How to breakout a .4mm BGA

This article was originally published on Sierra's blog.

One thing PCB designers will all agree on is that to breakout a BGA, you need precision and discipline. When you breakout a BGA, you basically apply a fanout solution and route traces from those fanouts to the perimeter of the device prior to general routing of the PCB.

Becoming a PCB master for HDI starts with learning how to breakout a BGA. BGAs have the highest density of I/O connections and array pins on a device, which is the most complex part of the layout.

Since there is not just one but multiple ways to breakout a BGA, I am giving you an example of how to breakout a .4mm BGA. A good design practice is to do the layout part by part and make sure that you will be able to fanout and connect all the pads under the BGA. Another good design practice is to draw short tracks.

When planning out how to breakout (or route) a .4mm BGA, the overall size and the pin out of the part need to be thought through so the most cost efficient technology can be used. It is not just about the fact that the pads are .4mm apart and figuring out what are the numbers of widths and gaps that work.

When fanning out a .4mm BGA, the geometry does not work out so that you can route a trace between the pins. The traces and gaps are just too small to go down enough layers to get all the pins fanned out.

So with a .4mm BGA, blind and buried vias are required. The pins on the outside row in the BGA are routed on Layer 1, the pins on the next row in the BGA are routed on Layer 2, etc. The way the BGA is pinned out will determine how to route the BGA.

In the configuration above, the problem is: What can you do with the GND pin? Drilling from Layer 2 to Layer 3 will cost more.

Example of how to breakout a .4mm BGA:

This part is a standard .4mm BGA. The best way to fan it out is to use blind and buried vias and a multi-lamination fabrication.

Start off by adding a Layer 1 to Layer 2 blind via on all of the GND pins of the BGA. (This is a hole in the BGA pad that will tie that pin to the GND plane on Layer 2.) Now all of the GND pins are done and do not need to follow the routing described below.

The routing plan:

Route the outside row or ring of pins (24 pins, row 1 and 7 and A and G) on the top side of the board, Layer 1. You may need to route out some of the GND pins on the outside row and not just add a 1 to 2 via in the BGA pad.

Add a Layer 1 to Layer 2 blind via to the next ring of pins (16 pins, row 2 and 6 and B and F), route this ring out on board Layer 2 – Layer 2 is a GND plane so these Layer 2 routes can only route out a short distance and then a standard top to bottom via is added. Do not cut off the GND plane with these short Layer 2 routes such that the GND plane is cut up and pins are not connected.

Then, on the next ring of pins (8 pins, row 3 and 5 and C and E), add a Layer 1 to Layer 2 blind via and a Layer 2 to Layer 3 buried via and route these nets out on board Layer 3.

The BGA pin in the very center of the chip will need to be worked out on Layer 2 or 3.

Pad and drill sizes for .4mm BGA:

There will be no traces between pads on the top layer so the pads can be 10 mils and have a 5.7 gap between them. For the larger pad, let the manufacturer drill a 4 to 6-mil laser drill for the buried and blind vias. The distance between the board layers will determine the best drill size so they can plate the hole shut and make a flat pad for the BGA, etc.

Sequential lamination:

Let’s pretend that this is an 8-layer board.

Board Layers 2 through 7 will be laminated together. Then, a laser drill will be done from Layer 2 down to Layer 3: This is your 2 to 3 buried via. Note that a Layer 7 to 6 buried via could also be used, and a Layer 2 to 7 could be used; but the 2 to 7 needs to have bigger drill and pad.

Board Layers 1 and 8 will now be laminated to the board and a laser drill will be done from Layer 1 down to Layer 2 (Layer 10 to 9 if needed).

Then, the standard through via is drilled top to bottom – here again, bigger drill and pad are required.

In your Gerber files, you have parts placed around the .4mm BGA. The fanning out and adding standard size vias and trace width and spacing will take up a lot of room around the BGA.

Technology level for .4:

Robert Feranec’s fanout strategy:

And because a picture - or in this case a video - is worth a thousand words, I highly recommend that you watch a demo by my good friend Robert Feranec.

Here is his strategy:

“Be careful about drawing the connections all at once. It is a good design practice to do the layout part by part. If you have a BGA, the first goal is to go out of the BGA. Be sure that you can go out with all the pads and all the pins under the BGA. Do not fully route all the interfaces, but instead just route a little bit of track, and then stop routing to place vias, or draw the tracks for different parts just a little bit out of the BGA area. Make sure that you will be able to fanout and connect all the pads under the BGA. You can then start connecting all the interfaces.

If you only draw short tracks, it is very easy to delete them and redo parts of the layout under the BGA because you did not fully route the tracks. So you can just delete them and try to find a way to fanout all the pins. This practice is much easier and will save you a lot of time since you only have to deal with a small part of your layout. You want to avoid finding yourself in a difficult position once you have fully routed your BGA and you need to take out only one pin, for example. If that happened, you would have to deal with a lot of tracks just to be able to remove one signal.

There are some exceptions, like memories or high-speed interfaces. In this case, you can fully route the connections between the BGA and the memories or the high-speed interfaces before you finish the fanout of all the pads.”

Now that you know how to breakout a BGA, learn more about the blind and buried via technology and download our free HDI design guide!