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

Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.


Voiding in LGA (LT) soldering

Views: 14786

We having issues installing the Linear Tech LTM8023. We have... - Aug 30, 2010 by SMT-manufacturer  

#62556

Voiding in LGA (LT) soldering | 30 August, 2010

We having issues installing the Linear Tech LTM8023. We have different soldering results in our PBfree testruns. The voiding and solderball rate is not stable (random positions and sizes). The pcb's (gold finish)are prebaked, the LGA's are stored in Drycabinets etc. We tried already different value's of the following parameters: Stencil thickness, Stencil apperture, stencil reduction, different PCB manufacturers, P&P pressure and reflowprofile. The only item we have not changed is solderpaste, we now use Almit paste. LT and Almit have also investigated the X-ray pictures in combination with the procesparameters and can't give us more suggestions to approve the results. Who can give some tips or advise to solve the soldering issues with this LGA.

reply »

#62561

Voiding in LGA (LT) soldering | 30 August, 2010

We currently use this devise in one of our designs with no issues. We are using a 3mil stencil and a 45% apperature reduction. It is virtually impossible to eliminate voids but they are in an acceptable range. Solder balls are caused on this component if you lay down to much paste. I have built using Multicore LF218 and Aim293+. The Aim paste is leaded and the Multicore is Leadfree.

reply »

#62572

Voiding in LGA (LT) soldering | 31 August, 2010

Hi Mark,

Thanks for your responce! The stencil thickness and especially the reduction of 45% are not recommend by LT, did you came to this combination experimentally? We never have the illusion to produce voiding free, but now its around the acceptable range. What's the quantity of the LGA's you produce. Do you also have X-ray pictures of your soldering results? Which Multicore paste do you use, I know the LF318 or do mean another Multicore paste?

reply »

#62581

Voiding in LGA (LT) soldering | 1 September, 2010

Dear SMT Manufacturer, I am wondering if the device could have so much moisture in it and storing it in a dry cabinet isn't going to remove it all. As a final test to see if moisture in the device is causing the voids/solder balls then I'd bake it at 125 C for 48 hours before assembly. This is Linear Tech recommendation and is inline with IPC JEDEC 33. Infact feedback form one of my other customers who assemble a similar device is that Linear recommend the device is always baked before assembly. Good luck.

reply »

#62583

Voiding in LGA (LT) soldering | 1 September, 2010

Dear Graham,

Moisture isn't the issue because we have done several tests. 1>We baked the parts just before assembly. 2>we baked the parts and store them in 4 days in drycab (60degC, Rh0,5%)and finally assembly 3> we baked the parts and refowed them on a leadfree profile and finally the assembly on the boards. The same we also did with the bareboards. Anyway, moisture in devices can't be the cause of voiding.

reply »

#62585

Voiding in LGA (LT) soldering | 1 September, 2010

Dear SMT Manufacturer, As moisture is eliminated as the cause then solder paste volume will probably be the main contributing factor. More paste means more chance of solder balls and more flux fumes that need to be evaporated to reduce voiding. Even though it is outside LTs recommendation maybe try 80 micron thick stencil with a small reduction (10% reduction off length and width of pad).

reply »

#62587

Voiding in LGA (LT) soldering | 1 September, 2010

Dear SMT-manufacturer, I would recommend to also have other device joints on board X-ray inspected to understand if void formations are seen across the board or confined only to one component. If seen across, your "Reflow Profile" could be the cause of such problem.

reply »

#62595

Voiding in LGA (LT) soldering | 1 September, 2010

Dear Sachu_70,

During the tests we also checkt other devices, there are a few small voids visible, but they are marginal. We only see large voids in this device. During the testruns we tried several reflow profiles, but without large succes.

reply »

#62615

Voiding in LGA (LT) soldering | 2 September, 2010

Seems interesting ... I am certain this could be resolved. But first one question: How did you monitor the reflow profile at this component? How and where was the thermocouple connected?

reply »

#62617

Voiding in LGA (LT) soldering | 2 September, 2010

We drilled a hole in the centre of the LGA and put the thermocouple inside, so we measure the exact temperature between PCB and device.

reply »

#62618

Voiding in LGA (LT) soldering | 2 September, 2010

Good. This is the right way to profile. Let me ask you, have you anytime soldered QFN componets on your boards? The concept of LGA soldering is similar to that used for a QFN package. QFN soldering also exhibits void formations, and is an inherent process drawback when using such devices. In my opinion, voids can only be reduced, but are not avoidable in solder joints of QFN and LGA devices.

reply »

#62619

Voiding in LGA (LT) soldering | 2 September, 2010

Just to add to my comments,solder joint voids do occur in LGA. Voids in LGA can be larger due to geometry and greater ratio of flux to solder. IPC-A-610D specifies a greater than 25% voided area is a defect for BGA, however, it does not specify the defect limit for LGA. This specification applies only to collapsing spheres, which LGA does not have. In an article, an LGA device manufacturer states that: "Thermal cycling with LGA solder joint voiding of greater than 25% repeatedly shows excellent solder joint reliability."

reply »

#62627

Voiding in LGA (LT) soldering | 2 September, 2010

Voiding is tricky, but some pastes have flux residue that can escape others don’t. It is normally called out. So I would change pastes. If you do not want to do that, try a slightly longer time above eutectic (give the flux a little more time to get out)

Frank www.dispensetips.com

reply »

#62632

Voiding in LGA (LT) soldering | 2 September, 2010

I ment LF318. We have a power supply board that has 3 different types of these LGA's and i have had great success with the reductions and stencil thickness mentioned. We had to go with 4 mil stencil due to the consideration of the other large components on the board. I have attached a picture. one thing that is critical is board size. I have been able to keep the board size down to below 4"sq.

Attachments:

reply »

#62639

Voiding in LGA (LT) soldering | 3 September, 2010

Yes, we assembled a lot of QFN in the last years, but we don't have a X-ray machine by ourself, so we cant't inspect each device. This type LGA was designed on a prototyping board of our customer so we would validate our LGA reflow process and a research-centre took the X-ray pics of all the testruns. You are right that IPC doesn't specify voiding rate at LGA's, we have contacted IPC last year about that, but there are still no specifications available. LT did their reliability tests with 30% voids and all the test where approved. But we will not accept voids of 25% or higher.

reply »

#62643

Voiding in LGA (LT) soldering | 3 September, 2010

Mark,

Can you send some X-ray foto's of your LGA solderjoints, I am curious about the results of a marginal paste volume in combination with the Multicore LF318. Did you follow the recommandation of LT and use also soldermask defined pads on your board?

Thanks in advance!

reply »

#62664

Voiding in LGA (LT) soldering | 9 September, 2010

You can reduce and almost eliminate voids and other paste related defects in LGA's BGA's, QFN's etc, by using solid solder deposit.

Its been around for 24 years, 13 in the US. It is now included in the new IPC 7093 specification for bottom termination components (BTC's)

Go to http://www.sipad.com for more information.

reply »

Reflow Oven

reflow oven profiler