3.1.15Boolean Algebra & Logic Gates

Logic gate propagation delay

1,943 words9 min readdifficulty · medium6 backlinks

WHAT is propagation delay?

WHY 50%? Because the "logic threshold" (the voltage the next gate decides is a 1 vs a 0) sits roughly halfway. Measuring at 50% is a consistent, technology-independent reference point.

Note the difference from transition (rise/fall) time: that is measured between the 10% and 90% points of one signal and describes how sharp the edge is — a different quantity.

Figure — Logic gate propagation delay

WHY does delay exist? (Derivation from first principles)

A CMOS gate output drives a load capacitance CLC_L (wire capacitance + input capacitances of the gates it feeds). When the output switches, a transistor must move charge onto or off that capacitor through its channel resistance RR.

Model the switching transistor as a resistor RR charging capacitor CLC_L from supply VDDV_{DD}. By KCL, the current charging the cap equals the current through the resistor:

CLdVdt=VDDVRC_L\frac{dV}{dt} = \frac{V_{DD}-V}{R}

Why this step? The same current flows through RR and into CLC_L (series path), and a capacitor's current is CdV/dtC\,dV/dt by definition.

Separate variables and integrate from V=0V=0:

dVVDDV=dtRCL    ln(VDDV)=tRCL+k\frac{dV}{V_{DD}-V} = \frac{dt}{RC_L} \;\Rightarrow\; -\ln(V_{DD}-V) = \frac{t}{RC_L} + k

Applying V(0)=0V(0)=0 gives k=lnVDDk=-\ln V_{DD}, so:

V(t)=VDD(1et/RCL)\boxed{V(t) = V_{DD}\left(1 - e^{-t/RC_L}\right)}

Why this step? This is the standard exponential charging curve — the output does not jump; it eases up to VDDV_{DD} with time constant τ=RCL\tau = RC_L.

Now find tpLHt_{pLH}: the time to reach the 50% point V=VDD/2V = V_{DD}/2:

VDD2=VDD(1et/RCL)    et/RCL=12    t=RCLln2\frac{V_{DD}}{2} = V_{DD}(1-e^{-t/RC_L}) \;\Rightarrow\; e^{-t/RC_L} = \tfrac12 \;\Rightarrow\; t = RC_L\ln 2

Consequences you can now predict (Forecast-then-Verify):

  • Drive more gates → larger CLC_Lmore delay. (Fan-out costs time.)
  • Use a bigger/stronger transistor → smaller RRless delay (but bigger input cap for the previous gate!).
  • Longer wires → more CLC_L → more delay.

Delays ADD along a path


Worked examples


Common mistakes


Active recall

Recall Try before revealing
  • Define tpdt_{pd} and the reference points used.
  • Derive tpd=0.69RCLt_{pd}=0.69RC_L from the RC charging equation.
  • Why does fan-out increase delay?
  • What causes a static hazard glitch?
What is propagation delay of a logic gate?
The time from the input crossing 50% of its voltage swing to the output crossing 50%, in response to that change.
Why is delay measured at the 50% point?
Because the logic threshold (where the next gate decides 0 vs 1) is ~halfway, giving a consistent technology-independent reference.
What are t_pLH and t_pHL?
Delays for the output going Low→High and High→Low respectively.
Derive the delay formula: what equation governs the output rising?
C_L dV/dt = (V_DD − V)/R, giving V(t)=V_DD(1−e^(−t/RC_L)).
What is the propagation delay in terms of R and C_L?
t_pd ≈ 0.69·R·C_L (since ln2 ≈ 0.693).
Why does delay increase with fan-out?
More gates driven = larger load capacitance C_L, and delay ∝ C_L.
How do delays combine through a chain of gates?
They ADD along a path; the longest (critical) path sets the max clock frequency f_max = 1/t_critical.
Difference between propagation delay and rise time?
Propagation delay is 50%(in)→50%(out) between two signals; rise/fall time is 10%→90% of ONE signal (edge sharpness).
What is a static hazard (glitch)?
A brief false output pulse caused by unequal path delays making inputs to a gate arrive at different times.
Does making the driving transistor bigger always speed the circuit?
No — smaller R helps, but its larger input capacitance slows the previous stage; it's a trade-off.

Recall Feynman: explain to a 12-year-old

Imagine filling a cup with water through a thin straw. Even if you decide instantly to fill it, the water still takes a moment to rise to the halfway line. A logic gate is like that: the "water" is electric charge, the "cup" is a tiny capacitor, and the "straw" is the transistor's resistance. The gate knows the answer immediately, but it takes a little time to actually push the voltage up to where the next gate can read it. Chain many cups and straws together, and the waiting times add up — that's why fast computers work hard to keep those straws fat and the cups small.

Connections

Concept Map

takes time

measured at

distinct from

averages

driven through

forms RC circuit

solve at 50%

grows with

from fan-out and wires

delays sum along

limits

Transistors charge capacitors

Propagation delay tpd

50% voltage level

Transition time 10%-90%

tpLH and tpHL

Load capacitance CL

Channel resistance R

Charging curve V=VDD 1-e^-t/RC

tpd approx 0.69 R CL

More gates or longer wires

Critical path

Max clock frequency

Hinglish (regional understanding)

Intuition Hinglish mein samjho

Dekho, logic gate ka truth table batata hai output kya hoga, lekin ye kabhi nahi batata ki output kab milega. Real hardware me har gate transistors se bana hota hai, aur transistor ke through ek chhoti capacitor charge karni padti hai. Capacitor ko charge hone me time lagta hai — bilkul jaise patli straw se glass bharne me time lagta hai. Isi waiting time ko hum propagation delay (tpdt_{pd}) kehte hain, aur ye input ke 50% se output ke 50% tak measure hota hai.

Formula derive karna simple hai: current RR aur CLC_L dono me se same behta hai, to CLdV/dt=(VDDV)/RC_L\,dV/dt=(V_{DD}-V)/R. Isko solve karo to output exponential tarah upar chadhta hai: V(t)=VDD(1et/RCL)V(t)=V_{DD}(1-e^{-t/RC_L}). Jab output 50% pe pahunchta hai tab t=RCLln2t=RC_L\ln2, yaani tpd0.69RCLt_{pd}\approx 0.69\,RC_L. Matlab jitni zyada load capacitance (zyada gates drive karo, lambe wires), utni zyada delay. Ye baat yaad rakho — fan-out free nahi hota, time lagta hai.

Aur ek important cheez: agar gates chain me lage hain, to delays add hoti hain, average nahi. Sabse lamba path (critical path) hi decide karta hai ki chip kitni fast clock chala sakti hai: fmax=1/tcriticalf_{max}=1/t_{critical}. Jab do inputs alag-alag time pe pahunchte hain (unequal delay), to ek chhota galat pulse aa sakta hai jise glitch ya static hazard kehte hain. Isliye engineers transistor sizing aur wiring bahut dhyan se karte hain — logic sahi ho ye kaafi nahi, timing bhi sahi honi chahiye.

Go deeper — visual, from zero

Test yourself — Boolean Algebra & Logic Gates

Connections