Delta-T (Thermal)

Delta-T (written ΔT) is the temperature difference between two points in a thermal system, calculated by subtracting one temperature from another. In mining it most often describes the gap between the cool air entering a miner or facility and the hot air leaving it, expressed in degrees Celsius or Fahrenheit. It is a direct, measurable indicator of how much heat the airflow is actually removing — and because it is a difference, it stays meaningful while absolute temperatures swing with weather and season.

The physics in one line

Heat removed by air is proportional to airflow multiplied by ΔT: the same watts can leave as a large volume of slightly warmed air or a small volume of strongly heated air. A multi-kilowatt ASIC moving air with high-static-pressure fans produces a characteristic exhaust rise, and that expected rise is the baseline an operator learns for each machine. Some hashboards even instrument the gradient directly, with temperature sensors placed at the inlet and outlet ends of the board, so firmware sees the same physics at chip scale that a facility designer sees at room scale.

Reading a single machine

For one miner, ΔT is exhaust temperature minus intake temperature, and its stability is the signal. A healthy, well-ventilated machine shows a consistent ΔT at a given power level; changes tell stories. A shrinking ΔT at constant power suggests air is passing through without collecting heat — often a bypass leak — or that hashrate has quietly dropped; paired with rising chip temperatures, it points at dust-choked heatsink fins, where air skims past clogged channels without contacting hot metal. A growing ΔT at constant power means less air is moving: a failing fan, a blocked filter, or restricted exhaust. Either drift, caught early, is a maintenance ticket; ignored, it matures into thermal throttling and lost hashrate.

Reading a facility

The same idea scales up. Operators track supply-to-return ΔT across rows to spot uneven airflow, recirculation, and stratified hot spots — a machine ingesting its neighbor's exhaust shows an inflated intake temperature that no amount of fan speed fixes. This is the arithmetic behind hot-aisle/cold-aisle discipline: separation keeps every intake at supply temperature, making the facility ΔT clean and the cooling predictable. Designers run the equation the other direction too, sizing ventilation from target ΔT and total watts to arrive at required CFM before a single machine is racked.

ΔT as an asset

For heat reuse, ΔT stops being a diagnostic and becomes the product: the temperature rise across the machine determines what the exhaust stream is good for — space heating, greenhouse air, pre-heating water — and hydro or immersion systems push coolant ΔT and outlet temperatures into ranges far more useful than warm air. A homestead miner heating a workshop is, in effect, managing ΔT twice: once for the silicon's health and once for the room's comfort.

Two thermometer readings and a subtraction — no tooling, no vendor, no cloud. ΔT complements rather than replaces direct chip-temperature monitoring, but as an early-warning instrument per watt of effort, nothing on the bench beats it.

Good ΔT numbers come from consistent measurement, and the errors are all avoidable. Measure intake where air actually enters the machine, not across the room; a sensor in recirculated exhaust wash reads high and hides real problems. Take exhaust in the airstream but shielded from radiant heat off the chassis. Log at matching load — ΔT at half power tells you nothing about ΔT at full — and record readings at the same points every time so trends are real rather than artifacts of a wandering thermometer. A pair of inexpensive probes and a spreadsheet is a fully adequate instrument stack; the discipline, not the equipment, is what produces the early warnings. The habit generalizes: nearly every quantity worth tracking in mining — efficiency, rejects, uptime — is a difference or ratio watched over time, and ΔT is the cheapest place to learn that craft.

Compare thermal performance in the cooling methods comparison.