Stencil performance

[davegravy]

My first time trying to use solder paste and a stencil.

Attached are a couple microscope pictures of how I did. I feel like it deposited well in some areas but not so well on some fine pitch (especially the 8-pin DFN). It's like it seeped under the stencil or something.

The stencil is stainless 4mil from OSH Stencil, and the solder paste is: http://www.chipquik.com/datasheets/SMD291AX.pdf

With the stencil securely taped in place I applied firm downward pressure and only made 2-3 passes with the applicator. Does it seem like there's too much applied? Is my paste maybe not sufficiently viscous?

[48X24X48X]

Paste not mixed well, should limit to 1 pass, and squeegee angle should be around 60 degree. It also looks like your stencil are not full flat. Is this a frameless stencil?

[davegravy]

Paste not mixed well, should limit to 1 pass, and squeegee angle should be around 60 degree. It also looks like your stencil are not full flat. Is this a frameless stencil?

It's a frameless stencil, yes.

Not mixed well? Hmm, it came in a syringe, I don't supposed I'm meant to take it out and stir it, am I?

Thanks for the other tips.

[mairo]

Yes, paste is too much. Still you can try reflow it and fix any issues post reflow, or clean the board from the paste and do it again. Try aiming for just a single pass. Paste and room temperatures is also important - if it is too warm the paste will be more runny.

[48X24X48X]

Paste not mixed well, should limit to 1 pass, and squeegee angle should be around 60 degree. It also looks like your stencil are not full flat. Is this a frameless stencil?

It's a frameless stencil, yes.

Not mixed well? Hmm, it came in a syringe, I don't supposed I'm meant to take it out and stir it, am I?

Thanks for the other tips.

If it is a frameless stencil, you had to ensure it's fully pressed onto your PCB during the stencilling process. This is one of the reason there are stencil printer that holds these frameless stencil in place, to give a proper tension across the entire sheet. I mentioned it's not mixed well because the paste looks too runny. Whether it's a paste in a jar or syringe, the liquid portion will always follow the gravity. Therefore it has to be mixed properly to get the correct consistency. The stencil thickness is okay for the IC pitch you are using.

[davegravy]

Yes, paste is too much. Still you can try reflow it and fix any issues post reflow, or clean the board from the paste and do it again. Try aiming for just a single pass. Paste and room temperatures is also important - if it is too warm the paste will be more runny.

I decided to reflow and fix after. Maybe this was a mistake - I can see solder bridges and fixed a few but have some stubborn ones. These are DFN with very little exposed pad. Dragging across with a soldering iron tip does not seem to do anything (I used lots of flux but my tip is perhaps not fine enough). I tried lots of flux and copper solder wick but can't seem to get it in contact to wick anything.

I guess the options are

a) buy a finer tip or
b) buy a hot air station and see if I can desolder these chips and then wick some solder before replacing them

?

Also I'm fairly sure there's bridging between the exposed pads and other pins, under the chips. Maybe option b) is best then.

[jpanhalt]

@Post #2 (48X24X48X) "single pass."

I have been reluctant to post this comment/question as it is a little OT.  Now, it appears the TS has a solution to try, so here goes.  My experience to date has been limited just to testing with scrap boards and getting the T-962 oven "tuned."  I have not paid much attention to spreading the paste.  I have also seen work done by 48X24X48X, and the quality is impressive.  Perhaps he will share his experience.

There appear to be several methods to apply solder paste manually, and there is no lack of examples on the web.  I have tried to summarize just a few of the methods:
Multiple passes, minimal paste on stencil:

   (JLCPCB, skip to 3:40)

Blob of paste on stencil, multiple passes:
https://www.sparkfun.com/tutorials/58

Blob of paste on squeegee (like drywall plasterers do):
https://learn.adafruit.com/smt-manufacturing/laser-cut-stencils?view=all   (about 2/3 way down)

There are also research papers focused on automated paste application.  They seem to advocate a SS, flexible (e.g., 0.007") squeegee and one pass, but their conditions are better controlled than one is likely to do manually.  Is there one preferred method for manually applying the paste with a stencil.  Multiple passes seem to provide opportunity to remove excess paste (e.g., from crowning in the apertures) and perhaps squeeze out any paste that has crept under the stencil.  Or, does that cause more problems?

With small pitch devices (0.5 mm), what is advised?   The stencil I am about to use is SS from OSHStencil like the TS is using.  I have both plastic spreaders, including the credit card provided by OSHStencil, and carbon steel semi-flexible blade putty knives (0.020" thick).

John

[Siwastaja]

Aim for a single pass. Any further passes increase the risk of smearing.

Without good "stencil printer", framed or frameless, if you just tape the stencil down, the risk of stencil lifting up slightly between the passes is significant.

With the "tape down" strategy, it's hugely important that the area surrounding your PCB is same thickness, i.e., flat at the same surface height. I use excess boards from the same fabrication pass (i.e., they have closely the same board thickness) around the victim board to be pasted. This way, wherever your squeegee goes outside of your board shape, it doesn't press down and warp the stencil. Thanks to this, you can apply enough paste outside of the active pad area, and apply decent pressure starting early, and make a long single drag, applying significant pressure all the time. I use old credit cards or equivalent for this.

At the end of this single drag, immediately and carefully lift up the stencil in a smooth motion.

If you feel some pad near the board edge has too little paste, don't worry, it's easier to rework that small issue (especially since "too little paste" is seldom a huge problem), than to go for second pass, which smears the paste up, adds too much everywhere, and makes the whole board fail.

Finally, if the result is crappy, just clean the board and the stencil, thoroughly, and redo the paste application. Amount of wasted paste is not big, and this takes maybe 5 minutes extra, much less than trying to fix it afterwards.

[48X24X48X]

@Post #2 (48X24X48X) "single pass."
I have also seen work done by 48X24X48X, and the quality is impressive.  Perhaps he will share his experience.

John

What I wanted to say has already been very well explained by @Siwastaja.
But, from the picture posted by the TS, I think the aperture of the opening is too big for center pad. It is should been reduced and segmented into several smaller opening.

On top of that, I totally don't agree with the above 2 videos' approach. I think if you have small amount of paste, if you can't avoid multiple pass, it's best to segment the application and never ever try to push or shove in the paste like what they were doing, this is where the over application and smearing comes from. The first video has smaller pitch components and it's very clear there's paste smearing all over the place. The 2nd video is more forgiving as the components is not that small and also only a few.

I have a video on this paste application with 0.4 mm and 0.5 mm pitch components but it's buried within one of my HWGC P&P machine video. Try to look up on that.

And like what @Siwastaja said, if the result is bad, start all over again. Don't waste your time and parts.

[jpanhalt]

Very helpful.  Thank you both.

I have been hand soldering SMD for more than 20 years, but as I approach 80, my hands are not as steady as they once were -- even for placing components.  I don't do a lot of boards.  What I found helpful was to tin one pad, place the part.  That allows adjusting placement without a mess.
 Add heat with my iron.  That stuck it well enough to solder the other pads, then resolder the pad that was just tinned. (For some components, I use a microdot of CA adhesive.  Some people have called that method crazy.)

I was wondering whether an extension of the tinning method, particularly for leadless components, would be to add solder paste with stencil, then reflow with no components.  Clean up any bridges or areas with too much or too little solder such as the TS showed.  Then clean the board, add a very thin layer of fresh flux, add components, and reflow again.  Tedious and slow, but for just a few boards a year, do you think that might work and avoid the rework the TS is now facing?

[davegravy]

I was wondering whether an extension of the tinning method, particularly for leadless components, would be to add solder paste with stencil, then reflow with no components.  Clean up any bridges or areas with too much or too little solder such as the TS showed.  Then clean the board, add a very thin layer of fresh flux, add components, and reflow again.  Tedious and slow, but for just a few boards a year, do you think that might work and avoid the rework the TS is now facing?

This sounds like a great approach, I'm curious to know any potential issues with it. I think I'll essentially be doing this process anyways with my DFN parts that I now need to rework,  just without paste being applied via stencil.

Having a domed (tinned) surface rather than flat to place parts on might make positioning a little more finicky? Reflow seems to be pretty forgiving at pulling parts into position, so maybe not a big concern. I'd rather spend a little extra time nudging parts than destroying them while trying to fix solder bridges.

I suppose having no bridges with the components absent isn't a guarantee you'll continue to have no bridges after adding components since they'll displace some solder. But it should be easier to see where solder needs to be added /removed compared with paste that hasn't been reflowed.

[jpanhalt]

Yep, balancing on a domed surface would be a potential problem.  But, I would use my trusty Kester 44 paste flux.  It really doesn't take much.  You can apply it, then wipe it with a towel.  That leaves enough if everything is clean.

With the single tinned pad approach, the whole pad is not tinned.  Just a tiny spot, and the parts align well.  That might be harder to control with stencil-applied paste.

My current project is the most complex I have done in several years.  I have OSHPark boards (3), stencil, and only enough IC's for two boards.  So, if I screw up one, then my back is against a wall.  Otherwise, I would have jumped into it much earlier.

[kylehunter]

Just as an FYI guys, solder paste sold in syringes (i'm not talking about the massive proflow cartridges) are not made for stencil application really. They have a much lower viscosity. So that's likely the main problem. Use a jar of paste, and this issue will be reduced.

[davegravy]

Just as an FYI guys, solder paste sold in syringes (i'm not talking about the massive proflow cartridges) are not made for stencil application really. They have a much lower viscosity. So that's likely the main problem. Use a jar of paste, and this issue will be reduced.

That's good to know. The jars I've seen have been way more product than I'll ever use in a several dozen lifetimes at my current rate of board production. If anyone's aware of a small volume of paste that's good for stenciling, please do share.

As a (crude) workaround could one not leave the syringe paste out to dry a bit before applying to the stencil? After applying through the stencil at night I didn't finish populating my board until the next morning, and the paste seemed much more viscous at that point.

[jpanhalt]

I am not sure that statement applies to all paste.  CMLSupply.com states explicitly that it simply repackages Kester paster,

CML Supply purchases Kester EP256 solder paste in bulk quantities from an authorized distributor and repackages this solder paste in smaller quantities for convenience.

Kester doesn't mention any change in viscosity either.  I would check with what you intend to use to confirm whether the paste in a syringe is "thinner" than that in a tub.

[dkonigs]

I've actually used ChipQuik SMD291AX out of the syringe for the majority of my own board assembly these past few years. I know that stuff is supposed to be kept in the refrigerator, and then taken out a few hours before use.  This is annoying for small-scale stuff, which is why I've switched to TS391AX more recently. (And am likely to finally try GC10 on my next round.)

My first major recommendation is to modify your PCB's paste layer, and hence the stencil design. Those large ground/thermal pads under the ICs are supposed to be pasted as a sort-of hatch-grid, so there isn't such a huge gob of paste applied. I'd also to a general small aperture reduction in the paste later, to simply have less paste on the other pads (I use a "-5%" setting for "Solder paste relative clearance" in KiCad's board settings.)

I know if you order a stencil from JLCPCB, they automatically "reshape" the 0805 paste footprints as well, but OSH Stencils obviously doesn't do this:
https://support.jlcpcb.com/article/77-solder-beading-treatment
(FWIW, I've seen the "solder beading" problem that article illustrates quite often.)

I've attached a picture that hopefully shows the hatch-grid under a QFN part. The paste still a bit runny, like what you had, but the result reflowed just fine.

[Brage]

If you feel some pad near the board edge has too little paste, don't worry, it's easier to rework that small issue (especially since "too little paste" is seldom a huge problem), than to go for second pass, which smears the paste up, adds too much everywhere, and makes the whole board fail.

I'm not quite sure I agree. For ICs without a power pad, I find having too little paste to be more annoying than too much. During visual inspection after reflowing, I find it easier to spot bridges than unsoldered/loose pins. Failuremodes are also something to consider; bridged pins will be a continuous fault while a loose pin on top of a pad might go through initial testing just fine. Then after you mount it in a case or a product the board might be subjected to a different flex than it was during testing. Or the product is subjected to vibrations, causing intermittent faults.
We are talking about ICs with a power pad, in this case however, and so removal after the fact is quite hard... I'd recommend smaller stencil pads for the power pads, like others here have suggested, but working with what OP got it might be the smart move to run the solder paste applications lean like you suggested.

The goal is of course to get as close as possible with the first application, but I have no qualms about running a second or partial pass on a board just to make sure it is all covered.

[Siwastaja]

Well yeah, if you have just a bit too much, but in consistent manner, it can be worked with.

My reply was to avoid multiple passes with the squeegee, as that results in smearing; possibly a lot too much solder, and uneven amounts, smeared between pads, instead of on pads.

[davegravy]

Very helpful.  Thank you both.

I have been hand soldering SMD for more than 20 years, but as I approach 80, my hands are not as steady as they once were -- even for placing components.  I don't do a lot of boards.  What I found helpful was to tin one pad, place the part.  That allows adjusting placement without a mess.
 Add heat with my iron.  That stuck it well enough to solder the other pads, then resolder the pad that was just tinned. (For some components, I use a microdot of CA adhesive.  Some people have called that method crazy.)

I was wondering whether an extension of the tinning method, particularly for leadless components, would be to add solder paste with stencil, then reflow with no components.  Clean up any bridges or areas with too much or too little solder such as the TS showed.  Then clean the board, add a very thin layer of fresh flux, add components, and reflow again.  Tedious and slow, but for just a few boards a year, do you think that might work and avoid the rework the TS is now facing?

I tried this approach the other day out of curiosity but sadly can't recommend it.

I did a single pass with the squeegie and after lifting the stencil the smearing was less than my last attempt but enough I was still apprehensive. So I decided to reflow with no components. After doing this there were no bridges and everything looked just fine. So I added components and re-reflowed but then had solder bridges all over. It turns out that ICs displace solder - I should have seen that coming.

So began the cycle of removing components, wicking solder, cleaning the board, replacing components, and reflowing. One DFN I think took about 20 or more cycles before I finally cleared all the bridges (mainly they are hidden ie exist between the exposed pad and pins where I rely on continuity test to detect them). I was amazed it functioned after that many cycles.

[jpanhalt]

I actually stopped practicing and did my first stenciled PCB Saturday.  Kester EP256 paste and T-962 oven.  Overall, I am pretty happy with the results but was so nervous, I banged the board lifting it from the table, and some of parts moved (attached).  The last three IC's added moved enough that I needed to unsolder and reseat them. I attribute that to taking more than an hour after applying the paste to getting it in the oven, and the paste was decidedly dryer at that point.  Note the singed Molex FFC connector (0.5 mm).  Others have noted Molex seems less heat stable than other brands.  Won't know for sure until I add a cable.

Right now, I am experimenting with adding dots of paste so those components (SOT23-6 packages) can be reflowed with hot air.  I can confirm that reflowing previously reflowed solder did not work well.

PS: I have got to get a better camera for macro.  Any advice that won't break the bank (<$1000)?

[tooki]

In my (admittedly not that large) experience, hand-applying paste with a stencil is much more likely to end up in excess solder paste than if it were being done in a paste printer. No surprises there. So here’s factors to consider, and how to mitigate them:

As others have said, avoid multiple passes. What happens, especially if the stencil is not 100% flat, is that paste gets squeezed under the stencil. Then on subsequent passes, the solder balls on the underside raise it up a tiny bit more, so even more paste gets squeezed. And the pressure of subsequent passes spread the paste.

What’s even easier to forget: any subsequent first passes will also be affected. Any time you have smearing, clean the stencil perfectly before using it again. And be sure to do this on a perfectly flat surface to make sure you don’t bend the stencil at all, as even the tiniest dent will cause smearing.

Make sure one edge of the stencil is taped down such that it acts like a hinge so you can lift the stencil up cleanly. Once you’ve lifted the stencil off a print, DO NOT lower it back down! This easily causes smearing.

Stencil thickness is another factor. For a given aperture, a thicker stencil will obviously deposit more paste.

Finally, aperture size: the openings in the stencil are NOT supposed to be the same size as the pads!! Typical PCB manufacturer recommendations are to reduce the apertures about 15% (by area). For hand stenciling, you might even want to reduce it a bit more.

On the most recent PCB I did, in Altium, I set to -15% as the global paste mask reduction setting, and for large ones like thermal pads, around -20%. Then I went through every damned footprint and reviewed the paste mask reduction for all the pads, because it turns out that many footprints are set up to override the global setting and use 0% reduction instead. 

The result was hands-down the best results I’ve had with a stencil. Not a single short, not a single incomplete joint. On the smallest components (0603) it could probably have even gone with -20%.

[tooki]

Just as an FYI guys, solder paste sold in syringes (i'm not talking about the massive proflow cartridges) are not made for stencil application really. They have a much lower viscosity. So that's likely the main problem. Use a jar of paste, and this issue will be reduced.

I am very, very skeptical of this claim. Solder paste is almost entirely made for stenciling, other than a few specialized for application with a dispensing needle. I don’t think any of them are designed for hand use, whether stenciling or dispensing.

What I think is much more likely is that:
a) the paste in jars dries out faster
b) the long shape of syringes warm up faster than the compact shape of jars
c) dispensing from a syringe causes the paste to temporarily go thin. Solder paste is thixotropic, which is why you’re supposed to stir it well in the jar before use.

All of these things likely make it seem as though the syringe paste is thinner, when in fact it is exactly the same.

[48X24X48X]

All of these things likely make it seem as though the syringe paste is thinner, when in fact it is exactly the same.

They are definitely the same thing except that it's best to take it out from the syringe and mix it well before using which most people don't do.
It has been ages since I use paste in syringe. In my experience, the more paste you have, the slower it dry off. I practically only buy paste in 250 & 500 gm jar.