How small can you go with DIY reflow soldering?

[newto]

I've got a project that I'd like to use a pretty small part on, 12-WFDFN Exposed Pad 3mm x 2mm, but it's also a very experimental prototype, so I don't want to commit to getting a shop to assemble it.

Is there any chance that I could successfully reflow using the hotplate+sand method , or even with a modified toaster oven? I'd obviously prefer to use the parts I have in mind, but I also don't want to commit a bunch of design time, delivery time, and component costs for something I have no chance of actually soldering...

So how small can you go? How small have you gone with DIY/homemade/custom reflow equipment? Most of the examples I've seen online for DIY reflow are using sample boards with fairly large components

[SiliconWizard]

You probably can. Never done it though. Be prepared for some trouble IMO. For this, you'll need of course a proper stencil.

For the record, the smallest I've done with hand soldering was a UDFN-6 package (1.2x1.0mm, 0.4mm pitch). Yes, hand-soldering. I succeeded, and it was kind of my record here. But it was pure hell. Hot air was a dead cause of course, so I actually used an iron with a very fine tip (JBC C210). A lot of flux. A good camera microscope. And lots of patience.

Maybe it would have been easier with a hotplate, but the stenciling would have been tricky already.

[langwadt]

I've done a small board with an LGA-14L (2.5*3mm) and a 16 pin QFN (3*3mm with a 1.7*1.7mm exposed pad) using a lab hot plate with the magnetic stirrer removed, with no big problems

[newto]

What's that part?  0.5mm or 0.4mm pad pitch?  Already too small for solder mask slivers between the pads, I assume.
If so that probably doesn't make it very different from any other QFN with a similar pad pitch.

It's 0.5mm pitch, so I'm probably overthinking it. I've managed SOT-363 which is 0.65 a few times by hand, and that wasn't too bad, 0.5 with some sort of reflow should be doable

I wouldn't worry about it too much.  Heck do it a couple of times with hot air manually under a microscope / magnifier.
You can always rework it if it doesn't work and you don't damage the board / parts in the process of assembly & test.

Maybe add some current limits or other test points to help assess lack of bridges and opens and correct operation non-destructively.

I like these ideas. The small chips will be samples, so I can get enough parts for 5x boards pretty easily, and I only really need 2-3 to be fully functional for testing. I can hand fix any other misaligned parts, and hot air would be a lot less likely to damage a board than a soldering iron if I need to pull it off a few times (last prototype suffered a few torn traces swapping out op-amps). I think trying both hot plate and hot plate + hot air is probably worth the risk (and opens a lot of opportunities if I get comfortable with it)

For the record, the smallest I've done with hand soldering was a UDFN-6 package (1.2x1.0mm, 0.4mm pitch). Yes, hand-soldering. I succeeded, and it was kind of my record here. But it was pure hell. Hot air was a dead cause of course, so I actually used an iron with a very fine tip (JBC C210). A lot of flux. A good camera microscope. And lots of patience.

  I don't think my hands are steady enough to even consider that...

[ajb]

You really don't need anything fancy, if you've got a hot air setup you've got what you need.  You actually don't even need a stencil or solder paste, you can install leadless or thermal pad parts by using an iron to apply a small pillow of solder to each pad, coating with a liberal amount of GOOD paste flux, and then putting the part on while heating it with hot air.  You may need to gently preheat the board a fair bit before placing the part.  The real trick is to not put too much solder on the pads, especially the center pad, as that can squish out and cause shorts, or just the surface tension can keep the part from seating properly.  It takes a bit of practice to get this right, but the correct amount is generally less than you think you need.  With the right amount of solder, adequate flux, and a carefully heated board, the part should snap into place once the part makes contact with the molten solder.

For cleanup, an iron with a pointy tip (preferably a cartridge type, the irons that take a hollow tip like the FX-888 etc have too much thermal resistance to really make a fine tip work well) can help as it's easier to get into the inside corner where the side of the package meets the board to retouch any joints there if necessary.  This is easier if the part has pads that extend up the sides a bit, although not all do. 

You will want some magnification to inspect your work.  A 10x Jeweler's loupe that can be had for a few bucks is great for post-soldering inspection and will remain useful no matter what fancier equipment you have access to.  I do not recommend trying to solder under it though  .  Some people like optivisors, which are a few bucks more, or for ~$80US you can get an okay USB microscope.  These can be soldered under, although the USB microscopes often have significant latency and/or poor dynamic range plus inherently no depth perception which makes this tricky.  A proper binocular optical setup with a stereo microscope, or a Mantis if you can afford it, is much better since there is no latency and you retain your depth perception.

[mjkuwp]

should be fairly easy.  I do a ton of soldering with a customized toaster oven.  You can do this with an unmodified toaster oven if you modulate the heat yourself.  I don't know how big your board is or what other equipment you have.  Hot air may also work and a hot plate may also do just fine.

I highly recommend you get a metal stencil.  These are so cheap it's silly.  I get mine from oshstencil.  use leaded solder paste and keep everything clean.  The only challenging thing on the jobs I have done is the stenciling.  sometimes I get the pressure and amount right and sometimes I have to wipe the board clean and start over.  if the stenciling is good the process can be very robust.

[Cerebus]

I successfully soldered an MSOP-10 (0.5mm pitch, 0.3mm pins, 0.2mm gap between pins) the other week just slathering the paste on the board with a wooden toothpick and reflowing it with hot air. I got one bridge that quickly cleaned up with a little chem-wik and a not particularly small tip on the Metcal. If that works doing it with a proper stencil ought to be a breeze.

[mawyatt]

Working on a complex project that requires multiple HV channels (~100, later more) created from a few AD5380 14 bit 40 output DACs. These are in 100 pin LQFP packages, and after a few attempts at finding a reasonable resource to do the surface mounting decided to give it a try. The AD5380 are ~$100 each, so not subject to trial and error!!

Anyway, studied a few videos of folks removing a replacing various small lead packages followed the general theme of using a quality paste/gel flux (Mechanic Uv559 Flux Paste) and a small tip iron. Aligned the package, applied the flux paste, carefully realigned under a microscope, then solder anchored a corner pin. Checked alignment and anchored the opposite corner pin. Then carefully went along each pin with the iron tip, very tedious but this worked and all the pins were mounted (checked with meter) and no solder bridges. Checked the circuit boards after attempted cleaning IPA (after a few passes with a toothbrush, still some residue between the pins) and everything worked!!!

BTW what cleaning method/solution is recommended for this type of Flux Paste?

Edit: Forgot to mention used the same technique to mount a couple 16 pin TSSOP packages, for the larger SOIC packages and 0603 passive elements used a toothpick to place a small amount of solder paste and used a heat gun.

Best,

[fourfathom]

I've done 0.5mm-pitch LCCs, running a small line of paste along the pads (toothpick applicator), and reflowed the board in my stock toaster oven using a home-made oven controller.  As long as I don't put down too much paste it works reasonably well.  But I use a stainless stencil when I can, the hand-stuff is for one-off jobs.

[bson]

Hot air to reflow and a 5x microscope to place parts works fine for me.  I don't even bother with stencils and just dot on paste with a toothpick.  That, however, takes some practice to get proficient at.  QFNs are pretty easy.  0201 discretes is harder, not because of pasting or placement per se, but they're more prone to tombstoning and sticking electrostatically.  But very doable.  I limit myself to 0402 though, but don't mind 0201 in a pinch to fit a particular board size.

[Bud]

To deposit soldering paste I use a small syringe and medical stainless steel needles shortened to 5-10mm and finished on a piece of 180 grit sandpaper. Pasting and placing is under a microscope - with a habit the process is not too slow. Fine pitch ICs come out of the reflow oven just fine. For tiny bodge wires to solder  to ICs afterwards it is a JBC soldering iron with a needle conical tip, again under the microscope.

[mawyatt]

What method/solution cleaning do you folks use? I've been using Mechanic Uv559 Flux Paste, and using a toothbrush and IPA doesn't seem to fully clean between the leads on small pitch ICs. Using various types of solder including ChipQuick SMD291AX50T3 which is Sn63/Pb37 No-Clean T3 and Mechanic XGSP30 (183C) solder paste.

May give ultrasonic cleaning a try.

Best and Happy 4th,

[Doctorandus_P]

I would not dare to go smaller then 0201 (Metric).


Those are plenty full insane enough for me.
In case you're wondering about the background. Yes that is a fingertip.

[SiliconWizard]

Hot air to reflow and a 5x microscope to place parts works fine for me.  I don't even bother with stencils and just dot on paste with a toothpick.  That, however, takes some practice to get proficient at.  QFNs are pretty easy.  0201 discretes is harder, not because of pasting or placement per se, but they're more prone to tombstoning and sticking electrostatically.  But very doable.  I limit myself to 0402 though, but don't mind 0201 in a pinch to fit a particular board size.

Well, if you can successfully reflow 0201 parts with hot air, I'd be curious to see that. They are so light and tiny that the smallest flow of air will blow them off, so I'm cuirious to know about your technique.

[ajb]

What method/solution cleaning do you folks use?

No-clean fluxes like that Mechanic brand stuff you mention are often harder to clean than conventional fluxes, so we usually use a combination of a dedicated flux remover like this and a rinse with IPA.  Once the cleaner/IPA have loosened/dissolved the flux it's important to flush it from the board, any dirty cleaning solution left on the board will leave behind the flux residue when it dries.  A squirt bottle with IPA helps to flush between pads and under parts.  You can follow up the IPA rinse with a water rinse and then a second IPA rinse (generally overkill, but helps to get the water off the board and the IPA will dry more quickly).  Or you can just blast off the remaining rinse liquid with air (preferably from a can or an air compressor with a proper filter/oil separator, not from your mouth).

A water flosser might also be worth trying, would be good for cleaning stubborn residue in hard to reach places, especially if there's one that will tolerate running IPA through it.  Looks like replacement wands are available pretty cheaply, I wonder if they'd work with typical utility water pressure, so you wouldn't even need the pump--would be nice to not have to worry about refilling a reservoir all the time, although you might need another rinse afterwards if your water has a lot of minerals or other junk. 

[newto]

Thanks for all the advice! The board is just going to be a cheap 100mmx100mm JLC board, so I'll spend the extra money for a frameless stencil (it's for a work project anyways, and they're amazed I can do a full prototype for a few hundred dollars, vs the literal multiple thousands the old engineers used to spend...). We're a very small shop (I'm literally the only engineer), so being able to do rapid design testing with small parts will be really useful.

I think I'll stay away from any 0201s, but I might try making a super tiny version of my main circuit in a corner of the board using 0402s and see if I can get some tiny package versions/alternatives to stuff I was already looking at.

Really opens up the possibilities, assuming I can actually manage it!

[MarkR42]

I've done 0.5mm-pitch LCCs, running a small line of paste along the pads (toothpick applicator), and reflowed the board in my stock toaster oven using a home-made oven controller.  As long as I don't put down too much paste it works reasonably well.  But I use a stainless stencil when I can, the hand-stuff is for one-off jobs.

I second that, I have done 0.5mm pitch parts using exactly that process.

If you want to make a one-off prototype just for a proof of concept (rather than actually prototyping the product where you will probably want several boards with lots of parts) then it's a fine way to go. Fiddly and somewhat time-consuming, but allows hand-assembly.

I've done breakout boards for DFN parts before, or used off the shelf breakout boards for qfn parts, then connect with breadboard wires or whatever.