Measure one transistor at its bias and write its small-signal model. The model must match the real stage's gain and −3 dB point in three surroundings.
Circuit context: A SKY130 common-source stage (W/L = 3/0.18 µm, 0.70 V on the gate, 50 kΩ to 1.8 V) and the linear model that replaces its transistor. Write the small-signal model a hand analysis of this stage would start from.
A common-source stage uses one SKY130 NMOS with W/L = 3/0.18 µm. Its gate sits at 0.70 V and a 50 kΩ resistor joins its drain to 1.8 V.
Write the transistor's small-signal model as the subcircuit nmodel. Use one G source, resistors and capacitors. The source and body are node 0.
The grader puts your model where the transistor was and compares the two circuits. It does this in three surroundings, listed below.
A model fitted to one surrounding fails the other two. Each element must hold the value the transistor has at this bias.
What the Grader Measures
Your subcircuit nmodel (ports g, d) under the grader's own bench, one run per corner (tt):
| Check | Corner | Rule |
|---|---|---|
| Gain, model ÷ stage (1 kΩ source) | tt | 0.98 to 1.02 |
| Gain, model ÷ stage (extra load) | tt | 0.98 to 1.02 |
| −3 dB point, model ÷ stage (1 kΩ source) | tt | 0.95 to 1.05 |
| −3 dB point, model ÷ stage (2 MΩ source) | tt | 0.95 to 1.05 |
| −3 dB point, model ÷ stage (2 MΩ source, extra load) | tt | 0.95 to 1.05 |
Full specification
.subckt nmodel g d
* G sources, resistors and capacitors only; source and body are node 0
.ends
| Surrounding | What changes | What it exposes |
|---|---|---|
| 1 | 1 kΩ source | The gain and the output node's −3 dB point |
| 2 | 2 MΩ source | The capacitance seen at the gate |
| 3 | 2 MΩ source, extra 25 kΩ load | How that capacitance changes with the gain |
| Check | Surrounding | Rule | How the grader measures it |
|---|---|---|---|
| Gain | 1 | 0.98 to 1.02 | Model gain divided by stage gain, at 1 kHz |
| Gain | 3 | 0.98 to 1.02 | As above, with the extra load |
| −3 dB point | 1 | 0.95 to 1.05 | Model frequency divided by stage frequency |
| −3 dB point | 2 | 0.95 to 1.05 | As above, with the 2 MΩ source |
| −3 dB point | 3 | 0.95 to 1.05 | As above, with source and load |
The extra load is 25 kΩ behind a large blocking capacitor. It loads the signal and leaves the bias alone.
The grader runs at the tt corner and 27 °C. The −3 dB point is where the gain falls to 1/√2 of its 1 kHz value.
Examples
Example 1
The Starter: Placeholder Values, No Capacitors- Design
.subckt nmodel g d G1 d 0 g 0 100u RO d 0 100k .ends
- Measured by the grader
Gain, model ÷ stage (1 kΩ source): 0.4652 (needs 0.98 to 1.02) Gain, model ÷ stage (extra load): 0.5124 (needs 0.98 to 1.02) −3 dB point, model ÷ stage (1 kΩ source): 1.192 (needs 0.95 to 1.05) −3 dB point, model ÷ stage (2 MΩ source): 6.291 (needs 0.95 to 1.05) −3 dB point, model ÷ stage (2 MΩ source, extra load): 8.811 (needs 0.95 to 1.05)
The gain is less than half of the stage's. Without capacitors the gate node has no pole at all.
Example 2
Right g_{m} and r_{o}, but No C_{gd}- Design
.subckt nmodel g d G1 d 0 g 0 183.5u RO d 0 181.6k CGS g 0 3.912f CDB d 0 1.102f .ends
- Measured by the grader
Gain, model ÷ stage (1 kΩ source): 1.004 (needs 0.98 to 1.02) Gain, model ÷ stage (extra load): 1.005 (needs 0.98 to 1.02) −3 dB point, model ÷ stage (1 kΩ source): 1.002 (needs 0.95 to 1.05) −3 dB point, model ÷ stage (2 MΩ source): 2.237 (needs 0.95 to 1.05) −3 dB point, model ÷ stage (2 MΩ source, extra load): 1.552 (needs 0.95 to 1.05)
Both gains match and so does the first −3 dB point. Behind 2 MΩ the model is more than twice too fast.
Example 3
The Full Model, Read at the Stage's Bias- Design
.subckt nmodel g d G1 d 0 g 0 183.5u RO d 0 181.6k CGS g 0 3.18f CGD g d 0.737f CDB d 0 0.366f .ends
- Measured by the grader
Gain, model ÷ stage (1 kΩ source): 1.004 (needs 0.98 to 1.02) Gain, model ÷ stage (extra load): 1.005 (needs 0.98 to 1.02) −3 dB point, model ÷ stage (1 kΩ source): 1.001 (needs 0.95 to 1.05) −3 dB point, model ÷ stage (2 MΩ source): 0.9974 (needs 0.95 to 1.05) −3 dB point, model ÷ stage (2 MΩ source, extra load): 0.9976 (needs 0.95 to 1.05)
All five ratios sit within 0.5 % of 1. The small gain error is the contact resistance inside the SKY130 device.
Constraints and Contract
- The netlist defines .subckt nmodel g d ... .ends exactly once, with these ports in this order.
- Inside nmodel: G sources, resistors and capacitors, with comments and .param lines. No transistor.
- The grader supplies the source resistor, the 50 kΩ resistor, the loads and the real stage.
Hints
Hint 1
Read gm and gds from the operating point of this stage. Check where its drain sits before you copy a bench from the lesson.
Hint 2
Surrounding 2 has a slow gate node. The gate sees more than Cgs: one capacitor joins it to a node that moves the other way.
Hint 3
Cgg is everything on the gate, Cgd included. Measure Cgd on its own and subtract it before you place Cgs.
What the grader checks: The grader compares your model with the transistor stage in three surroundings at tt and 27 °C. All numbers come from ngspice 42 with the pinned SKY130 models, measured 2026-11-12. It checks small-signal gain and the first −3 dB point only. It does not check large signals, noise, or frequencies above the first pole.
Run simulates your netlist as written, with your own analyses and measurements. Submit sends it to the grader, which keeps only the subcircuit nmodel 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 1 steps and the final capstone