Digital VLSI Simulation


   1.  The group of gates whose names begin with "V_" are intended to
       emulate digital CMOS transistors and structures in the Digital LOG
       simulator.  Naturally they are not as accurate a simulation as
       AnaLOG's transistors, but they have the advantages that the digital
       simulator is much faster, and that they can be used in combination
       with all the other digital gates in the library.

   2.  The basic VLSI gates are V_NFET and V_PFET, the CMOS transistors.
       These gates can model most digital CMOS circuits which do not
       involve charge sharing or other "analog" effects.  The transistors'
       gate pins have "capacitance" which holds the last charge stored on
       them, allowing you to build dynamic as well as static registers.
       The transistors are symmetric:  You don't have to worry about which
       end is the source and which is the drain.

   3.  V_NFET conducts zeros between its source and drain pins whenever its
       gate pin is one.  It does not conduct logic ones at all.  Similarly,
       V_PFET conducts logic ones when its gate is zero.  In other words,
       V_NFET and V_PFET act just like CMOS transistors, as far as most
       digital applications are concerned, and that's all you need to know.
       If you are interested in a detailed understanding of how the CMOS
       simulation works in LOG, read the following few paragraphs.

   4.  Because Digital LOG is a gate-level simulator, it needs to know at
       any given time which pins are "inputs" and which are "outputs".
       Since transistors are symmetrical devices which can conduct in
       either direction, the V_NFET and V_PFET gates are implemented as
       bidirectional buffers which can change their direction of transfer
       according to their environment.  Specifically, any time a V_NFET
       sees a zero on its output, and an undriven node on its input, it
       changes direction so that it can conduct the zero onto the undriven
       node.  This change of direction occurs only if the gate is switched
       on.  Likewise, the V_PFET switches direction in order to conduct a
       one onto an undriven node.  In most circuits, this all happens
       automatically and you can ignore it.  But if a particular V_NFET or
       V_PFET looks like it is getting confused, you might swap it for a
       V_NFETD or V_PFETD (described below).

   5.  The "gate capacitance" is implemented by driving a "weak" one or
       zero onto the gate, according to the last "strong" value seen there.
       A weak one or zero is the kind of signal produced by a TIE (resistor
       to Vdd) or TIEGND (resistor to Gnd) gate.  If another circuit drives
       an actual value onto the gate, this strong value overrides the weak
       "capacitative" value.  To avoid undesirable charge-sharing problems,
       the sources and drains of transistors treat weakly driven ones or
       zeros as if they were undriven.  This means, for example, that a
       value stored dynamically on the gate of a transistor will not leak
       back through any "pass" transistors connected to that gate.

   6.  There are also "directional" transistors, V_NFETD and V_PFETD.
       These gates are similar to V_NFET and V_PFET, respectively, except
       that they conduct in a fixed direction.  Information is always
       propagated *from* the pin with the arrow, *to* the other pin.  In
       terms of current, the directions of the arrows show the direction of
       current flow.

   7.  Finally, there are V_NFETX and V_PFETX.  They are directional
       transistors that do not "weakly" drive their gates to show the
       dynamically stored value there.  Instead, they record the value in
       a hidden internal state variable in the transistor.  V_NFETX and
       V_PFETX are provided as a minimal-frills transistor, in case the
       fancier transistors are too "smart" to work properly in your
       circuit.  Also, since the programs for V_NFETX and V_PFETX are
       simple they will simulate faster, especially in hierarchical
       designs.

   8.  V_NFETX and V_PFETX can pass weakly driven signals as well as
       strongly driven ones.  As a result, you can use them to simulate
       NMOS circuits.  Use V_NFETX for enhancement devices, and the
       standard TIE gate as a depletion (pull-up) device.  The fancier
       transistors will not work in this way because the the weak one from
       the TIE gate will not be able to overpower the weak one or zero
       being driven by the transistor's capacitance.

   9.  The following additional gates are provided in the VLSI library:

       V_BUF      VLSI buffer.  This gate acts like a non-inverting
                  buffer made out of V_NFETX and V_PFETX transistors.  That
                  is, it does not affect its input pin in any way, but the
                  output pin gets a copy of the most recently driven input
                  value.  You can use V_BUF to get a visible indication of
                  what value is stored on the hidden gate capacitances of
                  V_NFETX's and V_PFETX's on the same wire.

       V_INV      VLSI inverter.  This gate acts like a usual digital
                  inverter.  The difference between V_INV and the standard
                  INV gate is that V_INV's input is dynamic in the same way
                  as the V_NFET and V_PFET transistors' gates.  V_INV
                  weakly drives its input according to the capacitively
                  stored value.  It is provided simply as a shorthand to
                  drawing the two transistors.

       V_AND, V_OR, V_NAND, V_NOR
                  VLSI AND, OR, NAND, and NOR gates.  These gates have
                  capacitive inputs, just as in V_INV.

       V_TRANS, V_TRANSN
                  CMOS transmission gates (with active-high or active-low
                  control).  These gates act much like V_NFET's and
                  V_PFET's connected together to form the traditional CMOS
                  "pass" gate.  They require only a single control input,
                  generating the inverted control signal internally.  For
                  V_TRANS, the gate conducts ones or zeros in either
                  direction whenever the control is one, and is an open
                  circuit when the control is zero.  For V_TRANSN, the gate
                  conducts only when the control is zero.  If the gates get
                  confused and you need a directional transmission gate,
                  you can use the 74125 and 74126 "tri-state buffer" gates
                  from the library.

       V_CSRL     Complementary Set-Reset Logic element.  This is a one-bit
                  latch designed to emulate a CSRL gate.  If the "latch"
                  input is high, then a zero on the upper data input clears
                  the latch and a zero on the lower data input sets the
                  latch.  If the latch input is low, or if both data inputs
                  are high or undriven, the latch remembers its previous
                  value.  The result is undefined if the latch is high and
                  both data inputs are low.  The latched value and its
                  complement are driven on the outputs.

       V_CSRL0    To get a complete two-phase master-slave flip-flop,
                  connect two V_CSRL's in series.  The V_CSRL0 gate is
                  electrically equivalent to V_CSRL, but with the control
                  pin in a different place so that parallel "phase 1" and
                  "phase 2" clock busses can be used.

       V_CSRL2    Two-bit parallel CSRL latch.

       V_CSRL4    Four-bit parallel CSRL latch.

       V_CSRLN    CSRL latch with active-low control.


Email
lazzaro@cs.berkeley.edu
Phone
(510) 643 4005
SMail
UC Berkeley / CS Division / 387 Soda Hall / Berkeley CA 94720