Pads of SMD connector falling of PCB

[jusaca]

I have some trouble with a current manufacturing batch of some PCBs.
We use JST connectors handsoldered sideways as SMD connectors. We do this for years already and did some strength tests in the past. We always snapped the pins of the connector before we would lift the pads. Other connectors (including THT) are not an option because of size constraints.

Now with a new production batch we have about 10% (!!!) of the boards failing, because one of the connectors falls of (always the same connector).
As you can see in the image, the pads just came off.

We talked to the PCB manufacturer and the assembly house, but they just point fingers.
Is it possible to decide with the image what the most likely problem is? Is it a problem with PCB manufacture? Or more likely a problem with hand soldering? Maybe to hot soldering temperature?
What puzzels me is that it is always the same connector, while the other ones are quite strong. The layout is pretty identical for all 5 connectors next to each other...

[Psi]

The corner edge of a pcb has less heatsinking and so will get hotter and quicker with the same iron temp.
So that would be my guess. They're probably using a really hot tip to solder things quickly.


Could also be a pcb fab issue, maybe one side of each panel was not submerged enough and the bubbles/turbulence at the top of the etch bath made it etch faster than the rest of the board. Resulting in thinner copper along one edge.  But if that were the case you should be able to tell with some calipers.

Another option is a packaging/shipping issue. The corner of the pcb is subjected to more forces and more likely to get damaged by a knock.

[mon2]

Agree with the last post. SMD connectors just cannot sustain the force during a hit, especially in the corners like your pcb design. You can see that the nylon boardlock pegs used to secure this edge connector have broken off.

Highly recommend THT but aside from this suggestion, share more details of this part number and also the other JST connectors to the right side in your pic.

Which part numbers were used to the right side ?

Will share more details (when at the office in a few hours) of a mfr we have used in Asia that can customize the JST connectors to suit. They are dirt cheap but MOQ will apply. We customized a similar connector that JST does not mfr. It was a THT connector with body of 18 pins but stuffed with only what we requested for the male pins. This allowed us to build a galvanic isolation between interface ports due to the missing metal male pins.

Perhaps they can customize the smd version to allow for a single SMD connector to be stuffed for the entire horizontal plane.

Then the entire blade of a single nylon connector body will assist to absorb the forces.

Just a suggestion.

Update:

1) can the PCB layout be modified for future PCBAs ?

2) We have used with very positive results, LHE out of Asia:

https://www.lhecn.com/blogs/what-is-smd-connectors/

Note the extra SMD PCB pads to improve the strength of the mounting of this connector.

Here is our last sales contact:

Ms.Amy
Sales Assistant
Zhejiang Lianhe Electronics Co., Ltd.
Tel:86-577-62383212
Fax:86-0577-62382566
Email:lhecn@lhecn.net
Trade Manager:lhecn@lhecn.net
Skype:lhecn-littlebear
Address: No 8 Innovation Road, Gangwan area, Nanyue Town Yueqing Zhejiang.China  325609
Website:http://www.lhecn.com/      http://www.lhecn.com.cn/

Contact Amy to ask for their advice on the issue. Do keep the soldering temp down but do believe you need to reduce the force this 2 pin connector is facing during transit (more bubble wrap?) and/or insertion of the mating cables. This vendor can customize the JST style connectors to suit.

[thm_w]

OP, we can see glue holding the connectors together, I assume that was from the manufacturer?
So you've left some information out that the connectors are glued for reinforcement. Was the previous supplier using the same glue? Same amounts? Proper prep?

Agree with the last post. SMD connectors just cannot sustain the force during a hit, especially in the corners like your pcb design. You can see that the nylon boardlock pegs used to secure this edge connector have broken off.

What boardlock pegs?
These just look like regular through hole connectors, mounted on the edge of a PCB.

[jusaca]

We use JST B2B-ZR connectors, which are just mounted as SMD parts. All five connectors are the same, the ones at the edges to not have extra support.
That glue like stuff is actually nothing that we specified. I only noticed it when taking this picture. I assume the assembler glues five connectors together to increase speed for manual placement, I doubt that it will give any additional support.

A design change for future is possible (and something obviously has to be done), but we have no option to increase neither length nor width of the board. THT is not really an option, because the other side of the board is jsut to dense to make room.

One "solution" might be to just increase strength of the joints by pooring some epoxy over the pins after soldering...?

[Northy]

Can you add small vias either in the pads or at each side/end of the pads to provide some strength to the pad?

G

[jusaca]

There are actually three uVias in each pad to the second layer, but they don't seem to do much. A very tiny PTH (about 0.2mm) might be worth the shot.
Do you think that would help significantly?

My inital hope was that it would be possible to recognize what went wrong visually from the PCB. The white areas below the pad do not by any chance tell us that the FR material became to hot?

[Psi]

There are actually three uVias in each pad to the second layer, but they don't seem to do much. A very tiny PTH (about 0.2mm) might be worth the shot.

yeah, i noticed that. Makes me wonder if the PTH plating is too thin.

Or it didn't bond together properly for some reason.

You could try to measure the pealed off copper with a micrometer and see if its as thick as it should be.

(I don't really consider them uVia's until they are so small they have to be laser drilled.  So 0.15mm and smaller.  0.2mm is just a regular via)

[Psi]

Hm.. just reading your comments and noticed, are those vias blind?

So you have vias joining the top later to the next layer down only?

I'm wondering if what is happening is a thermal expansion issue made worse by the blind vias locking layers 1 and 2 together and providing stress against the other layers when heated from the top until something gives. And being on a corner of the pcb would make it even worse.

Do you have any ground fills (isolated from the connector pins) on the layers 3-n under those pads ?
As soon as heat is applied layers 1 and 2 will get hot and layers 3-n will stay cold from the ground fill pulling all the heat away. Which is going to create stress on the PCB layers trying to expand at different rates which might cause the resin to fail.

It's just a thought, i'm no expert in blind vias.

[Northy]

There are actually three uVias in each pad to the second layer, but they don't seem to do much. A very tiny PTH (about 0.2mm) might be worth the shot.
Do you think that would help significantly?

I've never actually had a need to use blind vias so I'm not sure how much they will be helping to be honest.

I've used PTHs (around the pad) a few times to try and beef up the strength of a SMD pad that I've felt could potentially get ripped off a board as it's for a high or easily knocked in storage part. AFAIK I've never had one fail, but I don't know if thats luck or not. In my mind it must add a significant amount of mechanical strength, especially if you can mirror the pad copper onto more layers under the connector?

G

[wraper]

As someone who professionally repaired cell phones in the past, and have seen this a lot, I can say this is a poor design which bit your ass once you had PCB with worse copper adhesion. PCB house is not necessarily at fault. IME different laminates often have vastly different actual copper adhesion even if spec is similar (more expensive laminate does not mean better adhesion). I suggest making those pads as large as possible, a lot of vias stitching around and make openings in solder resist for soldering.

What puzzels me is that it is always the same connector, while the other ones are quite strong. The layout is pretty identical for all 5 connectors next to each other...

Other connectors are longer and have more pads, thus hold better. Also they are not on the corner which helps against abuse too.

[jusaca]

@Psi
uVias are laser drilled from one layer to the one below. This is what we have here. Three uVias (150 um) from Top layer to the layer below.
Regarding the solid layers: Actually yes, inner layers 3 and 4 are GND and 3V3 as solid planes. That might possibly be a problem with heatsinking, but our assembly house told us, that solid copper is no problem, as long as it is symmetrical to the core (otherwise warpage might be a problem).

@wraper
The connectors are all the same, the picture might not make that clear. It is five times a 2-pin connector (JST B2B-ZR).

For a new layout I will try to somehow fit some tiny PTHs, but I'm not very confident they will fit. Maybe I can reduce some passives a bit further in footprint size... And despite that, it seems we are stuck with the idea to glue/epoxy the connectors together and to the board to increase strength.

[wraper]

As connector is on the edge, improvement from using adhesive between connector and PCB will be marginal. If you put adhesive between the connectors it may somewhat help. Or you may put the adhesive on top of them all and then place some strength bar on top which will unite all of the connectors into a single piece. It could be just a narrow piece of FR-4. Also I suggest using somewhat elastic adhesive rather than completely hard stuff.

[thm_w]

As connector is on the edge, improvement from using adhesive between connector and PCB will be marginal. If you put adhesive between the connectors it may somewhat help. Or you may put the adhesive on top of them all and then place some strength bar on top which will unite all of the connectors into a single piece. It could be just a narrow piece of FR-4. Also I suggest using somewhat elastic adhesive rather than completely hard stuff.

There was some glue between the connectors as mentioned, although poorly applied, and unknown type.

I think filleted epoxy between the PCB and connector can be quite strong, if the vias don't fix it.
Lightly flexible epoxy yeah (eg https://www.3mcanada.ca/3M/en_CA/p/d/b40066461/).

[mon2]

Consider to contact (ok pun intended) Loctite for advice on this case. They offer some nice products / adhesives. We love, yes love, their GC-10 solder paste !!

[jusaca]

I think filleted epoxy between the PCB and connector can be quite strong, if the vias don't fix it.
Lightly flexible epoxy yeah (eg https://www.3mcanada.ca/3M/en_CA/p/d/b40066461/).

I will at least test it, it seems like a nasty fix, but with the size constraints it might be the best short term solution.

[kylehunter]

Consider to contact (ok pun intended) Loctite for advice on this case. They offer some nice products / adhesives. We love, yes love, their GC-10 solder paste !!

Sadly it appears they are making the GC-10 obsolete. At least according to DK.

[SMTech]

Consider to contact (ok pun intended) Loctite for advice on this case. They offer some nice products / adhesives. We love, yes love, their GC-10 solder paste !!

Sadly it appears they are making the GC-10 obsolete. At least according to DK.

Henkel/Loctite/Multicore are reorganizing the brands a bit, I saw on LinkedIn, I forget the details but all the solder products and the staff were moving over so they all lined up under one roof. That could mean it ends up with a slightly different listing somewhere like DK. I don't see an EOL on the Henkel site but the GC pastes are a range so you could try one of the others.

[Gribo]

Depending on your current carrying requirements, I would go for 1mm pitch connectors with positive latch, from the usual suspects (Samtec, Molex, JST) For example the Samtec T1M series.

[mon2]

Sadly it appears they are making the GC-10 obsolete.

Just confirmed that GC-10 is alive and well. Contact Dave M --> adtool (canada) for more details.

No, GC10 is still available
Henkel sold their solder division to Harimatec and will still be producing all solder materials

Adtool Corp.
514-943-4857