Size one NMOS that turns 12 µA into at least 200 µS of transconductance. It must still reach 2 GHz, at tt and at ss.
Circuit context: The input transistor of an amplifier stage, fed 12 µA with its drain held at 0.9 V. Size an input device for a g_m target on a fixed current without giving up its speed.
An amplifier stage has 12 µA for its input transistor. From that current the device must give a transconductance of at least 200 µS. It must also stay fast: its transit frequency f_T must be at least 2 GHz.
The transit frequency is the frequency where the signal current in the drain equals the signal current into the gate. A larger gate has more capacitance, takes more gate current, and reaches that point sooner.
Choose W and L for the NMOS in the subcircuit gmc. The grader feeds it 12 µA and holds its drain at 0.9 V. It measures g_m and f_T at the tt and ss corners.
The starter was sized with the square law. Run it first. Then size the device from measured g_m/I_D against current density, as lesson E2.5 does. The length is yours to choose.
What the Grader Measures
Your subcircuit gmc (ports d, g) under the grader's own bench, one run per corner (tt, ss):
| Check | Corner | Rule |
|---|---|---|
| g_m at tt | tt | ≥ 200 µS |
| g_m at ss | ss | ≥ 200 µS |
| f_T at tt | tt | ≥ 2 GHz |
| f_T at ss | ss | ≥ 2 GHz |
Full specification
.subckt gmc d g
* sky130_fd_pr__nfet_01v8 instances only; the source and body are ground (node 0)
.ends
| Check | Rule | How the grader measures it |
|---|---|---|
| g_m at tt and at ss | ≥ 200 µS | ac at 1 kHz, 12 µA in the drain at 0.9 V: the drain signal current for 1 V on the gate |
| f_T at tt and at ss | ≥ 2 GHz | the same ac sweep, up to 1 THz: where the drain signal current equals the gate signal current |
The grader runs its bench twice, with .lib sky130 tt and .lib sky130 ss, at 27 °C. The bias current is 12 µA in both.
Examples
Example 1
The Starter: W = 4.2, L = 0.5 from the Square Law- Design
.subckt gmc d g XM1 d g 0 0 sky130_fd_pr__nfet_01v8 W=4.2 L=0.5 .ends
- Measured by the grader
g_m at tt at tt: 181.1 µS (needs ≥ 200 µS) g_m at ss at ss: 179.2 µS (needs ≥ 200 µS) f_T at tt at tt: 2.382 GHz (needs ≥ 2 GHz) f_T at ss at ss: 2.402 GHz (needs ≥ 2 GHz)
Fast enough, but 9.5 % short of 200 µS. Near threshold the square law promises more g_m than the device gives.
Example 2
W = 7, L = 0.5 from the L = 0.5 Curve- Design
.subckt gmc d g XM1 d g 0 0 sky130_fd_pr__nfet_01v8 W=7 L=0.5 .ends
- Measured by the grader
g_m at tt at tt: 209.4 µS (needs ≥ 200 µS) g_m at ss at ss: 208 µS (needs ≥ 200 µS) f_T at tt at tt: 1.724 GHz (needs ≥ 2 GHz) f_T at ss at ss: 1.725 GHz (needs ≥ 2 GHz)
The chart gives enough g_m, but the gate is now too large. At L = 0.5 µm no single width meets both limits.
Example 3
W = 60, L = 0.5- Design
.subckt gmc d g XM1 d g 0 0 sky130_fd_pr__nfet_01v8 W=60 L=0.5 .ends
- Measured by the grader
g_m at tt at tt: 266.9 µS (needs ≥ 200 µS) g_m at ss at ss: 266.5 µS (needs ≥ 200 µS) f_T at tt at tt: 291.3 MHz (needs ≥ 2 GHz) f_T at ss at ss: 292.9 MHz (needs ≥ 2 GHz)
Weak inversion gives the most g_m per ampere: 22 per volt here. The gate is 30 µm² and the device is seven times too slow.
Constraints and Contract
- The netlist defines .subckt gmc d g ... .ends exactly once, with these ports in this order.
- Inside gmc: X instances of sky130_fd_pr__nfet_01v8 with W= and L= (m= optional), comments and .param lines. No sources and no resistors.
- The grader writes the .lib line for each corner itself. Everything outside gmc is replaced by its bench.
Hints
Hint 1
Start with the ratio the spec fixes: g_m/I_D = 200 µS / 12 µA. Add some margin before you read a density.
Hint 2
At a fixed current, more width raises g_m/I_D and adds gate capacitance. Find the width where f_T crosses 2 GHz.
Hint 3
If no width meets both limits at one length, change the length. A shorter device reaches the same g_m/I_D at a higher density.
Hint 4
The lesson drew its chart for W = 2 µm at tt. Confirm your size on the bench at both corners.
What the grader checks: The grader places gmc under its own bench once per corner, tt and ss, at 27 °C. It feeds 12 µA, holds the drain at 0.9 V, and reads g_m and f_T from one ac sweep. Engine: ngspice 42 (libngspice0 42+ds-3build1) with the pinned SKY130 models, measured 2026-11-12 with tools/spice-judge.mjs. It proves those two numbers at this bias. It does not check output resistance, noise, mismatch, temperature or the other corners.
Run simulates your netlist as written, with your own analyses and measurements. Submit sends it to the grader, which keeps only the subcircuit gmc and makes 2 ngspice runs. 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 1 steps and the final capstone