Antminer T19 – Hashboard Not Detected

Power down the T19 at the breaker for 60 seconds, then power back up. Full mains disconnect - not soft-shutdown - lets control-board capacitors drain and forces a clean POST. Watch the first 30 seconds of kernel log via SSH if possible. Resolves transient handshake failures from wedged control-board state after OTA updates or brownouts. Note the event and watch for recurrence.

Open the web UI firmware page. If a previous firmware update was interrupted, re-install the correct T19 image from support.bitmain.com/downloads for your specific hardware revision. An incomplete flash can corrupt the chain-enumeration table and present as ERR_NO_HASHBOARD until a clean image is written. Verify hardware revision off the control-board silkscreen before flashing.

Pull the front panel, verify intake filter and fan airflow are clean, confirm nothing rattles when you tilt the chassis. Recent shipping can dislodge heatsinks or thermal pads; listen for loose components. This is a 2-minute sanity check that rules out the rare-but-real shipping-damage pattern before you invest time in deeper diagnostics.

Power off at the breaker. Re-seat both ends of the ribbon/data cable on the affected chain (control-board end and hashboard end). Inspect for bent pins, oxidation, blackening - wipe contacts with 99% IPA on a lint-free swab, let dry, re-seat firmly until the latch clicks. Do the same with the DC power cable feeding the hashboard. Power back up, re-check chain count. Resolves approximately 50% of ERR_NO_HASHBOARD tickets in the D-Central intake queue.

Label the three hashboard slots 0/1/2 with masking tape. Move the suspect board to a known-good slot, and move a known-good board into the suspect slot. Re-connect all ribbons and power, boot, read the kernel log. Fault follows the board = bad hashboard (move to Tier 3). Fault stays in the slot = bad control-board path, bad ribbon, or bad ribbon connector on the control board.

If Step 5 pinned the fault to a slot, swap the ribbon cable. Borrow from a known-good chain in the same miner or from a donor T19/S19 chassis - they are compatible. Re-boot. Fault moves with the ribbon = replace the ribbon ($15-30 CAD, D-Central stocks them). Fault stays in the slot with a known-good ribbon = the control-board ribbon connector is the next suspect.

Multimeter on DC, probe at the PSU-to-hashboard power connector while the miner is attempting to start. Expect 12.5-14.0 V sustained from the APW12 under boot current. A sagging rail below 12.0 V can stop the hashboard boost stage from waking up, which presents as ERR_NO_HASHBOARD. Measure all three chains - if all sag, the PSU is tired. Swap with a known-good APW12 or APW17 as a test.

If your miner has lived in a garage, basement, or humid environment, ribbon contacts oxidize. Clean both the ribbon and the hashboard-side connector with DeoxIT D5 or equivalent contact cleaner. Let dry fully before re-seating. We have seen T19s recover from a 9-month warehouse shelf where oxidized ribbon contacts were the only fault.

Swap the PSU. T19 ships with an APW12-12-3600-A3. A tired APW12 with internal capacitor degradation can output within spec at idle and sag under the boot current spike, leaving the affected chain dark. Borrow a known-good APW12 or APW17, swap, re-test. If the missing chain returns with a fresh PSU, the PSU - not the hashboard - is the fault.

Flash DCENT_OS (D-Central's open-source Antminer firmware) for per-chip diagnostics that stock firmware does not provide - per-chip enumeration, PIC firmware revision, per-chip HW%, full kernel-log viewer in UI. Alternatives: Braiins OS+, LuxOS, Vnish. Back up original firmware first. DCENT_OS is the Mining Hackers' option: open-source, no licensing, maintained in public. This single step is the biggest diagnostic upgrade on a T19.

If per-chip diagnostics show the chain would enumerate beyond chip 0 if chip 0 came up, reflow chip 0 on the affected board. Remove heatsink, flux the BGA with SMD291 or equivalent no-clean flux, preheat bottom-side to ~150 C, top-side hot air at 310-330 C for ~30 seconds. Cool naturally, clean flux with IPA, re-paste, reassemble. BM1398 packages tolerate one or two reflow cycles well.

If DCENT_OS PIC-handshake test shows the PIC dead, hot-air the old PIC16F-family part off the hashboard and solder in a pre-programmed replacement. D-Central keeps a stock of pre-flashed PICs. Surface-mount work at 300-320 C, fine-tip iron or hot air. Match pin-1 orientation (dot marker). Buying pre-flashed is faster and cheaper than trying to source and program a blank PIC yourself.

If thermal inspection shows localised heat on the boost stage (inductor, MOSFET, boost controller IC) feeding the first voltage domain, replace the failed part. Match the T19 BOM exactly - substitutions on the boost stage turn amateur repairs into permanent damage. Hot-air the inductor and MOSFET at 300-320 C; the boost controller is a finer-pitch part that may need paste + preheat.

If the slot-swap test (Step 5) pinned the fault to the slot and a ribbon swap (Step 6) did not clear it, reflow the ribbon connector on the control-board side. Hot-air at 280-300 C for 15-20 seconds to re-flow existing solder. If joints are visibly damaged, replace the connector with a matching part off Digikey or from a donor control board.

If stock firmware is the suspect after hardware has been cleared, roll to the last-known-good Bitmain image for your specific T19 hardware revision. Verify hardware revision printed on the control-board silkscreen against the hardware table on Bitmain's support portal before flashing - wrong firmware for a late-rev board bricks the control board. Alternatively flash DCENT_OS and bypass stock entirely.

Stop DIY and book D-Central repair when: PIC handshake fails and no replacement on hand; thermal inspection shows damage on voltage-domain ICs or PMIC; Step 14 reflow did not clear a slot-specific fault; visible cap bulging or burnt-component smell; a Tier 3 chip reflow failed within 30 days. These need a test fixture with programmable load and per-domain isolation capability - equipment most home operators do not have.

D-Central bench process on your T19 board: inbound inspection against your reported symptoms, structured isolation on the test fixture, PIC reflash or replacement, per-chip enumeration at rated voltage, per-domain current draw check, BM1398 chip replacement for isolated failed positions from graded salvage inventory, full reflow and reseal, thermal paste refresh, 24-hour nameplate burn-in before shipping back. Typical turnaround: 5-10 business days. Book at https://d-central.tech/services/asic-repair/.

Ship the hashboard (not the whole miner unless necessary) in an anti-static bag, double-boxed with at least 5 cm of foam on every side. Include a printed note with: observed symptoms, kernel-log excerpt, firmware version at failure, contact info. Each piece of that information shaves diagnostic time off our end and repair cost off yours. Canada-wide shipping is routine; US and international accepted.