EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: johnnyfp on March 09, 2013, 08:18:10 am
-
Is working with BGA even doable for a hobbyist? Before I start looking at BGA devices I would like to know if, without expensive ss stencils, pnp machines, and IR ovens, a hobbyist will be able to put a functioning circuit together(using a commercial pcb manufacture like pcbzone or pcbcart)
Cheers
Jfp
-
It's possible for small BGA's but requires quite a bit of skill.
Have a look at some of the reballing videos on youtube.
-
Just search youtube for "reballing ps3/xbox".
Routing a PCB on 2 or even 4 layers is another matter. I'm not aware of any places that do batch PCBs more than 4 layers, and I'm not sure Laen does 4 layers anymore.
-
Yes, possible.
A cheap mylar/kapton stencil will do, and at a pinch, you could do it with a hot air gun. But you are much better off using a toaster oven.
Not many usable BGA's can be routed out on cheap 2 layer boards though.
Dave.
-
Yes... but it depends on what kind of bga you're talking about.
Pitch, total balls, ballout pattern, total ball usage all affect what pcb specs you will need.
Design considerations:
Devices designed for low cost embedded applications (say, TI arm cpus) have pinouts optimized for breakout on 4 layer and with some effort, even 2 layers.
These devices will have voids in the ballouts to allow you to place bypass caps and breakout routing. For IO constrained applications like FPGAs you will need 4 layer and go up from there.
For proper bga fanout you will want to be able to put a via in between the ball pads. The deeper the distance to the center of the bga, the more layers you need to break it out.
(http://img14.imageshack.us/img14/9083/23754241.png)
Here is a pcb I did with 6mil trace/space and 13mil drills. It is sufficient for a modestly broken out 1mm pitch BGA. 0.8mm, not so much. The pad land diameter affects the height the package will sit at, and also the size and geometry of the resulting joint. I used non-soldermask defined pads (NSMD). This allows the solder to grab around the edges of the copper etching.
(http://img213.imageshack.us/img213/5512/croppedbga.jpg)
Assembly considerations:
Reflowing the bga is a process in itself. You can't just blast heat on it. The requirements are: preheat the entire board to the first step of the bga reflow profile. Then apply local heat as evenly as possible to the bga solder balls. That's the rough view.
You can reflow smaller (<256) bgas that are MONOLITHIC flip chip (that is, solid black square) with a preheat and a hot air station. Substrate-mounted chips do NOT like this, you will warp the substrate and ruin the chip. The bigger the package, the lower maximum temperature the package can handle. Always look at the part datasheet for soldering guidelines.
I use a $600 convection oven with shielded elements and local thermocouple. You will want good, even heat that is well monitored. You can fudge dry runs to get the results you need, but it's hopeless if you can't read the results.
Profile for a large board with 484pin 1.0mm bga:
(http://img801.imageshack.us/img801/7289/reflowprofile.png)
Important parameters are rate of temperature ramp up, length of pre-heat (which is almost zero here), length of time spent above liquidus (solder molten temperature) and cooloff timing.
There is lots more info than I have room to write about here.
Heres the result, pardon the glare, it's hard to take pictures at such an extreme angle.
(http://img43.imageshack.us/img43/8841/bgal.jpg)
All BGAs nowadays come preballed with leadfree balls. Melting point as about 218 C. I mount the bga first on empty pcb, then go back and hand solder everything else with leaded.
You can either use a paste stencil and deposit solder paste as your means of "starting" the connection, or just flux. Laying down paste is a better idea but if you don't have a stencil, you can flux the pcb pads, lay down the BGA and let'er rip. Alignment is a non-issue. You can be off the bga pads by 49% and it still works. Rotated chip? No problem.
Flux-only method means you need a FLAT pad surface. HASL is unusable. I use ENIG as you can see above. Immersion silver would also work.
Flux type is important. You don't want the flux to completely burn off and be useless by the time the balls start melting. DO NOT use water soluble or no-clean flux. You will have no more vias underneath your BGA. Always tent vias with soldermask. I use MG Chemicals ROSIN flux pen. It may not be the best choice but it works. Pastey/less runny flux is better.
Here is one last pic.
(http://img811.imageshack.us/img811/7453/img0576ta.jpg)
You can see the small soldermask allowance around the pad for NSMD pads. Keep your annular ring on vias as big as allowable. You can see the drill wander on some of the holes here. Always neck down the dogbones to 2/3 or less of the pad diameter to keep the ball shape from becoming abnormal and stressing the joint. That fpga on the left shows what you will get from the factory.
One last thing: You MUST observe moisture rules on the parts. Don't remove parts from the bags with silica gel until you are ready to reflow. Moisture will work into the porous bga package and the water pockets can popcorn and cause unseen damage. JEDEC procedure for removing latent moisture is something like 125 C for 12 hours. Same goes for the pcbs themselves.
If you have more specific questions let me know. There is a lot of stuff I probably missed.
-
The main barrier is the cost of the number of PCB layers & small track/via sizes you need to route it.
Super-fine pitch stuff is probably a nonstarter, but as long as you have a good quality paste print, a steady hand, alignment marks on the PCB and leaded paste it should be doable for prototype quantities.
The only differences from other packages are the need to drop it down right first time as any sideways nudging risks getting paste bridges, and inability to rework.
lf you have a process which reliably does QFNs, QFPs etc. without rework, then BGA is probably not a big step.
I have seen someone (with a very steady hand) at an assembly house hand-placing boards with several large BGAs. They did have an x-ray facility though.
-
The equipment is cheap enough, but you have to invent a method of accurately place the part.
You would need an oven and a controller for it so you can do a proper temperature profile. That is actually not a big investment, less than 300€. With that, you want to use a stencil to print paste, which adds to the cost of the PCB, 20€ or so for a plastic one or around 50€ for steel. There are places that offer "free" stencil wi the PCB, but last board I did, the PCB would have costed more that the board and stencil billed separately elsewhere. So, there is some added cost; for one BGA chip that might not make sense, but the cabability to do SMD is a very nice addition to your lab anyway.
Once you have the oven, the issue is how to place the part. You need to drop it in place pretty much at pne go in the right place, and you can't see what you are doing. Pock and place machines are still out for hobbyists, the lowest cost available is like 4000$ or so, and without vision, I'm doubtful about if the accuracy can be good enough.
If you are doing the PCB design yourself, one idea might be putting in some guides, like hand soldering some guide pins so that when you put the part between the pins it drops in the right place?
-
If you have more specific questions let me know. There is a lot of stuff I probably missed.
huh ? your answer is already really impressive... !
-
Soldering BGA chips is not so hard once you gain some experience, and with right kind of equipment. Equipment can be built for not too much money.
I do BGA rework on a daily basis, reflowing, reballing and changing chips on xbox360, PS3 game consoles and notebook motherboards.
This is my homemade IR rework station during initial testing (pulled massive RSX chip from PS3).
(https://dl.dropbox.com/u/21387397/slike/eevblog/heater_2.jpg)
-
Soldering BGA chips is not so hard once you gain some experience, and with right kind of equipment. Equipment can be built for not too much money.
I do BGA rework on a daily basis, reflowing, reballing and changing chips on xbox360, PS3 game consoles and notebook motherboards.
This is my homemade IR rework station during initial testing (pulled massive RSX chip from PS3).
https://dl.dropbox.com/u/21387397/slike/eevblog/heater_2.jpg (https://dl.dropbox.com/u/21387397/slike/eevblog/heater_2.jpg)
Wow, that is seriously impressive! :-+ Do you have any more details on it?
I found a broken PS3 dumpster diving a few weeks ago and was going to try to reball it with a heat gun. Anyone done this before?
-
Thanks for that, quite an impressive response from you guys.
The problem I see here is that during prototyping I would make one board, test it refine any issues with it and usually with my hobbyists tendency find that my pitches are all wrong, and then either fix the issue or re order the pcb. This is fine with soic, ,dils, qfns even leadless, because I can see any shorts, dry joints, dodgy vias or routes, but with a bga its all hidden.
So how can I tell if my circuit doesn't work that its because of my dodgy circuit design or my dodgy soldering technique, short of borrowing Mikes airport Xray machine.
It is certainly going to be a lot more challenging especially when working with Android or Linux based SOC devices, like Ti or AllWinner chips, that have a lot of pins and require external memory which tend to also be BGA based chips.
-
DO NOT use water soluble or no-clean flux. You will have no more vias underneath your BGA. Always tent vias with soldermask. I use MG Chemicals ROSIN flux pen.
Isn't that a no clean flux?
-
So how can I tell if my circuit doesn't work that its because of my dodgy circuit design or my dodgy soldering technique, short of borrowing Mikes airport Xray machine.
As said, you need to have an oven with a controller. You do a couple of monitored runs with it; after that, you will trust your soldering process. After two real boards, I trust my process, it really is that easy. After 30 years I still don't trust my circuit desings as much...
-
Yeah, I figured as much. I just find that if a circuit doesn't work then I often think it's my soldering, I suppose it's just a matter of practice and confidence in my ability.
But anyway thanks for the advice. Now off to design the next best bga circuit with my new found knowledge. >:D
-
I found a broken PS3 dumpster diving a few weeks ago and was going to try to reball it with a heat gun. Anyone done this before?
Don't even try it. You need to preheat the whole board to 160-180C and then just finish heating chip to reflow temperature with hot air or IR, carefully watching for temperature.
If you attack the chip with just a heat gun, you will kill it. Don't ask me how I know that... :D
Reworking PS3 boards is hardest thing I ever did. Board is thick, multilayer and with strong heatsinking effect. If your equipment can cope with that, every other board is piece of cake ;)
-
Here is a "DIY BGA rework station" project that just last week was reported on Hack a Day:
http://edmarhobby.blogspot.de/2012/07/diy-bga-rework-station.html (http://edmarhobby.blogspot.de/2012/07/diy-bga-rework-station.html)
-
DO NOT use water soluble or no-clean flux. You will have no more vias underneath your BGA. Always tent vias with soldermask. I use MG Chemicals ROSIN flux pen.
Isn't that a no clean flux?
I reflowed a Spartan 6 BGA in a toaster oven using a water soluble flux pen, and it turned out ok.
-
I found a broken PS3 dumpster diving a few weeks ago and was going to try to reball it with a heat gun. Anyone done this before?
Don't even try it. You need to preheat the whole board to 160-180C and then just finish heating chip to reflow temperature with hot air or IR, carefully watching for temperature.
If you attack the chip with just a heat gun, you will kill it. Don't ask me how I know that... :D
Reworking PS3 boards is hardest thing I ever did. Board is thick, multilayer and with strong heatsinking effect. If your equipment can cope with that, every other board is piece of cake ;)
I've got nothing to lose by breaking it, but it would be cool to fix. Maybe I can find a reflow oven at my university.
-
Last time I replaced a BGA I simply tinned the pads, applied flux and heated the board with a heat gun until the solder melts.
-
i attempted to reflow a dead PS3 using a painting hotair gun on the bottom of the pcb as preheater and a soldering rework gun on the top.
It fixed the problem but only lasted 3 weeks before failing again.
-
If you dont need a custom pcb for the project maybe try one of these?
http://www.schmartboard.com/index.asp?page=products_bga (http://www.schmartboard.com/index.asp?page=products_bga)
-
Wow, I like the idea. Shame that it cost as much as getting a custom PCB made.
-
If these are just one off designs and you can't get what you want in other chip forms then start emailing and calling some of the PCB fab places. Sometimes you luck out, I have one not too far from me that will do BGA for $10 a chip and that isn't the first time I have found someone who will do it.
Also don't forget forums like this one, there may be other hobbyist who have access to the right gear to be able to do a one off chip for you.
-
Here's my 2 cents:
I work at a contract manufacturer and do testing.(functional and burn-in) One of the most difficult to troubleshoot soldering issues is with BGAs. I'm not saying I know everything about BGAs, soldering or anything like that. I have worked alongside a troubleshooting engineer with the X-ray machine to diagnose BGA issues and it is really hard to know what to even look for on the X-ray.
It may be possible, yet difficult to design, stencil and reflow BGAs at home.
-Tim
-
Or you can do like I do if you have trouble (at the beginning) with very small features: go to the guys that have been doing this for years. I have a few friends back home that do cellphone repair and they do this stuff on a regular basis.
Because if it's a one-off design it's worth do have someone more experienced help you. If it's already a series run you would go with automated placement.
Now, these cellphone experts have very little theoretical knowledge in how modern electronic circuits work, but they are much better handymen than most of the professionals I have seen. Cost-per-hour is also a magnitude lower.