Cooling Math for a 10-Inch Cabinet: Vents, Fans, and Stack-Effect

A 10-inch cabinet is a small sealed box, and the laws of thermodynamics do not care that it is a tidy one. Whatever heat your gear produces has to leave the enclosure at the same rate it enters, or the internal air temperature climbs until something throttles, ages faster than it should, or in the worst case trips a thermal shutdown. The useful part is that heat is one of the few homelab variables that obeys arithmetic you can do on the back of an envelope. You do not need a CFD model to decide whether your cabinet needs a fan. You need one equation, a rough parts count, and a willingness to measure the result.

This post is that arithmetic. How much heat a 10-inch cabinet can shed on its own, when you actually need to add a fan, and how to think about the stack effect and vent placement inside an enclosure that is only about 300 mm on a side. The numbers below are order-of-magnitude tools, not datasheet guarantees — the point is to get you within a fan size of the right answer before you spend money.

The one equation that matters

Cooling a rack comes down to moving heat out with airflow. The relationship between the heat you are dumping, the airflow you push through the box, and the temperature rise that results is a single equation. In imperial units:

CFM = 3.16 × Watts ÷ ΔT(°F)

In metric, if you prefer cubic meters per hour and Celsius:

m³/h ≈ 3 × Watts ÷ ΔT(°C)

ΔT is the temperature difference between the air leaving the cabinet and the air coming in. Both forms come from the same physics — the heat capacity of air is about 1005 J/kg·K and its density is roughly 1.2 kg/m³ at room temperature — so they agree with each other. Rearranged, the equation answers the question you actually have: for a given heat load and a temperature rise you are willing to tolerate, how much airflow do you need.

Work an example. Say the gear in your cabinet dissipates 80 W, and you are willing to let the exhaust air run 10 °C (18 °F) warmer than the room. Plug it in: 3.16 × 80 ÷ 18 = about 14 CFM, or equivalently 3 × 80 ÷ 10 = 24 m³/h, which is the same 14 CFM. That is not much air. A single quiet 80 mm fan can move it. The equation is telling you something important before you have bought anything: an 80 W homelab does not have a cooling problem, it has a ventilation problem, and those are cheap to solve.

How much heat is actually in your cabinet

Before you can use the equation you need the wattage, and the honest way to get it is to add up idle-to-typical draw for each device rather than nameplate maximums. Nameplate figures are sized for worst-case inrush and legal margin; they will have you cooling a fire that is not burning. A representative 10-inch homelab load looks something like this:

• A fanless gateway (a UCG-class device or similar): roughly 8–15 W typical.
• A small managed switch, no PoE draw: roughly 5–10 W. Add whatever PoE you are actually delivering on top — a couple of access points and a camera might add 25–45 W of pass-through, most of which leaves via the powered device, not the switch.
• A mini PC node (Optiplex Micro, ThinkCentre Tiny, EliteDesk Mini): 8–15 W at idle, 35–65 W under sustained CPU load.
• A 2.5-inch SSD or a single laptop-class NVMe: 2–5 W. Spinning 3.5-inch drives are the exception that ruins the budget, at 6–9 W each under load.

Add a realistic small build — gateway, switch, one mini PC, an SSD — and you land somewhere between 40 W at idle and 100 W under load, with the mini PC's CPU being the swing factor. The gotcha is PoE. If your switch is delivering power to devices that live outside the cabinet, that wattage passes through the switch but the heat mostly does not; it leaves with the access point on the ceiling. Count the switch's own conversion loss (a few watts) and ignore the PoE budget that exits the building. Get this wrong and you will size a fan for 150 W when the box is really shedding 70.

What a sealed box can shed on its own

Some of your heat leaves without any airflow at all, straight through the cabinet walls by conduction and off the outside surfaces by natural convection and radiation. For a small enclosure you can estimate this with Q = h × A × ΔT, where h is a natural-convection-plus-radiation coefficient of roughly 6–10 W/m²·K for a vertical surface in still room air, A is the effective external surface area, and ΔT is how much hotter the cabinet skin is than the room.

A 300 mm cube has a total surface area of 6 × 0.09 = 0.54 m², but the bottom sits on a shelf and one side may be against a wall, so call the effective radiating area 0.4–0.45 m². At a 15 °C skin-to-room difference with h of about 7, that is Q = 7 × 0.45 × 15 = roughly 47 W. In other words, a small cabinet with its skin running warm to the touch passively sheds heat in the tens of watts, not the hundreds.

That is the whole reason the question "do I need a fan" has a real answer instead of an automatic yes. If your load is 40–50 W and you are content with a warm cabinet, passive dissipation plus a bit of leakage through the seams may carry it. If your load is 90 W and climbing, passive shedding covers half and the rest has to be ventilated out. The equation from the first section tells you how much air the remainder needs.

Stack effect: the chimney you already own

Warm air is less dense than cool air, so it rises, and if you give it a low opening to enter and a high opening to leave, a cabinet becomes a weak chimney with no moving parts. This is the stack effect, and it is the reason a passively cooled enclosure should always have vents low on one face and high on the opposite face rather than a single grille in the middle.

The driving pressure is small, and it is worth seeing how small. The stack pressure scales with the height between your inlet and outlet and with the temperature difference. For a 10-inch cabinet, the usable height between a low intake and a high exhaust might be 0.3 m, and with a 10 °C inside-to-outside difference the buoyancy pressure works out to roughly 0.1 Pa. That is a tenth of a Pascal. Push that through a pair of vents with realistic free area and you get on the order of 2–4 CFM of passive flow.

Here is the gotcha worth internalizing: stack effect in a box this short is real but tiny. It is a bonus of a few CFM, not a cooling strategy. In a two-meter 42U cabinet the same physics moves serious air because the height term is seven times larger; in a 300 mm box it barely registers. Design your vents to cooperate with it — low in, high out, unobstructed — but do not expect it to replace a fan once your load climbs past what passive dissipation can carry.

When to add a fan

Put the two pieces together and a decision rule falls out. Estimate your load. Subtract the tens of watts the box sheds passively. If what remains needs more than a handful of CFM to stay within your target temperature rise, add a fan. In practice, most 10-inch builds cross that line somewhere between 60 and 90 W of internal heat, depending on how well vented the cabinet is and how warm you are willing to let it run.

The good news from the airflow equation is that the fan does not have to be aggressive. Even 100 W at a comfortable 10 °C rise needs only about 18 CFM. A single quiet 120 mm fan moves several times that in free air, and even after you derate for the restriction of grilles and the pressure it has to work against inside a cramped box — call it a 40–60 percent haircut — you have margin to spare. This is why oversizing the fan and then slowing it down is the right move: a big fan turning slowly moves the same air as a small fan screaming, and does it at a fraction of the noise.

Concrete choices that work in a 10-inch cabinet:

• An 80 mm quiet fan (a Noctua NF-A8 runs about 17.7 dB(A) at full tilt) for loads under about 70 W. Small enough to fit a single-U vent panel.
• A 120 mm quiet fan (a Noctua NF-A12x25 is about 22.6 dB(A) at full speed, far less when slowed) for anything heavier, if the cabinet has the face area for it.
• A low-noise adapter or a PWM thermostat so the fan idles nearly silent and only ramps when the internal temperature actually rises. A fan that runs flat-out 24/7 to handle a load that only appears during a nightly backup is a fan you will eventually unplug in annoyance.

Orient the fan to exhaust warm air out the top, and let the intake be passive and low. Pulling air out the top puts the cabinet under slight negative pressure and lets cool room air get drawn in through the low vents, which is exactly the direction the stack effect already wants to push. Blowing in from the top fights the buoyancy and tends to leave a warm pocket at the top of the cabinet where your hottest device usually lives.

Vent free area, and the intake you forgot

A fan can only move the air its openings allow. The rated free area of a vent grille is the actual open area, which for a typical hex or slot pattern is only 40–60 percent of the hole you cut. If you fit a 120 mm fan capable of 60 CFM but feed it through an intake with 20 cm² of real free area, the intake chokes the flow and the fan spends its energy generating noise instead of moving air.

The rule of thumb that keeps you out of trouble: make total intake free area at least equal to, and ideally larger than, the exhaust area. A common mistake is a cabinet with a generous fan on top and almost no intake at the bottom, which starves the fan and creates a partial vacuum that pulls warm air back in through every seam. If you are printing your own vent panels, err toward more open area than you think you need — you can always block some off, but you cannot conjure free area that is not there.

A worked example, start to finish

Say the cabinet holds a fanless gateway (12 W), a small switch with modest PoE conversion loss (8 W of its own heat), and a mini PC that idles at 12 W and peaks near 55 W during a nightly Proxmox backup. Idle internal heat is about 32 W; peak is about 75 W. The cabinet is a 300 mm box that passively sheds perhaps 40 W at a 15 °C skin rise.

At idle, 32 W is under the passive budget, and with decent low-and-high vents the box handles itself; the stack effect's couple of CFM is enough. At peak, 75 W exceeds passive shedding by roughly 35 W, and that surplus needs ventilating. Using the equation for a 10 °C rise: 3.16 × 35 ÷ 18 = about 6 CFM of forced air to carry the overflow. A single 80 mm fan on a thermostat, idle most of the day and ramping only during the backup window, covers it with margin and stays inaudible across the room. Total cost of solving the "cooling problem": one fan and a printed vent panel.

Wrap-up

Cooling a 10-inch cabinet is three numbers: the heat going in, the heat the box sheds on its own, and the airflow needed to carry the difference at a temperature rise you can live with. Get those and the fan decision makes itself — and more often than not, the honest answer for a small homelab is a single slow fan, good low-and-high venting, and letting the stack effect chip in its few free CFM.

Measure after you build. A cheap probe thermometer taped to the exhaust vent for a day tells you whether your arithmetic matched reality, and it almost always lands close. If the exhaust is only a few degrees over ambient, you have headroom to add gear. If it is climbing past 15 °C over the room under load, revisit the vents before you reach for a bigger fan — the intake you forgot is the usual culprit.

Building something like this? Browse 10" mini rack mounts — every one is designed, printed, and test-fitted here.