G9.7.1 Size the Buffer Chain

HardOptimizengspice 42 + SKY130 · serverReview the topic

Size a non-inverting SKY130 buffer so that a unit inverter can drive 216 unit inverters through it within 240 ps.

Circuit context: A buffer chain that carries a clock enable from one small gate to 64 flip-flop inputs. Size the buffer between a small gate and a large load so that it meets a delay limit.

A clock enable leaves a small gate and must reach 216 flip-flop inputs. The small gate is a unit inverter, and it cannot drive that load fast on its own. You place a buffer between them.

This is an Optimize problem. Build the block buf from inverters, each a whole number of unit inverters. The buffer must not invert: out follows in. Choose how many inverters it has, and their sizes.

The check drives in through a fixed unit inverter and loads out with 216 unit inverters. The mean delay of both edges, from the source to out, must be 240 ps or less.

What the Grader Measures

Your subcircuit buf (ports in, out, vdd) under the grader's own bench, one run per corner (tt):

CheckCornerRule
Mean delay, source to outtt≤ 240 ps
out after the source risestt≤ 180 mV
out after the source fallstt≥ 1.62 V
Full specification
The block that is checked (keep the name and the ports)
.subckt buf in out vdd
* inverters of sky130_fd_pr__nfet_01v8 (W=1) and sky130_fd_pr__pfet_01v8 (W=2), with m= for size
.ends
CheckRuleHow it is measured
Mean delay≤ 240 psthe mean of two delays: source rising to out falling, and source falling to out rising, at 0.9 V
out after the source rises≤ 0.18 Vthe level of out 2.35 ns after the source rises
out after the source falls≥ 1.62 Vthe level of out 2.4 ns after the source falls

The fixed driver inverts, so out falls when the source rises. The check runs once, at the SKY130 typical corner (tt), 27 °C and a 1.8 V supply.

Examples

Example 1

The Starter: Two Unit Inverters
Design
.subckt buf in out vdd
* stage 1: 1 unit inverter in parallel
XN1 n1 in 0 0 sky130_fd_pr__nfet_01v8 W=1 L=0.15
XP1 n1 in vdd vdd sky130_fd_pr__pfet_01v8 W=2 L=0.15
* stage 2: 1 unit inverter in parallel
XN2 out n1 0 0 sky130_fd_pr__nfet_01v8 W=1 L=0.15
XP2 out n1 vdd vdd sky130_fd_pr__pfet_01v8 W=2 L=0.15
.ends
Measured by the grader
Mean delay, source to out: 1.486 ns (needs ≤ 240 ps)
out after the source rises: 595.1 mV (needs ≤ 180 mV)
out after the source falls: 1.194 V (needs ≥ 1.62 V)

The second inverter drives 216 times its own input. The mean delay is 1486.0 ps, and out has not settled when it is read.

Example 2

Sizes 1 and 216
Design
.subckt buf in out vdd
* stage 1: 1 unit inverter in parallel
XN1 n1 in 0 0 sky130_fd_pr__nfet_01v8 W=1 L=0.15
XP1 n1 in vdd vdd sky130_fd_pr__pfet_01v8 W=2 L=0.15
* stage 2: 216 unit inverters in parallel
XN2 out n1 0 0 sky130_fd_pr__nfet_01v8 W=1 L=0.15 m=216
XP2 out n1 vdd vdd sky130_fd_pr__pfet_01v8 W=2 L=0.15 m=216
.ends
Measured by the grader
Mean delay, source to out: 1.345 ns (needs ≤ 240 ps)
out after the source rises: 978.9 mV (needs ≤ 180 mV)
out after the source falls: 1.8 V (needs ≥ 1.62 V)

The last inverter is as large as the load, but the unit inverter before it must drive 216: 1345.2 ps.

Example 3

Two Inverters of Sizes 6 and 36
Design
.subckt buf in out vdd
* stage 1: 6 unit inverters in parallel
XN1 n1 in 0 0 sky130_fd_pr__nfet_01v8 W=1 L=0.15 m=6
XP1 n1 in vdd vdd sky130_fd_pr__pfet_01v8 W=2 L=0.15 m=6
* stage 2: 36 unit inverters in parallel
XN2 out n1 0 0 sky130_fd_pr__nfet_01v8 W=1 L=0.15 m=36
XP2 out n1 vdd vdd sky130_fd_pr__pfet_01v8 W=2 L=0.15 m=36
.ends
Measured by the grader
Mean delay, source to out: 250.1 ps (needs ≤ 240 ps)
out after the source rises: 346 nV (needs ≤ 180 mV)
out after the source falls: 1.8 V (needs ≥ 1.62 V)

Three stages with the driver, each with a fan-out of 6: 250.1 ps.

Constraints and Contract

  • The netlist defines .subckt buf in out vdd ... .ends exactly once, with these ports in this order.
  • Inside buf: X instances of sky130_fd_pr__nfet_01v8 or sky130_fd_pr__pfet_01v8, with comments and .param lines. No sources, resistors or capacitors. Use m= for parallel copies; mult= is refused.
  • The check writes the .lib line itself. Everything outside buf is replaced by its own bench.

Hints

Hint 1

Count the path from the fixed driver to the load. Its electrical effort H is 216, and each inverter has g = 1.

Hint 2

The buffer must have an even number of inverters. With the driver, two inverters give three stages and four give five.

Hint 3

Find the stage count that gives a stage effort closest to 4, among the counts the polarity allows. Then size each stage from the load backwards.

What the grader checks: The check places buf between a fixed unit inverter and 216 unit inverters: 1.8 V, tt, 27 °C, 50 ps source edges. It reads both delays at 0.9 V and the level of out after each edge. The tool is ngspice 42 (libngspice0 42+ds-3build1) with the pinned SKY130 models, through tools/spice-judge.mjs. Other corners and wire loads are outside this problem.

Run simulates your netlist as written, with your own analyses and measurements. Submit sends it to the grader, which keeps only the subcircuit buf and makes 1 ngspice run. The numbers come from ngspice and the SKY130 models the site pins; they prove the measured cases, not every condition a chip meets.

See all 2 steps and the final capstone

Report a problem with this page

Your netlist SPICESaved locally
Ln 1, Col 1Tab indents · Ctrl/⌘ + Enter runs
Not run yet
Run your netlist to see its measurements and the ngspice log here.
ngspice 42 + SKY130 · server