PCBA lecture hall: comparing SMD and NSMD pad design

2021-10-30

Author：Downs

In order to prevent solder ball cracking, should the BGA pads be designed as SMD or NSMD?

Why discuss the SMD (Solder-Mask Defined) and NSMD (Non-Solder-Mask-Defined) design of BGA pads/pads in the PCBA processing process? It is to allow BGA to increase the resistance to external stress and impact caused by tin cracking. Although the final conclusion is that BGA should be designed as SMD or NSMD, there is no significant difference, but the BGA pad of the circuit board uses [NSMD+plugged-via] Still our design direction has not changed.

The purpose of this experiment is to use SMD or NSMD to withstand greater stress when verifying the design of the BGA pad.

Before the experiment, I consulted some PCB experts. The answer I got was that the error in the results of such an experiment is actually very large. Can it be used as a reference for doubt, because many parameters will affect the results.

BGA solder ball thrust (shear) and pull (Pull) test conditions and parameters:

BGA solder ball thrust (shear) and pull (Pull) test preparation

▪ ball diameter: 0.4mm

▪ Laminate: FR4, TG150

▪ Thickness: 1.6mm

▪ Circuit board surface treatment (Finished): ENIG (nickel immersion gold)

▪ Ball solder alloy: SAC305

▪ Solder paste alloy: SAC305

▪ >Shear at speeds: 5000um/sec

▪ shear tool standoff: 10%

pcb board

▪ NSMD pad size (diameter): 0.35mm(pad), 0.40mm(S/M)

▪ SMD pad size (diameter): 0.35mm(S/M), 0.40mm(pad)

BGA solder ball thrust (shear) and pull (Pull) test conditions setting problems:

▪ In this experiment, the solder balls are directly soldered on the FR4 circuit board of our own design, instead of the BGA carrier board. Solder paste must be printed before ball planting to avoid displacement when passing through the reflow oven. Also because the temperature of the reflow oven is difficult to control, many solder balls are found to be deformed after going through the reflow oven, but the spherical shape is still there. In this test, a total of four boards were made, two of which were designed for SMD solder pads, and two were designed for NSMD solder pads. Each board was selectively soldered with 20 solder balls, and the via-in-pad had 11 solder balls. No-via has 9 solder balls each.

BGA solder ball thrust (shear) and pull (Pull) test results

The average thrust (shear) and average pull (Pull) after the test both show that NSMD is better than SMD, but the difference in pull is not very obvious, and the difference in pull (Pull) is considered significant. (If you have time, let's study whether the ANOVA judgment is significant. At present, we only judge whether it is significant from experience)

▪ Pull: NSMD (884.63gf), standard deviation 57.0gf> SMD (882.33gf), standard deviation 75.1gf. The difference is only 2.3fg.

▪ Shear: NSMD (694.75g), standard deviation 45.8gf> SMD (639.21g), standard deviation 54.5gf. The difference is 55.54fg.

▪ Regardless of the pull or thrust SMD and NSMD pad designs, it is shown that the pads with through-holes and plugged-via have better ability to withstand the push-pull stress, but it is not as obvious as expected. Under the thrust (Shear) test item, [NSMD+plugged-via] performed the best, which was in line with expectations. However, under the pull test item, the [SMD+plugged-via (plug hole)] performed best. This requires further discussion.

BGA solder ball thrust (shear) and pull (Pull) test conclusions and observed phenomena and bad phenomena after experimental failure (Failure Mode):

Pull: NSMD No-via pad

▪ Observing the test samples of the NSMD pad design under the tensile test item, it is found that almost most of the No-Via pads have been peeled off after the tensile test, and 7 of the 9 pads have peeled off, only 2 The pads are not peeled off. One solder ball on the pad failed before the experiment.

Pull: NSMD + plugged-via (plug hole) pad

▪ The tensile results of via-in-pad solder pads in the test samples designed by NSMD solder pads are rather messy. 2 of the 10 solder pads are completely undamaged, and there are still tips in the middle of the broken solder ball solder pads. Shaped solder material (945.4gf), the other 5 solder pads are pulled up, but the solder pads are only partially peeled off, the fracture surface is in the IMC layer of the solder (863.8gf), and the remaining 3 solder pads are completely pulled up (903.9gf).

Pull: NSMD + plugged-via (plug hole) pad

Pull: SMD

▪ The 10 solder pads with via plugs and 10 without vias are all left on the circuit board without being pulled up, and there are sharp solder residues on the pulled section. This result also proves our past knowledge that the bonding force of the SMD pad will be stronger, so the crack will appear on the solder surface.

Thrust (Shear): NSMD

▪ One of the solder pads without vias was completely removed, and the remaining 18 solder pads were not pulled up, all of them were broken at the thrust. One solder ball on the pad failed before the experiment.

Thrust (Shear): SMD

▪ 20 solder pads are all left intact and undamaged, leaving sharp solder residues.

▪ Comparing the phenomenon of SMD and NSMD pad pull-up, it can still vaguely prove that the bonding force of SMD is stronger.

Failure Mode after BGA Solder Ball Push-Pull Force Test

In summary, the pad design of [NSMD+plugged-via] actually has a certain effect of strengthening the bonding force of the pad. Although 3/10 of the pad is pulled up by the entire peeling, it is compared with [NSMD No-via] There are 7/9 solder pads that are completely peeled off, which is considered an improvement, but the improvement is not as significant as expected. It may be related to the depth and size of the via.

Possible residual problems:

▪ When the fracture surface appears in the IMC layer, the tensile stress it can withstand is the worst. What does this mean? The via-in-pad did not achieve the expected ground chrysanthemum effect?

▪ The IMC layer is actually the weakest place in the entire solder structure?

postscript:

Although the above conclusions suggest that [NSMD+plugged-via] pad design is recommended to enhance the ability of BGA solder to withstand stress, it is undeniable that if the PCB factory only wants to rely on these tiny pad design changes to achieve Solving the problem of cracking or falling of the solder of the BGA solder ball seems to be a source of fate, which is impractical! Imagine how the small solder ball can withstand the bending stress caused by the external force of the circuit board? To completely solve the problem of BGA tin cracking, it is necessary to return to the essence of mechanism design.