Soldering a QFN (Quad Flat No-Lead) Package by Hand
an upside-down QFN chip showing alignment markers on its side
intro: QFNs are difficult because all the connections are on the bottom of the chip. Some versions have small extensions
of these connections that wrap around the bottom corner and come up the edge a little. The perimeter connections
(not the heat sink in the middle) on these can actually be soldered with a regular iron by applying lots of flux and
touching each side and pad with a tinned iron. However, the version demonstrated in this guide only has small markers
on the side, so hot air must be used to melt the bottom connections. Note that the best way to solder this chip is to use
solder paste, a stencil and hot air (or a toaster oven or skillet), but we'll be demonstrating a technique that doesn't require
either paste or a stencil.
We're going to show a hot air station ($250-$1000+) and an embossing tool ($25 arts and crafts tool for
making raised ink decorations), as well as a hot-air pre-heater. You may be able to get by without a pre-heater,
but it makes the job faster and less risky to the components and board. If you were doing a small one-sided board,
you could also use a coffee cup heater, or, with a larger chip, a skillet set to about 100 degrees C.
A 24 pin QFN from a
SI Laboratories radio dev kit will be removed and then replaced.
Original QFN Si4701 on a SiLabs Radio Dev Kit
First, the original chip is removed by pre-heating the board and applying hot air using the same methods that will
be detailed below. The pads on the removed chip and board are then cleaned by adding flux and using solder wick.
Using solder wick to clean the pads
The PCB with QFN removed
The cleaned QFN pads--note the vias in the heat sink pad
The heat sink has several vias (small holes meant to connect different layers on the board) that are used here to help
draw heat away from the component and into the ground planes. While these help to keep the chip cool during use,
they make it difficult to solder because heat from a hot air tool quickly dissipates into the board. A pre-heater
is especially helpful if you're going to solder the middle heat sink. However, many times it's not necessary
to solder this middle heat sink, and this is what makes soldering this chip especially challenging. So if you don't need it,
don't add solder to it. And, if it's your first time soldering a QFN, it's probably best to skip the middle heat sink.
Adding flux to the bottom of a clean QFN chip
Before applying solder, clean off any old flux residues with alcohol and a brush, and then apply fresh flux.
A small pillow of solder on the middle heat sink of a QFN
Apply solder to middle heat sink: To do this, tap a very lightly tinned tip against the fluxed pad until a small
amount of solder wicks out onto the pad. The pillow of solder that forms should be very small, no more than 10 thousandths of an inch
(1/3 of a 1/32") tall. If you have calipers, you can check the pillow height by first measuring the original thickness
and then comparing it the soldered thickness. If there's too much solder on this pad, it will short to the outer connections
during reflow. It's better to have too little. Remove excess by applying flux and touching the solder pillow with a clean tip.
Tinning the outer connections of a QFN
now tin the outer connections. Again, clean off old flux residue, add fresh flux, and touch a lightly tinned
iron tip to the perimeter connections. Small beads of solder should collect on each pad. A microscope or loupe can
be used to make sure each pad received solder.
QFN with center heat sink and outer connections tinned.
Perimeter connections on the PCB tinned.
tin the outer connections on the PCB: Do the same process for the PCB's outer connections. Leave the middle heat sink
free of solder.
Pre-heating the board
Pre-heat the board: Turn the pre-heater on and wait a few minutes. If you have a thermocouple or other temperature display,
you want the board to be about 212-250 degrees F before continuing.
Using hot air from an embossing tool or hot air station
apply hot air: Holding the chip with tweezers, first apply hot air from a few inches away, and then move in to about
3/4" away from the chip. Move the hot air in small circles. When the solder reflows, you should feel the chip drop
into place. Let go with the tweezers. For this size of chip, surface tension will actually snap it perfectly into place unless there are shorts in
the solder. Test to make sure it's located by gently nudging the chip with tweezers--it should spring back into place.
Check to make sure alignment markers on the side are above the pads
Check to make sure the chip is lined up correctly: A loupe or microscope can be used to make sure the markers on the side
are directly over their corresponding pads. You should also be able to barely see connections under the chip.
Still works!
test your circuit: You can't hear it, but our dev kit works once again with the re-soldered QFN chip.
Most boards actually have the pads extending beyond the chip as standard practice. See this document for recommendations on the pad footprint.
Whether those pads help you depends on the design of the chip. The packages described in the linked document have pads that wrap around from the bottom of the chip up the side a little. You can solder to these sides by applying ample flux and playing a ball of molten solder against them. Unfortunately, the chip shown in this guide only has markers on the side, and solder wouldn’t stick to them, or wick underneath. I’d love to hear if others manage to solder a chip like this without hot air.
One other thing: if you design your own board, you can put a large, plated hole in the middle of the heat sink and then flood it with solder from the back after finishing the perimeter connections. This is how the schmartboards work.
I am soldering on my QFN right now onto a test board . . . totally messed up my first one so I have a question that might help me.
What recommended air flow rate and heat rate would you recommend me to use? I have a few more test boards to mess up =P
Also for other people, be careful with the solder wick! It ripped up two of my QFN trace pads somehow . . . And burnt PCB leaks black ooz, never knew that before, lol . . .
I think wick is frowned on in industry for exactly that reason… cotton swabs are an alternative, but it can a little tricky to avoid leaving lint. With either of those options, I’ve found that loads of flux solves a lot of problems. It distributes heat better so you’re less likely to have a piece of the wick freeze to a pad while you’re pulling on the rest. A bunch of solder can also help, even though you’re actually trying to remove solder…
About temp and flow rates…
I doubt there’s an ideal flow and temperature to use since so much depends on how far away you hold the hot air tool. High flow rates can spray small components and paste, but you also don’t want such a low rate that you have to max out the temperature. The max temperature I’d use would be around 800 degree F for lead-free stuff, probably lower for lead-based solder. If you preheat the board first, which helps to reduce the amount of heat that escapes away, you can get well below 600 degrees F.
I’m assuming you’re using lead-free… I’d try out something around 750 degree F and a fairly low flow rate and see where that gets you. What kind of hot air station are you using? If you’re heating too much more than 30 seconds close up you probably need more velocity or temp.
I did two more attempts yesterday (one on a damaged chip just for practice). In your tutorial you said I will feel the QFN go down and settle, but I’m not feeling this. But after the solder is melted, the QFN no longer blows away with the hot air so I assumed that was good enough.
Unfortunately, the chip didn’t work when powered up so I’m guessing it wasn’t soldered properly.
I sadly got incredibly shaky hands, and no preheater, making this a bit difficult, but I’ll remove the chip and try again today. Ill do a higher air/temp rate. I’m using a decent $250 hot air station with the same small tube attachment as you, and lead free solder.
maybe try a slower air rate, but higher temperature? Lead-free’s going to be a bit harder to do, although knowing that probably doesn’t help.
Are you applying a good amount of liquid flux before starting? Is there a big heat sink in the middle that’s sucking away lots of heat? I wouldn’t put any solder on a middle heat sink for a first run at things. When you think it’s molten, try every so lightly tapping the side to see if if springs back into place. If it’s centered, it should snap back into position.
Do you have a scope or loupe to check the sides to see if it looks like it’s positioned correctly?
I usually hold the chip with tweezers until it reflows, I don’t just lay there if that wasn’t clear.
good luck!
if worst comes to worst my next best advice is to get a schmartboard or to use solder paste and a skillet / toaster oven.
ok I finally got it to work . . . spent another half an hour, board started turning black, took like 20 attempts . . . but somehow my QFN never fried through all this abuse! =)
I warmed the board from a distance with my hot air station for a bit. Subjectively, I think low air flow and high heat is best. Its also better to blow the hot air from all angles by moving the gun around . . . I didn’t do this before because my magnifier lamp got in the way . . . so I just did it without the magnification this time.
I didn’t use liquid flux, I used that tube kind that dries instantly. I applied lots. It is a no-clean type. Would liquid flux work better?
As for scopes, I have access to any scope you can imagine . . . I work in a biochem lab . . . I find between 5x and 10x optimal. Underside microscope lighting through the PCB I found best to check for shorts, while using above lighting best for making sure all pins are soldered.
nice. I don’t know if I’ve ever use the tube kind of flux… what’s the brand?
So you can actually see under the chip with those scopes? That’d be an awesome picture if so…
The last couple pages of this guide talk about soldering QFN’s by hand, and I believe they recommend doing some pre-heating like you were talking about before moving in closer, especially with no bottom side pre-heater and lead-free.
I usually only use the mic to check my work, too, not during the actually soldering.
I recently mounted a pair of QFN-8 chips using just a soldering iron. Started off very similarly to your suggested technique: put a tiny dab of solder on each of the pads on the chip and the pads on the board (it’s best to keep the dabs as equal as possible). Then I attempted to line up the chip on the board and use the soldering iron to remelt the solder on a pad which in turn remelts the solder on the chip and the two solder dabs join. The unmelted dabs of solder make the chip want to naturally line up in between the pads on the board, which makes the initial alignment a bit tricky and took a couple tries to get correct. Once the alignment is correct, I repeated the process on the other pads. In a few cases, the solder on the pad on the board didn’t quite touch the solder on the chip’s pad, so I had to add a little after the fact. In the end, the chip sits a little higher on the board than it would have otherwise, but this just makes it easier to verify the connections with a Loupe. This was on a board I designed, so I made sure the chip wouldn’t entirely cover the pads on the board; I don’t think this method would work without part of the pad exposed.
Thanks for posting the tutorials and videos; it’s given me the confidence to successfully mount a few TQFPs that otherwise I would never have attempted otherwise.
feedback and corrections are appreciated
Awesome Tutorial, I did not know this was possible.
If you’re designing the board yourself, is it possible to solder these with a soldering iron if you extend the traces beyond the chip?
Most boards actually have the pads extending beyond the chip as standard practice. See this document for recommendations on the pad footprint.
Whether those pads help you depends on the design of the chip. The packages described in the linked document have pads that wrap around from the bottom of the chip up the side a little. You can solder to these sides by applying ample flux and playing a ball of molten solder against them. Unfortunately, the chip shown in this guide only has markers on the side, and solder wouldn’t stick to them, or wick underneath. I’d love to hear if others manage to solder a chip like this without hot air.
One other thing: if you design your own board, you can put a large, plated hole in the middle of the heat sink and then flood it with solder from the back after finishing the perimeter connections. This is how the schmartboards work.
I am soldering on my QFN right now onto a test board . . . totally messed up my first one so I have a question that might help me.
What recommended air flow rate and heat rate would you recommend me to use? I have a few more test boards to mess up =P
Also for other people, be careful with the solder wick! It ripped up two of my QFN trace pads somehow . . . And burnt PCB leaks black ooz, never knew that before, lol . . .
(admin from http://www.societyofrobots.com)
I think wick is frowned on in industry for exactly that reason… cotton swabs are an alternative, but it can a little tricky to avoid leaving lint. With either of those options, I’ve found that loads of flux solves a lot of problems. It distributes heat better so you’re less likely to have a piece of the wick freeze to a pad while you’re pulling on the rest. A bunch of solder can also help, even though you’re actually trying to remove solder…
About temp and flow rates…
I doubt there’s an ideal flow and temperature to use since so much depends on how far away you hold the hot air tool. High flow rates can spray small components and paste, but you also don’t want such a low rate that you have to max out the temperature. The max temperature I’d use would be around 800 degree F for lead-free stuff, probably lower for lead-based solder. If you preheat the board first, which helps to reduce the amount of heat that escapes away, you can get well below 600 degrees F.
I’m assuming you’re using lead-free… I’d try out something around 750 degree F and a fairly low flow rate and see where that gets you. What kind of hot air station are you using? If you’re heating too much more than 30 seconds close up you probably need more velocity or temp.
I did two more attempts yesterday (one on a damaged chip just for practice). In your tutorial you said I will feel the QFN go down and settle, but I’m not feeling this. But after the solder is melted, the QFN no longer blows away with the hot air so I assumed that was good enough.
Unfortunately, the chip didn’t work when powered up so I’m guessing it wasn’t soldered properly.
I sadly got incredibly shaky hands, and no preheater, making this a bit difficult, but I’ll remove the chip and try again today. Ill do a higher air/temp rate. I’m using a decent $250 hot air station with the same small tube attachment as you, and lead free solder.
(admin from SoR)
maybe try a slower air rate, but higher temperature? Lead-free’s going to be a bit harder to do, although knowing that probably doesn’t help.
Are you applying a good amount of liquid flux before starting? Is there a big heat sink in the middle that’s sucking away lots of heat? I wouldn’t put any solder on a middle heat sink for a first run at things. When you think it’s molten, try every so lightly tapping the side to see if if springs back into place. If it’s centered, it should snap back into position.
Do you have a scope or loupe to check the sides to see if it looks like it’s positioned correctly?
I usually hold the chip with tweezers until it reflows, I don’t just lay there if that wasn’t clear.
good luck!
if worst comes to worst my next best advice is to get a schmartboard or to use solder paste and a skillet / toaster oven.
ok I finally got it to work . . . spent another half an hour, board started turning black, took like 20 attempts . . . but somehow my QFN never fried through all this abuse! =)
I warmed the board from a distance with my hot air station for a bit. Subjectively, I think low air flow and high heat is best. Its also better to blow the hot air from all angles by moving the gun around . . . I didn’t do this before because my magnifier lamp got in the way . . . so I just did it without the magnification this time.
I didn’t use liquid flux, I used that tube kind that dries instantly. I applied lots. It is a no-clean type. Would liquid flux work better?
As for scopes, I have access to any scope you can imagine . . . I work in a biochem lab . . . I find between 5x and 10x optimal. Underside microscope lighting through the PCB I found best to check for shorts, while using above lighting best for making sure all pins are soldered.
(admin from SoR)
nice. I don’t know if I’ve ever use the tube kind of flux… what’s the brand?
So you can actually see under the chip with those scopes? That’d be an awesome picture if so…
The last couple pages of this guide talk about soldering QFN’s by hand, and I believe they recommend doing some pre-heating like you were talking about before moving in closer, especially with no bottom side pre-heater and lead-free.
I usually only use the mic to check my work, too, not during the actually soldering.
I recently mounted a pair of QFN-8 chips using just a soldering iron. Started off very similarly to your suggested technique: put a tiny dab of solder on each of the pads on the chip and the pads on the board (it’s best to keep the dabs as equal as possible). Then I attempted to line up the chip on the board and use the soldering iron to remelt the solder on a pad which in turn remelts the solder on the chip and the two solder dabs join. The unmelted dabs of solder make the chip want to naturally line up in between the pads on the board, which makes the initial alignment a bit tricky and took a couple tries to get correct. Once the alignment is correct, I repeated the process on the other pads. In a few cases, the solder on the pad on the board didn’t quite touch the solder on the chip’s pad, so I had to add a little after the fact. In the end, the chip sits a little higher on the board than it would have otherwise, but this just makes it easier to verify the connections with a Loupe. This was on a board I designed, so I made sure the chip wouldn’t entirely cover the pads on the board; I don’t think this method would work without part of the pad exposed.
Thanks for posting the tutorials and videos; it’s given me the confidence to successfully mount a few TQFPs that otherwise I would never have attempted otherwise.
Erik
(complete instructions)