What do you do when you have temperature tolerant components on your PCB that require a different reflow profile?
To subscribe to my Podcast for iTunes (click here).
What do you do when you have temperature tolerant components on your PCB that require a different reflow profile?
To subscribe to my Podcast for iTunes (click here).
Reflow Process Inspection is catching on as the next advancement in-line inspection systems for SMT Reflow. I have pulled together the various aspects of RPI to better explain how it works and what are its benefits.
Unlike SPI and AOI that are defect inspection systems specifically designed for viewing solder deposition and component assembly respectively, RPI complements these systems by inspecting the performance of the thermal process IN-LINE. RPI inspects the thermal process for any joint, including those that are not visible to the AOI system such as BGA components.
RPI charts the thermal Process Yield and DPMO
RPI provides information on the “health” of the thermal process over time. The Yield and DPMO charts provide instant understanding of detrimental changes in the process. The following format is easy to read and understand and often used by management as well as engineers.
A question was posted on Circuitnet (May 18, 2009) asking if there is a standard test board that can be used for profiling/calibration of a reflow oven. Answers were provided by profiling companies, oven and rework station manufacturers.
The consensus from all groups was:
Here is a summary of panelist replies, including from yours truly (for a full transcript go to http://www.circuitnet.com/articles/article_59131.shtml).
The short answer to your question is no. There is no industry standard test board.
Test boards, also sometimes called “golden boards,” are an imperfect measure. Often, they are used for calibration purposes, but keep in mind every time you run the same PCB through an oven, some mass of the board is lost. For this reason, a true GOLD standard that is identical to your production board is difficult to achieve, unless you can somehow recreate the exact same conditions each and every time you profile your standard test board.
Since PCBs lose mass, some manufacturers will create calibration tools out of plates of stainless steal and use metal slugs to simulate components. Of course, a hunk of metal is no closer to a production board than a golden board, but at least it gives you a relative measure that is repeatable.
So what is the best answer if there is no perfect tool? There is no better representation of what is going on with your Reflow process than running an actual profile of a production board. The good news is that there are tools available that do not necessarily mean running a profile equals destruction of a sellable product nor does it mean that you need to waste the next few hours profiling.
Both oven manufacturers and profiling companies have developed onboard databases that allow you to develop in-spec profiles before you even profile (see this link) so when you run a verification profile you can at least do so knowing that the PCB being used can still be sold!
Another method of ensuring your process is continuously in spec and can serve as an early warning if things are going astray is the use of systems designed to monitor your oven.
For example, KIC’s 24-7 and Vision will create virtual representations of your PCB all based off of a true “golden board,” since the PCBs used to set up the system to create these virtual profiles are run through your process as actual profiles. As an added bonus, these same boards do not suffer from the repeated use problem described above with golden boards.
Oven manufacturers normally use custom designed test fixtures to simulate a board but their real purpose is to measure uniformity across the oven and confirm that the oven is working correctly. The test board might match a small percentage of boards actually being produced but is not close to many more and is not intended for calibration.
….I have personally seen companies place unrealistic performance specifications on reflow oven testing with boards that have little to do with actual production needs. For example, we once were required to show that an oven could reproduce an inspect ramp soak spike profile on two 12 X 18 inch aluminum sheets that were 0.040 and 0.080 inches thick without changing any recipe parameters….
….From a surface mount manufacturing point of view – single board oven performance testing has little benefit. The real answers are to use actual boards with TCs placed on the critical components….
First of all, nothing can take the place of running profiles of your actual PCB’s…..
…There is really no industry standard test board available……to suggest otherwise would be dangerous whereas this would assume that all assemblies are identical and this is not the case. If you set the oven up to the test board, it would invariably be different than your own assemblies. This is not a risk worth taking.
Test boards can be created to illustrate specific characteristics of a reflow system, be it heating / cooling capacity, thermal repeatability, thermal uniformity across a conveyor system or designed to emulate a particular type of product.
It’s very difficult for one test vehicle to do it all well. A test board supplied by an oven manufacturer or independent supplier will likely address one or two of the aforementioned.
For example, a test vehicle designed to compare several ovens across multiple lines can be vastly different from a test vehicle designed to measure cross belt uniformity. Similarly, a test vehicle designed to gauge percent infrared, may not be well suited for CpK measurement.
Mike Buetow of Circuits Assembly magazine moderates a discussion panel on soldering and thermal profiling at APEX 2009. Panelists include Keith Howell of Nihon Superior, Fred Dimock of BTU and Michael Limberg from KIC.
Much of the 30 min discussion hits upon how customers often confuse an oven’s recipe with a PCB’s profile/recipe. Factors such as density, delta Ts, belt speed, different components and extraction are used as examples as to why the oven’s set points don’t always match the temperatures on the PCB. All panelists agree that a fair amount of customers do not understand these important concepts.
Fred Dimock of BTU cites an interesting study he conducted to highlight the difference mass has on the peak temperatures a board experiences without changing the oven set points. The example he gives is a 100gram board that achieves a 231 C peak when compared with a 230gram board only reaching a 225C peak with everything else being equal. Panelists agree that customers often expect to see the same profile at a given oven set up, when obviously factors such as mass play such a critical role!
All panelists talked at length about how to minimize delta Ts as an important factor in producing quality PCBs. The PCB design and layout of components was discussed by Keith and Mike.
Fred cited a study that higher convection rates also yield a lower delta T, taking into account the need to maintain a stable environment early on in the reflow process before components have had a chance to take hold. Starting at low convection allowing the flux to become tacky (thus keeping components in position) and eventually raising convection in the peak zone can minimize large deltas.
Fred also shared a profiling trick with Ramp Soak Spike profiles he likes to use when trying to minimize the delta Ts at peak. In RSS profiles, one would run as close to the edge of the top of the spec of soak and get as high as you can in temp early before you hit the spike, but you need a quality profiler and good ThermoCouple attachment to pull this off, Fred added.
The session also covered briefly upon topics such as:
To watch a video of the session, click here: http://blip.tv/file/1969267/
What is go-no-go profiling? Well basically you get a pass/fail, green light/red light indication at the end of your profile letting you know whether you are in spec or not.
How often are most of us, when we get the “go” or green light at the end of the profile indicating that we finally got the profile in spec, ready to button it up and turn the thermal process over to production?
If I have and in-spec profile does this mean I’m ready for production?
Well, if you think about it, if you are just relying on a go-no-go indication for your profile results you are never sure if your process is at the very limits or somewhere deep within spec. In the case of the former, depending on the demands of your production on your thermal process you could be operating at an out of spec condition during production, even though your profile gave you a green light!
If you are like most production facilities being in spec during production not only matters but having an indication of how far within spec is your process is just as important since depth of being in spec is the only way to ensure continuous quality production. A common tool used for not only determining whether you are in or out of spec as well as how far in or how far out (depth) is using PWI (process window index).
See the blog entry PWI.
Perhaps the most challenging components to thermocouple are BGAs, since the area of reflow is hidden underneath the component. The most accurate methodologies are destructive. The simplest way to thermocouple a BGA may be to drill a hole on the underside of the BGA and thread the TC bead into the drill-out hole that allows access to the target area without having to remove and re-attach the BGA.
Another method is to use a very thin gauge TC wire (40 AWG) and separate the two dissimilar wires, as shown below. Then attach the TC and place the BGA on top of the TCs. Again, the point of this exercise is to achieve an accurate “direct” reading.
Special thanks to Scott Nelson at Harris Corporation for providing this example.
A non-destructive method for thermocoupling BGAs is to simply mount the TC on top of the BGA and, perhaps, to the underside of the PCB directly below the BGA, and develop an offset. There is no right or wrong answer; much depends on your production tolerances and a whole host of other variables that have been discussed in this guide. The point is that only you know your process and limitations with respect to product and tools.
Note: This is an area of particular interest to me since it is a concern that just about everyone has an opinion on and no true right or wrong way has been developed. Stay tuned more to come soon.
Baseline profiles are used purely for set-up purposes. They are manually-fed into the reflow oven and are run in tandem with a monitoring system in order to create what is called a “virtual profile.” Baseline profiles, thus, become the standard by which all automated profiles are compared. If there are any deviations between the baseline profile and the automated profile, this information is recorded and can trigger alarms and stop production temporarily. Baseline profiles also have the added advantage of becoming a true “gold standard” for all of your profiles, since the same profiling board is not being used repeatedly or subject to degradation of the PCB, thermocouples or operator error.
The set-up of a baseline profile is performed by running a normal profile, per recipe. Meanwhile, 1-2 fixed mounted probes (typical an array of 15-30 thermocouples) collect their own set of data as the PCB runs the length of the Reflow oven with a profiler. This information is then compared and processed by the monitoring system. This array of internally-mounted thermocouples is able to reconstruct a virtual model of a profile. When a production board enters, without the profiler, the same TC readings that were present when your oven was profiled are now re-mapped onto each and every PCB.
Since it is not always possible to run all products under a single profile, most factories require product change over time in order to reset the oven to a new set of temperatures, per product class or board. Depending on the variations of PCBs, changeover time can add up to hours, days or weeks of lost production time.
So, what can you do to minimize this dead period? Determine, in advance, the best grouping of products that have shared profiles and then schedule product runs over available equipment that minimizes changeover. It is worth every minute of company effort and time to review daily production in order to run products that require small set point changes.
It is entirely possible, in an organized methodical way to consolidate groups of PCBs under the umbrella of one or more Thermal profile, I like to think of putting them into small, medium and large buckets. Testing can be done to prove to yourself and/or your customers that the profile is both producing a product within specification and, quantifiably, how deep within spec. Yes, it is possible to have your cake and eat it too!
Profiling software can do some of the heavy lifting for you. You can reload past profiles to see whether or not they can be produced at given oven set points. Every time you ask the profiling software to run a prediction analysis, it searches billions of possible combinations. If you want the deepest in spec process, it will pick the one closest to the center of all variable specs. If you want to maximize conveyor speed, it will search for that one profile that gives you the highest throughput without violating any of your parameters. As it is crunching through these possible combinations, there are millions of other profiles in spec that you don’t see because they are not the optimal profile for a given request. This is not to say that they are any worse. When you reload a recorded profile to determined fixed oven set points, there may very well be profiles that are still deep within spec that can be run under your new Reflow oven configuration. Perhaps, instead of changing your oven over 4-5 times a day, it now only needs to be changed 1-2 times.
Many of you will have an issue with “bottlenecks” in your process. This can happen at any point in the SMT Reflow process. Depending on the product that you are manufacturing, it is also likely that the “bottleneck” will jump from equipment set to equipment set. For our purposes, let’s look at the reflow process when it is identified as your bottleneck.
I have seen several methods that address a reflow bottleneck. The obvious solution is to increase the conveyor speed of the reflow oven. This is a task that requires a bit of skill. Your profiler becomes the single most effective tool to improve the Throughput Time (TPT) of the reflow process.
Let’s look at the fundamental changes of your profile with an increase in conveyor speed. First, the PCB will spend less time in each zone. Also, your process will move toward the shorter end of your spec as defined in seconds. For example, if soak time is defined as 30 to 90 seconds, your actual process will be perhaps in the 30-50 second range as opposed to a comfortable 50-70 second range that was established at slower conveyor speeds.
The longer the oven, the more wiggle room you have for increasing conveyor speed without having to make significant changes to your profile. Having a longer oven suggests that the PCB stays in the reflow process for a longer period of time, but keep in mind that it really comes down to how many products per minute exit the oven. Whether the oven is 10 feet or 30 feet in length, a higher conveyor speed will increase the number of products that exit the oven per minute.
Work In Process (WIP) is determined from the time the lot enters the SMT process to the time it is ready for shipment as a finished product. This duration is stated normally in hours and can add up to a few days to weeks, depending on the product. Having a longer oven does not mean increased TPT nor does it violate your WIP objectives.
It is tricky working with shorter ovens with fewer zones since they do require higher temperatures per zone to reach the desired specifications, as compared to longer ovens. I have found that using solder paste that uses a Ramp to Spike (RTS) profile works better in ovens with fewer zones. The RTS pays less attention to the soak and more to the overall length of the profile (the soak, of course, is often not listed in the solder spec for an RTS profile). Also, shorter ovens impact the slope in the RTS profile. For example, if you have a 5 zone oven, the first zones will need to be set at a fairly high set point in order to process the rest of the profile. At this point, the slope becomes very steep as the PCB moves from ambient to 160°C. 160°C could be the set point of the first zone! Also, in a distance of just a few feet, the product will need to rise in temperature to greater than 217°C in PB Free and above 183°C in eutectic solder. Many specifications will call for a peak of 200 to 240°C, which puts further demands on your shorter oven. In this instance, it is desirable to calculate the slope over the entire profile, setting it from 130°C to 180°C over a shorter period of time. Again, you need to look at how long (in seconds) the PCB stays in each area of the profile when designing your spec.
Now that we have looked at oven length when considering an increased throughput, how do we develop a profile that will remain within specification at higher conveyor speeds? Some profiling software will allow you to make a desired change to conveyor speed and then return a predicted profile. This typically can be completed in seconds, before even running your first profile.
A clever feature of some profiling software is that you can set a range of allowable conveyor speeds while maintaining acceptable limits to your process window. For example, when factoring the variability (drift) in your oven, you can comfortably run your process using only up to 70% of your available process window. In practice, anything over 70% is risky due to drift that can push your process out of spec. To eliminate this concern, run a “what if” scenario, where you define your minimum and maximum range for conveyor speed and maximum allowable PWI (in this case, set to 70%).
The profiling software will literally search billions of possible combinations, giving you the maximum possible conveyor speed without violating your 70% PWI threshold. Of course, you may find the conveyor speed to be too slow still. What can you do? Bump up your allowable process window or buy a new oven. In either case, you are in control of your predictive modeling (this sure beats hundreds of hours of trial and error and possible board destruction). This process is similar to the prior section on Getting your Product Deeper in Spec and it will take as long as it takes to heat up or cool down your oven and re-profile if verification of your new predicted profile is required. In practice, I find that it takes less than one hour.
Q: What happens when defects occur when the thermal load on the oven increases? Do you slow down production? Do you change the oven set points by cranking up the heat to compensate for the increased load?
A: The answer is to establish a NEW profile that is deeper in spec., a profile that is able to better stand up to the daily variations of a reflow process.
Today, profiling software allows you to establish these new deep-in-spec profiles with relative ease. You can precisely define your specifications and run various predictive scenarios. For example, you know that you can’t slow down your conveyor speed, but you can change your oven set points. The profiling software can give you a predictive result that puts you as deep in spec as possible before ever having to run a profile.
In practice, how much work needs to be done to take this out-of-spec process and bring it within spec depends on a lot of factors. How far out of spec are you to start with? What inputs can be changed? How tight are your specs?
If your process is already taking up most of your process window or not far out of spec, then only minor changes will most likely be needed to bring your profile much deeper into spec. In this case, only one additional profile is likely required to bring your profile very deep within spec. In my experience, this profiling process takes about 30 minutes, most of which is waiting for the oven to cool. If your profile is far out-of-spec., you may need up to an hour to bring it within spec.
Each time you re-profile, it is an opportunity to further improve your profile, bringing it further into spec with each effort. Profiling software will tell you a possible scenario for improvement each time, which takes your excellent deep-in-spec profile still deeper within spec. Each one of these changes, on average, takes about 30 minutes.
A word of caution: having a profile in the center of the spec or at 0% PWI, is not always the optimal improvement. While “0%” PWI is statistically desirable, there are other factors to consider. For example, though 30% PWI indicates that you are only utilizing 30% of the allowable process window of your solder paste, in practice, when you find that a PWI of 65% produces a physically better connection, which is better? Specs are just that: specs. They have a range for a reason. In this case, at the upper end of the spec (opposed to the center of the range), a joint may solder better. Advantageous about most profiling software is that you can go back and re-define your specs to see what your new profile will look like without having to rerun the profile. The allowed range can be further narrowed to a particular spec, which results in a better joint. In essence, you are now re-defining your spec, and all future profiles will only consider this new range.