3
\$\begingroup\$

I was searching for a method to use a BGA based MCU on PCB which could be easier to change frequently. There are sockets for this purpose but I want to ask what if I simply put MCU on the right place and then hold it with a 3D printed holder with some force. This idea is storming in my mind but a question is raising frequently if it is possible or not, will it work??

Any help with this?? Thank you.

\$\endgroup\$
3
  • \$\begingroup\$ Ironwood sockets run to 10+ GHz. \$\endgroup\$ Commented Aug 24, 2020 at 21:25
  • 3
    \$\begingroup\$ The potential problem, just to make it totally explicit, is that the solder balls don't all protrude by the exact same amount. Some will touch and others will have a small air gap. If you push hard enough to eliminate the air gap, you will have to deform the other balls quite a bit. And balls are not springy. So this doesn't seem reliable. But I have never tried it. I have had a hard time trying to temporarily hold 4 or 5 pin connectors against solder pads. So trying to do it with a BGA sounds hard. You need some kind of spring in the stackup somewhere. \$\endgroup\$
    – user57037
    Commented Aug 25, 2020 at 6:19
  • 2
    \$\begingroup\$ The coplanarity of many BGAs is specified at 100 microns or better (Xilinx app notes like brag they easily meet the JEDEC standard, but don't offer any specifics). From what I can tell, SAC205 solder is harder than FR4, but softer than copper. It wouldn't surprise me, that if you apply enough pressure, you could deform/straighten out the BGA/PCB/Copper/Balls enough to make a connection. It might only be a one time connection, if things gets squashed too much though! Not saying it's a good idea (in fact it's a bad idea), but if you get lucky maybe you'll have 99% of your pins connected. \$\endgroup\$
    – Adam
    Commented Aug 25, 2020 at 7:55

3 Answers 3

9
\$\begingroup\$

The Ironwood sockets work like this. They use an elastomer interposer with embedded wires. Depending on the type they’re good for hundreds to thousands of mating cycles. They use a slightly modified version of the BGA footprint, and are compatible with soldering down the BGA later.

The other BGA socket type uses pogo spring pins. These are more common for production test boards.

That said, it’s not feasible to just push the BGA down onto the board with a clamp. Something has to compensate for co-planarity as well as make a gas-tight contact. Not only that, but pushing the chip down with that much force will damage the solder balls, making it impossible to solder the chip later.

\$\endgroup\$
4
\$\begingroup\$

I strongly suspect the answer is "not reliably if at all".

Electrical contacts need to be have pressure applied to break through surface oxidation and prevent the contacts from separating with vibration/thermal changes.

The problem is that without some "springy" components to take up the slack, small variations in the surface dimensions will translate to massive variations in the contact force or even some pins that are not in contact at all.

\$\endgroup\$
1
\$\begingroup\$

Probably not very reliably, but I can imagine it working for prototypes. Whenever I reflow BGA chips, I apply tacky flux, then push down on the BGA and sort of slide it around until it 'locks' into position. The locking mechanism comes from the fact that there is a ~25 micron thin film of soldermask 'cupping' the solder ball contacts (the solder contacts probably need to be gold for planarity reasons). Multiply this tiny force by 400 odd balls, and it becomes quite significant with applied downforce and doesn't allow it to slide around.

I don't really understand the purpose having a socketable BGA though, unless the pinouts between chips are compatible. If you're looking to do it so you can easier time soldering (eg, by breaking it out to header pins), you're defeating the purpose of the BGA package, which is to reduce inductance, especially to decoupling capacitors.

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.