{ "meta": { "title": "D-Central — ASIC BGA Reball & Rework Thermal Reference", "description": "The lead-free reflow thermal profile (JEDEC J-STD-020 envelope adapted for hot-air rework) as a staged temperature/ramp table, plus a per-chip reference (solder-ball spec, package, thermal mass, caveats) for the Bitmain/Whatsminer ASICs a repair bench reballs.", "generated": "2026-07-18T20:14:26+00:00", "version": "1.0", "license": "https://creativecommons.org/licenses/by/4.0/", "license_name": "CC BY 4.0", "source": "https://d-central.tech/asic-bga-reball-reference/", "method": "Grounded in the D-Central Laval repair bench (0.4 mm SAC lead-free solder balls, ~350-380°C hot-air rework, 861DW-class station — Mining Bible HASHBOARD_DIAGNOSTICS) + the public JEDEC J-STD-020E lead-free reflow envelope. Every figure is a RANGE flagged by basis (bench-practice / JEDEC-standard / practitioner-typical); per-chip ball count and exact pitch vary by hashboard PCB revision and are flagged \"verify\" rather than asserted with false precision. SAC liquidus is 217°C; the nozzle SET temperature is NOT the joint temperature.", "provenance": "D-Central Mining Bible HASHBOARD_DIAGNOSTICS (§8 test equipment, §9 repair techniques) + D-Central Laval bench practice; JEDEC J-STD-020E moisture/reflow classification (public standard). BM1397-vs-BM1396 chip split per D-Central owner canon.", "disclaimer": "🔴 SAFETY: BGA reball/reflow is expert-level hot-air rework that permanently destroys boards and chips when done wrong (cracked pads, lifted traces, cooked silicon, warped PCB). Peak temperatures here are JOINT targets, not station display settings — air cools between nozzle and pad. Always full-board preheat, measure the joint (thermocouple/thermal camera), and verify ball spec against the specific PCB revision. This is a reference for trained technicians, not a substitute for training. If you do not run a hot-air bench, this is exactly what D-Central chip-level repair is for." }, "reflow_profile": [ { "stage": "Preheat", "temp_c": "25 → 150", "time_s": "60-90", "ramp": "≤ 3 °C/s", "basis": "JEDEC J-STD-020", "what": "Bring the board and chip up gradually to drive off moisture and reduce thermal shock. Rushing this is the #1 cause of cracked pads and lifted traces on a hashboard's thick copper.", "caveat": "On a large multi-layer hashboard, preheat the WHOLE board from below (hot plate / IR preheater) so the hot-air nozzle isn't fighting a cold copper heatsink." }, { "stage": "Soak", "temp_c": "150 → 200", "time_s": "60-120", "ramp": "~0.5-1 °C/s", "basis": "JEDEC J-STD-020", "what": "Hold in the soak window to activate flux and equalize temperature across the package before the balls melt. Too short and the balls melt unevenly; too long and the flux exhausts.", "caveat": "Soak time is the knob you actually tune per board thermal mass — a dense S19-gen board needs the longer end." }, { "stage": "Reflow (above liquidus, TAL)", "temp_c": "217 → peak", "time_s": "60-90 above 217 °C", "ramp": "≤ 3 °C/s to peak", "basis": "JEDEC J-STD-020 (SAC liquidus 217 °C)", "what": "Time-above-liquidus: the balls are molten and self-align to the pads. SAC305-class lead-free solder melts at ~217-220 °C; the joint needs enough time molten to wet fully but not so long it grows brittle intermetallics.", "caveat": "217 °C is the SAC melting point — this is why lead-free rework runs hotter than old leaded (183 °C) profiles." }, { "stage": "Peak", "temp_c": "235 → 250", "time_s": "seconds at peak", "ramp": "n/a", "basis": "JEDEC J-STD-020 (by package volume)", "what": "Peak reflow temperature. JEDEC caps peak by package thickness/volume — larger, thicker packages tolerate less overshoot. Aim for the lowest peak that fully reflows (~235-245 °C at the joint) rather than the maximum.", "caveat": "🔴 The nozzle SET temperature (~350-380 °C hot air on the D-Central bench) is NOT the joint temperature — air cools between nozzle and pad. Measure the JOINT with a thermocouple/thermal camera, not the station display." }, { "stage": "Cooldown", "temp_c": "peak → 25", "time_s": "controlled", "ramp": "≤ 6 °C/s", "basis": "JEDEC J-STD-020", "what": "Cool at a controlled rate. Too fast (a cold-air blast or moving the board) cracks joints and warps the PCB; let it settle on the preheater.", "caveat": "Never quench a hot board. Uneven cooldown is a leading cause of the intermittent BGA cracks you were trying to fix." } ], "per_chip": [ { "chip": "BM1387", "models": "Antminer S9 / S9i / S9j / T9", "ball_spec": "0.4 mm SAC lead-free (Sn/Ag/Cu) — D-Central bench standard", "package": "QFN-class daisy-chained ASIC; exact ball count per revision — verify on the board", "thermal_mass": "low-moderate (small die, thin S9 board)", "reflow_note": "The most forgiving generation to practice on — small chips, robust board. Standard lead-free profile.", "grade": "A (bench-verified alloy/ball; package geometry practitioner-typical)" }, { "chip": "BM1397", "models": "Antminer S17 / S17 Pro / T17 / S17+ / T17+", "ball_spec": "0.4 mm SAC lead-free — bench standard", "package": "BGA-class; ball count/pitch varies by hashboard revision — verify (do NOT assume across the 17 family)", "thermal_mass": "moderate-high (denser board, more copper)", "reflow_note": "17-gen boards run hot and warp — full-board preheat is mandatory, not optional. Longer soak.", "grade": "B (alloy bench-verified; per-revision geometry unverified)" }, { "chip": "BM1396", "models": "Antminer S17e / T17e ONLY", "ball_spec": "0.4 mm SAC lead-free — bench standard", "package": "BGA-class; distinct silicon from BM1397 despite the '17' name (per D-Central chip canon)", "thermal_mass": "moderate-high", "reflow_note": "Same profile family as BM1397 but a DIFFERENT chip — order the correct part; not interchangeable in a repair.", "grade": "B" }, { "chip": "BM1362 / BM1360 family", "models": "Antminer S19 / S19 Pro / S19j Pro", "ball_spec": "0.4 mm SAC lead-free — bench standard", "package": "fine-pitch BGA; ball count high; exact pitch per PCB revision (BHB number) — verify", "thermal_mass": "high (dense multi-domain board, heavy copper)", "reflow_note": "The bread-and-butter modern reball. Dense boards demand aggressive full-board preheat and the longer soak; wrong test file at bring-up can over-volt a freshly reballed chip.", "grade": "B" }, { "chip": "BM1366", "models": "Antminer S19 XP / S19 XP Hyd", "ball_spec": "0.4 mm SAC lead-free — bench standard", "package": "fine-pitch BGA", "thermal_mass": "high", "reflow_note": "XP boards are the densest S19-gen — thermal management on the bench is the whole game.", "grade": "B" }, { "chip": "BM1368", "models": "Antminer S21 / T21 / S21 Pro / S21 Hyd", "ball_spec": "0.4 mm SAC lead-free — bench standard (verify; newest gen)", "package": "fine-pitch BGA; 108 chips / 12 domains of 9", "thermal_mass": "high", "reflow_note": "Newest mainstream reball target. No PIC on this generation (removed at BM1368). Confirm ball spec against the specific board before committing.", "grade": "C (newest gen; ball geometry least-publicly-confirmed — verify first-party)" }, { "chip": "BM1370", "models": "Antminer S21 XP / Bitaxe Gamma", "ball_spec": "0.4 mm SAC lead-free — bench standard (verify)", "package": "fine-pitch BGA (industrial) / single-chip on Bitaxe", "thermal_mass": "high (S21 XP) / low (Bitaxe single chip)", "reflow_note": "On a Bitaxe the BM1370 is a single accessible chip — the easiest modern hot-air practice target; on S21 XP it is a dense industrial board.", "grade": "C" }, { "chip": "Whatsminer (MicroBT) ASICs", "models": "M30 / M50 / M60 series", "ball_spec": "0.4 mm-class SAC lead-free — verify (MicroBT geometry differs from Bitmain)", "package": "BGA-class; MicroBT board architecture differs from Antminer", "thermal_mass": "high", "reflow_note": "Same lead-free physics, but MicroBT hashboards use a different layout and test fixture — do not assume Bitmain values transfer. Verify per model.", "grade": "C (MicroBT-specific geometry not bench-canon here)" } ] }