Building a Thermal Resistance Network Model
Short answer: A thermal resistance network model treats every layer between a heat source and ambient air as a resistor in series and parallel, measured in °C/W. You sum junction-to-case, interface, spreading, base conduction, fin conduction, and convection resistances to get total θJA. For a typical 50 mm extruded aluminum heat sink in still air, the convection term dominates at roughly 1.5–3.0 °C/W, while a good thermal interface adds only 0.1–0.4 °C/W. Build the model first, then size the heat sink — BQUQ quotes custom heat sinks in 12 working hours.
Why Bother With a Thermal Resistance Network?
Most thermal problems on a project schedule are not solved by buying a bigger heat sink. They are solved by knowing which layer of the stack is actually hot. A thermal resistance network turns a vague "it runs warm" complaint into a number you can attack.
The concept is borrowed from electrical circuits. Heat flow (watts) is current, temperature difference (°C) is voltage, and thermal resistance (°C/W) is resistance. Ohm's law becomes ΔT = Q × θ. Once you draw the network, you can see immediately whether the bottleneck is the interface material, the heat sink base, the fins, or the air around them.
Engineers who skip this step usually over-specify the heat sink and under-specify the interface, or vice versa. Both mistakes cost money. A network model takes an hour to build and typically saves weeks of prototype iteration.
The Basic Series Network
The simplest useful model is a series chain from junction to ambient:
| Element | Symbol | Typical value | Notes |
|---|---|---|---|
| Junction to case | θJC | 0.2–1.5 °C/W | From the device datasheet |
| Case to heat sink (TIM) | θCS | 0.1–0.8 °C/W | Depends on grease, pad, or adhesive |
| Heat sink base spreading | θspread | 0.05–0.5 °C/W | Grows fast with small source area |
| Heat sink to ambient | θSA | 0.3–5.0 °C/W | The dominant term in most designs |
Total junction-to-ambient resistance is the sum:
θJA = θJC + θCS + θspread + θSA
Then junction temperature is:
TJ = TA + (Q × θJA)
If your device dissipates 15 W, ambient is 40 °C, and θJA is 3.0 °C/W, the junction sits at 85 °C. If the datasheet limit is 125 °C, you have margin — but only if every resistance in the chain is realistic. Optimistic interface numbers are the most common source of error.
Where the Series Model Breaks Down
The series model assumes all heat flows through one path. In reality, some heat leaves through the PCB copper, some radiates, and some conducts into the enclosure. For a heat sink mounted on a small package with a poor PCB thermal path, the series model is conservative — it overestimates θJA, which is usually acceptable. For a device soldered to a large copper plane, the PCB is a parallel path and ignoring it makes the model pessimistic by 10–30%.
Modeling the Heat Sink Itself
The heat sink resistance θSA is not a single number. It is a sub-network of base conduction, spreading, fin conduction, and convection. Splitting it out is where the model earns its keep.
Spreading Resistance
Heat enters the base through a small footprint, then spreads laterally. If the source is much smaller than the base, spreading resistance becomes significant. As a rule of thumb, when the source area is less than about 20% of the base area, spreading resistance can exceed 0.3 °C/W on an aluminum base.
Two fixes exist. First, thicken the base locally under the source — a coin, boss, or pedestal. Second, switch to copper for the base. Copper's thermal conductivity is roughly 400 W/m·K versus 200 W/m·K for common 6063 aluminum alloy, so a copper base or copper insert cuts spreading resistance roughly in half for the same geometry. Copper heat sinks and copper-base aluminum-fin designs exist precisely for this reason. You can review the options on our heat sinks product page.
Fin Conduction and Fin Efficiency
Along each fin, temperature drops from base to tip. Fin efficiency is the ratio of actual heat dissipated to the heat that would be dissipated if the whole fin were at base temperature. Long, thin fins have low efficiency; short, thick fins have high efficiency but fewer of them fit.
| Fin parameter | Effect on efficiency | Practical limit |
|---|---|---|
| Fin height | Decreases with height | Efficiency drops below 80% past ~30–40 mm in still air |
| Fin thickness | Increases with thickness | 1.0–2.0 mm typical for extruded profiles |
| Fin conductivity | Increases with k | Aluminum 200 W/m·K, copper ~400 W/m·K |
| Fin spacing | Affects convection, not conduction | 6–12 mm for natural convection |
For natural convection, fin spacing is the parameter that most often gets optimized badly. Too tight and the boundary layers merge, choking flow. Too loose and you waste base area. Our article on natural convection fin spacing covers the trade-off in detail.
Convection and Radiation Resistance
Convection resistance is the inverse of h × A, where h is the heat transfer coefficient and A is the effective surface area. In still air, h is typically 5–10 W/m²·K. With forced air at 2–3 m/s, h rises to 25–50 W/m²·K. This single coefficient often changes θSA by a factor of three or more.
Radiation contributes roughly 10–25% of total dissipation for anodized or painted surfaces at moderate temperatures, and less for bare shiny aluminum, which has low emissivity. Black anodizing raises emissivity to around 0.8 and is one reason many heat sinks are finished that way even when corrosion is not a concern.
A Worked Example
Consider a 20 W load, 45 °C ambient, target junction below 110 °C. That leaves 65 °C of budget.
| Layer | Resistance (°C/W) | ΔT at 20 W (°C) |
|---|---|---|
| θJC | 0.40 | 8.0 |
| θCS (grease, 0.1 mm) | 0.15 | 3.0 |
| θspread (base 80 × 80 mm) | 0.20 | 4.0 |
| θSA (natural convection) | 2.50 | 50.0 |
| Total | 3.25 | 65.0 |
The result lands exactly at the 110 °C target, which means no margin. To gain margin you either reduce θSA with forced air, enlarge the heat sink, or reduce the load. Reducing θCS further is nearly pointless — it is already only 4.6% of the total.
This is the practical value of the model: it tells you where not to spend effort. Thermal grease selection matters, but only up to a point. Our guide on thermal grease selection explains why a 0.05 °C/W improvement in θCS rarely changes the outcome, while a 0.5 °C/W improvement in θSA usually does.
Parallel Paths and Real-World Corrections
Real assemblies have parallel heat paths. The most common are the PCB, the mounting hardware, and the enclosure.
The PCB as a Parallel Resistor
A device soldered to a copper plane conducts heat into the board. That path can carry 10–40% of the total heat in surface-mount designs. In the network, it appears as a resistor in parallel with the heat sink branch. Ignoring it makes your model conservative; including it requires estimating board thermal conductivity, which is anisotropic and awkward to measure.
A pragmatic approach: build the series model first, then apply a correction factor of 0.8–0.9 if the device has a substantial copper pad and the board is not thermally isolated.
Interface Resistance Is Not Constant
Thermal interface resistance depends on pressure, bond line thickness, and surface flatness. A grease layer squeezed to 0.05 mm might give 0.08 °C/W; the same grease at 0.15 mm might give 0.25 °C/W. That is a three-fold swing from assembly pressure alone.
For bonded or adhesive-mounted heat sinks, the adhesive itself becomes the interface. Bonded fin and adhesive-back designs trade some thermal performance for mechanical simplicity and lower cost. If your model assumes 0.1 °C/W but the adhesive delivers 0.5 °C/W, the whole calculation is invalid.
From Model to Manufactured Part
A thermal model is only useful if the part you receive matches the geometry you modeled. Three manufacturing details routinely break models:
Fin thickness tolerance. Extruded profiles hold wall thickness within roughly ±0.1 mm, which is fine. Skived and bonded fins can vary more. Our skiving process overview explains where the limits sit.
Base flatness. A base that is not flat creates a thick, uneven interface layer. CNC-machined bases can hold flatness in the 0.05 mm range, which keeps the interface thin and predictable. See CNC machined heat sinks for how this is controlled.
Surface finish and emissivity. Anodized surfaces radiate far better than bare mill finish. If your model includes a radiation term, the finish must match the assumption.
BQUQ runs CNC machining to ±0.005 mm, metal stamping, custom springs, and heat sink production across four lines in one Dongguan factory, all under ISO9001. That matters for thermal modeling because the geometry you simulate is the geometry you receive. Extruded profiles are available through our extruded heat sinks range, and MOQ is flexible for prototype and pilot builds.
Frequently Asked Questions
Q: What is a typical θJA for a small extruded heat sink?
A: For a 50 × 50 × 25 mm extruded aluminum heat sink in still air, total θSA typically falls between 2.5 and 4.5 °C/W. Adding a small fan at 2 m/s can bring it to 0.8–1.5 °C/W. The exact value depends heavily on fin spacing, base thickness, and whether the surrounding enclosure restricts airflow. Always measure or simulate rather than assuming a catalog number.
Q: How accurate is a hand-calculated thermal resistance network?
A: Hand calculations are usually within 15–25% of measured results when geometry is simple and airflow is well defined. Accuracy degrades in enclosures, with multiple heat sources, or when radiation is significant. Treat the network as a sizing tool, not a final answer, and validate the chosen design with a thermocouple or thermal camera before committing to production tooling.
Q: Does copper always beat aluminum for heat sinks?
A: No. Copper conducts heat about twice as well, but it is roughly three times denser and considerably more expensive. Copper wins when spreading resistance dominates, such as with small heat sources on large bases. Aluminum wins on cost, weight, and extrudability. Copper-base aluminum-fin designs capture most of the spreading benefit at a fraction of the weight.
Q: How do I model a heat sink with a fan?
A: Replace the natural convection coefficient with a forced convection value, typically 25–50 W/m²·K at 2–3 m/s airflow. Then add the fan's own thermal and reliability considerations. The fin spacing that optimizes natural convection is usually too wide for forced air; tighter fins with more surface area perform better when air is moving.
Q: What information do I need to send for a heat sink quote?
A: Send the thermal load, ambient temperature, available envelope, mounting method, and any airflow data. A sketch or STEP file helps. BQUQ returns a quote within 12 working hours, and flexible MOQ means you can order prototype quantities before committing to volume. Email sc@bquq.com with your requirements.
Related Resources
- About BQUQ and our Dongguan manufacturing footprint: /about/
- Heat sink product families and finishes: /heat-sinks/
- Extruded aluminum profiles for thermal applications: /extruded-heat-sinks/
- Industry trends in electronics cooling: /industry-dynamics/
- Full technical article library: /bquq-blog/
- Common sourcing and engineering questions: /faq/
- Contact the engineering team: /contact/
Authored by the BQUQ Engineering Team. BQUQ (Dongguan) runs CNC machining (±0.005 mm), metal stamping, custom springs, and heat sink production in one ISO9001 factory. Source-direct from Dongguan, China — quote in 12 hours: sc@bquq.com | WhatsApp +86 13713157787 | www.bquq.com
