|Ask the Experts
July 13, 2023 - Updated
August 2, 2010 - Originally Posted
Need for Reflow Profiles
How important is it to run a profile on every different assembly?
I have been told choosing an established low, medium or high mass oven setting is sufficient. Just choose the low, medium or high profile based on the size of the board.
Do you agree or disagree?
|Expert Panel Responses
I think that there are a lot of SMT facilities that have an optimized profile for every board they are running. This can be tedious and it certainly may end up costing valuable time to the engineering team tasked with collecting and optimizing the reflow profile data.
If you are a high-mix facility, however, there could be hundreds (if not thousands) of boards that are run on a semi-regular basis. If this is the case, I think that it is wise to do some simplification work up front so that you aren't chasing after hundreds of reflow profiles.
The simplification approach that I would recommend taking is dividing up the boards that you build into groups based on mass and (potentially) mix of component types. You may be able to easily assign each board into a "family" of products that can be handled with common reflow profile settings. You may be able to simplify down to 3 different groups, but don't be discouraged if you end up with 10 of them. (It's better than "hundreds" right?)
Once you have your product family groups, start with a proposed "optimal profile" for the group, then run the highest mass and lowest mass boards within the family and see what kind of reflow profile results that you get. Be sure to thermocouple the highest and lowest mass components on the board when doing this.
If the profiles end up being close enough to each other (which should be the case if the variance in mass is narrow enough), then all boards in that family should be ok to run on said profile. Any new board coming into the plant with a mass in the same range can be run on the same profile without hesitation.
I'm assuming a single solder paste alloy for the entire facility but a mix of leaded (Sn63/Pb37) and lead-free (SAC305) alloys would complicate this by a factor of two.
With all of this said, some customers may expect that you have a specific profile for each of their boards. If this is the case, then the answer is pretty simple.
You should also be running calibration tests on your oven periodically to verify that there is no drift in the oven's temperature or heat transfer over time.
General Manager - Electronic Assembly Americas
Thermal profiling is the only way to prove to your customers that your oven settings (the recipe) are such that the solder experiences the required thermal excursion (thermal profile) AND that any thermally sensitive components have not experienced any damaging thermal limits.
So yes, you should profile each new assembly to have this proof. You need only do it in the beginning to establish the correct oven recipe for that assembly, and then use oven capability measurement tools (oven checking pallets) to show and maintain the repeatability of your oven over time.
Keep in mind that the solder paste defines the needed thermal profile (time vs temp graph) and the components may limit it to protect them from damage. So for a given solder paste, the required thermal profile is the same, no matter what assembly you are soldering. The solder MUST experience the same thermal profile in order to solder correctly, and of course protect the components.
So even though the required thermal profile is the same, assuming you are using the same solder paste, the oven recipe needed to achieve that thermal profile will vary from assembly to assembly for many reasons like:
* Component's thermal conductivity (how fast heat flows through)
* Assembly's component density (components/cm^2)
* Assembly's Size (length x width)
* Assembly's Mass
* Board's thickness
* Component's specific heat (amount of heat needed to raise its temperature)
* Oven's load capacity (boards per minute)
To size up all these possible factors and determine which oven recipe is best would take much longer than running the profile and being sure.
You may require only a few thermal profiles, because you are only using a few different solder pastes on all your assemblies. However, the oven recipes required to achieve those few profiles can be far more varied than a simple low, medium, high set of settings. If the diversity of the product you assemble does not change much, then a few recipes may cover the majority.
However, this does not release your from being able to show to your customers that a given recipe indeed meet the required thermal profile on that assembly and did not risk any components. This can only be done through thermal profiling of each assembly at the oven recipe settings characterized for that assembly.
Senior Project Engineer
Electronic Controls Design Inc
Even with the most forgiving reflow systems available, developing a "golden recipe" that produces adequate thermal results for all products is a long shot, especially considering the reduced Lead-Free process window and knowing that all reflow systems are not equal. With popular convection reflow systems, multiple recipes are typically required.
The number of recipes can be reduced through proper practice and tools. Some of the latest reflow software reduces the actual number of unique recipes required to a half a dozen or so, however your success is proportional to the effort applied and inversely proportional to thermal mass variances.
Regional Sales Manager
Whether you need to run a profile on every different assembly depends on your requirements, and what you would like to achieve.
If you are required to document that a particular assembly is in spec, you need to profile that particular PCB. If there are different process windows, for example when the assembly includes sensitive components with unique tolerances, then again you are recommended to profile each unique product.
Some electronic manufacturers that only need to process their products in the vicinity of their process windows will occasionally organize their assemblies in groups with similar characteristics such as PCB size, weight, and process specs.
The right way to manage your thermal process is to profile each unique assembly. Practically speaking there are occasions when this is not possible, such as for those clients that require the return of all the assemblies and hence do not accept sacrificing a single PCB for profiling purposes.
There are, however, fixtures on the market now that very accurately emulates the profile on the PCB. By using non destructive profiling methods for an initial profile on the PCB, the fixture can be used for all subsequent profile runs. This method eliminates the need to sacrifice the PCB. Go to www.kicthermal.com for a copy of RIT's non destructive profiling research paper.
inspīre solutions LLC
Many companies can be very successful using 3 or 4 profiles and grouping boards by size or mass. Where things get tricky is when thermally challenging components like high-mass, slow-heating, or heat-sensitive parts are involved.
A bit of common sense should be applied to choosing the profile for each new board. If a board is similar in thermal requirements to others you make, it is probably unnecessary to profile it, and your first piece inspection should tell you if that assumption was correct.
However, if you can predict thermal challenges with a new board that are not addressed by an existing profile, it will be safer and more cost-effective in the long run to develop a new profile, rather than end up manufacturing a bunch of scrap.
Application Engineering Supervisor
Many of my KIC customers do exactly what you are suggesting. Here is how it is done:
Step 1 - Cluster your boards into two-three categories or more like you suggested.
Step 2 - Develop a profile for each group.
Step 3 - Run Navigator prediction software to drive the profile as deep into specification as possible.
Step 4 - Test run actual profiles for a few boards from each group to the same recipe as your representative profile.
Step 5 - Re-run the representative profile and sample test profiles periodically to ensure your assumptions are still valid.
The key to success is Step 3. It is critically important your profile be as deep in spec as possible. Since this task is difficult to achieve, having intelligent software such as Navigator to help you to these ends will not only save you time, effort and frustration, but your results will more likely be better than then what can be done by trial and error.
It is important to drive your representative profile as deep or to the center of your process specifications so your grouping of boards have as much wiggle room that they remain in specification. If you choose a good representative PCB for your profile and your group is fairly homogenous, all other test sample PCBs would reveal profiles that cluster around your representative board.
This is why you want to be close to center of your specification. You don't want to be close to any of your process limits and therefore have some of your samples fall out of spec. Performing step 4 will make sure you have set up your model correctly.
Now before you get carried away with this plan, you need to think long and hard about the equipment you have to run this on. Don't forget the best reflow ovens and process environments can have several degrees of variability from profile to profile.
This makes it even more critically important step 3 and to periodically test your assumptions with step 5. This variability is due to a whole host of reasons some of which you may even have no control or limited control over, such as tolerances in the process controller, TCs, across the board uniformity, external factors such as exhaust to name a few.
You can find excellent discussions on this forum addressing this issue of variability and therefore the need to profile.
Give you a great example from a customer who I met today. The customer had an unacceptable level of rework on a PCB. Turns out they only have five degrees at peak to play with in profiling a BGA to spec, while not cooking other smaller mass components in close proximity, while trying to bring the BGA's joint to liquidus.
You really need total visibility as to what is going on with the reflow process for this board, wouldn't you agree? To compound matters, the customer happed to be using an oven characterization fixture to measure instability, but it was revealing there was no instability in the reflow process.
I bet the customer the variability on this oven exceeded several degrees at peak and the changes were so subtle the device was not picking it up. We ran a profile to prove my point. The profile run last week compared to the one run today showed just this! Same profile, same board, yielded a different profile result. Customer was able to see the source of their cold solder joints.
Take a deep breath; the best modeling may be flawed if you don't factor in process control. You could be spending more time created unreliable modeling and chasing problems than if you had just run profiles in the first place. This is not an exact science, thus a wide range of opinions you will find on this subject. So when in doubt what do you do? You guessed it, PROFILE! Hope this helps.
Global Account Manager
Completely disagree. I consider this is the perfect recipe for disaster - even if there are 3, it is like 1 good for all. The other experts gave you multiple reasons or things to take in consideration that are more than enough to make you understand that you will need 1 oven profile for each assembly.
Now there are going to be profiles that are very close or maybe even identical for 2 or more assemblies. From here to the 3 profiles for all is a very long way...
Engineering and Operations Management
Ask yourself this question:Bernie Marks, Avidyne
The last time there was a board problem, how long did it take you to convince them (and yourself) the process was in control? It happens too every process engineer, the easiest place to point the finger at is board build.
Having the evidence on hand eliminates the fishing expedition and gives everyone a head start at finding the real problem. On a side note: If it has not happened too you yet, give it time.
What is lost to the "experts" in this field is that a single profile cannot work for multiple CCAs for the simple reason that nearly all CCAs go through the oven under various oven loading situations. You profile each CCA so that it's profile is near the center of the window of the required profile. That way if you are running a single CCA it will see its hottest profile, but if you are running 7 or 8 CCAs through the oven at once each will see its coolest profile due to the oven loading.
Both the hottest profile and the coolest profile must be sufficient to produce good wetting without any cold or overheated solder joints. This means that you should profile every board, because some similar boards will fall outside of either the hottest or coolest profile points during real production conditions within the oven. All profiles should be tested in a single board running alone, and the same single board running amongst several others, so you know your profile works under different circumstances of real production.
Strongly disagree. While you may be able to lump some builds together by mass what you need to focus on with each assembly are the parts with the largest mass that can vary greatly from board to board. These parts are the ones that need extended dwell times above liquideous to not only create an acceptable solder joint there is the matter of complexing the flux activators.
We have seen a difference of only 5 degrees below what is required is enough to leave active/hygroscopic residues which will certainly increase your risk of electrical leakage leading to dendrite growth. Thermal profiles don't take near as long to complete when comparing against a failure analysis project.
Take the time to profile each board based on overall mass, but look specifically at the largest thermal mass components and components tied to larger grounds. It's worth the effort.
After 30 years of profiling boards, I am still surprised that this question comes-up. Every board is different. How do you know you did it right if you don't profile? And while you are at it, set it up right, don't tape thermocouples down to a bare board. Every tIme I see that I just shake my head. You need to be at the joint interface, which usually means drilling into the board if your doing BGA's or LGA's. I drive customers nuts when I tell them I will be scrapping out one of their boards and all the parts, especially if it is a board worth several thousand dollars. But better one scrap up front then a pile of them because the profile was wrong. And keep that board around to validate from time to time.Alan Woodford, NeoTech
I find most manufacturers have low, mid, and high density board recipes and that is where they start but that is only your starting point. At that point you need to profile the board because each pcb design will act slightly different within the oven. Taking into account what parts you put on the board due to mass or temperature limitations you can then fine tune your profile into what it needs to be. Just remember, you are profiling how your paste reacts on that specific board and every board design is unique. Some boards may overlap on recipes but without proper profiling you don't know how good your profile is and that could lead to unpleasant conversations with your customer.Clinton Buldrini, Matthew Associates
|Submit A Comment
Comments are reviewed prior to posting. You must include your full name to have your comments posted. We will not post your email address.
Free Newsletter Subscription
Circuitnet is built for professionals who bear the responsibility of looking ahead, imagining the future, and preparing for it.
Insert Your Email Address