How to solder BGAs with reflow oven, and what footprint to use?

I am not sure about how to get the footprint for this chip but I would look around for a similar pitch/size BGA and see if someone provided the fp for it. Also, most commercial layout packages, such as Eagle have footprint libraries that can choose the footprint for you. I do not think that there is anything special to the pad layout (other than the inevitable signal integrity issues), so you can put a via under a pad.

Finally, I saw with my very own eyes how a BGA was soldered on in a toaster oven controlled with a laptop (it is important to follow the temp curve recommended by the manufacturer). Surface tension essentially positions the chip where it has to go. Check out this guy's page where he describes a similar procedure:

Hopefully, more knowledgeable people in this group will provide more insight for you.

Good luck!

Alex

Yes round pads are typically used. I ended up using 0.52mm diameter circles for the adxrs BGA pads (just under the average 0.55mm ball size) Although I think some people use 75% or less for the pad size.

For vias on the pads, the via's will suck solder into them unless they are filled. They can be filled by you manually or by using a board with buried/blind vias. If you don't fill them before putting on the gyro, you may lose a connection as a solderball may get "sucked down the drain". I think this is why people usually put vias in between the pads, but this requires small drilling to make the vias small enough so they don't short out to the BGA pads.

I've used a toaster oven to solder these gyro's and it works fine. I have eagle cad files for the PCB footprints that I used here:

I wouldn't recommend using solder paste for the adxrs parts, just put a very light and smooth coating of flux on the PCB before you set the gyro on and then bake it. For other BGA parts you may or may not want to use solder flux, depending on if the part has solder balls or non-melting balls. (I think most parts have melting solder balls though).

If you do want to put paste on the pads, it is much easier to use a stencil than to manually place small drops of solder by hand.

Also maybe check out other newer gyro's like this one:

cheers, Jamie

Only if you plan for solder wicking into the via during reflow, which would starve the solder joint. I've seen recommendations to use tented vias (on the other side), or fill them with conductive polymer. Perhaps applying paste and reflowing the bare board (to pre-fill the vias with solder) beforehand would help?

Random thought: If you had a piece of glass the size of the BGA, you could rest it over the paste when pre-filling the vias, to help get a nice coplanar surface to put the BGA on.

I wonder if it would be better to use a hotplate for BGAs, as the solder would melt before the chip itself got cooked, since the heat comes up through the board. Not that my hotplate is controlled in any way (it heats at 0.8°C/sec, regardless of the knob setting), but it did well with non-bga parts. Mostly, I'm curious about which part of the assembly is most sensitive to the profile, and why; perhaps a change in procedure could make it less sensitive and thus easier to do for hobbiests?

More likely, I just got lucky and would crack the BGAs due to thermal stress :-P The chips I used so far were SOJ and TSOP styles, more flex in the pins.

Good point, prefilling is probably the best course of action if that via is desired under a pad, I stand corrected.

This is also a great idea. I saw people using superglued `guides' made out of smd resistors to hold the chip in place but a nice flat surface would make it easier to position.

...

I would be more worried about the traces coming off on the side of the PCB that is in direct contact with the hotplate. Also, when you heat with a hotplate, the solder balls on the chip that have poor contact with the PCB below will melt last and might make a poor connection. Another possibility is that surface tension will pull the chip at one end and lift the opposite end without creating a soldered connection. Your `flattening off' idea might solve this problem though. Thanks for good ideas, in any case!

Alex

The three boards I've done so far didn't have any problems with that. The board is only solder-hot for a few seconds, there wasn't even any discoloration. Ovens get both sides of the board that hot, and they don't melt the copper off.

I've put a thermometer right on the hotplate before, the plate itself is not much hotter than the board, at least the solder melts at about the temperature I'd expect. It's not like it's 50 degrees hotter on that side.

The trick is to move the board around so it heats evenly, otherwise it burns in one spot before it's hot enough to melt in some other spot.

I'd probably want to put solder paste on the pads. They'll help wick heat and ensure that they all make contact, plus the flux would help a little too, both soldering and sticking the part down.

From what I've read, that's a problem with the smaller parts, like

Does the gyro part require any pins to escape? The one I worked with was in a BGA package but only the outer row of pins was unique. The internal pins were duplicates.

I believe that is the case with the Analog gyros - but I'd also like to be able to do other chips as well. I mean if I can get a reliable process for soldering BGAs it would open up a whole world of chips that I hadn't been able to use before (like alot of high end ARM parts).

-Mike