Content originally posted in LPCWare by ktownsend on Wed Dec 21 14:27:39 MST 2011
Quote: vasanth
Hai rmteo, Have you noticed the speed of the LPC4350 in those links? (Digikey->180MHz & Futureelectronics->150MHz) :confused:
Early silicon was specced for those speeds, but this was later increased because the chips were able to support the higher speed. You should always use the manufacturers sites and latest datasheet for authoritative data. The people entering data on sites like Digikey do a good job, but mistakes still happen and they're more likely to get corrected on the manufacturer's site than in a distributor's database.
As with any new devices, they go through various revisions, and improvements are made along the way, though, including the discovery that these chips can apparently supported a higher maximum clock frequency than expected.
As for the packages, the BGA256 packages actually isn't that hard to work with since it's 1.0mm pitch, which is very forgiving even if the placement is less than surgically precise. If you put down a good paste layer (the most critical step for BGA) I can't imagine anyone getting much bridging, and there's still plenty of room for the vias doing the fanout. You'll still need at least a 6 layer board, though. Maybe (???) you could do it in 4 but the quality will be mediocre without a decent GND and power plane, so I wouldn't bother with less than 6.
But the only real issue is good PCB design. The actually assembly isn't that scary. BGA reflows well (with a good paste layer) down to 0.8mm pitch ... smaller than that it becomes more of a challenge to do the fanout, and you risk getting bridges underneath if you're sloppy (and obviously not many people have in house xray machines to test for this :) ).
