Goldshell SC-BOX II Repair Service

Goldshell SC-BOX II repair fixes a Siacoin miner that reports fewer than 36 chips, stops on a chip overheat, or blows cold air while the pool stays empty. A healthy SC-BOX II board reports 36 chips in 18 pairs; any other count is a fault. Component-level repair in the USA, validated in a running miner under load. Goldshell SC-BOX II repair starts from what the miner says about itself: how many chips the board finds at startup, and which of them are still answering on the chips page. This is a complete, self-contained Siacoin miner rather than a removable board from a larger chassis, so the fault can sit in the board, in the controller, or in the power path, and those are separated during incoming diagnostics before anything is quoted. You do not have to work out which one it is before contacting us. The first-generation SC-BOX is covered here as well — same service, smaller board — and it is offered as the second option on this page. Our lab is in the USA, so the machine never has to leave the country to be repaired. Technical Diagnostics and Common Issues with Goldshell SC-BOX II Quick Symptom Checklist Fans spin, the case blows cold air, and the pool shows nothing arriving. The SYS indicator sits steady red while the green or blue light flashes — the label Goldshell gives this combination is Hashboard startup failure. The dashboard reads Failed to startup, Overheat (chip), Overheat (PCB) or Exception. The startup line in the log reports a chip count below the full set, for example C0: SCBOX Init sucessed. 36 chips, N Total goodcores — it is the shape of the line and the number in it that matter, not the exact digits your machine prints. Hashrate drops in a visible step rather than drifting down, and the chips page shows a neighbouring pair no longer answering. The miner reports Oops! Fan can not start normally, or stops with CPB0 Overheat, So Shutdown It in the log. Try This Before You Ship It In Several of the faults that look terminal on this miner are not, and each of these takes a few minutes. If none of them changes anything, the fault is in the hardware, and the rest of this page describes what we do about it. Power the miner down completely for 30 seconds — unplug it, rather than rebooting from the interface. Check the power supply against the requirement. This miner is fed by an external 12 V supply through 6-pin connectors. Goldshell asks for 16 AWG cable and warns that 18 AWG can overheat and damage the connector, and it asks that the supply carry a clear margin over what the miner draws. Every connector on the miner has to be populated, not just one — a 400 W machine on a single connector is the classic way to melt one. Confirm both fans turn and nothing blocks the grilles. Two fans cool this case and there is no spare capacity. Check the power plan. The SC-BOX II has a published low-power mode that runs it at 1.45 Th/s on 260 W instead of 1.9 Th/s on 400 W, so a fully healthy miner can look like it has lost a quarter of its output. The setting is called Powerplan in the interface, and a second-hand machine can arrive already switched. Run the built-in benchmark. Entering benchmark in place of the pool address puts the miner into a self-test that ignores the network and the pool. A normal result points at the pool or the network; an abnormal one points at the hardware. Verify the pool address, port and worker name before assuming the miner is at fault. Reset the miner to stock firmware at stock settings if it has been reflashed or retuned. Model-Specific Patterns We See on Goldshell SC-BOX II Chips fail in pairs, not one at a time. The chips on this board are powered in series in groups of two, so a fault inside one chip takes its partner down with it. That is why the loss arrives as a step rather than a slow decline, why the chips page shows two neighbours dark instead of one, and why the goal of diagnosis is to find the failed pair and prove the other seventeen are sound — not to hunt a single bad chip. A Goldshell SC-BOX board out of its case, every chip carrying its own heatsink. This is the first-generation board, 16 chips in 8 pairs; the SC-BOX II uses the same chassis and carries 36 chips in 18 pairs. Either way the pair, not the chip, is the unit that fails. An overheat message that is not about the room. The firmware reads a temperature from each chip individually and will shut the miner down on one chip that has run away, while the board average still looks unremarkable. Goldshell reports this as Overheat (chip), and it means something different from Overheat (PCB), which comes from the sensors on the board itself and usually does follow dust, blocked airflow or a hot room. Owners who see the word overheat clean the fans; the miner is pointing at a chip. A dead controller mistaken for a dead board. The controller in this miner boots from a memory card, and a machine that will not answer on the network, will not show a dashboard or sits with its lights in a strange combination frequently has nothing wrong with the hashing side at all. Goldshell says as much about miners that do not respond after being powered on. The controller is also the cheapest part of the miner, so establishing which of the two failed changes what the repair costs. The control board lifted in its card-edge slot. It carries the network port, the antenna and the interface you log into, and it can fail on its own while the hashing side is perfectly healthy — which is why the two are separated before any repair is quoted. Burnt power connectors from a supply that was never quite right. The 6-pin input is the part that fails when a supply is undersized, second-hand, wired with thin cable, or connected through only one of the connectors the miner expects. It arrives discoloured or melted, and the miner starts intermittently or not at all. Goldshell excludes damage caused by a non-standard supply from its warranty, which makes this a repair rather than a claim. Hardware Notes Specification Details Device type Complete self-contained compact miner. The board is not a removable plug-in part and Goldshell does not sell it separately. Algorithm Blake2B-Sia Coins targeted Siacoin. The first-generation SC-BOX was also sold for ScPrime. Enclosure format Enclosed case, 178 by 150 by 84 mm, about 2 kg. Both generations share the case. Rated hashrate 1.9 Th/s at 400 W, with a published low-power mode of 1.45 Th/s at 260 W. The first-generation SC-BOX is rated 900 Gh/s at 200 W. Chips per board 36, reported by the miner at startup. The first-generation SC-BOX carries 16. Chip grouping 18 series-powered pairs of 2 chips. The first-generation SC-BOX has 8 pairs. ASIC marking ICT580, designed by Intchains, on both generations. Goldshell publishes no data on the part in its own specification pages. Controller platform Goldshell MCB V6.3.5. The first-generation SC-BOX uses MCB V5.4, and the two generations are not interchangeable. Firmware Vendor-proprietary, signed and installed by the owner from the web interface. The interface calls the board a computing board, which is the same part the wider market calls a hashboard. Normal board temperature 77 C, shutting down at 86 C. The first-generation SC-BOX runs at 65 to 75 C and shuts down at 85 C. Cooling Air, two fans rated 4500 rpm Power input External 12 V supply through 6-pin connectors, user-supplied. Goldshell specifies 16 AWG cable and a supply rated well above the miner's draw. What the Log Is Telling You None of this is something you need to act on. It is here so you know what state the miner is in before you decide to send it. What you see What it usually means What we do with it A startup line reporting fewer chips than the full count One or more pairs stopped answering when the miner came up We read the board pair by pair on the bench and establish which pair failed and whether the rest are sound Overheat (chip) with the board temperature looking normal A single chip is reporting a runaway temperature and shutting the miner down We identify the chip that reports it and check its thermal contact and its pair before deciding what is replaced Overheat (PCB) The sensors on the board itself are reading high — usually airflow, dust or ambient temperature We inspect the airflow path and the thermal interface, and look for heat damage that has already been done Failed to startup with nothing else on the dashboard The board did not initialise at all — which the board, the controller or the power path can each cause We separate the three before quoting, because two of them are not a board repair A barcode error such as Get CPB0 Barcode Error The controller cannot read the board's identity, which often accompanies a communication fault rather than a chip fault We check the link between board and controller before treating it as a hashing failure Diagnostics Focus The controller, the supply and the memory card come first, before the board. On this miner the dead machine is very often not the hashing side, and the vendor says the same about units that do not respond after power-up. The controller is also the least expensive part in the box, so establishing which part failed changes the price of the job rather than just its description. We work power and controller first, board second. We read the per-chip picture, not the average temperature. The chips page reports each chip separately, including its built-in self-test result, so it shows which pairs have dropped and where they sit relative to one another. That distinguishes a failed pair from a miner that is overheating as a whole, and the front dashboard average shows neither. Our Professional Repair Process Gotchas This is a whole miner, not a plug-in board, and that changes how it is tested. Bench fixtures built for loose boards from rack miners do not apply here, and there is no independent test rig for this hardware on the market at all. We validate through the miner's own diagnostic surface — its interface, its log and its chips page — and then under real mining load in the assembled machine, which is the only place a repair on this hardware can honestly be proven. Replacing the failed part is not the end of the job. The chips are powered in series pairs, so after component-level work the operating point has to be re-established and confirmed. The question is not whether the miner powers on afterwards, but whether it returns to its rated output and holds it warm. Corrosion is a boundary, and the vendor says so too. Goldshell treats boards with corrosion or extensive burn damage as beyond repair and states that a repaired corroded board may corrode and fail again. We say the same before starting rather than after: where a board arrives with corrosion, we tell you what can honestly be promised. Typical Service Scenario A quiet miner living indoors. These were bought as home machines at around 35 dB and they sit on shelves, in cupboards and in alcoves. Goldshell lists the causes of overheating plainly: an air inlet or outlet that is blocked, fans fitted the wrong way round, several miners sharing one small room, or a room above 35 C. Two fans and a small case leave nothing in reserve, and the SC-BOX II puts twice the heat of the first generation into the same box. A cheap or second-hand power supply. The miner is fed externally, so an undersized supply, thin cable or a half-populated connector is a fault waiting to happen, and the vendor asks for a supply rated far above what the miner actually draws. A machine long out of warranty. Goldshell's coverage runs 180 days from shipment, and this generation shipped in 2023, so effectively every one of these miners is outside it. Both generations have also been withdrawn from the vendor's own catalogue. A cold garage in winter. Goldshell states that low ambient temperatures raise the hardware error rate and gives 10 to 35 C as the working range, so a miner that reports errors on a cold morning may have nothing wrong with it. What Happens After Intake Intake: we record the condition of the case, connectors and fans, and what the miner was doing when it stopped. Incoming inspection: we look for heat damage, contamination, corrosion and connector damage before anything is powered. Diagnostics: we check the 12 V input and the on-board power, the communication with the chips, the per-chip readings and the behaviour of the fans and sensors, and we separate a board fault from a controller or supply fault. Component-level repair: failed chips, power components, sensors, connectors and damaged traces are repaired or replaced where the board allows it. Cleaning and thermal service: contamination is removed and degraded thermal interface material replaced. Validation: the repaired miner is checked on its own interface and chips page, then run for a minimum of two hours under real mining load. Extended burn-in beyond that is available as a separate service. Diagnostics and Validation Equipment Bench power diagnostics, multimeter fault isolation, microscope work and thermal inspection for the repair itself; the miner's own interface, log and chips page for what the machine reports about itself; and a live run under load for the result. Because this is a complete miner with no board redundancy, a fault that survives the repair takes the whole machine offline rather than a fraction of it, which is why the load run is the longest step and not the shortest. All repairs are performed personally by Alex, who has been servicing ASIC miners since 2019. See credentials → Contact our repair team today and get your miner back to full power.