CMOS Inverter Delay Calculator
Calculate switching time, propagation delay, and power dissipation for a CMOS inverter driving a capacitive load.
How to Use
- Enter the supply voltage.
- Enter the load capacitance the output drives: input pins, traces and any cable.
- Enter the driver's on-resistance: a few tens of ohms for typical logic outputs, kilohms for a weak drive or pull-up.
- Enter the switching frequency and read the delay, edge time, dynamic power and the highest frequency the output can swing fully at.
Worked Example
A 3.3 V logic output with about 50 Ω of on-resistance drives 10 pF at 50 MHz. τ = 50 Ω × 10 pF = 0.5 ns, so the propagation delay is 0.69 × 0.5 = 0.35 ns and the 10–90% edge 2.2 × 0.5 = 1.1 ns. Dynamic power: 10 pF × 3.3² × 50 MHz = 5.4 mW.
Highest full-swing frequency: 1 ÷ (2 × 1.1 ns) ≈ 455 MHz, so 50 MHz is comfortable. Swap in a 1 kΩ pull-up and the edge stretches to 22 ns, limiting the line to about 23 MHz.
The common mistake: forgetting that power scales with the square of voltage. The same load at 5 V costs 2.3 times as much as at 3.3 V, not 1.5 times.
Show Work
Formulas
History of the CMOS Inverter
Frank Wanlass invented the CMOS inverter at Fairchild Semiconductor in 1963, pairing an N-channel and P-channel MOSFET in a complementary configuration. His key insight: only one transistor conducts at a time in steady state, so the static current is essentially zero — just leakage. Dynamic power (charging and discharging capacitance) dominates instead, which is why CMOS became the foundation of every low-power digital IC from the 1970s onward.
The classic delay formula tp = 0.69·R_on·CL comes from first-order RC analysis: the inverter's pull-up or pull-down transistor acts as a resistor driving the load capacitance, and 0.69 = ln(2) is the time to reach the 50% crossing point. In practice, R_on varies with gate-source voltage (hence saturation and triode regions), and more accurate α-power-law models (Sakurai 1990) are used in modern EDA tools.
Dennard scaling (1974) predicted that as MOSFETs shrink, gate capacitance shrinks proportionally while speed increases — enabling Moore's Law's doubling cadence through the 1990s. Dennard scaling broke down around 2005 when leakage currents (sub-threshold, gate-oxide tunneling) stopped shrinking, forcing the industry to multi-core architectures and FinFET transistors. But the fundamental tp = 0.69·Ron·CL relationship still governs every internal logic cell in a modern SoC.
About This Calculator
Enter supply voltage Vdd, load capacitance CL, transistor on-resistance R_on, and switching frequency. The tool returns propagation delay tp = 0.69·R_on·CL (50% to 50%), 10-90% rise/fall time 2.2·R_on·CL, dynamic power P = CL·Vdd²·f (the classic CMOS dynamic-power formula), and a rough maximum frequency estimate.
Note: this is an idealized single-stage analysis. Real chip power includes short-circuit current (during transitions when both transistors conduct briefly), leakage current (significant in modern small-geometry processes), and clock-tree distribution power. For full SoC power analysis, use EDA tools like Synopsys PrimePower or Cadence Voltus. Everything runs client-side; no values leave your browser.
Frequently Asked Questions
What sets the delay?
The driver's on-resistance charging the load capacitance. The output crosses 50% after 0.69·R·C (ln 2) and takes 2.2·R·C (ln 9) to go from 10% to 90%.
How is the dynamic power worked out?
Each full cycle charges the load to Vdd and empties it again, costing C·Vdd² of energy, so power is C·Vdd²·f. 10 pF at 3.3 V and 50 MHz is 5.4 mW. Data lines that toggle less often use proportionally less.
Why does supply voltage matter so much?
Power goes with the square of the voltage. Dropping from 3.3 V to 1.8 V cuts the dynamic power of the same load by 70%, which is why chip core voltages keep falling.
What on-resistance should I use?
Check the output's drive strength in the datasheet: output voltage drop ÷ output current gives a rough figure. Typical push-pull logic outputs are tens of ohms; a 1 kΩ pull-up into 10 pF gives 22 ns edges.
Is this the whole power?
No. Real chips also draw short-circuit current while both transistors conduct during an edge, leakage current, and the charging of their own internal capacitance, which datasheets list as CPD.
How do I use the CMOS Inverter Delay Calculator?
Simply type your numbers and read the result, which refreshes the instant you change something. There is nothing to submit and nothing to wait for.
Do I need to install or sign up for anything?
Not at all — it runs in the browser with nothing to install and no account. After it loads once, it even works without an internet connection.
Is my information private?
Yes. Everything happens in your browser. Nothing you type is sent to a server or saved anywhere.
Common Use Cases
Driving a Cable
A 25 Ω output into 100 pF of cable gives 5.5 ns edges and, at 20 MHz and 3.3 V, 21.8 mW just for that load.
Pull-Up Lines
A 1 kΩ pull-up into 10 pF has 22 ns edges, so it cannot swing fully above about 23 MHz.
Low-Voltage Logic
A 1.8 V output with 200 Ω into 2 pF switches in 0.88 ns and uses 3.2 mW at 500 MHz.
Clock Distribution
A clock toggles every cycle, so its CV²f power is the full figure; clock buffers are sized so edges stay a small fraction of the period.
Last updated: