is usually very close.

Some U.S. oven manufacturers have completed this work. These reflow oven makers ship ovens with a fully functional database that essentially allows their customers to set up for new production runs without the need for profiling and sacrificing PCBs.

These systems do have limitations. The first was alluded to above, namely that there will be a small percentage of the applications that will not be processed in-spec. The fail-safe method is to wait for the oven to stabilize on the suggested recipe and then run an old-fashioned profile to verify whether it is in-spec. If out of spec, it should, in the vast majority of the cases, be close enough to achieve an in-spec profile on the second try. One profiling pass through the reflow oven, with aluminum tape used for TC attach, should not damage the PCB assembly.

Another limitation is that an oven will, given enough time, eventually change its thermal properties. Wear and tear, changes in exhaust conditions, preventive maintenance, and a host of other factors will have an accumulative effect on the behavior of the reflow oven. Therefore, the initial database will need to be updated. This can be achieved by running some real-life profiles from time to time, and feeding this fresh information back into the database.

The final limitation is the fact that a system that eliminates profiling, by definition, does not have a profile recorded for the specific assembly. This means that there is no documentation or evidence that the PCB was indeed processed in-spec. Some customers will accept this, while others will not and require reflow profile documentation.

Conclusion

The correct method for reflow oven set up with a new application is to profile the PCB and dial the processed deep in-spec using prediction software. If the electronics manufacturer either cannot or will not perform this task, there are now thermal process tools available that achieve a more than 80% effective solution. Oven-inherent programming produces an in-spec recipe in the vast majority of the situations with no need to profile or sacrifice a circuit board. This technology also saves set up time and associated labor.

Using a profiling technology without an actual PCB profile run is also far better than doing nothing. Many manufacturers in our industry currently do not profile at all, or they limit their profiling to a single application a few times a year. If you do not want to do traditional profiling at all, oven-generated recipes can be an intermediary, rather than blindly reflowing.