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

Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.


QFN soldering

Views: 12066

We have started using QFN packages in large volume and are h... - Nov 09, 2007 by devajj  

Thanks! ... - Jul 11, 2008 by thughes  

#52430

QFN soldering | 9 November, 2007

We have started using QFN packages in large volume and are having problems with no-solders. We are using 6mil stencil with 1x1 paste print to pad. These also have thermal pads with thru vias that we reduce the stencil aperture by 1/4. Board also has BGA's, SMD connectors, inductors many large parts that need solder volume.

Any help would be appreciated. JD

reply »

#52432

QFN soldering | 9 November, 2007

What is the real problem? No-solder can mean a variety of things. No-solder due to lack of paste? No wetting to the part? Co-planer part? Please specify......

reply »

#52434

QFN soldering | 9 November, 2007

Best I can tell (Factory is in Mexico) is no solder from paste print. I'm ordering e-fab type stencils for them hoping to get better release. Also mandating they clean after each print, rotate their paste to keep it fresh, do LSM checks each hour for SPC check, Basic SMT 101 process control.

reply »

#52438

QFN soldering | 9 November, 2007

Hopefully this will help you. I happen to have a few similar problems with PN# MC33982BPNA. I had the stencil made at 6mil thick and had a step-down for 5mil in the problem areas. That seems to take care of the problem we had at the time. ALSO FORGOT TO MENTION the center was stepped down to help as well.

reply »

#52448

QFN soldering | 10 November, 2007

Hello, there are some questions first to be answered: 1. What is the body size? 2. What is the pitch? 3. What is the terminal size? 4. What is the land pattern size? 5. The Thermal square pad in the middle, with how many vias?

6. Keep in mind, that to much solder in the center pad will cause 1, 2 or 3 sides to be lifted, to less solder in the middle cooling pad - not enough heat transfer will occur, which in turn make the QFN failing. If The designer don't use 100 % of the capacity of the QNF, normally you will be safe to say 70 % of the ground pad should be soldered and still be cooled enough. A tip is to to ask the designer whether or not if they are going to use the full extents or capacity of the QFN.

For ex. QFN 9x9mm or QFN 5x5mm with 9 vias or (whatever nr) in the ground pad, pcb thickness 1.6mm => make 4 square rounded corner apertures around the ground pad and the total reduction of 20 % (80 % solder paste). Then you still be able to solder the terminals around.

We SMT-mount & reflow soldering various QFN packages onto our customers pcb:s without any problem. We also make our own stencil apertures always with the very same stencil thickness = 127 microns = 0.13mm (approx)

If you have any questions about this, dont' hesitate. Best Regards, Mika Ps. Hey, the designer doesn't know of how much cooling it is needed. He/She only looks at the effect of the QFN and calculate with the ground layer of the pcb, of how many layers of the pcb, single sided/double sided, how many mass components and so on Ds.

reply »

#52451

QFN soldering | 10 November, 2007

1. What is the body size? Who cares. 2. What is the pitch? Doesn't amtter. 3. What is the terminal size? Doesn't matter 4. What is the land pattern size? Really doesn't matter. 5. The Thermal square pad in the middle, with how many vias? If you know your solder paste - it doesn't matter.

Sounds like you under cut your stencil. Go 1:1 on leads and 50% reduction on the body pad. Works every time no matter what QFN.

reply »

#52453

QFN soldering | 10 November, 2007

Hi Real Chunks, Nice to meet you. Well, I certainly not agree with you. Sorry. Can You be a little more specific about your claims, about my statements? Do you know what you are talking about? I wish no polemic about this. Best Regards, Mika

reply »

#52454

QFN soldering | 10 November, 2007

BTW, The surrounded QFN fine pitch terminals should have the normal aperture reduction of 7 % in case of RoHS Senju solder paste. This works for our Telecom customer.

The vias in the ground pad will be somewhat filled with solder and makes it difficult to calculate the apertures. All of this is if you are looking for an optimum solder joint; both on the side terminals and the ground pad. We produce double sided pcba:s with a variety of components, such as 0402, 0603, QFP:, TSSOP and BGA. /Mika

reply »


RDR

#52460

QFN soldering | 12 November, 2007

Chunks is right, reduce center pad by 50% print the signal pads at 1:1. this will ensure you solder QFNs fine regardless of the actual part.

If you are not printing then it is process capability issue and has nothing to do with QFN but rathjer small aperture.

R

Russ

reply »

#52461

QFN soldering | 12 November, 2007

> BTW, The surrounded QFN fine pitch terminals > should have the normal aperture reduction of 7 % > in case of RoHS Senju solder paste. This works > for our Telecom customer. > > The vias in the > ground pad will be somewhat filled with solder > and makes it difficult to calculate the > apertures. All of this is if you are looking for > an optimum solder joint; both on the side > terminals and the ground pad. We produce double > sided pcba:s with a variety of components, such > as 0402, 0603, QFP:, TSSOP and BGA. /Mika

Thanks everyone for your input. It looks like a 5mil stencil, 1 to 1 for the pads and some level of reduction for the center thermal pad. What about process? 1.Clean stencil after each pass 2.Use 2d inspection each pass 3.Change paste every hour or two 4.SPC 2 boards every hour (Auto LSM) They don't have AOI or 5dx xray yet. Dose this sound like good SMT process control? Thanks!

reply »


RDR

#52463

QFN soldering | 12 November, 2007

You need to perform process a capability study prior to putting these controls into place.

Cleaning a stencil after each pass will get ugly, solvent residues remaining in apertures etc.

2D inspection is good.

Change paste every hour will get very expensive, normally it is around 4 hours, Check the paste data sheet. I have had good no cleans last 8 hours easily on 20 mil pitch.

Time should not be used for validation of print process (LSM), both time (stencil life) and number of prints are key here. An aperture can clog on the next print after LSM reading, you have a defect and the process is in control technically.

use the LSM to determine when and how often you clean stencil, add paste, etc...

I would make sure date code and storage procedures are in place as well.

To be honest, QFNs are very easy packages to process, you need to really determine what the defect is, it is either the paste not being printed due to poor process or stencil design, OR the center heat pad is lifting the part out of the solder during reflow.

Frankly if you the customer have to tell the manufacturer how to print and how to control it you really need a new supplier, no ifs ands or buts about it.

R

reply »

#52511

QFN soldering | 14 November, 2007

recently, we also have problem with QFN around 3% defect related to soldering. from the cross section it turns out to be solder crack. it could be the paste volume on the center pad is too much causing QFN to float that eventually will cause crack on the perimeter pad. by the way perimeter pads are solder mask defined and with micro vias.

reply »

#52527

QFN soldering | 15 November, 2007

Just to add to the subject, we experienced insufficient solder on our QFN's at our prototype level. To correct it, we went to a 5 mil stencil, 20% reduction on QFN pertures and used QFN's with solder bumps on the pads. We had great results using this method.

reply »

#52555

QFN soldering | 16 November, 2007

So you are saying that 0.13 mm stencil thichknes and 20% reduction of the middle cooling pad is working for you? Just as I said.. Best Regards, Mika Ps. Keep in mind that the QFN is designed to do some heavy work, while enough "heat transfer" occur. Ds.

reply »

#52572

QFN soldering | 19 November, 2007

We have found differing results based on the pattern used when reducing to total % of solder paste on the center pad. We arrived at 68% coverage on the center pad, and 1:1 for the perimeter pads.

To get the 68% coverage we used a number of different print patterns such as dots, stripes, squares, and other variants, settling on a "4 leaf clover" pattern that helped to minimize voiding in the center pad area.

We find these to be fairly forgiving parts using this methods.

reply »

#52577

QFN soldering | 19 November, 2007

BINGO! Now that's engineering!

reply »

#55409

QFN soldering | 7 July, 2008

I'm curious to know what the pitch of the outer pads is/was on your device. Also, is this a general formula you use on all of your QFN devices or was this for one specific device?

reply »

#55481

QFN soldering | 11 July, 2008

We used these same type of parameters regardless of the QFN package size. We used componentes that ranged from 3x3mm to 12x12mm Same idea. The reduction in solder paste for the center pad is the key. Too much, and you wind up with no connects at the perimeter. Too little and maybe your heat or signal transfer is not so good. Just guidelines that have worked for us... 'Hege

reply »

#55482

QFN soldering | 11 July, 2008

Thanks!

reply »

PCB Soldering Tools

SMT Replacement Nozzles