Dear List:
Now that we are in the deep submicron era, complex digital IC's
frequently
come with multiple sets of power pins, and sometimes multiple sets of ground
pins. A commonly occuring example is Vcc-output and Vcc-internal.
Occasionally, there is a separate power supply for specialty I/O such as
LVDS. And accompanying all of these IC's is varying quantities of
information of unknown quality about how these high-speed devices should be
powered and decoupled. Depending on process technology and application,
various voltages may be different, or may be the same.
(Note that there is little confusion about analog power supplies, or
power
for internal PLL's or oscillators. These should be isolated and separately
decoupled as best as one can.)
At a minimum there is an identified set of power supply pins that is
connected to the digital outputs. These should connect to the power plane,
and many traces leading away from the device will be closely coupled to the
plane. Very good.
If the devices is in a peripherally leaded package such as 240 pin PQFP,
the many power pins of some other type of power (Vcc-internal, for example)
will be very far apart, but they have to be connected to one node. So there
are two choices.
One could trace-wire them together. But then trace inductance will
appear
in series with whatever decoupling is installed. And the operating
frequency could be 100MHz or more, so every nanohenry counts.
Or one could put in a plane split. The usual warnings about what
happens
when your high-speed signals cross the plane split apply.
And, of course, the "internal" of the IC has to communicate with the
"outputs", if it does anything useful. And it has to communicate at 100's
of megahertz.
So:
Does one really want to keep the "internal" of an IC decoupled from the
I/O's at high frequencies?
If the process technology permits all the voltages to be the same, is it
better to lay down one low-impedance continous plane?
How important is it to keep the 'internal' of an IC supplied with quiet
power--especially if it is at the expense of the output drivers?
If one attempts to separate "internal" from "output" or "specialty I/O" from
"internal", will signals crossing the split within the IC then make lots of
noise because their return currents must go around a circle to be completed?
Best regards,
Boris
This message is from the IEEE EMC Society Product Safety
Technical Committee emc-pstc discussion list.
Visit our web site at: http://www.ewh.ieee.org/soc/emcs/pstc/
To cancel your subscription, send mail to:
[email protected]
with the single line:
unsubscribe emc-pstc
For help, send mail to the list administrators:
Ron Pickard: [email protected]
Dave Heald: [email protected]
For policy questions, send mail to:
Richard Nute: [email protected]
Jim Bacher: [email protected]
Archive is being moved, we will announce when it is back on-line.
All emc-pstc postings are archived and searchable on the web at:
http://www.ieeecommunities.org/emc-pstc
