SMD leadless soldering with a Butane Torch

[cde 334]

So those PCBs I got made? I got antsy, and took a hit. The main chip I wanted on is the LM4970 from Texas Instruments/National, which comes in a LLP (Leadless Leadframe Package), no external legs and not even a small bit of it on the edges. I tried this dead bug style before, but the first time was a disaster, and the second time worked, but the wires for the Mid/Red led flaked off (not a bad thing actually) and the i2c lines as well (Definitely not a good thing). I could try again, with some 34awg wires, but no.

 

Don't have a reflow oven, or toaster over I could hack (or the relays or anything), nor a hot air gun, but I did have a micro portable butane soldering iron. Cheap, at most hardware stores. The soldering iron tip can be removed, turning it into a torch, or a hot air blower. I was worried that it might blow the chips away, but I tested it on a led and a resistor, and nothing. It relies on the pressure of the butane to move the hot air, and that's very little. Of course, I didn't use any flux on the led or resistor, so it didn't work as well, but it reflowed the solder enough that it could be seen. Did brown the led some, and started bubbling the pcb edge, but not the face. Turns out I had the torch at 100%, and since there was no flux to help things along, I was just cooking the board.

 

On to the actual chip (or how to solder with a hot air burner for newbies):

1. I used a soldering iron to tin the board and the lm4970
2. Used a wick to take the excess solder off
3. Cleaned the cooked flux off (70% alcohol, water, and a toothbrush)
4. And refluxed the chip and board.
5. Put the torch at 75%, and lowered it near the chip.
6. The torch area is bigger than the chip (5mmx2mm) so the heat went around nicely (Scrapped a little of the ground fill for the center pad, since it was not exposed, something to fix on the next revision).
7. Since I couldn't really see the solder reflow except on one 10mil trace, I waited for the flux to boil off and then some.
8. AND IT WORKED. Chip was straight, nice and firm. Cleaned off the flux again (Not all came off, between the board and the chip, hard to reach, but it is no clean flux).
9. Tested for shorts (I had modified the LM4970 footprint for longer traces, which I am glad, made testing easier. Will make them longer again, as well as the ground center pad for this next time). No shorts between pins.
10. Threw on the led, bypass cap and extended header (and fixed the broken trace), plugged it in, and tested it out (With my trusty i2c explorer).

 

 

Notice, the board isn't browned at all, as long as you don't keep the heat too high. Used normal radioshack 60/40 rosin core solder (not lead free iirc). And the jumper wire to fix the broken ground path. The LM4970's address jumper pad is on the backside, with just enough solder to bridge it to ground (the other pad would bridge it to V+). Lesson Learned as well, once you use Locking header footprints, non-locking is a joke. Trying to keep a header straight and level with a helping hand while soldering it on is a pain. The Adafruit (or Sparkfun) locking header is so much easier.

 

This makes leadless soldering available to anyone with 12 bucks and a hardware store nearby (compared to 15 bucks for the breakout board for this chip) . Sure, it's not efficient for an entire board, considering the burner is about 8mm wide, but for a handful of chips for one off boards, this is simple. Next try will be a wl-csp 3mmx2mm chip, basically bga.

[roadrunner84 466]

Cool, I've seen one video on soldering a QFN package, I'd love to try it myself.

 

ne sily question though, what's a (non-)locking header? I do see a pinheader, but how is it different from other headers?

[cde 334]

Cool, I've seen one video on soldering a QFN package, I'd love to try it myself.

 

ne sily question though, what's a (non-)locking header? I do see a pinheader, but how is it different from other headers?

The Locking headers have every other pin hole offset by a small amount up or down. So it is a bit zig zagged on the board.

 

You could (kinda) see them on the middle five boards I had made (click to enlarge). See how every other pin in the header is sitting on the silkscreen?

 

 

 

Well, it provides just enough of a offset that it holds the header in place with friction. The regular header, like I used on the bottom left board, has all the holes equal, so the header falls out unless you hold it.

 

Here's where I first found out about these, with better pictures: http://www.sparkfun.com/tutorials/114

[username 198]

Haha, thats certainly 1 way to do it! Impressive that you got it to work! 

As for other methods,

Conventional ovens or burners work as well. Might be abit safer than a torch depending on how you do it. In addition, there is always the option of a 30 usd toaster oven

[cde 334]

Haha, thats certainly 1 way to do it! Impressive that you got it to work! 

As for other methods,

Conventional ovens or burners work as well. Might be abit safer than a torch depending on how you do it. In addition, there is always the option of a 30 usd toaster oven ;-)

It's not the torch function I used, but the hot air function. It has an attachment that turns the flame into heat. Same attachment is used to screw on the soldering tip. So no fire ever touched the chip or the board.