Fixing PCBs with dry-jointed SMD components?

[ocset]

Hello,

We have some all-SMD, FR4  PCBs which comprise LEDs and Driver components,  or some PCBs which have  just driver components. These PCBs are either double sided or 4 layer. They are 1.6mm in thickness. The components range from D2PAK FETs down to 0402 resistors, and 4mm by 4mm SMD LEDs.

We are  sometimes  finding that  these PCBs, when delivered to us, don't work because of dry-jointed LEDs or other components.

We find this out because we  can often make PCBs work by simply 'dunking' on/near the component pads with a big , wetted soldering iron.

However, we really need a better way to reflow these SMD components. Do you believe a hot-plate would be best?, or some kind of little SMD oven?, or an IR oven?
Sometimes the dry-jointed components have connection to a thermal copper pour  on the PCB bottom. -But sometimes this is not the case.

We are actually wondering if the best & quickest way of doing it could be to simply put the PCB on a cheap hot-plate and waft over the suspect dry-joint components with a  "gas gun" soldering tool?..(blowing scorching hot air on the suspect components)

[RobK_NL]

Send them back to the assembly house and tell them: "Fix this and get your shit straight because we won't tolerate this in the future."

[wraper]

These are certainly not dry joints. I's either tombstoning, pads under the parts not making connection because of wrong amount of solder paste or faulty components which become alive when heated. Tombstoning may happen because of incorrect pad layout, amount of solder paste or inferior reflow profile.

[KL27x]

However, we really need a better way to reflow these SMD components. Do you believe a hot-plate would be best?, or some kind of little SMD oven?, or an IR oven?

IF the faults are completely random and greater than a few percent, then it sounds like an assembly house fail. But I think it best to spend some time to better characterize and localize the issue while you have the boards. It is more likely that there are specific parts that tend to be faulty for local issues like layout/heatsinking or IR reflectivity, etc.

If the boards just weren't cooked long or hot enough or they were pasted or placed badly, the assembly house should probably fix them. Of course, shipping and customs and turnaround might be an issue.

A cheap USB microscope for like 30.00 is a no-brainer for taking pictures of the problem components and for communicating with your board house. Posting some pics on the forums could be helpful, as well.

If the faults are truly dry joints, as in not enough solder paste, and if an iron can fix them, then iron rework IS probably the easiest way to fix them. You can't stencil paste an already-populated board. If you have to add more solder (with compatible flux) you might as well use solderwire and an iron, barring big heatsinks, or otherwise, which might require more than an iron.

[RobK_NL]

A cautionary note with regard to reworking with an iron: Many components do not take kindly to the stresses caused by uneven heating. They may develop cracks that will cause reliability issues or, if you're lucky, instant failure on power-up.

A hotplate can greatly reduce this risk and is to be highly recommended. Simple hot-air stations are available for not much money and are a way better option than a gas gun tool.

[Conjure]

A cautionary note with regard to reworking with an iron: Many components do not take kindly to the stresses caused by uneven heating. They may develop cracks that will cause reliability issues or, if you're lucky, instant failure on power-up.

A hotplate can greatly reduce this risk and is to be highly recommended. Simple hot-air stations are available for not much money and are a way better option than a gas gun tool.

No disrespect, but your first claim is hogwash. Show me one verifiable claim of unreliable or damaged passives caused by only heating one pad at a time.

The ONLY exception I can think of is if you're heating highly temperature sensitive diodes in this way. Then of course you may run into some issues.

OP: something like the Weller WTA 50 may help your passive reflowing. Otherwise, tell your fab house to fix their solder reflow profiles.

[wraper]

No disrespect, but your first claim is hogwash. Show me one verifiable claim of unreliable or damaged passives caused by only heating one pad at a time.

MLCCs certainly don't like it. NASA had issues with hand soldered MLCCs when they start failing in the space.
https://nepp.nasa.gov/files/16346/08_002_01%20GSFC%20Teverovsky.pdf

[Bassman59]

A cautionary note with regard to reworking with an iron: Many components do not take kindly to the stresses caused by uneven heating. They may develop cracks that will cause reliability issues or, if you're lucky, instant failure on power-up.

A hotplate can greatly reduce this risk and is to be highly recommended. Simple hot-air stations are available for not much money and are a way better option than a gas gun tool.

No disrespect, but your first claim is hogwash. Show me one verifiable claim of unreliable or damaged passives caused by only heating one pad at a time.

Chip capacitors easily break when reworked with a soldering iron. I can’t remember off-hand which vendor’s application notes said basically, “Do not hand solder these parts,” but one of our techs came across it when we were trying to sort out why boards were failing.

[RobK_NL]

No disrespect, but your first claim is hogwash. Show me one verifiable claim of unreliable or damaged passives caused by only heating one pad at a time.

I was going to link to that NASA document, but wraper beat me to it 

The ONLY exception I can think of is if you're heating highly temperature sensitive diodes in this way. Then of course you may run into some issues.

The problem is not with the temperature sensitivity of the device as such, but with the mechanical stresses caused by uneven heating. This may cause cracks in the package which
ultimately can lead to device failure.

As noted above, MLCCs are especially prone, but also some LEDs , and I have personally experienced problems with a voltage regulator in a SOT-223 package.

[AndyC_772]

Send them back to the assembly house and tell them: "Fix this and get your shit straight because we won't tolerate this in the future."

^^^ This!!

The last thing you want to do is actually fix any of these boards. Inspect them very carefully under a good microscope. Take detailed, clear photos of the problem. Leave no doubt whatsoever what the root causes are of a good sample of failures.

Then, label them up, send back the *whole batch*, and refuse to accept further deliveries until you've received a report explaining exactly what went wrong in the assembly process, and what has been done to ensure it doesn't happen again.

[TJ232]

I have almost missed the new episode of the "treez" saga,my bad.

Treez, really, are you for real ? this is a question that has only one answer even if you are using a PCBA house for your own small hobby projects, not commercial products!

But I must agree with you,where is the fun to just act normal and take the only actions that can be taken in this kind of situations.  So, for the sake of the whole story, just choose your favourite blowtorch and cook them well on the both sides. And for that fancy SMD caps issue, it's well known, but kept as the best secret in the industry: use the angular bi-focal blowtorching technique. This is the best way to do a proper high quality Made in UK product these days.

[ocset]

The last thing you want to do is actually fix any of these boards.

Thanks, thing is we have to fix the boards to  by way of reflowing  them to proove that  its a dry joint problem.
Today i had a led driver ic that wouldnt work and had weird voltages on its pins......its pins are actually beneath it as it is  QFN type  package.......ie we cannot  visually inspect the joints.
I actually  got two large wetted soldering irons and kind of "smelted" all over  and around the IC to get it hot enough to reflow again...then i mopped up the excess solder...and hey presto, the product worked fine......there is no way i could have  prooved that this was a dry joint  issue with any other means.
Also our leds are the same, they have pads beneath the component body and we cant inspect them. We cannot proove that the problem is dry joints until we actually reflow them.

We are thinking, with   some of your kind advice, that we will have to reflow the boards by putting them on a hot plate and then blow-torching the ICs with a gas gun. For the leds.....we will just have to put the pcb on a hot plate again, and dunk a big wetted solder iron on the copper pour thats connected to its pads.
Then we can see if its a dry joint problem or not.

Then i guess we will have to give a  test jig to the PCBA house so they can see if the boards are working, and then if they are not, they can adjust their solder bath temperature profiling. What do you think?

[Gyro]

Then i guess we will have to give a  test jig to the PCBA house so they can see if the boards are working, and then if they are not, they can adjust their solder bath temperature profiling. What do you think?

So your volume assembly house has no way to test the boards before they ship them to you??     This is a commercial product isn't it? - Just checking. 

Yes, I think you'd probably better send them a test jig!   

P.S. Would you really have the nerve to ship product to customers that had been reflowed twice, once in an uncontrolled manner!

[ogden]

we will have to reflow the boards by putting them on a hot plate and then blow-torching the ICs with a gas gun.

Make sure to adjust temperature of your torch accordingly so components on the board are crispy & well-done.

p.s. This is british electronics humour or what?

[Gyro]

p.s. This is british electronics humour or what?

I wish! Please don't think that this is representative of British engineering, it's just embarassing! 

[dmills]

I think Treez went to the "Lucas" school of engineering! [1]

Treez, go and do some study of DFM and reliability engineering, QFN (And even BGA) are inspectable either by optical methods (if you leave a ~5mm moat around the BGA) or by xray (Will cost you a few hundred quid to get a set of dud boards done), it gives you AND the assembly house information about the problem you cannot get any other way.

You need to get with your assembly house and figure out the issue (Could be footprint, paste or even solder mask openings), fixing this sort of thing yourself is expensive and gives 'repaired' boards that I would be very reluctant to send to customers.

Regards, Dan.

[1] Lucas were a major manufacturer of automotive electrical components of such poor reputation that there were widely knows as "The prince of darkness" back in the 1970s.