Build a self-biased 10 µA current sink. It must hold within 3 % from 1.62 V to 1.98 V, move little with the supply, and start from 0 V.
Circuit context: A self-biased 10 µA current reference with its start-up circuit, feeding an amplifier's bias node held at 0.9 V. Design the bias reference of an amplifier that runs from a 1.8 V ± 10 % supply.
An amplifier needs a 10 µA bias that does not follow the supply. The supply is 1.8 V ± 10 %.
Build the subcircuit iref. Its node out sinks the bias current, and the grader holds that node at 0.9 V.
The current must stay between 9.7 and 10.3 µA at 1.62, 1.8 and 1.98 V. The supply sensitivity S must be at most 0.15.
The reference must also power up. The supply ramps from 0 V in 10 µs, and every node starts at 0 V.
The starter is the base design of lesson E4.4 with an output device. It gives 9.2 µA, its S is 0.28, and it stays off.
Retuning the lesson's designs to 10 µA is not enough. None of them reaches S = 0.15.
What the Grader Measures
Your subcircuit iref (ports vdd, out) under the grader's own bench, one run per corner (tt, ss, ff):
| Check | Corner | Rule |
|---|---|---|
| Output current at 1.62 V | tt | 9.7 µA to 10.3 µA |
| Output current at 1.8 V | tt | 9.7 µA to 10.3 µA |
| Output current at 1.98 V | tt | 9.7 µA to 10.3 µA |
| Supply sensitivity S | tt | ≤ 0.15 |
| Supply current at 1.8 V | tt | ≤ 25 µA |
| Current after the ramp ÷ DC value | tt | 0.98 to 1.02 |
| Current after the ramp ÷ DC value | ss | 0.98 to 1.02 |
| Current after the ramp ÷ DC value | ff | 0.98 to 1.02 |
Full specification
.subckt iref vdd out
* resistors, capacitors and SKY130 1.8 V FETs (sky130_fd_pr__nfet_01v8, sky130_fd_pr__pfet_01v8)
.ends
| Check | Corner | Rule | How the grader measures it |
|---|---|---|---|
| Output current | tt | 9.7 µA to 10.3 µA | dc: the current into out at 1.62, 1.8 and 1.98 V of supply |
| Supply sensitivity S | tt | ≤ 0.15 | (I at 1.98 V − I at 1.62 V) / (I at 1.8 V) / 0.2 |
| Start-up | tt, ss, ff | 0.98 to 1.02 | tran: the current 100 µs after the ramp, over the DC value at 1.8 V |
| Supply current | tt | ≤ 25 µA | dc at 1.8 V: the current out of the supply |
The start-up run is tran 0.1u 120u uic with the supply written as PWL(0 0 10u 1.8). The grader reads the current at 110 µs.
Each corner compares the started current with its own DC value. The grader runs at 27 °C.
Examples
Example 1
Long Devices, K = 16 and a Wide PMOS Mirror, with a Start-Up- Design
.subckt iref vdd out XP1 a b vdd vdd sky130_fd_pr__pfet_01v8 W=40 L=4 XP2 b b vdd vdd sky130_fd_pr__pfet_01v8 W=40 L=4 XN1 a a 0 0 sky130_fd_pr__nfet_01v8 W=8 L=4 XN2 b a s 0 sky130_fd_pr__nfet_01v8 W=128 L=4 RS s 0 14.5k XNO out a 0 0 sky130_fd_pr__nfet_01v8 W=8 L=4 XS1 x 0 vdd vdd sky130_fd_pr__pfet_01v8 W=0.5 L=20 XS2 b x 0 0 sky130_fd_pr__nfet_01v8 W=1 L=1 XS3 x a 0 0 sky130_fd_pr__nfet_01v8 W=2 L=1 .ends
- Measured by the grader
Output current at 1.62 V: 9.945 µA (needs 9.7 µA to 10.3 µA) Output current at 1.8 V: 10.03 µA (needs 9.7 µA to 10.3 µA) Output current at 1.98 V: 10.11 µA (needs 9.7 µA to 10.3 µA) Supply sensitivity S: 0.08257 (needs ≤ 0.15) Supply current at 1.8 V: 20.6 µA (needs ≤ 25 µA) Current after the ramp ÷ DC value: 1 (needs 0.98 to 1.02) Current after the ramp ÷ DC value at ss: 1 (needs 0.98 to 1.02) Current after the ramp ÷ DC value at ff: 1 (needs 0.98 to 1.02)
The current moves 1.7 % over the whole supply range, and the reference is running 100 µs after power-up in every corner.
Example 2
The Same Reference Without Its Start-Up- Design
.subckt iref vdd out XP1 a b vdd vdd sky130_fd_pr__pfet_01v8 W=40 L=4 XP2 b b vdd vdd sky130_fd_pr__pfet_01v8 W=40 L=4 XN1 a a 0 0 sky130_fd_pr__nfet_01v8 W=8 L=4 XN2 b a s 0 sky130_fd_pr__nfet_01v8 W=128 L=4 RS s 0 14.5k XNO out a 0 0 sky130_fd_pr__nfet_01v8 W=8 L=4 .ends
- Measured by the grader
Output current at 1.62 V: 9.945 µA (needs 9.7 µA to 10.3 µA) Output current at 1.8 V: 10.03 µA (needs 9.7 µA to 10.3 µA) Output current at 1.98 V: 10.11 µA (needs 9.7 µA to 10.3 µA) Supply sensitivity S: 0.08257 (needs ≤ 0.15) Supply current at 1.8 V: 19.96 µA (needs ≤ 25 µA) Current after the ramp ÷ DC value: 268.2 n (needs 0.98 to 1.02) Current after the ramp ÷ DC value at ss: 201 n (needs 0.98 to 1.02) Current after the ramp ÷ DC value at ff: 374 n (needs 0.98 to 1.02)
Every dc check passes, because dc finds the working state. After the ramp the output carries almost nothing.
Example 3
A Resistor Bias Trimmed to 10 µA- Design
.subckt iref vdd out RB vdd a 106.5k XN1 a a 0 0 sky130_fd_pr__nfet_01v8 W=8 L=4 XNO out a 0 0 sky130_fd_pr__nfet_01v8 W=8 L=4 .ends
- Measured by the grader
Output current at 1.62 V: 8.513 µA (needs 9.7 µA to 10.3 µA) Output current at 1.8 V: 10.05 µA (needs 9.7 µA to 10.3 µA) Output current at 1.98 V: 11.6 µA (needs 9.7 µA to 10.3 µA) Supply sensitivity S: 1.536 (needs ≤ 0.15) Supply current at 1.8 V: 9.987 µA (needs ≤ 25 µA) Current after the ramp ÷ DC value: 1.001 (needs 0.98 to 1.02) Current after the ramp ÷ DC value at ss: 1.001 (needs 0.98 to 1.02) Current after the ramp ÷ DC value at ff: 1.001 (needs 0.98 to 1.02)
It starts by itself and is right at 1.8 V. The current follows the supply, so both ends of the range fail.
Constraints and Contract
- The netlist defines .subckt iref vdd out ... .ends exactly once, with these ports in this order.
- Inside iref: resistors, capacitors and X instances of sky130_fd_pr__nfet_01v8 or sky130_fd_pr__pfet_01v8. No sources.
- The grader supplies vdd and holds out at 0.9 V. It writes the .lib line for each corner itself.
Hints
Hint 1
The supply reaches the current through finite output resistance. Which devices have a drain voltage that follows the supply?
Hint 2
An error in the mirror goes around the loop and comes back larger. More voltage across RS lowers the gain around the loop.
Hint 3
Check the lowest supply. A PMOS diode with a large VSG leaves little voltage for the NMOS below it at 1.62 V.
Hint 4
An op or dc run finds the working state. Only a transient from 0 V shows whether the reference reaches it.
What the grader checks: The grader measures the output current at three supply voltages and after one 10 µs power-up ramp, at 27 °C. All numbers come from ngspice 42 with the pinned SKY130 models, measured 2026-11-12. It does not check slow ramps, temperature, resistor spread or mismatch. A real design review checks all of them.
Run simulates your netlist as written, with your own analyses and measurements. Submit sends it to the grader, which keeps only the subcircuit iref and makes 3 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