When operators bring used S19 units to our Daegu bench, the first question is always about hashrate: does it match the seller’s claim? That number matters, but we have found that hashboard thermal variance — the temperature difference between the hottest and coolest board in a unit — is often a better early indicator of remaining useful life.
What We Measure
During a 48-hour burn-in, we log inlet temperature, exhaust temperature, and per-hashboard thermal readings using external probes placed on the heatsink surface. On a healthy S19 running at nominal voltage, we typically see a spread of 1.5–2.5°C between boards.
When the spread exceeds 4°C, our historical data shows a 70% probability of measurable hashrate decline within 60 days. The mechanism is usually a degrading thermal interface on one board, causing it to throttle earlier than its neighbors under sustained load.
Why Sellers Miss This
Brokers test units for minutes, not days. A hashboard with a failing thermal pad may produce full hashrate during a 10-minute showroom test. Under our 48-hour protocol, the variance becomes visible as the unit heat-soaks.
What Operators Can Do
If you are evaluating a used lot, ask for thermal imaging or per-board temperature logs — not just a hashrate screenshot. If the seller cannot provide them, factor a ₩200,000–₩400,000 hashboard replacement into your cost model for any unit showing above 3.5°C spread.
We include thermal variance charts in every Miner Longevity Evaluation report. For operators buying blind from remote sellers, this single metric has prevented more bad purchases than any other data point on our bench.