besides ITRS outlined roadmap pin counts (which all industry tend to
bench mark and try to keep in line), there is a diminish of return
when you goes to too high pin counts - not only causes layout
problem, but also production yield issues - e.g. >1000 interconnect
better have very flat on both components and PWB... you might need
to go for LGA instead of BGA... as for prototype, those components
are killers to testing initial functionality of the system (too
complex). rather than block set of components instead - that is why
initial customer chip is SPGA, until the funcationality are defined
to set into ASIC to speed things up and save the real estate... (more
integrate on chip the better - anything walk off chip is a problem...
impedance, fan out, etc.etc. you want as simple as possible - as
lower pin count as possibly, not more... IMHO... if you give
designer max pin count can be achieved for I/O, I bet he/she will
used to max +10%... leave you little or no room but many headache to
try to layout on PWB... )... By the way, you didn't say what PWB
tech you intende to use - 4/4? or 3/3? without knowing that, it is
missing half of the selection criteria... IMHO. One more thing,
assume you have deep pocket - e.g. you can custom make high density
ACF if you have enough volume or intend to use next gen pin count
outlined in ITRS... (although ITRS is no longer publish roadmap, I
believe it old version still cover next few years).
my 1.4 cents.
jk
On Mar 22, 2018, at 12:07 PM, Ahmad, Syed wrote:
> We are trying to determine limits on solder or other interconnects
> for a high pin count prototype and need help in identifying
> suppliers for solders, stencil manufacturers, ACAs, ACFs and other
> options that may help us achieve lowest pitch interconnects. All
> help is welcome. Thank you for your assistance. Suppliers may
> contact me directly on my email.
>
> Syed Sajid Ahmad,
> Research Staff, ECE, NDSU
> Cell 701-200-1674
|