Innosilicon T2T – PSU Compatibility

Read both stickers before plugging anything in. PSU sticker: brand, model, rated continuous wattage, rated DC output voltage. Miner sticker: T2T sub-model and rated input wattage. If rated PSU wattage is below `2000 W` continuous, the PSU is undersized — stop. If voltage is not `12 V DC`, stop. These two 60-second checks prevent roughly 40% of all T2T PSU accidents before they happen.

Power off at the wall (not the miner UI) for a full 60 seconds before any PSU work. Mining-PSU capacitors hold charge well past the moment the LED dies. Hot-swapping a PSU while the PDU is still live is the single most-documented cause of 'dead APW' in Bitmain support forums, and the same discipline applies to Innosilicon PSUs.

Inspect the PSU-to-hashboard harness for heat damage. Look at each 6-pin PCIe connector for discoloration, melted plastic, blackened pins, or brittle insulation. One burnt pin means the PSU or cable delivered more current than the connector could handle — replace the cable (and possibly the PSU) before diagnosing further. Do not re-power a harness with visible heat damage.

Verify cable run length. Stock T2T harness is 30 cm. Any PCIe extenders longer than 50 cm at 18 AWG start to sag voltage measurably under 40-60 A hashboard draw. If you are using a remote-PSU setup, shorten the cables or upgrade to 16 AWG. The miner does not care about aesthetics — it cares about the rail voltage at the PMIC input.

Confirm the electrical circuit. T2T at stock pulls `1800-2200 W` continuous. On a North American 120 V 15 A circuit that violates the NEC 80% continuous-duty rule and will trip eventually. Move the miner to a dedicated 240 V 20 A circuit — not another 15 A outlet on the same wall, a dedicated 240 V drop. This is non-negotiable for a T2T at stock.

Multimeter the PSU output unloaded. DC scale, 20 V range. Disconnect PSU from miner. Power PSU on. Probe every 6-pin PCIe output (pins 1/2/3 = `+12 V`, pins 4/5/6 = ground). Expect `12.0 V` ± `0.2 V`. For Bitmain `APW9` / `APW12`, the default factory config outputs `14.5 V` — verify before connection. Any reading outside `11.8-12.2 V` unloaded means the PSU is out of spec — do not connect to miner.

Multimeter the PSU output under load. Reconnect to miner, power on, let it hash at steady state. Probe at the hashboard-side of the 6-pin connector while the miner is drawing full power. Expect `11.8-12.4 V` sustained. Below `11.8 V` means the PSU cannot maintain rail — swap for a higher-capacity unit. This single measurement diagnoses roughly 60% of all 'my T2T is weird' tickets we see in the queue.

Swap PSU with a known-good Innosilicon-OEM unit (`P20` / `P21E` / `P221E` / `P222C` family). If you do not have a second T2T PSU, D-Central stocks these in salvaged-graded condition. Swap, power on, burn in for 4 hours. If symptoms clear, the original PSU was the problem. If they persist, the problem is miner-side (harness, PMIC, or control board) — move to Tier 3.

Cross-reference against the compatibility matrix. If you are using a Bitmain `APW` series: `APW3++` is undersized for sustained T2T load, `APW9` / `APW12` default output is `14.5 V` which abuses the T2T PMIC input. Only `APW9` / `APW12` units that have been pin-strapped to `12 V` output and verified with a multimeter are safe. Most grey-market `APW9` units ship `14.5 V` — never assume, always verify.

Measure line voltage at the wall under load. On 240 V split-phase expect `235-245 V`; on 208 V commercial expect `202-212 V`; on 120 V expect `115-125 V`. Low line voltage forces the PSU to pull more current to maintain its DC rail, which accelerates sag and trips `OCP` on cold-starts. Evening line sag from neighborhood peak load is the #1 undocumented cause of random evening T2T shutdowns — put a voltage logger on the outlet for a week before you ship the miner.

Re-seat every PSU connector. Power off at the wall. Disconnect every 6-pin PCIe plug. Visually inspect each for oxidation, blackening, or bent pins. Reconnect firmly and listen for the click. Vibration and thermal cycling loosen connectors over months of hashing, and a high-resistance contact on one of nine PCIe pins will silently sag that hashboard's rail without tripping any error.

Recap an old Innosilicon OEM PSU. If the original PSU is 6+ years old, audibly buzzes under load, or sags below `11.8 V` at the connector, the secondary-side electrolytic caps are at end-of-life. A bench recap at D-Central (or a competent repair bench) is cheaper than replacement and extends service life by 3-5 years. Typical cost `CAD $75-$150`. Not a Tier-1 DIY — requires soldering iron and ESR meter competence.

Oscilloscope the rail under transient load. Scope probe on a 6-pin PCIe output, miner hashing. Target: ripple under `100 mV` peak-to-peak, no droop greater than `500 mV` on step-loads. Ripple above `200 mV` p-p or droop over `1 V` on step-loads means the PSU output filter caps are dried or undersized for the application. Recap, or replace the PSU. Most home miners will not have a scope — a D-Central bench test covers this measurement.

Load-bank test the PSU. Programmable DC load, step from 50 A to 150 A (25%-75% of a 2000 W `12 V` PSU). Hold at full load for 30 minutes. Pass: rail stays within `11.8-12.2 V`, ripple under `150 mV`. Fail: replace PSU. If you do not own a load bank, a D-Central bench test runs `CAD $45-$85` and returns a PASS/FAIL with measurement data attached — strongly recommended before you commit a suspect PSU to another production cycle.

Swap the PSU between two T2T chassis to confirm PSU vs miner. If the symptom follows the PSU — bad PSU. If the symptom stays with the chassis — miner-side problem (hashboard PMIC, control board, or internal cable). This is the fastest isolation test when you have a second chassis available and should always precede chip-level diagnosis.

Rewire a server-PSU breakout correctly if you are intentionally running HP / Dell / Delta common-slot server PSUs on a T2T. Verify the breakout delivers `12 V` on pins 1/2/3 and ground on pins 4/5/6 of every 6-pin PCIe output (PCIe spec). Verify the PSU is rated for ≥`2000 W` continuous at the voltage you are feeding it (most common-slot server PSUs derate at 120 V vs 240 V). Verify PS_ON is strapped so the PSU stays on under load, not just at idle. 70% of grey-market breakouts fail one of these checks.

Stop DIY when you see capacitor bulging on the PSU secondary, discoloration or smell from the PSU enclosure, or you measured over `12.8 V` on a hashboard rail before power-off. Any of those mean the PSU has already harmed the miner. Do not power on again. Book a D-Central ASIC Repair slot. PMIC post-mortem + PSU bench-test is the right next step, not another power-on — each retry risks cascading damage into adjacent hashboards.

Ship safely to D-Central. Miner and PSU in separate boxes, or heavily foamed in the same box. Anti-static wrap on hashboards if disassembled. Include a note with observed symptoms, PSU sticker info (brand / model / W / V), miner sticker info (T2T sub-model), circuit voltage (120 V / 208 V / 240 V), and critically — whether the miner was ever powered on with the suspect PSU and for how long. Notes save us 30-60 minutes of diagnostic time, which directly saves you money on the invoice.