Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
07/31/06 18:13
Modified:
  07/31/06 18:16

Read: times


 
#121430 - He will need multiple layers anyway
Responding to: ???'s previous message
Routing an escape path for each of 1148 pins (balls) on a fpPBGA (fine-pitch plastic ball grid array) will require 6 or 8 layers anyway, particularly if the pitch is 0.5 mm or 0.8 mm. Fewer, perhaps, if blind and buried vias are used. Once one is in multilayer boards, however, the incremental cost of manufacture is not so bad going from 6 to 8 or thereabouts. Testing, with fpPBGA's is always a hassle. Specific provision for testing connectivity between critical paths needs to be designed into the board so that JTAG or other automated testing can be implemented.

Adding a few layers, since these devices seem to want their core supply (~1.2V) and at least one I/O supply(2.5 or 3.3V), and the external peripherals will want whatever they want (1.5, 1.8, 2.5, 3.3, or 5.0 volts), supply layers make the board easier and all-around better for supply distribution. Sometimes a dedicated GND layer for part of the circuit is better than sharing the common. This is not always the case, however.

The examples I've seen have routed the GND connections to the annulus on the inside of the fpPBGA, where bypass members can be connected easily, thereby relieving the congestion that often occurs when all the bypass is attached close to the outside.

Escape routing from 1148 pins is no picnic! The signals are of no use, though, if you can't get them to where they're needed.

RE


List of 25 messages in thread
TopicAuthorDate
Decoupling Capacitors            01/01/70 00:00      
   Paralleling...            01/01/70 00:00      
      Well have a look at the design            01/01/70 00:00      
         I will...            01/01/70 00:00      
            No layout available at the moment            01/01/70 00:00      
               May I ask why you don't follow the reference?            01/01/70 00:00      
                  Because            01/01/70 00:00      
                     No relevant difference, I think            01/01/70 00:00      
                        I do not think so            01/01/70 00:00      
                        try one poured layer for each supply            01/01/70 00:00      
                           Exactly!            01/01/70 00:00      
                              He will need multiple layers anyway            01/01/70 00:00      
                                 Fast clock signals            01/01/70 00:00      
                                 Multiple layers is no joke            01/01/70 00:00      
                                    That's a problem, all right ...            01/01/70 00:00      
                                       Same song second verse            01/01/70 00:00      
                                          Miker, it sounds as if you need to buy a book by D            01/01/70 00:00      
                                          Discontinuities and other mess            01/01/70 00:00      
                                    I wouldn't do that            01/01/70 00:00      
                     This stuff basically hasn't changed in 30 years            01/01/70 00:00      
      I never used individual ferrites ..            01/01/70 00:00      
         You WILL in mixed circuits...            01/01/70 00:00      
   never seen that, but have seen the opposite            01/01/70 00:00      
      Maybe somewhere in between will work            01/01/70 00:00      
   No longer interested, Mike??            01/01/70 00:00      

Back to Subject List