Re: BIRD #??? (yet to be issued)

Y'all,

  Say we have:

  Pins: 11 12 13 21 22 23
           + + + + + +
           | | | | | |
           | | | | | |
  Buses: +-----+------+-----> to a few +-----+------+-----> to a few
              GNDBUS1 drivers GNDBUS2 more

In Kumar's suggested format we have:

[Pin] signal_name model_name gnd pwr
1 RAS0# Buffer1 10 18
2 RAS1# Buffer2 12 20
....
...
...
11 GND GNDBUS1
12 GND GNDBUS1
13 GND GNDBUS1
....
21 GND GNDBUS2
22 GND GNDBUS2
23 GND GNDBUS2

One problem with this (as I see it) is that the *model* names GND and POWER
are "special" (they indicate that the pin is NOT connected to an ordinary
nonlinear I-V curve model, and so we don't go looking for that model in
the .ibs file.) In the suggested format above, we don't have any way to
tell that GNDBUS1 isn't an I-V model.

The *signal* names, of course, are sacred because they appear in the
schematic symbol for the part. Pins that connect to separate internal
buses might just be given the same name (e.g. GND) in the schematic, and
so we can't rely on these to distinguish buses.

I suggest we instead use the two remaining columns to specify the associated
bus names; for example,

[Pin] signal_name model_name gnd pwr
1 RAS0# Buffer1 GNDBUS1 PWRBUS1
2 RAS1# Buffer2 GNDBUS2 PWRBUS2
....
...
...
11 GND GND GNDBUS1 NA
12 GND GND GNDBUS1 NA
13 GND GND GNDBUS1 NA
....
21 GND GND GNDBUS2 NA
22 GND GND GNDBUS2 NA
23 GND GND GNDBUS2 NA

It's now clear which pin is a lowly ground pin (and not an I-V model),
and it's also clear to which bus every GND pin and every driver is
connected.

Now there's no ambiguity--I think...

But what would we do with buffers that have connections to multiple
supply voltages?

--Eric Bracken
  bracken@performance.com
Received on Thu Dec 2 07:41:46 1993