Electronics Assembly Knowledge, Vision & Wisdom
Why Do Our Boards Warp During Reflow?
Why Do Our Boards Warp During Reflow?
Board flexing has recently started to occur on one of our circuit board assemblies, yet our process has remained the same. We are using a three-zone reflow oven. Phil Zarrow and Jim Hall, The Assembly Brothers, discuss this.
Board Talk
Board Talk is presented by ITM Consulting

Phil Zarrow
Phil Zarrow, ITM Consulting
With over 35 years experience in PCB assembly, Phil is one of the leading experts in SMT process failure analysis. He has vast experience in SMT equipment, materials and processes.


Jim Hall
Jim Hall, ITM Consulting
A Lean Six-Sigma Master Blackbelt, Jim has a wealth of knowledge in soldering, thermal technology, equipment and process basics. He is a pioneer in the science of reflow.

ITM Consulting
* EMS Qualification, Evaluation and Selection
* SMT Process Consulting and Troubleshooting
* SMT Process Development and Set-up
* SMT Process Audits
* Lead-free Process Readiness Audits
* SMT Process Optimization
* On-Site Workshops
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.

Your Name


Your Company


Your E-mail


Your Country


Your Comment



Transcript
Phil
Welcome to Board Talk. You're with Phil Zarrow and Jim Hall. By day ITM Consulting, but with you now is Board Talk. And we're here to work on and solve and comment on process, problems, situations, methodologies and all things electronic assembly.

Jim, what's today's exciting question?  

Jim
Well, we, the Assembly Brothers, Pick and Place are going to answer a question sent to us by A.M.

Board flexing has recently started to occur on one of our circuit board assemblies. The assembly has a 40 millimeter package-on-package BGA component. The BGA has an integrated heat sink. 

We are using a three-zone reflow oven. The profile worked without issue for at least six months. The problem with boards bowing or curving upwards towards the middle of the board has just started recently. Yet our process has remained the same and there have been no changes.

Any help is appreciated.

Phil
First you have to take your dish of derision - a three-zone oven! Obviously you bought it out of a catalog.

Second, last time I saw a three-zone oven used was when it was to make toast at the hotel buffet.

And thirdly - something changed.

If it's not your process, it's probably your application because that's a pretty sophisticated application. To basically be using process equipment you bought at a restaurant supply place is not a good idea.

Let's talk about the mechanics.  

Jim
I agree, a three-zone reflow, assuming they are three equally sized zones, is really not very flexible in controlling the time temperature profile. So most people use at least four or five zones for tin-lead, and preferably six or seven if you're doing lead-free.

The idea of something changes though that you said, Phil, in your infinite wisdom is true. There's two things.

You say the process hasn't changed but all of a sudden you're getting warping. 

Is this warping on an assembly that you have been building with this big PoP component for all this time, or is this assembly a new one that you're trying to run in the same "process", i.e. the same recipe on your reflow oven, which could obviously be a source of the problem where a reflow recipe was adequate for one board and along comes a heavier board and you have to change the process to achieve acceptable soldering.  

The other thing is that unbeknownst to you something could have changed in the oven.

And getting back to my brother's comment, three-zone ovens tend not to be of the highest sophistication and quality, so there may have been a degradation in the oven, fan degradation, heaters, temperature sensors getting off balance.

Or something that's beyond the internal control system of the oven to diagnose.

So you think you're getting the same thermal profile on the board and you're not.   

Have you run a confirmation profile with a good calibrated profiler and a dedicated test board which we always recommend? Thermal couples on the board permanently mounted, hooked up to a reliable calibrated data logger, run it through the oven, are you sure you're getting the same process you were getting six months ago, assuming it's the same board.  

You know I just hate to see people skimp on equipment and just going below bare minimum like in this situation and going on the assumption that reflow is commoditized.

But that's even below commoditization.

Jim
A 40 millimeter PoP, that's almost over an inch and a half square. A package on package with a heatsink.
 
Most people I know wouldn't attempt that without a 10 or 12 zone oven to make sure they had the heating capacity to heat that assembly uniformly and avoid warping and all the other bad things that can happen.

Maybe you can do it with this oven but wow, I'd be really surprised if you get consistent results over a long period of time.   

I think you're interpretation that the assembly has changed, that the other boards were simpler and now this is a heavier one that isn't hacking it with this process that they have set up.  

Phil
It brings to mind what the rest of the line looks like. I don't even want to think what they're using for printing and Pick and Place, but I'll bet you a dollar that they're using a paper slicer for singulation.

We beat up on this poor guy enough. We feel your pain and you're just going to have to hunker down and get some real equipment. It's probably not what you wanted to hear.   

I've always said along the lines with my brother, Jim, the only absolute is that there are no absolutes. You can go anything. We've seen people reflow in a toaster oven which is pretty close to what you're doing.

With the consistency, repeatability, the practicality, everything else may leave a lot to be desired.   

I will say that regardless of whether you're using a 12-zone full force convection, convection dominant with nitrogen atmosphere, you know continuous feedback, conveyor control oven or using Mattel Easy Bake, whatever you do, don't solder like my brother.   

Jim
And don't solder like my brother. 

Comments
Based on the information provided by the requestor and the number of possible causes, it is time for a good old fashioned problem solving activity.

Creating a fishbone diagram to help track and eliminate possible causes would be a great start.
  • Man
  • Materials
  • Method
  • Mother Nature
  • Machine
  • Measurement

Many possible causes have been mentioned but they require investigation.
Brian Chandler, Schlumberger
If running on the rails/chain, make sure the width is set to allow for sufficient expansion of the PCB. The board is going to "grow". If constrained from expansion across its width, the middle is going to go somewhere- in this case, up.
Rick Wyman, Benchmark Electronics
I haven't seen much discussion on the effect of the 40mm PoP on the board warpage. A package this size with a heat sink may be strong enough to partially "drive" your board shape. We've seen this first hand more often with large packages, especially as boards get thinner. Whatever relative warpage happens between package and board after solder solidification is going to come out in either shape change of the assembly or stress in the joints, either of which can be a problem. You may want to ask your component supplier if they have thermal warpage data on the component. You can also compare the shape if the PoP is excluded from the assembly and taken through reflow to see if this is a main source of the board flex.
Neil Hubble, Akrometrix
Buying new updated equipment is not the answer. Physics will not suddenly change just because money is spent - it will remain the same (at least on this planet). Assuming that nothing has really changed in your process - I would bet that the board has changed. This is very common these days. Proto PCBs run perfect and then warp during production. Your board manufacturer probably is using a different facility to have the board made due to qty. The only way to really verify this is to ask them. I would assume they changed something in the stack-up. Did your Purchasing dept have to change board houses due to higher quantities? You have to ask them!

You can start to protect yourself by asking to view your board houses gerber data. Do you ask to view it before they make the board? If not, maybe you should. This will force them to tell you if they want to change copper weight, stack-up thickness, etc. Once you do agree on all the PCB changes, make sure you record it for future orders.

As for your process, can you verify your process has not changed? Do you have an actual board profile or do the temps and conveyor speed on the oven look about the same? You should have an actual board profile made during the proto stages of manufacturing. I would also check the width of the oven from start to finish if you are using an edge conveyor. Maybe your Maintenance guy decided to "calibrate" your oven and now all your widths are slightly off.

I've seen board warp occur over my 30 plus years of PCB manufacturing, and it's always gone back to the way the board was made unless something severe has happened to your process.
Kevin Husssey, The Morey Corporation
I have gone through all the views and comments and everyone is right in their own way. My submission is that instead of approaching the solution to the problem in a quick fix way, it would be better if a systematic/scientific Statistical Problem Solving (SPS)approach is adopted. All the probable causes be listed and Possible/Root causes be identified and confirmed through a data collection / validation process.

A few Possible solutions could be arrived at and feasibility of the best possible one could be identified through Trial Implementation with results tabulated. Then the best possible solution eliminating the failure could be standardised and practiced for follow-up. These will definitely involve some time and effort and expense by a Team, but would be I think would be best possible method.
Ranganath Mandayam, TQM Consultancy
One more thought is to check the oven loading (number of boards in the reflow oven at the same time). This might be less of an issue for a 3-Zone oven (especially if it very short) but a change in the number of boards processed at the same time can impact the thermal load and also the thermal demand to maintain a 'typical' reflow profile.
Rick Arnold, Continental Automotive
Are the bare boards vacuum sealed with desiccant? Are they resealed with desiccant if the complete package wasn't consumed in a previous run?

Has the board supplier changed their materials or process?

Go over the oven with a fine tooth comb. Are the orifices clogged? Has the air pressure changed? What is the environment in the factory and has it changed? Winter in many areas has lower humidity...
Jeff Kalish, Omron Automotive
I don't think 3-zone is the root cause of problem, as this is a sudden change with a same oven, also we have ever seen similar problem on 8-zone or 9-zone ovens in our facatory. In terms of our own experiences and other reader comments, I would propose to check following potential factors. They are listed in a sequence of probability:

(1)Copper imbalance of pcb. Check to see if the PCB supplier was changed at some point. A change to copper crosshatch can vastly improve warping.;

(2) PCB material changes. Check to see if the PCB supplier has changed material or is trying new material ,or has they changed raw material supplier?

(3) PCB humidity absorbing (Moisture in the boards). Try to bake boards for at least 2 hours in 80-100-degrees. After that let them cool to room temperature and try to build.;

(4) Reflow oven issue. Check the temp-time profile if same as before. Check air flow outlets of oven is clogged or not. Flux accumulation in the orifice plates may be limiting the forced convection flow, causing the boards to now heat unevenly ;

(5) Board orientation (large ones) when entering reflow oven. Try to go with the larger dimension parallel to the oven's conveyor to avoid uneven heating between pcb edge and middle, and also try to add supporting to the center of the board (e.g., add carriers);

(6) Bad PCB panelization method. Use a v-score solution instead of breakaway;

YD, In-Tech Elect.
Well for starters, hopefully there is a established profile that was run when this process was being set up. At any rate the profile should be checked to see if it's different. It wouldn't take much in a 3 zone oven to throw the process off and possibly cause some warpage. I like the bake out idea too. But seriously, with this sophisticated package an investment in better reflow equipment should be considered. I would almost say a Vapor Phase might be a good approach.
Bill Kasprzak, Moog Inc.
There is an asymmetrical residual stress with respect to the center-line of the PCB. It might look good at room temperature but the heating during assembly exposes what was there all along. The PCB vendor should be able to control this to a certain degree.
Jeff Lewis, Holaday Circuits Inc
Jim and Phil's comments on this oven possibly changing are spot on. Flux accumulation in the orifice plates may be limiting the forced convection flow, causing the boards to now heat unevenly. I recommend heating the oven to 90C and then cleaning it thoroughly with MicroCare's Reflow Oven Cleaner, a very safe product, aqueous-based, that my customers tell me works wonders on a clogged oven. Basically, you are seeing that the thermal transfer rate has changed, if as the experts say, the board assembly has not changed. I also agree that three zones is testing the limits of reflow profilability, but I have sold bench top ovens in the past in which a customers board was so big, it was in all three zones simultaneously. I know it can be done, but it is pushing the boundaries of good process.
Russell Claybrook, MicroCare Corporation
I agree with the reader comments here - A 3-zone oven is probably not the best choice, however it is what they have and there are many other contributing factors to warping as detailed by the other readers. PCB materials (especially copper) is one of the first things I would look at. If you are not performing copper balancing in-house for those designs and stack-ups that need it, check to see if the PCB supplier was changed at some point. A very good job of copper balancing by one supplier can be quickly undone by another. Good luck!
Kevin Syverson, Silicon Forest Electronics
Sorry Guys but this question is focusing for the potential reason of change: "profile worked without issue for at least six months". I fully agree with the comments of Readers about listed potential reasons, e.g. humidity absorbing, V-scoring. I would check the raw material of PCB and also supporting the center of the board in oven instead of roasting the questioner.
Akos Keseru, Videoton, Hungary
In a previous life I also saw issues with copper imbalance. Bottom layer was mostly solid copper and top layer sparse tracks. A change to copper crosshatch vastly improved warp and gave better EMC. It was an RF board with shield metal that needed a flat board during soldering so we also twist locked the shield pins into the PTHs.

I'd also agree moisture and material changes seem more likely than machine issues but then most companies do a profile as a first step, it seems this company with the 3 zone oven is not up to speed on process Engineering for SMT.
Tony Stanley, Tyco Security Product, UK
Take a good sample of that boards and bake them for at least 2 hours in 100-110 degrees. If you do not have dedicated equipment bake them in that three zone oven on 100/100/100 settings. After that let them cool to room temperature and try to build. This should help if they are "wet.

Also check with boards supplier on that. Make sure they give you on paper (email) that the boards you receive are from same material, factory and so on as the other good batches.

Like everyone else mentioned check humidity. If you have data from the inside of the shop floor try to look into that. If you don't, check your machines for logs. Many paste printers are equipped with that (manufacturer/model dependent).
Michal, Mackwell, UK
I have seen boards warping inside oven due to the following factors: (1) bad PCB panelization choice, breakaway used for cost reduction where should be better a v-score solution, (2) rectangular PCBs (large ones) entering in reflow oven with the larger dimension perpendicular to the oven's conveyor (bad choice of board orientation to manufacturing), and also (3) PCB humidity absorbing as it was mentioned in other comments.
Glayson Figueiredo, Philips Medical Systems, Brazil
We are a board fab company, nothing was mentioned about a material change in the manufacturer of the blank circuit board. I feel when we switch a supplier things happen, things can change. I would make sure, which is hard, if the company will not let you know if they tried or change raw material suppliers. There are new ones now out there and the cost is reduced. We buy almost 100% from the same supplier.
Richard Kincaid, K & F Electronics, USA
Good comments about the reflow oven equipment. In my 35 years of manufacturing PCB assemblies for automotive and marine electronics the PCB's themselves could be an issue for warping. Moisture in the boards cause various issues with one of them warping.
Pete Walker, Johnson Outdoors Marine Electronics, USA
What about humidity? Winter is wetter than Spring. This can also affect the warping.
A Upton-Huang, Schlumberger, UK
I was quite disappointed with the discussion on why boards warp. All you did was criticize a 3-zone oven and didn't say anything about what parts of the process could have changed (except oven failure) that would "cause" warped boards!

I think oven failure causes parts not to solder properly, but that's not was was being asked. Boards can warp from a number of factors. Copper imbalance - if copper fill is left to the board vendor and not properly designed, it may change run to run. Boards also might not be stored properly and may have absorbed moisture, or maybe the local humidity took a wild swing, or maybe the vendor send "green" boards not fully cured. Maybe the vendor switched the type of fiberglass. Lots of reasons.

Am I off base? Do boards warp for other reasons than 3-zone ovens?
Richard Huziak, SED Systems, Canada
Free Newsletter Subscription
Every issue of the Circuit Insight email newsletter will bring you the latest information on the issues affecting you and your company.

Insert Your Email Address

Directory Search


Program Search
Related Programs
bullet What is the Life Span of a Profile Board?
bullet Evaluating Accuracy of Thermocouple Attach Methods
bullet How to Reduce Voiding on QFN Components
bullet Double Reflow-Induced Brittle Interfacial Failures
bullet Why Do Our Boards Warp During Reflow?
bullet Tips When Moving a Reflow Oven
bullet 5 vs 8-Zone Ovens
bullet Methods to Get Volatile Compounds Out of Reflow Process
bullet Vapor Phase Quality Improvement
bullet Understanding PCB Design Variables That Contribute to Warpage
More Related Programs