Printed Circuit Board Assembly & PCB Design SMT Electronics Assembly Manufacturing Forum

Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.


BGA Placement Process

Views: 7963

#61464

BGA Placement Process | 25 March, 2010

Hello,

I have been having some troubling issues when it has come to BGA Placement and I was hoping if there have been others who have been able to resolve similar issues.

We are curently doing reworks on a variety of BGA, but seem to be having issues with only one particular PCB. The issue I am currently having is that we seem to be having intermittent failures when it comes to placing our 38mm x 38mm BGA onto the large PCB (nearly 40cm X 25cm). Our success rate seems to be rather low (around 60% to 70%).

Naturally the first thing I went after was my profile on the APRXL5000 OKI Device, but after numerous and expensive X-Ray and Barascopic analysis, we could not tell that there were any issues with the placement itself. Furthermore, the amount of warping that is occuring seems to be controlled to a minimum, which was extremly difficult to get to considering the size of the PCB and component.

We are currently baking the BGA on a weekly basis, and depending on the moisture level on the BGA, we bake them every 72 hours. We are also buying the components directly from prominent sellers. Also the fact that we are able to get a success rate higher than 50% shows to me that the components themselevs seem to be reliable.

The issues we are seeing after the placement itself is that the component exhibits symptoms as if it were shorted internally. As we do X-Ray analysis, we found to be no bridges or any other such issues with warping, open or voids. We were even concerned with the Head-in-Pillow effect that are discussed in other discussion forums. This was ruled out upon when we removed some of these faulty components. We found that the pads were sufficently connected by examining the remaining solder residue on them. We are currently at a loss on explaining what the possible causes of our issue is.

We are using a Dual Heater which has 5 zones to rework the component. A scrap board was used to develop the profile, where holes were drilled at different points in the area. I made the profile to ensure that the Temperature Differential does not exceed 15 Degrees C. The profile along with the support bars ensure that the PCB does not overly warp.

We finally ensured that the issue was related to the component itself by lifting the BGA and testing the board for shorts and other issues. These tests showed no issues with the PCB once the BGA was lifted. We can tell that the BGA is the root cause of the problem, but not why.

I am not sure at which temperature the wire bonding in a BGA geos critical and breaks. Would anyone know if that is a possibility, because the Broadcom BGA which we are working on currently has a built in thermal conduit from the top of the BGA where an external heatsink can be attached.

Have others come across similar issues? If so, have there been resolutions to such issues?

reply »

#61476

BGA Placement Process | 26 March, 2010

HI, regarding the baking....why are you doing it every 72 hours ? How many times has each device been baked ? If you are worried about moisture in the devices the best policy is to store them correctly in dry packs or dry cabinets or bake them once just before they are assembled. Follow IPC JEDEC 33 for storage / baking recommendations. Repeat baking will cause the solder spheres on the BGA to become very heavily oxidised and you'll probably get soldering problems and solder defects. This could be the root cause of your issues. Are you baking the devices at the correct temperature/time as specified on the moisture barrier bag or in IPC JEDEC 33 ? What moisture sensitivity level is the device..(I guess it could be level 4 if you are baking it every 72 hours). Your identification of the defect is confusing....you say it seems as if the BGA is shorted internally and then you look at possibility of head in pillow defect...head in pillow is where the solder paste does not mix completely with the bga solder sphere alloy....the pcb pad will show good signs of soldering but the solder paste to solder sphere joint is where the problem usually is. Generally the only way to prove you have head in pillow is by cross-sectioning joints. The fact is that repeat baking is a large cause of head in pillow.

reply »

#61478

BGA Placement Process | 26 March, 2010

Thank you Graham for the response.

The particular BGA that we currently Bake is a Level 3 BGA, which means the BGA will expire in 7 days after removal. We are currently baking the BGAs once a week for the problem described above. These devices are under constant use and therefore each batch is being constantly baked prior to usage. We are currently purchasing a Sealer that can ensure the BGA are stored in an air tight seal ESD bag.

I am currently Baking the BGA components at 125 Deg C for 24 hours, which is what I thought was the standard practise in removing moisture from the component. I will confirm this through IPC JEDEC 33.

My reasoning for looking into the Head-in-Pillow effect was because on some occasions, we had noticed that some of our ground pins were not soldered correctly, eventhough 2D X-Rays showed the ball to be connected. I was under the impression that if the Ground pins are not being soldered correctly the BGA would overload itself and thus creates the short. The connection was weak because the heat is being dissipated through the ground plane (which is huge on the current board). I have changed the profile to pre-heat the board with our bottom heaters and I measured the board temperature to be 130 Deg C once the pre-heat cycle is complete. I seem to be having limited issues with ground pins not being connected, but I still cannot tell if the pins are being connected correctly. I will try to get some sample boards to be corss sectioned and have a look at the results.

We currently do not use any form of solder paste to assist in the connection between the solder sphere and the PCB pad. Is this something that you think would be important for the replacement process. This is currently a tin leaded BGA and thus I did not think of it being an issue.

I was not aware that I would be reducing the integerity of the BGA spheres by repeated baking. I will be sure not to bake the BGAs if they are not being used. Is there a particular type of cabinet that you use to store the BGA? We were also looking into purchasing an Nitrogen cabinet, but thought it to be an expensive solution.

Regards

reply »

#61480

BGA Placement Process | 26 March, 2010

Hi Krish, thanks for your detailed reply. Sorry I assumed you may be using a solder paste to solder the BGA to the pcb. Either solder paste or Gel flux is best for resoldering BGA to pcb. Are you using Gel flux ? Most popular long term storage for moisture sensitive devices is a dry cabinet, like those offered by Totech and others. You do not need to have nitrogen. It does have advantages but cost is normally an issue. A dry cabinet at @1% humidity will keep your BGAs dry and even the Totech ones will remove moisture from them, thus resetting the floor life. You'll never need to bake another BGA again. But as this is a tin/lead BGA then even repeat baking isn't such an issue as it is with lead free BGAs. What peak temperature are you getting the BGA to when soldering it ? As the BGA is big and has a heat sink I think concentrating on getting a good thermal profile will be important. Good luck.

reply »

#61484

BGA Placement Process | 26 March, 2010

Hello Graham,

We are currently using Tacky Gel Flux during our replacement process. We dip the BGA component on the flux prior to placement. Once it is dipped, the alignment and placement is done.

As for the profile, I believe I am hitting the correct peak temperatures. The requirement for the Component is 183 Deg C as you most likely know. The Boiling point for the tacky flux however is 213 Deg C, which is what I try and hit. I am at about 220 to 230 max temperature and I try stay above reflow for atleast 1 minute.

The current profile preheats the board for approximately 4.5 minutes before taking the BGA above reflow. While the board is being preheated to 130 Deg C, the BGA itself is being warmed through the top heater. At the end of the pre-heat zone, it reaches a temperature of 170 Deg C where it will be taken into reflow. Since I have a 4 Zone machine, I use the first two zones to pre-heat the PCB, the third zone to take the BGA above reflow and the final zone to maintain the time above 213 Deg C so that the flux residue is removed from the component.

I know typical profiles have 3 minutes to pre-heat the board and then complete the reflow cycle in another 2 minutes, but I found this profile to work best for my board. I have attached a picture of my profile for your reference. Please comment on the profile if you think it is incorrect in any way.

Regards

Attachments:

reply »

#61485

BGA Placement Process | 26 March, 2010

Krish,

Looks like your peak temperature is around 230C and time above 183C is about 120-150 seconds. This is excessive for a leaded BGA. You should be aiming at 200 - 215C as the peak joint temperature with time above 183C of 60 -90 seconds. The problems you are seeing can also be due to the component being overheated internally. Attach a thermocouple to the BGA's top surface. If it exceeds 245C then there is a good chance thats your problem.

reply »

#61489

BGA Placement Process | 27 March, 2010

I agree with Steve. Your are too hot and too hot for too long. What is the device spec for Peak Package Temperature (PPT)? (this is usually on the moisture barrier bag) Many leaded BGAs have PPT of 225 Deg C. Even if it is higher, like Steve suggested maybe 245 C then with your profile your probably getting the body really close to it, and hence internal damage is a real possible. The solder balls melt at 183 C, but to help them flow better you need to get them to 200~215 C. Time above liquidus @ 60 secs...even 45 secs is usually ok. This will be much kinder to the package so your yields should improve. Have you seen the device manufacturers recommended reflow profile for the device ? Good luck.

reply »

#61495

BGA Placement Process | 29 March, 2010

Thank you Graham and Steve,

I realized I was running a bit too hot and that is easy for me to change.

The problem however was that I have 14 other profiles that I had created that still give me varied results. I am still facing issues with each different profile that I had made where the profile would work in some instances and then would not even power up. In those instances we did not have any shorts. In these profiles that I made, I followed the standard specs (like not going over 220 Deg C and etc.), but yet I could not get a consistent success rate.

To answer your questions, the PPT for component is 225 which I am clearly exceeding and I will correct for. The time above reflow recommended by the manufacturer is 60 - 150 S which I believe I am within.

The profile that I had initially made is attached as a jpeg on this E-Mail. From the Purple Reading, you can tell that the Board Temperature is around 140 Deg C at the end of Stage 2. Stage 3 begins with the bottom heater focusing on the BGA rework site itself, which means the board itself will maintain a temperature close to 100 Deg C for the duration of the profile. The purple TC was placed in close proximity to the rework site. I had taken these readings today to ensure nothing has changed with the profile since it was made. Is there any issues with this profile? I could not find anything wrong with this profile, but I was still unsure why we were having failures.

My main concern is if I am overlooking some other issues because after 14 tries over a 3 month span, I am unable to make a stable profile for this board. I was hoping if there is something else that I might be overlooking. I do not think Alignment is an issue since the machine is using a standard optical prism to align the BGA with the Pads. I will continue to do profiling on this board to see if I can perfect it, but is there something other than profiling and moisture control that I should be looking for?

Regards

Attachments:

reply »

#61523

BGA Placement Process | 31 March, 2010

One thing I would suggest right off the top is to go to a paste printing process. This will get first pass yields higher (usually 10+% higher than gel flux).

If you are suggesting that you have good collapse everywhere, the profile is adequate (sounds like too adequate) and you are using a paste print, I would look to see if you couls isolate the issue even further to a ball location (or is it random?).

Next I would suggest a die and pry to determine if there is a good connection between the balls and the pads.

Finally, I would look at several cross sections in/around the board.

Let us know!

Bob Wettermann BEST Inc

reply »

#61555

BGA Placement Process | 5 April, 2010

Hello Bob,

What makes this problem even harder for us is that we are not the manufacturer of this particular board. We are just performing re-works on them based on the issues they come in for, and thus refurbishing the unit.

We are currently in the process of ordering the stencils for this BGA and I will give you an answer based on what we find for this case.

We used to see that it was not a particular point but more random than anything. From what I noticed, it was mainly some of the ground pins were not making contact. With the newer profile that I had attached on an older post, I do not seem to have any connectivity issues (from a settling point of view).

Those two options (dye and pry and cross sectioning) are also something that I am looking into pursuing. One of the other posts had mentioned a problem where an issue of connectivity can arise if the flux gel is heated for too long. If this is heated for an extended period of time, it would dry out and thus create a glass type material which can prevent connectivity between the solder ball and pad. I want to try running a few samples without flux, but I am scared that I will oxidize the connection and thus make it electrically ineffective.

For the paste printing process, is there any particular paste that I should be using? What is the exact process when we switch to paste printing (i.e. Is there any particular steps that I need to follow and Is there anything in the profile that I need to look for)? We have not done any paste printing before and it will be a new process for us.

Regards,

Krish

reply »

PCB Soldering Tools

ii-feed SMD Intelligent Feeder