Any reason why one should be preferred?
Apart from the obvious ie BGA is smaller, but one can get at the pins with a
scope with LQFP
Dirk
BGA v LQFP
Started by ●July 14, 2008
Reply by ●July 14, 20082008-07-14
Dirk Bruere wrote:
>
> Any reason why one should be preferred?
> Apart from the obvious ie BGA is smaller, but one can get at the pins
> with a scope with LQFP
>
> Dirk
>
It first depends on your manufacturing process. Its not easy to do BGAs
without special equipment and it can be fairly expensive if you don't
need many boards.
BGAs are harder to use for prototypes since you can't solder the pins
directly. If you have a short or no connect, you are probably in trouble
with the BGA.
OTOH, you probably have fewer soldering problems with a 0.8 to 1.27
pitch BGAs than with 0.5mm QFPs. BGAs tend to self align.
From a pc layout point of view, BGAs are usually easier to route
assuming that the BGA is not a lot smaller than the QFP. For example,
the ADSP-21369 in BGA is much easier to route than the QFP variant. They
are close to the same size. The 0.8mm pitch Blackfins are also easy to
route. You do need to use a good board house that keeps tight
tolerances. My last BGA board used 0.1mm traces width and spacing (4/4)
with small vias. This costs about twice as much per sq inch (or mm)
than a 7/7 (0.2/0.2) board which might be appropriate for a QFP design.
BGA Layout gets a lot more difficult as the pitch becomes very small
(0.5mm for example) or when the pin matrix is full. BGA memory devices
can be difficult for this reason.
It's easier to get good high frequency performance from a BGA since lead
inductance is much lower. The part usually has better heat dissipation
as well. This is why many DSPs and FPGAs are not even offered in QFP
packages.
When routing BGAs, I place decoupling caps on the bottom side of my pcb,
usually in the center region of the part where most parts concentrate
power and ground pins. QFPs generally have decoupling caps positioned
around the outside perimeter of the part on the top layer. This is less
effective and also complicates routing. Skimping on decoupling
capacitors, series Rs, etc is a really bad idea. Most of the bad layout
and design that I see tend to be on boards that also avoid BGAs. This
may be partly due to the layout difficulty of QFPs, but I suspect that
its probably more due to the level of expertise of the designer. In all
cases, you should place fiducials on opposite corners of both BGAs and
QFPs.
Of course, you might avoid all these issues by buying boards or modules
from us. ;-)
Al Clark
Danville Signal Processing
>
> Any reason why one should be preferred?
> Apart from the obvious ie BGA is smaller, but one can get at the pins
> with a scope with LQFP
>
> Dirk
>
It first depends on your manufacturing process. Its not easy to do BGAs
without special equipment and it can be fairly expensive if you don't
need many boards.
BGAs are harder to use for prototypes since you can't solder the pins
directly. If you have a short or no connect, you are probably in trouble
with the BGA.
OTOH, you probably have fewer soldering problems with a 0.8 to 1.27
pitch BGAs than with 0.5mm QFPs. BGAs tend to self align.
From a pc layout point of view, BGAs are usually easier to route
assuming that the BGA is not a lot smaller than the QFP. For example,
the ADSP-21369 in BGA is much easier to route than the QFP variant. They
are close to the same size. The 0.8mm pitch Blackfins are also easy to
route. You do need to use a good board house that keeps tight
tolerances. My last BGA board used 0.1mm traces width and spacing (4/4)
with small vias. This costs about twice as much per sq inch (or mm)
than a 7/7 (0.2/0.2) board which might be appropriate for a QFP design.
BGA Layout gets a lot more difficult as the pitch becomes very small
(0.5mm for example) or when the pin matrix is full. BGA memory devices
can be difficult for this reason.
It's easier to get good high frequency performance from a BGA since lead
inductance is much lower. The part usually has better heat dissipation
as well. This is why many DSPs and FPGAs are not even offered in QFP
packages.
When routing BGAs, I place decoupling caps on the bottom side of my pcb,
usually in the center region of the part where most parts concentrate
power and ground pins. QFPs generally have decoupling caps positioned
around the outside perimeter of the part on the top layer. This is less
effective and also complicates routing. Skimping on decoupling
capacitors, series Rs, etc is a really bad idea. Most of the bad layout
and design that I see tend to be on boards that also avoid BGAs. This
may be partly due to the layout difficulty of QFPs, but I suspect that
its probably more due to the level of expertise of the designer. In all
cases, you should place fiducials on opposite corners of both BGAs and
QFPs.
Of course, you might avoid all these issues by buying boards or modules
from us. ;-)
Al Clark
Danville Signal Processing
