Power-delay product: Foundations — Hardware deep dive

This page builds every symbol the parent note leans on, from absolute zero. If you have never seen a capacitor, an integral, or the letter $α$ , start at line one and walk down. Each tool is earned before it is used.

0 · Before any symbols — what is a "gate flipping"?

A logic gate is a tiny switch that outputs either a HIGH voltage (call it logic-1) or a LOW voltage (logic-0). "Flipping" means its output changes: $0 \to 1$ or $1 \to 0$ . Every time it flips, it moves electric charge around, and moving charge costs energy. The whole topic is about how much energy per flip.

Look at the picture: the output node is a wire. Hanging off that wire is always a little bit of capacitance (next section). Flipping the output = charging or draining that capacitance. Hold this image; everything else decorates it.

1 · Voltage $V$ — the "electrical height"

Picture: a water tank. The water level is the voltage.
Why the topic needs it: logic-1 is a full tank at voltage $V_{D D}$ ; logic-0 is empty at $0$ . Flipping the gate = raising or lowering this water level.

The subscript $D D$ is just historical naming (from the "drain" terminal of transistors). Read it as "the supply rail."

2 · Charge $Q$ and current $i (t)$ — the "water itself" and its "flow rate"

Picture: current is the width of the stream from the tap; charge is the total water collected in the bucket.
The link: total water = flow rate summed over time. Written with the tool of section 6: $Q = \int_{0}^{\infty} i (t) d t$
Why the topic needs it: the parent computes energy from the supply as "voltage times all the charge delivered," so it must first know how much charge moved to fill the output to $V_{D D}$ .

The notation $i (t)$ — the " $(t)$ " — means "current is a function that changes with time." Right after a flip, current is large (tank filling fast); once full, current is zero (nothing more to pour).

3 · Capacitance $C_{L}$ — the "width of the bucket"

Picture: a wide bathtub vs. a thin test tube. Both filled to the same height (voltage), but the wide tub holds more water (charge). Capacitance = the tub's cross-sectional width.
The equation $Q = C_{L} V$ in words: charge stored = (width of tub) × (water height). Wider tub or higher water → more charge.
Why the topic needs it: to flip a gate you fill this tub up to $V_{D D}$ . A bigger $C_{L}$ = bigger tub = more charge to move = more energy per flip. This is the thing that sets the "cup size."

Typical values are tiny: femtofarads, $1 fF = 1 0^{- 15}$ F. See section 8 on these prefixes.

4 · Energy $E$ and Power $P$ — "cup of water" vs "flow rate of water"

This is the pair the whole topic is built on. Keep them separate in your head.

Picture: $E$ = the total litres you used to fill the balloon. $P$ = litres-per-second while the tap is open. Same water, one is a total and one is a rate.
Why the topic needs it: PDP $= P_{a v g} \times t_{p}$ . Multiplying a rate (power) by a time (delay) cancels the "per second" and leaves a pure total: energy. That is the entire reason PDP has units of joules.

5 · Propagation delay $t_{p}$ — "how long the flip takes"

Picture: you open the tap; the water level takes a moment to rise past the "half-full" mark. That moment is $t_{p}$ . See Propagation delay for the full charging curve.
Why the topic needs it: PDP multiplies power by exactly this $t_{p}$ . A slower gate (bigger $t_{p}$ ) usually draws less power, so the product tends to stay put — which is the "honesty" the parent celebrates.

6 · The integral $\int_{0}^{\infty} (\dots) d t$ — "add up all the little bits over time"

The parent uses $\int_{0}^{\infty} V_{D D} i (t) d t$ . Here is that symbol, from zero.

Picture: the current $i (t)$ traces a curve that starts high and fades to zero as the tub fills. Chop the time axis into thin strips of width $d t$ ; each strip is a sliver of charge (flow rate × tiny time). The integral sums all slivers = total charge $Q$ .
Why this tool and not another? We want the total charge delivered, but the flow rate keeps changing. Simple multiplication "rate × time" only works for a constant rate. When the rate varies, the honest way to total it up is to add infinitely many tiny constant-rate pieces — and that is precisely what an integral does.
The one line the parent needs: $\int_{0}^{\infty} i (t) d t = Q = C_{L} V_{D D}$ This just says: the whole area under the current curve equals the total charge that filled the tub to $V_{D D}$ . Because $V_{D D}$ is a constant, it slides outside the sum: $\int_{0}^{\infty} V_{D D} i (t) d t = V_{D D} \int_{0}^{\infty} i (t) d t = V_{D D} Q = C_{L} V_{D D}^{2} .$

7 · Switching frequency $f$ and activity factor $α$ — "how often it flips"

Picture: the clock is a metronome. On some beats you pop the balloon, on others you don't. $α$ is the fraction of beats that get a pop.
Why the topic needs them: average power (a rate) needs to know how often flips happen: $P_{a v g} = α C_{L} V_{D D}^{2} f$ . But PDP deliberately strips $f$ and $α$ away to talk about one single flip — this is the whole point of section 4's "per-event" idea.

Recall Why does PDP throw away

$f$ ? Because PDP is energy per event, not power over time — removing "how often" stops a designer from faking a better number just by running slower. ::: PDP = energy per single transition, frequency-free by design.

8 · The metric prefixes — femto, pico, nano, micro, giga

The parent tosses around $fF$ , $ps$ , $fJ$ , $μ W$ , $GHz$ . These are just size labels:

Picture: a ruler zooming down: nano is a billionth, pico a trillionth, femto a thousand-trillionth.
Why the topic needs it: capacitors are femtofarads, delays are picoseconds, energies are femtojoules. If you can't convert these, the worked examples turn to nonsense. Practice: $10 fF = 10 \times 1 0^{- 15} F$ .

9 · The two derived formulas, now fully legible

Every symbol above is now defined, so the parent's key results read cleanly:

Prerequisite map

Read top to bottom: voltage and charge define the capacitor; the integral turns changing current into total supply energy; energy-vs-power plus delay assemble the final PDP.

Equipment checklist

Test yourself — reveal only after answering.

What does

V_{D D}

mean and what picture goes with it?

The single fixed supply voltage of the chip; the maximum water height the tap can fill the tub to (logic-1).

What does capacitance

C_{L}

physically say, in one relation?

How much charge is stored per volt:

Q = C_{L} V

; it is the "width of the bucket."

State the difference between energy and power in one sentence each.

Energy (J) is a total amount spent, no clock; power (W) is energy per second, a rate that needs a clock.

Why does power × time give energy units?

P = E / t

, so

P \times t = E

— the "per second" cancels, leaving a pure total in joules.

What does

\int_{0}^{\infty} i (t) d t

compute, and why an integral not a multiplication?

The total charge

Q

= area under the current curve; an integral is needed because the current changes over time, so plain rate×time fails.

What is

t_{p}

and where does it enter PDP?

Propagation delay, the time the output takes to flip; PDP

= P_{a v g} \times t_{p}

What do

f

and

α

mean, and why is

f

absent from PDP?

f

= switching cycles per second,

α

= fraction of ticks the gate actually flips; PDP is per-event energy so both "how often" terms are stripped out.

Convert

10 fF

and

50 ps

to SI.

10 fF = 10 \times 1 0^{- 15} F

;

50 ps = 50 \times 1 0^{- 12} s

What are the units of

\frac{1}{2} C_{L} V_{D D}^{2}

Farad × volt² = joules (energy).

Connections

3.2.13 Power-delay product (Hinglish)
Dynamic power dissipation
CMOS Inverter
Propagation delay
Energy-delay product (EDP)
Voltage scaling / DVFS
Switching activity factor
RC charging energy