Lifting and tilting 0402 SMD LEDs by solder paste while soldering

[kanaron]

Hi
So I have been trying to solder 0402 SMD LEDs to a PCB using solder paste and I was wondering if there is a way to prevent LEDs from lifting and tilting. So they would stay flat on PCB

This is my process:
1. I'm putting solder paste 


2. Then I'm placing LEDs. On the photo left one is just placed on top of solder paste and the right one is placed and pushed to PCB. So there is almost no paste under LED, just aroud.


3. Then I'm placing PCB in mini toster oven (wihtout any modifications). PCB is sitting there for 5 minutes in about 130*C. I may add that PCB has surface finish LeadFree HASL (from jlcpcb) and solder paste is Sn42Bi58 with melting point of 138*C. (I tried lead free Sn96,5Ag3Cu0,5 before but 250*C didn't reflow solder properly without adding more flux which created smoke machine and on top of that LEDs were burned and melted and PCB was boiled)


4. Up to this point LEDs are sitting as straight as I placed them. Then I'm cranking up heat for a few seconds till temperature of a PCB reach about 160*C so solder melts and I'm turning heat off. Now LEDs are lifted and tilted, sometimes even LED is on its side.


5. And results under the microscope. One LED tilted to the right and one to the left. Maybe not a big deal but on LED matrix slight missaligments probably could be visible


Also one more thing: these 0402 LEDs have pads under them, so I understand that solder have to be under and probably have to be lifted a bit. But is there something that  I could  be doing to prevent tilting? I know I could try to correct them by hand and I tried that before but there is not much space between LEDs to not knock down nearby LED and also I have to make 6 PCB with 140 LEDs each so that would take weeks

[ebastler]

In the final picture it looks like the solder has not wetted the LEDs at all. It's not flowing around them, but they are just sitting on top of the solder drops.

My guess is that you are letting the PCB soak at 130°C for too long, and that most of the flux evaporates during that time. There is no need to stay at that temperature for such a long time. It's just meant to give the board and components some time to reach an even temperature. Typical soldering protocols require 30s to 1 minute of soaking time. If you have not done so yet, please try with a shorter time.

You might also have a bit more solder paste than required, and the pads look a bit large for the component size. But I don't think either is the cause for the problems you are observing.

[Gyro]

I would say much to much, uncontrolled, solder paste application and pads that are too big.

The excess solder lifts the LEDs and surface tension causes the biggest amount to lift the LED  vertical (tombstoning). Such large pads might be ok for hand soldering but not for reflow, apart from the soldering problem it will cause registration issues too. Ideally you would use a mask for consistent paste application.

As ebastler says, the LEDs don't appear to be wetting properly (although the main contact area is underneath). Maybe not enough placement pressure, old paste, soak time, or contaminated packages (sorry that's too long a list to be helpful).

[ebastler]

Good points by Gyro; there are more potential reasons for the lack of good wetting. You already tried varying the placement pressure, but old solder paste and old/tarnished components remain in the race. Nevertheless -- before you toss out and repurchase anything, start with trying a shorter soak time. And if you can, reduce the amount of paste you apply.

[kanaron]

My guess is that you are letting the PCB soak at 130°C for too long, and that most of the flux evaporates during that time.

Well I guess soak time was too short this time. I just stoped on about 120*C for about 20 seconds and I cranked heat to max till solder melted (180*C) and then LEDs tilted. I will try again tomorrow. Also I hope that flux won't evaporate if it is on the PCB for 30 minutes? I need about 30 minutes to place all LEDs

I placed less paste on the right LED but again probably too short soak time

And paste is brand new, opened about 2 weeks ago and stored in the fridge as it is recomended. LEDs are opened from sealed transport package about 3 weeks ago. So I hope that paste and LEDs are not old

And for me solder pads are also a bit too big but they are standard footprint from KiCad Diode_SMD:D_0402_1005Metric. Not enlarged for hand soldering. So I just had to trust that this footprint will be correct for 0402 LEDs

[ebastler]

Yes, this time it looks like the paste did not want to wet the PCB either. Since it did that nicely in your prior attempts, maybe the PCB did not get up to soldering temperature.

As an experiment you could try to melt some solder paste onto an LED with a regular hand soldering iron. Just to check whether the LED contacts get wetted nicely or whether they don't "like" the solder paste.

[tooki]

3. Then I'm placing PCB in mini toster oven (wihtout any modifications). PCB is sitting there for 5 minutes in about 130*C. I may add that PCB has surface finish LeadFree HASL (from jlcpcb) and solder paste is Sn42Bi58 with melting point of 138*C. (I tried lead free Sn96,5Ag3Cu0,5 before but 250*C didn't reflow solder properly without adding more flux which created smoke machine and on top of that LEDs were burned and melted and PCB was boiled)
(Attachment Link)

You should put the PCB in the center of the oven.

The edges of the oven are the worst places to use. The oven walls and door act as a huge heat sink, and the edges have the board in a place where one side is closer to the heating element.

The result is that one side of the board heats faster than the other, and that means the solder on one pad will melt before the other, which also contributes to tombstoning. (That is the official name of the problem you have.)

I also agree with all the other advice you’ve gotten. I’ll also add that unequal amounts of solder paste on the two pads will contribute to tombstoning. It’s practically impossible to dispense identical amounts of paste by hand — you either need an automatic dispenser (which I wouldn’t bother with) or stencils, which I absolutely recommend you use. With a bit of skill/practice, that produces very consistent amounts. 

KiCad’s default pads are huge, presumably designed for hand-soldering. There are various industry standards for pads for reflow soldering. When designing the stencil, do not make the stencil apertures the same size as the pads — that’s way too much paste. Most PCB manufacturers will automatically reduce the apertures by some percentage (e.g. 10-25%), but for using stencils by hand (which tends to apply a bit more paste than when it’s done with an automatic stencil printer), I’ve had better luck reducing them by 40%.

[BennoG]

Unless you want to learn yourself reflow soldering. If you order the pcb from JLC order them with the leds soldered on. This will add only
$2 per pcb or so including the parts if you order 5 pcb's  (and 2 or 5 assembled)

You can leave the expensive unpopulated and solder it yourself if you like.
I usually order the PCB's over there with all the passives assembled, most times everything without the true holes assembled.
Done from kicad this is a simple process (you need to chante 1 line in the bom file)

Benno

[forrestc]

Just a peeve of mine to start:  Can you try using the camera on your phone to take the picture of the board?  Most modern cameras produce a far better picture by themselves than you're getting with the picture of the microsocope image.  The way you're taking the picture reduces the information available so it is harder to tell what is going on.

I agree with others about likely causes.

Have you looked at the datasheet for the LEDs you're using to see what the manufacturer's recommended footprint size is?   

Also, what solder paste are you using?  i.e. brand, metal type (sac305 vs sn63pb37), and most importantly expiration date, and how has it been treated since you received it?   I.E. is it refrigerated.  If it is refrigerated, do you let it warm up before use.   Have you looked at the datasheet for the solder paste to see what the recommended ramp time/recipe/envelope is?

As others have pointed out - the problem is almost always related to the way you're heating the board, and where it isn't it's a paste quantity issue - often too much instead of too little is the cause here.

[kanaron]

Ok, so...

Yes, this time it looks like the paste did not want to wet the PCB either. Since it did that nicely in your prior attempts, maybe the PCB did not get up to soldering temperature.

As an experiment you could try to melt some solder paste onto an LED with a regular hand soldering iron. Just to check whether the LED contacts get wetted nicely or whether they don't "like" the solder paste.

I tried again, this time I waited about 30 seconds on 120-130*C and then heat slowly rised to about 150*C. Then leds again tipped to the side. And they very easy came off from PCB so again: PCB temp probably too low

Also I tried hand soldering with iron and I melted and deformed led. So I tried again with hot air set to 300*C. Results are here:

Bottom right is the one I soldered with hand... Lets say that I'm not impressed with this result.
Left leds are result of my other experiment: I placed a bit of flux on that leds. Also I put flux on 3 other leds that were lifted and tilted and also I put flux on that 3 leds from previous post that were on the side. These are the results:
Before: (this was few attempts ago but I remember that I tried putting a bit less paste)

After:


And leds that were on its side:
Before:

After:


Still leds are lifted and tilted but solder joints looks better and that leds that were on the side are in a bit better position now. However still far from perfect...

You should put the PCB in the center of the oven.

The edges of the oven are the worst places to use. The oven walls and door act as a huge heat sink, and the edges have the board in a place where one side is closer to the heating element.

The result is that one side of the board heats faster than the other, and that means the solder on one pad will melt before the other, which also contributes to tombstoning. (That is the official name of the problem you have.)

I also agree with all the other advice you’ve gotten. I’ll also add that unequal amounts of solder paste on the two pads will contribute to tombstoning. It’s practically impossible to dispense identical amounts of paste by hand — you either need an automatic dispenser (which I wouldn’t bother with) or stencils, which I absolutely recommend you use. With a bit of skill/practice, that produces very consistent amounts. 

KiCad’s default pads are huge, presumably designed for hand-soldering. There are various industry standards for pads for reflow soldering. When designing the stencil, do not make the stencil apertures the same size as the pads — that’s way too much paste. Most PCB manufacturers will automatically reduce the apertures by some percentage (e.g. 10-25%), but for using stencils by hand (which tends to apply a bit more paste than when it’s done with an automatic stencil printer), I’ve had better luck reducing them by 40%.

Thanks, now I placing PCB in the center

And Im using stencil. It would be impossible to place so much paste by hand

KiCad have two types of footprints: default and enlarged for hand soldering. Good thing that I used default but they are still a bit too big. Also my stencil that jlcpcb made looks like its holes are the same size as pads. But as in the second picture of this post when I tried adding less paste to pads (I just scooped about 20% from edges) it looks like there is not enough paste. But after adding flux and heating it again then it looks better.

Unless you want to learn yourself reflow soldering. If you order the pcb from JLC order them with the leds soldered on. This will add only
$2 per pcb or so including the parts if you order 5 pcb's  (and 2 or 5 assembled)

You can leave the expensive unpopulated and solder it yourself if you like.
I usually order the PCB's over there with all the passives assembled, most times everything without the true holes assembled.
Done from kicad this is a simple process (you need to chante 1 line in the bom file)

Benno

I kind of want to learn it. Also I tried ordering assembly from jlcpcb but there were few issues: They would not provide any stencil when ordering assembly and I kind of need stencil for a few other components. Also the biggest issue was that they didn't have any pure green 0402 leds. And also 10 PCBs, 4 layers, 140 leds per PCB and for 6 assembled PCBs I would need to order 10 assembled PCBs, so 1400 leds. That was not $2 per PCB but a lot more. Cost was high enough that I'm willing to place all leds by hand (and Im getting better at this) and I'm willing to learn reflow soldering at all cost

Just a peeve of mine to start:  Can you try using the camera on your phone to take the picture of the board?  Most modern cameras produce a far better picture by themselves than you're getting with the picture of the microsocope image.  The way you're taking the picture reduces the information available so it is harder to tell what is going on.

I agree with others about likely causes.

Have you looked at the datasheet for the LEDs you're using to see what the manufacturer's recommended footprint size is?   

Also, what solder paste are you using?  i.e. brand, metal type (sac305 vs sn63pb37), and most importantly expiration date, and how has it been treated since you received it?   I.E. is it refrigerated.  If it is refrigerated, do you let it warm up before use.   Have you looked at the datasheet for the solder paste to see what the recommended ramp time/recipe/envelope is?

As others have pointed out - the problem is almost always related to the way you're heating the board, and where it isn't it's a paste quantity issue - often too much instead of too little is the cause here.

I tried using phone but results are less readable:

But I found SD card so now Im taking pictures by microscope. Hope that it is better

About footprint recomended from led datasheet... I just checked and they recommend 0.6x0.4 and spacing 0.5. All with +-0.1 tolerance. KiCad's pads are 0.65x0.6 and spacing 0.45. So one dimension is 0.1mm larger than recomended with added tolerance.

Solder paste is "classical mechanic" (Chinese) type V4B48, alloy Sn42Bi58, melting point 138*C. No visible expiration date but I bought it 2 weeks ago. And after opening I stored it in the fridge at 5*C as recomended. It warms itself before I put it in oven. And I can't find any datasheet of this paste


Anyway I will try again. This time less paste, a bit more preheat time (close to 60 seconds) and we wil see

[kanaron]

So I tried few more times. I tried shorter heating time, then longer. I found reflow profiles for Sn42Bi58 solder paste and I tried to folow that one. So about 1 minute to 100*C, then about 2 minutes in range of 110-120*C and then heat up to 160*C for less than a minute. And it always end up the same. Tilting and falling and looking like pads are not wetted properly.

Last try looks like this:


Also once again I just put a bit of AMTECH flux RMA-223-UV and again in oven for about 3 minutes up to 160*C. Then I cleaned flux with Nitro and it looks like this:


Again not perfect but it kind of corrected itself. Also pads looks like they are wetted this time.

So my understanding is that solder paste just don't want to wet and properly stick to pads for some reason.
Should I maybe cleand PCB with Nitro and then put some no clean alcohol based flux onto PCB and then spread solder paste?

[ebastler]

Agreed, in all of your recent experiments the solder paste was wetting neither the LEDs nor the PCB pads properly. I don't think cleaning with an organic solvent will help much. If you were dealing with oxide layers on the metallic surfaces, a solvent would not be effective. And given the fact that both, the pads and the LEDs, are not wetted properly, I would rather think the solder paste is to blame.

You mentioned that you are using the low-melting-point solder -- which I don't have any experience with in reflow soldering -- since you had problems with excessive temperatures while trying regular lead-free solder. That should not be the case, really -- PCBs as well as components are designed and specified to withstand those temperatures (for obvious reasons ).

Could you try again with your regular lead-free solder paste? Now that you place the PCBs in the center of the oven, you should have a more even temperature distribution without gradients. Keep an eye on the temperature-over-time curve and try to stay reasonably close to the component manufacturer's guideline. Although in my experience you can easily get away with eyeballing that:

In my manually controlled pizza oven I keep the heat on until 160 °C are reached, switch it off and wait for the overshoot to settle and the temperature to start dropping again (~45 seconds soak time), turn the heat back on until it reaches 240 °C, turn off, wait for overshoot and temperature drop (giving ~20 seconds at 240..245 °C). Then open the door a tad (small slit) to accelerate the cooling a bit.

I saw in one of the initial pictures that you are using a temperature probe right on the PCB to be soldered; that's good. Do you trust that probe, or could it read too low, causing excessive temperatures?

[kanaron]

Well, maybe LEDs and PCB survived heating up to 260*C once or for short period of time. But when I initially tried lead free solder paste then I was heating boards for quite a long time and few times. Then PCBs were turning brown and LEDs were burned (but still worked).

Also I found LED datasheet with reflow profiles for lead free solder so I tried to sta close to that.
https://v4.cecdn.yun300.cn/100001_2003185297/1%20FYLS-0402CXXC-D.pdf
However it took me more than 5 minutes to make full cycle up to 260*C. Then I let it cooled down.

Anyway: my lead free solder paste is Easy Print no clean solder paste Sn96,5 Ag3 Cu0,5.

1. First of all I spreaded paste using stencil. This paste is a bit closer to liquid than that Sn42Bi58 so it spreaded all over the pads and between pads. From my previous attempts it really is not a big deal but about that later.


2. I placed few LEDs and cleaned exces paste a bit


3. I placed it in the oven. 120 seconds till it reached 160*C. Then about 70 seconds between 150-180*C. Then about 90 seconds till it reached 250*C and I turned heating off. Peak was at 265*C. Few minutes to cool down.
Results:




I had the same results with my initial attempts. I had 2 PCBs fully populated (so 280 LEDs) that looked like this. I was trying to fix that by placing it in oven again for about 5 minutes in 260*C. It turned PCBs brown and LEDs also brown but solder was looking the same. I tried fixing it by hand with solderiing iron and I burned off few LEDs.
Then I tried soaking whole LED matrix in that AMTECH flux and I placed it again in oven for about a minute in 260*C. Flux turned my oven into smoke machine, smell was really bad, LEDs turned really brown and black. And after that I had LEDs soaked in hard shell of burned flux. I placed whole PCB in a Nitro for a day and this is result:


So now that I have that 4 LEDs which looks horrible I could put a bit of flux and heat it up again but it will create smoke machnie. And I kind of can not open windows right now so Im not doing that

Anyway I figured that maybe I need higher temperature for this solder which my oven can't achieve. Or maybe something is wrong with that paste. Or maybe flux from that paste evaporates quicly. In this attempt with 4 LEDs flux did not evaporate as time from putting paste to placing it in oven was less than 10 minutes, but when I was placing all 140 LEDs then after like 20 minutes paste was looking much dryier than when spreaded it. And placing 140 LEDs took me about 40 minutes. So thats why I opted for solder paste with lower melting point and I bought this Sn42Bi58 one as it has the lowest melting point.

And also I can not trust this temperature probe. I have nothing to check it with. This is my only probe that was sold with Uni-T UT33C multimeter. Which is also cheap. So no idea if readings are correct

[BennoG]

Sometimes I mix 50% solder paste and 50% flux together (mix them well) and then use it on the stencil.
This makes the paste a lot thinner and added benefit you have less solder on the pads.

Can you post the kicad files. My experience at JLC is that it is not expensive (I have looked green led 0402 les than $0.01 per led) so 1400 leds is $14
Plus 1 time $2 setup fee.
https://jlcpcb.com/partdetail/Xinglight-XL1005SYGC/C965794

Benno

[ebastler]

I am running out of ideas. Those microscope photos from the lead-free solder paste (except for the last one) seem to show that the paste did get beyond its melting point, causing the original small solder balls to coagulate into somewhat larger ones -- but not forming a contiguous solder drop which could flow onto the pads and LED terminals. What's going on there?

Found a datasheet for your solder paste which indicates rather tame soldering temperatures -- it melts at 180 °C:
https://termopasty.com/wp-content/uploads/2019/07/Soldering-profile-Sn965Ag3Cu051-1-1-new.pdf

Maybe others can chime in?

[kanaron]

I will try to mix them and we will see the results.

And I placed production files in attachments. When I upload this zip to jlcpcb then costs are very low. But probably they did not detected that my vias are 0.2mm holes with 0.35mm diameter. So I changed manually "Min via hole size/diameter" to 0.2mm which changed costs from 1.80€ to 33.52€ for 5 pcs. For 10 pcs this is 38.22€. And that is before assembly costs.

Also that url to LEDs that you posted is to yellow green LEDs. Im looking for pure green LEDs. If all my attempts would fail then I could settle with that yellow green ones but I kind of would prefer pure green ones

I am running out of ideas. Those microscope photos from the lead-free solder paste (except for the last one) seem to show that the paste did get beyond its melting point, causing the original small solder balls to coagulate into somewhat larger ones -- but not forming a contiguous solder drop which could flow onto the pads and LED terminals. What's going on there?

Found a datasheet for your solder paste which indicates rather tame soldering temperatures -- it melts at 180 °C:
https://termopasty.com/wp-content/uploads/2019/07/Soldering-profile-Sn965Ag3Cu051-1-1-new.pdf

Maybe others can chime in?

Wait what???
It can do those things when above melting temperature?? I will try in a minute again with 190*C max

[ebastler]

Wait what???
It can do those things when above melting temperature?? I will try in a minute again with 190*C max

I had another "what?!" moment too... On second thought, the temperatures stated in that data sheet appear too low. And suspiciously, they are exactly the same as in the data sheet for the leaded paste. So probably a copy-and-paste error?

https://termopasty.com/wp-content/uploads/2019/07/Profil-karta-tehniczna-Pasta-olowiowa-ang-2-1-new-1.pdf

[kanaron]

Okay, so... As always I placed paste


I placed LEDs


I heated oven as in that datasheet which in fact apper to be from leaded paste
I heated to 160*C for 90 seconds. Then to 190-200*C for 60 seconds. Then it was quite strange to me that preheating phase states "or max. 210- 220˚C "... So just to be sure I heated from 180*C to 210*C and then I let it cool down... I did not make a photo. I completly forgot when I saw that paste still looks like paste. Is a bit harder in touch and under"paste" there is a bit harder part with a bit larger balls. But still... That was a paste. So I loaded it to oven again. I heated it to 220*C and turned heateing off. It overshooted to 230*C. Then it coolde down and looks like this:


I think that tomorrow I will make a call and send a mail to one company demanding explenations...

Anyway... It looks like I will need a few recomendations regarding solder paste. Preferably one that I would be able to get in Poland or at least in Europe...

[RogerThat]

Haven't read all posts but.....looks like your solder is not wetting/attaching to the board, causing your component to slide around on top of the lead-free solder (photo in your second post). I would order cards with snbi surface finish or bare copper.

[tooki]

I am running out of ideas. Those microscope photos from the lead-free solder paste (except for the last one) seem to show that the paste did get beyond its melting point, causing the original small solder balls to coagulate into somewhat larger ones -- but not forming a contiguous solder drop which could flow onto the pads and LED terminals. What's going on there?

Found a datasheet for your solder paste which indicates rather tame soldering temperatures -- it melts at 180 °C:
https://termopasty.com/wp-content/uploads/2019/07/Soldering-profile-Sn965Ag3Cu051-1-1-new.pdf

Maybe others can chime in?

It doesn’t actually say that. I suspect the “x seconds above 180C” is a leftover from an old leaded datasheet, and is describing which phase of the profile they mean, not the temperature it reaches.

That is SAC305 solder, which has a melting point of 227C. It will not be melting at 180C.

[kanaron]

That is SAC305 solder, which has a melting point of 227C. It will not be melting at 180C.

However this paste still does not melt completly at 230C and above. I tested this paste up to 270C. My oven cant reach anything higher than that. And this paste still looks like on the last photo that I uploaded. If others have different results then me then my only explanation is that there is something wrong with my paste. I dont know what, maybe reseller did not stored it properly. Maybe they had it in storage for years (I cant find expiration date anywhere so I cant tell how old it could be).

One thing that I may add about this Sn96, Ag3 Cu0,5 paste is that its "syringe" (no idea how this container is called) was very sticky when I opened the box. There were some transparent orange kind of solid liquid substance all over "syringe" and it also was inside, where Im placing piston.

It couldnt be cleaned by IPA or acetone. Only by Nitro and also not completly as it is still sticky. Also I tried cleaning that liquid inside visible on the photo and I removed some of it with tissue but it appeared again after some time. As if this paste is producing this liquid. It kind of smell like flux. So probably it is flux but why it is outside...

Anyway Im waiting for a reply from manufacturer... I doubt that I will get any reply anytime soon so if anybody have any recomendations regarding solder paste then I would kindly buy them and check if I would get proper results

[Jeroen3]

I have found significantly less tombstoning of parts when using a 100um stencil instead of 130um.
What stencil, if any, are you using?

For solder paste TS391SNL50 because it needs no fridge.

[kanaron]

Im using stencil from jlcpcb. 100um

[eleguy]

As someone already stated I also think you have pushed in too much paste.

[tooki]

That is SAC305 solder, which has a melting point of 227C. It will not be melting at 180C.

However this paste still does not melt completly at 230C and above. I tested this paste up to 270C. My oven cant reach anything higher than that. And this paste still looks like on the last photo that I uploaded. If others have different results then me then my only explanation is that there is something wrong with my paste. I dont know what, maybe reseller did not stored it properly. Maybe they had it in storage for years (I cant find expiration date anywhere so I cant tell how old it could be).

One thing that I may add about this Sn96, Ag3 Cu0,5 paste is that its "syringe" (no idea how this container is called) was very sticky when I opened the box. There were some transparent orange kind of solid liquid substance all over "syringe" and it also was inside, where Im placing piston.
(Attachment Link)
It couldnt be cleaned by IPA or acetone. Only by Nitro and also not completly as it is still sticky. Also I tried cleaning that liquid inside visible on the photo and I removed some of it with tissue but it appeared again after some time. As if this paste is producing this liquid. It kind of smell like flux. So probably it is flux but why it is outside...

I was just saying that SAC305 solder absolutely cannot melt at 180C, and that that isn’t what the datasheet claimed.

Anyhow.

It seems to me like that paste is really old.

Refusing to melt is a classic symptom of very old paste. I found some paste at work that expired something like 15 years ago, and for fun I tried applying some to a board and heating it with hot air. It did not melt properly.

Solder paste is basically solder powder suspended in gel flux. This means there is a huge amount of surface area of solder that is in constant contact with the flux. Eventually this oxidizes the surface of the solder powder, making it extremely hard to melt. That’s why solder paste has a limited shelf life.