Okay, we think it’s questionable when people say they have no problem soldering QFN packages, but BGA? Granted this chip has far fewer balls on it than many, but it’s still quite impressive that [Xevel] was able to solder this BGA breakout by hand.
The chip you see above is a TMP006 infrared temperature sensor from TI. [Xevel] picked up the part but didn’t want to break the bank when prototyping by buying a proper PCB to host it. There are only eight conductors on it, arranged in a grid with 0.5mm pitch. That didn’t seem to scare him off, as the video after the break shows him connecting each to a conductor on a hunk of stripboard.
[Xevel] mentions that this is a dead-bug style project. Usually you glue the part upside down when using that technique, but it needs line of sight to get an accurate temperature reading so he first cut a hole in the substrate. We’d bet he’s using wire-wrapping wire to make the connections. It’s a very fine solid core wire which is perfect for this kind of work.
46 thoughts on “Hand Soldering BGA Parts Should Be A Circus Act”
“Unfortunately, this EMI-music-content is not available in Germany because GEMA has not granted the respective music publishing rights.” :-(
Still wondering why Youtube/Google don’t just strip that music for those places where we have to suffer the blindness of such fine ass. and let people watch the video.
Better still: give a play-by-play!
What not to do: overdub a song you like but that (1) other people probably won’t like and (2) will make YouTube prevent embedding.
Really, I want more play-by-play walk-throughs. Like a cooking show. A cooking show where they cook and just play bad music loud would be a terrible cooking show.
Otherwise… wow, great soldering skills.
Just use some open licensed track. The music industry is digging its own grave with this because people will become much more exposed to good music coming from outside the industry.
I’ve soldered components onto micro 1/64 scale rc pcb’s, which are about as small as this or sometimes smaller(ir havoc heli boards). nice work though.
That’s impressive! Does he also program with a needle and a steady hand?!
Pshhh! Every one knows real programmers use butterflies!
Nobody ever suspects the butterfly!
He makes that look easy
WTF the video is not available in Germany :/
Ha cr*p, that’s probably due to the music playing in the background :( Sorry about that :s
youtube copywrong detecor hears mono only. make the music mono, then stereo again by cloning one channel and then inverse one channel – if the copywrong detector listens to the music it hears nothing simply because the mixed signal (right+left)/2 is zero
Indeed, it seems to work with just inverting one of the channels, no need to down mix to mono first. That depends on the type of track played, I imagine there are tracks that would still get picked up and thus require the mono treatment.
Fun with YouTube’s Audio Content ID System
by Scott Smitelli
I would have soldered the wires to the board first, then get them all in place, kinda spring loaded pressure on the chip. It’s be hard and tedious, but it would make the final soldering to the chip easier. Just place the chip, and heat it up and they’re all soldered at once.
Pretty good work. I’ve done some tricky stuff, like mounting a chip, on resistor packs that were stood up on end.
SMD is not hard. Pays to have good eyes though.
That honestly sounds way way way harder to solder. You’d be trying to maneuver the tiny chip, instead of a long wire, and trying to line up all 8 connections at once, instead of one at a time.
I think this guy’s method is almost perfect, I would just put a little solder on the iron so it has more thermal contact with the wire to heat it up.
I soldered the same chip. I made a PCB with a presensitized board. I needed two tries before getting it correctly though! Then, I soldered it simply by putting it on my vitroceramic cooktop until the chip moved slightly, meaning that the balls melted.
When I tried it and it worked, I really couldn’t believe it!
I’m most surprised that he’s able to transport enough heat through that tiny wire to make the balls flow.
Also, which aisle has the steady hands? I need to pick up a pair.
They’re on hangover cures and rocket surgery.
I tried my hand at this just the other day with a denser array (and failed horribly). It is surprisingly easy to make the connection to the ball…just heat the wire until the solder melts and grabs on. The trick is only getting one ball at a time and not snapping off wires once you have them set.
I’m going to have to pick up one of these chips and try the drilled out PCB method.
Wire-wrapping wire core is thicker then what is seen in the picture. This is more likely a magnet wire, e.g. from ear buds’ voice coils.
Not sure exactly about the “real” use of this wire, but what I use here is 36 awg Teflon coated multi-stranded wire sold in RC stores to wire extremely lightweight RC planes (motors, servos). It is very flexible , I love it :)
36awg! damn man, that is tiny!
for reference, wire wrap wire is 30awg.
I got a Casio watch with an IR thermal sensor in it a long time ago, accidentally broke the FFC between the sensor and the watch when changing the battery so I set to with a sewing pin attached to a soldering iron for a ‘fine tip’. I had to use individual strands from some multistrand wire.
Still works to this day, but does need a new strap :)
Yes, well done, sir!
It should be a general rule that if you want your skills seen in Germany, don’t use copyrighted music in your video. (Probably a good idea to not use copyrighted music at all given that Youtube may delete your video for it, there’s more than enough suitable music under Creative Commons out there.)
I should have replaced the soundtrack, here it only picked up what I was listening at the time…
Or just don’t include music because seriously, chances are almost everyone is gonna mute it anyway.
Second time I see something like that with a hole in the pc board. What’s the point with punching the pc board? I did something similar latety with a tiny DFN-8 (0,5mm contact spacing) I glued the IC on the board on its back with Loctite fast glue and then soldered the wires. Suspending the chip just put stress on wires.
Here, the chip is a basically an IR temperature sensor: it needs a direct line of sight between its top and the object it measures the temperature of. Hence the hole.
Otherwise I agree, gluing the chip beforehand would have been a good idea.
Ok! I checked what tmp006 is, now I know why you punched a hole… the back is the sensor window.
Nice soldering job Xevel! I have used that sensor before with the dev kit and it performed terribly when compared to a PIR sensor in a TO can like the MLX90614.
So if you get terrible readings it’s probably not your mounting dead-bug style, it’s just a pain in the ass sensor!
At this point, I would take some epoxy and cover the soldered side. Making sure the wires don’t get moved.
that doesn’t look so impressive. i’ve soldered way bigger things myself.
That is crazy small.
One thing that helps when doing this type of soldering job is to use a high-density parallel hard drive cable, UDMA, etc, which tend to be a series of very small single conductor wires individually insulated.
Having a row of tiny wires at a small pitch makes soldering these types of parts slightly easier, depending on the situation.
I’ve done this with many types of surface mount chips, as well as high density flexible PC board connectors when no socket is available on-hand and it’s either not worth it to find/buy one or when under time constraints.
IDE / pata uses relatively large wires compared to this, I think 28awg. Wire wrap wire is 30awg, and I have some scsi cables that I think are 30awg. Xevel said he used 36awg, so a bit smaller.
Plus, IDE / PATA are stranded usually (the scsi ones I’ve got are solid), which wouldn’t work so well for this.
ATA100, which most people refer to as ide, uses 80 conductors of 30awg. You can separate them with a needle (a safety pin worked great). Slightly too big pitch to use on 0.5mm chips if you need more than maybe 3 or 4 conductor together. 0.5mm pitch ribbon tends to come only in FPC.
yep, these ribbon cables have a 0.635 mm pitch, which is close to what is used on these small chips. The hardest part is to remove the plastic insulation evenly on all wires without separating them. How did you achieve that?
While I don’t use ribbon cable for this kind of job in particular, there is a way to strip all cables evenly effortlessly. Just do a gentle cut (profound enough to reach the copper on most cables but ligh enough not to cut any copper strand – takes a little practice) on each side of the robbon cable, 2 or 3 mm from its edge. The plastic insulation of these 2-3mm will then separate by just gently pulling on it with pliers, leaving each individual cable ready for tinning/soldering.
Well, on the ribbon cable I tried it was effortless, results may vary.
Why do people still insist on using AWG? Use mm so the unit is linear and easily comprehended without a reference table.
Also this type of soldering isn’t that hard, there is no center ball so this is basically a fine pitch QFN or LGA which i have done this type of operation on. He actually makes it much harder than it has to be. The easiest way to do this is to fasten the chip temporarily with double sided tape, solder on all wires, then solder them to the board. With a board this small i would also tape the board down to hold it in place.
Most SMD is pretty easy, people are just misinformed and/or don’t use good enough equipment.
I wholeheartedly agree about strange obsolete units.
Not sure your technique would be that easier, but I’ll be sure to try it next time!
And finally, while I haven’t ever soldered 01005 with an iron (or ever), I again agree with you: when I show people how I do it they usually think it’s really hard, and after they try it once or twice (with the appropriate tools, which I did not have at hand for this video :s) they always conclude with something along the lines of “Meh, that was OK after all”. ;)
Please be kind and respectful to help make the comments section excellent. (Comment Policy)