Whatsminer Hydro – Quick Connector Seal Broken Alarm

Emergency kill at the PDU. Trip the breaker or pull the C19 — not the web-UI shutdown, not the chassis power switch. A leaking hydro server with the PSU still energized can flash a chip in the time it takes to read this sentence. Wait 5 minutes after power-off for coolant pressure to bleed back to the reservoir and for PSU bulk capacitors to discharge before any physical inspection. This single action prevents the catastrophic half of the failure mode (wet silicon while energized).

Photograph the leak before touching anything. Bright phone flashlight at a side-light angle, capture the QD, the chassis surface beneath, the drip tray (if equipped), and the colour and pattern of any residue. Note: clear droplets = fresh active weep, white crust = evaporated weep with mineral deposit, green tint = glycol-additive package. This documentation saves diagnostic time at the bench and at D-Central if the server ships.

Reservoir level check against operator log. Read the CDU / bulk reservoir level mark, compare to last week's reading. A drop of ≥ 100 mL per week on a 4-10 L loop with no scheduled top-ups confirms a real leak. No measurable drop points to a possible false-positive — read the sibling Water Leak Alarm guide and rule out condensation or sensor contamination before proceeding to teardown.

Do not power-cycle to clear the alarm and resume mining. MicroBT firmware does not let you override the moisture-sensor latch, and re-pressurizing a compromised seal repeatedly accelerates the failure. Server stays off until the seal is repaired. This is the single most-violated rule on home-mining forums; resist the urge.

Walk the operator-side hose runs from the CDU back to the server. Inspect every clamp, fitting, elbow. A leak upstream can pool and wick back to a server-side QD, presenting as a server-side leak when the actual problem is on the cooler. Free 10-minute check that has prevented dozens of unnecessary teardowns; do this before draining the server.

Drain and remove the server from the loop. Close isolation valves if equipped, uncouple QDs over a drip tray or 10 L bucket, expect 0.5-1.5 L of back-drained coolant. Cap both QD ports with the supplied dust caps. Move the server to a dry bench. Dry exterior surfaces with lint-free cloth. This is the prerequisite for every deeper diagnostic — do not try to inspect a leaking QD with the loop pressurized.

External low-pressure compressed-air test. With QDs capped on the CDU side and the server uncoupled, apply low-pressure compressed air (0.10 MPa / ~15 PSI maximum — do NOT exceed nameplate) to the server's coolant inlet. Submerge each QD coupler in a shallow tray of water and watch for bubble streams. This is the field version of the fluorescein test and catches roughly 70% of seal weeps without dye. Mark any bubbling QDs for Tier 3 seal replacement.

Coolant chemistry sample analysis. Draw 10 mL from the reservoir. Check three parameters: clarity (cloudy = biological growth), colour (rust-tinted = corrosion and inhibitor depletion), and pH (target 7-9 on most propylene-glycol formulations). Test strips are $15 CAD on Amazon. Acidic or rust-tinted coolant means every O-ring in the loop is degraded — plan a full chemistry replacement alongside the seal repair.

Re-grease and re-torque the suspect QD without changing seals (diagnostic only). If the O-rings look clean under a 10× loupe (no crush marks, no crust, no extrusion) and the coupler body looks clean, sometimes the seal failure is just dried-out grease on a high-cycle coupler. Strip the coupler, clean both seal surfaces with a nylon brush and 99% IPA, apply a thin film of food-grade silicone grease (Super Lube 92003 — never petroleum-based grease, which swells FKM 5-15% within hours and creates a brand-new failure 30 days from now), reassemble to torque spec. Re-test on the bench at 0.10 MPa. If it holds, you are done. If it weeps again within a 30-minute hold, escalate to seal replacement.

Inspect adjacent QDs proactively. If one O-ring in a 12-month-old coolant fill has failed, the others are aging on the same chemistry exposure. While on the bench, strip the second-most-suspect coupler (typically the supply-side QD with the most thermal cycles) and inspect. Replace seals proactively if early-stage crystallization or surface drying is visible. A single hour of preventive work now beats a weekend emergency repair when the next QD fails in three months.

Replace the primary and secondary O-rings with fresh FKM 75A stock. Pull the QD per the chassis service procedure. Extract the primary face seal and secondary radial seal with a non-marring O-ring pick. Clean the seal seats with 99% IPA and a nylon brush — never cotton swabs (fibres bridge into the seal area and cause first-day weeps). Verify the seat surface is unpitted under 10× magnification. Install new FKM 75A O-rings sized to the specific coupler — confirm the exact size against the model's service manual; common Whatsminer Hydro sizes are 2 × 10 mm and 2 × 12 mm but vary per chassis revision. Lubricate with food-grade silicone (Super Lube 92003). Reassemble. Torque the QD body per MicroBT's hydro service manual — typically 8-12 N·m, but verify per model with a calibrated wrench.

Fluorescein UV pressure test (the bench gold standard). Drain the loop, add 5-10 ppm food-grade fluorescein dye, re-pressurize the server-only section to nameplate (0.15-0.25 MPa typical — verify against your model's spec sheet), hold 15 minutes. Inspect the entire server under a 365 nm UV lamp. Any weep at any seal, fitting, or chassis port glows under UV. This catches: invisible coupler-body cracks, damaged seal seats, threaded-port leaks at the chassis, and confirms the fresh O-rings actually sealed correctly. Complete kit is approximately $40 CAD.

Coupler body inspection and replacement decision. If the fluorescein test shows a leak that follows the coupler body (not a seal), the QD itself is damaged. Common causes: freeze damage from off-season cold storage with coolant in the loop, brazing failure at the inner sleeve, thread damage at the chassis port from over-torquing during a previous service. Order a replacement QD with the exact MicroBT part number for the model. Do not substitute a generic stainless QD — flow coefficient and burst rating must match the loop's design pressure. Lead time from authorized parts channels is typically 2-4 weeks; D-Central can usually source faster from internal parts inventory.

Loop-pressure verification. Before re-deploying a freshly-sealed server, verify the loop's operating pressure under load is within QD nameplate. Many farms have grown CDU pump capacity beyond original Whatsminer hydro spec, then wonder why the QDs keep failing. Measure pressure at the CDU outlet under full server count + nameplate hashrate; if it exceeds the QD nameplate (typically 0.25 MPa continuous, 1.0 MPa burst), the seals will fail again — replace the pump or add a bypass-and-relief loop.

Coolant chemistry replacement (proactive). If Step 8 showed degraded chemistry: drain the entire loop, flush with distilled water until the discharge runs clear, refill with MicroBT-spec coolant — propylene-glycol / distilled-water blend with corrosion inhibitor at the concentration specified for your model. Never substitute automotive antifreeze (wrong inhibitor package, attacks FKM), never use tap water (mineral content trashes coldplates), never reuse drained coolant. Document the refill date and set a 12-month replacement interval on the calendar.

Stop DIY and ship to D-Central when: fluorescein test shows weeps at the coldplate gasket (not the QDs), coolant residue visible on the hashboard PCB, server power-cycled multiple times while wet, QD body shows freeze damage or pitting without a verified MicroBT replacement on hand, no calibrated torque wrench in 1-20 N·m range, or this is the second seal failure on the same server in 12 months. These conditions require bench-level diagnostics and repair capability.

D-Central bench process for hydro QC seal repairs: full server teardown, board-level fluorescein retest at operating temperature (the leak that doesn't show at 25 °C often shows at 75 °C), coldplate re-gasketing to OEM spec with calibrated torque if indicated, coolant-exposed chip inspection under microscope, PMIC and voltage-domain verification under programmable load, per-chip SHA-256d validation on the test fixture, full re-seal and pressure-test, 24-hour burn-in at nameplate hashrate. Every QD torque value and every coldplate bolt torque value gets recorded in the service record. Hydro turnaround is typically 7-14 business days.

Shipping prep. Drain the server completely. Blow out residual coolant with low-pressure compressed air (0.05 MPa maximum). Cap both QDs with dust caps. Double-box with at least 5 cm of foam on every side. Never ship a hydro server with coolant in the loop — transit vibration can damage hoses and turn a localized seal repair into a chassis-wide flood. Label the outer box 'HYDRO ASIC — DRAINED — FRAGILE'. Include a written diagnostic note: observed symptoms, log excerpts, firmware version, coolant age, operator-side maintenance history, photos of the visible weep before disassembly.