Has anyone got it right yet to solder BGA devices at home, and would share this information with us?
I’ve successfully used a Lattice CPLD in a home project. I only needed to connect 20 or so signals so gluing the chip upside down and soldering link wires from the balls to the PCB worked a treat. That device had 1mm ball pitch; I also tried a 0.7mm pitch device but gave up.
Unfortunately, your chances of mounting a BGA the correct way are very slim. We have a manual BGA pick and place machine at work with an optical alignment system (a video camera mounted on a microscope) and that gives a pretty poor success rate.
Good luck!
David.
Unfortunately, your chances of mounting a BGA the correct way are very slim. We have a manual BGA pick and place machine at work with an optical alignment system (a video camera mounted on a microscope) and that gives a pretty poor success rate.
Good luck!
David.
Attachments
Thanks, but no thanks. The device in question would be either 256FBGA or a QFN108 and it would have to go on a pcb.
I guess my best bet would be to either 'stay away from BGA' (hey that has a nice ring to it!) or cultivate a contact at some place like MTech or Walters with the necessary facilities.
Stinking IC manufacturers: 1
Determined hobbyists:0
Can we allow this to continue?
Cheers
John Hope
I guess my best bet would be to either 'stay away from BGA' (hey that has a nice ring to it!) or cultivate a contact at some place like MTech or Walters with the necessary facilities.
Stinking IC manufacturers: 1
Determined hobbyists:0
Can we allow this to continue?
Cheers
John Hope
Its not the IC manufacturers to blame, its the customers. With my work hat on I want the smallest device package I can get and BGAs fill that need. So I buy BGA package devices over leaded ones.
But BGAs are not the only non-DIY friendly package. Chipscale packages, 0.5 or smaller pitch leaded packages etc.. are all beyond 99.9% of DIYers. Even with a nice metcal and a mantis they are a nightmare to solder.
If you really have to use a BGA I'd start getting friendly with a PCB assembler.
But BGAs are not the only non-DIY friendly package. Chipscale packages, 0.5 or smaller pitch leaded packages etc.. are all beyond 99.9% of DIYers. Even with a nice metcal and a mantis they are a nightmare to solder.
If you really have to use a BGA I'd start getting friendly with a PCB assembler.
Thanks Dave. Very useful info. I knew somebody must have done it.
One of the articles mentions a couple of 'opens and bridges' on fine pitch devices that had to be cleared manually. I guess with BGA FPGA's s there is no recourse to fixups, and it would be a good idea to have as much 'redundant I/O breakouts' as possible so that opens could be fixed by using another pin on the FPGA.
With shorts, I guess one can similarly reroute the I/O and tristate the shorted pins. Assuming the short is not between power rails, or JTAG pins or anywhere else really tragic.
This all sounds like a good challenge!
Cheers
John
One of the articles mentions a couple of 'opens and bridges' on fine pitch devices that had to be cleared manually. I guess with BGA FPGA's s there is no recourse to fixups, and it would be a good idea to have as much 'redundant I/O breakouts' as possible so that opens could be fixed by using another pin on the FPGA.
With shorts, I guess one can similarly reroute the I/O and tristate the shorted pins. Assuming the short is not between power rails, or JTAG pins or anywhere else really tragic.
This all sounds like a good challenge!
Cheers
John
Im making my living by placing BGA's, and other components, so it's a simple thing to do for me 😎
Perhaps when faced with a mid life crisis instead of a sports car with a six figure price tag, one could buy an Ersa instead. It's a lot cheaper.
rfbrw: Perhaps I'm missing something here, but I don't see the connection between sports cars, midlife crisis and diy'ers figuring how to get BGA devices soldered. Perhaps you have the wrong thread?
CJ900RR: Do you offer a service to guys wanting one or two BGA's soldered? Tell us more.
Cheers
John
CJ900RR: Do you offer a service to guys wanting one or two BGA's soldered? Tell us more.
Cheers
John
I have learned to solder the small pitch stuff... against my wishes. And yes, Metcal is your friend.
I don't wish to discourage anyone... but BGAs?
🙁
I don't wish to discourage anyone... but BGAs?
🙁
anyone tried the hot air desoldering tool?
honestly, I haven't tried yet since I didn't have a need to work with those...for now.
honestly, I haven't tried yet since I didn't have a need to work with those...for now.
the only experience that I had so far is the tripath chipset for amp2 in 41hz.
had "luckily" been able to solder the heatslug under the driver IC. not sure if I'm still going to be that lucky when I get amp8. 🙁
had "luckily" been able to solder the heatslug under the driver IC. not sure if I'm still going to be that lucky when I get amp8. 🙁
John Hope said:One of the articles mentions a couple of 'opens and bridges' on fine pitch devices that had to be cleared manually. I guess with BGA FPGA's s there is no recourse to fixups, and it would be a good idea to have as much 'redundant I/O breakouts' as possible so that opens could be fixed by using another pin on the FPGA.
With shorts, I guess one can similarly reroute the I/O and tristate the shorted pins. Assuming the short is not between power rails, or JTAG pins or anywhere else really tragic.
Back when I worked in the industry which was about 5 years ago, we scrapped the boards if a BGA got screwed up in the assembly process. There were some exceptions, like the $10,000 boards we were making for Sun Microsystems where we'd try to rework the faulty BGAs, but let's just say the PACE rework stations we were using was worth about as much as the aforementioned boards.
In any case, reworking a BGA is something I wouldn't even think about trying without the special (and $$$) tools, and even with the tools it's usually easier and faster to just build a new board.
Fine pitch packages (eg PQFP208, TQFP144 etc) are probably the best compromise for a hobbyist who needs a fair sized FPGA. I've soldered them before in similar manner to the technique posted by Dave: Align with fingers, tack the corner pins in place with a fine tip slodering iron, slosh everything liberally with flux, and use the iron to 'roll' a big blob of solder along each row of pins while holding the board at a slight angle. This works quite well for me. About the only thing that can go seriously wrong is the big blob of solder falls off and splatters all over other parts, tracks, pins etc., so its wise to shield the rest of the board with some kind of material that won't melt if splattered with solder; I use simple cardboard .
The problem today is that BGAs have kind of insinuated themselves into the product lines for even low pincount devices (eg QFN108) to the point where the manufacturer doesn't always make the device in an alternative package. For example, take a look at Actel's Fusion range of FPGAs. The only non-BGA type package is the AFS600 in PQFP208, which is a great big clumsy thing.
This trend will not stop, so if we hobbyists don't want to become technologically insular and avail ourselves of the amazing repertoire of silicon out there, we have to find a simple, cheap and successful way of soldering BGA packages. Failing that we need to identify small businesses or individuals who have the wherewithall, and who are prepared to solder one-off or two-off BGAs for a few dollars. Most industry SMD assy establishments get snotty and disparaging if you ask them to do less than 100 of anything, so I believe this is something worth pursuing.
John
The problem today is that BGAs have kind of insinuated themselves into the product lines for even low pincount devices (eg QFN108) to the point where the manufacturer doesn't always make the device in an alternative package. For example, take a look at Actel's Fusion range of FPGAs. The only non-BGA type package is the AFS600 in PQFP208, which is a great big clumsy thing.
This trend will not stop, so if we hobbyists don't want to become technologically insular and avail ourselves of the amazing repertoire of silicon out there, we have to find a simple, cheap and successful way of soldering BGA packages. Failing that we need to identify small businesses or individuals who have the wherewithall, and who are prepared to solder one-off or two-off BGAs for a few dollars. Most industry SMD assy establishments get snotty and disparaging if you ask them to do less than 100 of anything, so I believe this is something worth pursuing.
John
Adheasive stencils
I'll second the toaster oven recommendation.
(Actually I use an old second hand industrial oven that I got for next to nothing).
Stencils unlimited has a great product that addresses a number of issues:
http://www.stencilsunlimited.com/bga_stencils.php?source=google&gclid=CMTVr5mz2IcCFSVVYQodbx9ApA
(I have no affiliation with this company - just a customer.)
These are adhesive stencils that stick in place between the board and the BGA device, becoming a permanent part of the assembly. Besides acting as a stencil, they help align the device and they act as a physical barrier to prevent the solder from shorting between balls.
They offer some standard sizes off the shelf. They also make custom sizes for $70 for a pack of 10.
**************
For unsoldering BGAs, I epoxy a bolt to the center of the BGA so that I can hang the board from the top of the oven with the board just a couple of millimeters above the rack. I heat the entire board and the weight of the board gently pulls it away from the BGA just as the solder liquefies.
Heating the entire board works much better than attempts I've made to just heat around the BGA.
I haven't yet tried to reuse a BGA, but there are quite a few places that make reballing kits.
********************************
There are other important reasons besides saving space that manufacturers are switching to BGAs and Chip Scale packages:
Better thermal dissipation and signal integrity.
The internal bond wires and pins of a leaded device are a major source of thermal and electrical impedance. Lead inductance is getting more detrimental as digital signal rise times speed up.
I still use leaded surface mount packages as much as possible for assembly convenience, but it appears that more new parts will only come with leadless packaging.
The biggest obstacle I see with DIY BGAs isn't soldering. It's board layout. If you're only using a limited number of I/O pins and power connections, a large FPGA is doable with six layers (or maybe four if you're lucky). For full I/O utilization, eight layers becomes almost mandatory. I'm not aware of any economical sources for prototype eight layer boards. (I had to laugh a while ago when I was reading an Altera app-note that described a 24 layer board stack-up as ideal, and then went on to outline a good 'compromise' 18 layer design.)
Regards,
Brian.
I'll second the toaster oven recommendation.
(Actually I use an old second hand industrial oven that I got for next to nothing).
Stencils unlimited has a great product that addresses a number of issues:
http://www.stencilsunlimited.com/bga_stencils.php?source=google&gclid=CMTVr5mz2IcCFSVVYQodbx9ApA
(I have no affiliation with this company - just a customer.)
These are adhesive stencils that stick in place between the board and the BGA device, becoming a permanent part of the assembly. Besides acting as a stencil, they help align the device and they act as a physical barrier to prevent the solder from shorting between balls.
They offer some standard sizes off the shelf. They also make custom sizes for $70 for a pack of 10.
**************
For unsoldering BGAs, I epoxy a bolt to the center of the BGA so that I can hang the board from the top of the oven with the board just a couple of millimeters above the rack. I heat the entire board and the weight of the board gently pulls it away from the BGA just as the solder liquefies.
Heating the entire board works much better than attempts I've made to just heat around the BGA.
I haven't yet tried to reuse a BGA, but there are quite a few places that make reballing kits.
********************************
There are other important reasons besides saving space that manufacturers are switching to BGAs and Chip Scale packages:
Better thermal dissipation and signal integrity.
The internal bond wires and pins of a leaded device are a major source of thermal and electrical impedance. Lead inductance is getting more detrimental as digital signal rise times speed up.
I still use leaded surface mount packages as much as possible for assembly convenience, but it appears that more new parts will only come with leadless packaging.
The biggest obstacle I see with DIY BGAs isn't soldering. It's board layout. If you're only using a limited number of I/O pins and power connections, a large FPGA is doable with six layers (or maybe four if you're lucky). For full I/O utilization, eight layers becomes almost mandatory. I'm not aware of any economical sources for prototype eight layer boards. (I had to laugh a while ago when I was reading an Altera app-note that described a 24 layer board stack-up as ideal, and then went on to outline a good 'compromise' 18 layer design.)
Regards,
Brian.

Hi Brian,
What sort of temperature controller do you use on the oven?
Elektor Electronics have a design here http://www.elektor-electronics.co.u...oup=<< empty >>&PN=On&SearchText=toaster oven
for an oven controller which they recommend for BGAs.
What sort of temperature controller do you use on the oven?
Elektor Electronics have a design here http://www.elektor-electronics.co.u...oup=<< empty >>&PN=On&SearchText=toaster oven
for an oven controller which they recommend for BGAs.
Dave said:Hi Brian,
What sort of temperature controller do you use on the oven?
Actually, I haven't been using a closed-loop temperature control system with a specific reflow temperature profile. It's been on my to-do list of projects.
For reflowing new parts on the board, I mark the large parts with a 210C color-change temperature indicating crayon (I have an old set from years ago, I think it was from Omega Engineering.) I preheat the oven to about 250C using the oven's temp controller. I place the board on the rack and crank up the oven setting to maximum until the marks change. They usually seem to all turn within a few seconds of each other. I then cut the power and gently open the door to let it cool.
Yeah, it's pretty crude. I initially tried it this way as an initial experiment, intending to follow up with a controller. It's worked out well enough that I haven't been in a rush to make a controller.
Brian.

- Status
- Not open for further replies.
- Home
- Source & Line
- Digital Source
- Anyone got it right with BGA's yet?