Visual walkthrough — Thermal design power (TDP)

You have probably seen the headline formula on the Thermal design power (TDP) parent note:

$$\text{TDP}=\frac{T_{\text{j}}-T_{\text{ambient}}}{R_{\theta ,\text{total}}}$$

But where does it come from? Why does a temperature difference divided by a resistance give you a flow of heat? In this page we build that equation from nothing — no prior formula assumed — using one picture per idea. By the end you will be able to draw the whole cooling law on a napkin.

Let us agree on the very first word we will use.

---

Step 1 — A chip is a leaky bucket of heat

WHAT. Picture the CPU die (the silicon square) as a tiny stove. Every second it produces a fixed amount of heat — call that number $P$ watts. That heat has to go somewhere, or it piles up.

WHY. Before we talk about cooling, we must be crystal clear that the chip is a constant heat source. Almost all the electrical power a chip draws turns into heat (it does no lifting, no motion) — so "power in" equals "heat out" once things settle. This single fact is what lets us later write "TDP" (a heat number) using electrical-power reasoning.

PICTURE. The red die pours heat upward. If nothing carries it away, temperature climbs forever.

---

Step 2 — Temperature is a "height," and heat flows downhill

WHAT. Give every point a temperature: the hot chip die sits at $T_{\text{j}}$ (the "$\text{j}$" means junction, the working guts of the silicon), and the room air far away sits at $T_{\text{ambient}}$. Heat always slides from the high number to the low number.

WHY. Heat flow needs a reason to move, exactly like water needs a downhill slope. That reason is the temperature difference:

$$\Delta T=\underset{\text{hot chip}}{\underbrace{T_{\text{j}}}}-\underset{\text{cool room air}}{\underbrace{T_{\text{ambient}}}}$$

• $\Delta T$ (read "delta-T") — the drop in temperature from chip to air. This is the push driving heat outward.
• $T_{\text{j}}$ — the top of the hill.
• $T_{\text{ambient}}$ — the bottom of the hill.

We choose subtraction here (not a ratio, not a sum) because what makes heat move is purely how much hotter the chip is than the room — a difference of heights.

PICTURE. The taller the step from $T_{\text{j}}$ down to $T_{\text{ambient}}$, the harder heat is pushed out.

Recall

What single quantity is the "push" that drives heat out of a chip? ::: The temperature difference $\Delta T=T_{\text{j}}-T_{\text{ambient}}$.

---

Step 3 — Something resists the flow: thermal resistance

WHAT. Heat does not teleport out. It must ooze through silicon, through a metal case, through paste, through a heatsink, into the air. Each layer fights the flow. We bundle all that fighting into one number $R_{\theta }$, the thermal resistance, measured in °C per watt (°C/W).

WHY. Why invent a resistance at all? Because experiment shows a beautifully simple pattern: to push more heat through the same material, you need a proportionally bigger temperature drop. That "how many degrees of drop per watt of flow" is exactly what $R_{\theta }$ counts.

$$R_{\theta }=\frac{\Delta T}{P}⟺\underset{\text{degrees}}{\underbrace{\Delta T}}=\underset{\text{watts of heat}}{\underbrace{P}}\times \underset{\text{°C per watt}}{\underbrace{R_{\theta }}}$$

Read the units to feel it: $\text{watts}\times \frac{\text{°C}}{\text{watt}}=\text{°C}$. Watts cancel, degrees remain — the arithmetic literally builds a temperature.

• A small $R_{\theta }$ = easy path = heat escapes with only a tiny temperature step. Good cooler.
• A large $R_{\theta }$ = clogged path = you need a huge (dangerous) temperature step to force the same heat out. Bad cooler.

PICTURE. Two pipes carry the same water flow: the fat pipe (low resistance) needs barely any slope; the thin pipe (high resistance) needs a steep slope.

---

Step 4 — Balance: heat in must equal heat out

WHAT. The chip pours in $P$ watts. The path drains out $\Delta T/R_{\theta }$ watts. Wait a moment and these two settle into a tie — a steady state — where the temperature stops changing.

WHY. If the chip made more heat than the path could drain, temperature would climb, which raises $\Delta T$, which increases the drain, until they match. If the path drained too fast, temperature would fall, shrinking $\Delta T$, slowing the drain, until they match. The system self-corrects to the point:

$$\underset{\text{heat produced/sec}}{\underbrace{P}}=\underset{\text{heat drained/sec}}{\underbrace{\frac{\Delta T}{R_{\theta }}}}$$

This equilibrium is the whole reason a chip reaches a stable temperature instead of exploding.

PICTURE. The bucket now has both an inflow ($P$) and a drain whose speed grows with water height. The level parks itself where inflow = outflow.

Recall

At steady state, what balances the heat generated by the chip? ::: The heat drained through the path, $\Delta T/R_{\theta }$ — they are equal, so temperature stops rising.

---

Step 5 — The path is many resistances in a row

WHAT. The escape route is not one wall but three stacked in series: junction→case, case→sink, sink→air. Series resistances add.

WHY. Heat must cross all three one after another — like walking through three doorways in a hallway. The total delay is the sum of each doorway's delay. Nothing is skipped, so nothing is dropped from the sum:

$$R_{\theta ,\text{total}}=\underset{\text{silicon}\to \text{metal case}}{\underbrace{R_{\theta ,\text{JC}}}}+\underset{\text{paste layer}}{\underbrace{R_{\theta ,\text{CS}}}}+\underset{\text{heatsink}\to \text{air}}{\underbrace{R_{\theta ,\text{SA}}}}$$

• $R_{\theta ,\text{JC}}$ — Junction-to-Case, baked into the chip; you can't change it.
• $R_{\theta ,\text{CS}}$ — Case-to-Sink, the thermal paste. Cheap paste = higher = worse.
• $R_{\theta ,\text{SA}}$ — Sink-to-Ambient, the fan-and-fins part you buy.

PICTURE. Three temperature steps in a staircase, each step's height set by its own resistance carrying the same heat $P$.

---

Step 6 — Rearrange the balance into the TDP cooling law

WHAT. Take the steady-state balance from Step 4, replace $R_{\theta }$ with the total from Step 5, and rename the sustained heat number $P$ as TDP — the heat the manufacturer promises the chip will make under a realistic heavy load.

WHY. TDP is defined as that sustained heat rate. Plugging it into the balance turns physics into a design contract: "cooling, you must be able to drain this many watts."

$$\boxed{\text{TDP}=\frac{T_{\text{j}}-T_{\text{ambient}}}{R_{\theta ,\text{total}}}}$$

Term by term, this is just Step 4 with prettier labels:

• top $T_{\text{j}}-T_{\text{ambient}}$ = the temperature push (Step 2),
• bottom $R_{\theta ,\text{total}}$ = the total path resistance (Step 5),
• left $\text{TDP}$ = the heat that must flow (Steps 1 & 4).

PICTURE. The finished flow diagram: push on top, resistance on the bottom, heat streaming out the side.

Now flip it to get the design form. We usually know TDP (from the box), $T_{\text{j,max}}$ (from the datasheet, ~90–105 °C), and room air $T_{\text{ambient}}$. We solve for the resistance we are allowed to have:

$$R_{\theta ,\text{total}}\le \frac{T_{\text{j,max}}-T_{\text{ambient}}}{\text{TDP}}$$

We write ≤ (not =) because a lower resistance runs the chip cooler than the limit — always welcome. Anything above the limit pushes $T_{\text{j}}$ past $T_{\text{j,max}}$ and triggers Thermal throttling.

---

Step 7 — Edge cases: the formula at its breaking points

Every law must be tested where it looks like it might break. Three degenerate cases:

Case A — Zero resistance ($R_{\theta ,\text{total}}\to 0$). Dividing by a tiny number gives a huge allowed TDP. Makes sense: a perfect, resistance-free cooler could drain any amount of heat with no temperature rise. This is the unreachable dream — you can approach it (liquid nitrogen), never touch it.

Case B — Hot room ($T_{\text{ambient}}\to T_{\text{j,max}}$). The top $T_{\text{j,max}}-T_{\text{ambient}}\to 0$, so the allowed resistance $\to 0$. If the room is already as hot as the chip's limit, there is no push left — no cooler on Earth helps. This is why Data center cooling fights so hard to keep intake air cold: every degree of hotter air steals from your $\Delta T$ budget.

Case C — Room hotter than the limit. Then $T_{\text{j,max}}-T_{\text{ambient}}<0$; the formula spits out a negative resistance, which is physically impossible. The honest reading: there is no cooler that works. Heat cannot flow uphill by itself, so the chip is doomed to overheat. The math flagging "negative" is its way of shouting "give up."

PICTURE. The allowed-resistance budget shrinking to zero and going negative as the room heats up.

Recall

Why does a hot room make cooling impossible even with a giant heatsink? ::: Because cooling runs on the temperature difference $\Delta T$. When ambient reaches $T_{\text{j,max}}$, $\Delta T\to 0$, so no heat is pushed out no matter how good the heatsink.

---

The one-picture summary

Everything above collapses into a single "voltage-divider style" picture: one heat current $\text{TDP}$ flowing through three stacked resistances, dropping the temperature from $T_{\text{j}}$ at the top down to $T_{\text{ambient}}$ at the bottom, one step per resistance.

Read it top to bottom: start at the hot junction, subtract a temperature step for each resistance ($P\times R$ per layer), and you must land at or below the room by the time you exit — otherwise the chip cooks. Choosing a smaller heatsink resistance $R_{\theta ,\text{SA}}$ shrinks its step, leaving more headroom.

Recall Feynman retelling — say it like you'd explain to a friend

The chip is a stove making a fixed number of watts of heat. Heat only moves when one side is hotter than the other, and how much it moves depends on how "clogged" the escape path is — that clog is the thermal resistance. String the clogs (silicon, paste, heatsink) in a row and they add up. At balance, the heat made equals the heat drained, so the chip's temperature is just the room temperature plus (watts × total resistance). Turn it around and it tells cooler-builders the biggest resistance they're allowed. And if the room ever gets as hot as the chip's limit, the push vanishes and no cooler can save you — the equation warns you by going to zero and then negative.

---

See also: Heatsink design and thermal resistance · Power consumption in CMOS circuits · Thermal throttling · Turbo Boost and power states · Laptop thermal design