Why Acceptance Testing Matters for Every ASIC Purchase

Whether you are buying new machines from a manufacturer, purchasing used units from the secondary market, or receiving hardware back from a repair facility, every ASIC miner should pass a structured acceptance test before it joins your production fleet. Skipping this step is one of the most common and costly mistakes in mining operations.

The stakes are straightforward. A single defective hash board in an Antminer S23 reduces that machine’s output by roughly one-third. If you deploy 100 machines without testing and 8-12 arrive with underperforming boards (a realistic defect rate for new shipments, higher for used equipment), you are losing 3-4% of your total hashrate from day one. At scale, that translates to thousands of dollars in lost revenue per month, quietly compounding until someone notices the discrepancy between expected and actual output.

A systematic incoming inspection protocol catches these issues before deployment, when they can still be addressed through warranty claims, return shipping, or targeted repairs rather than expensive field troubleshooting.

Phase 1: Physical Inspection Before Powering On

The first phase of acceptance testing requires no electricity. It is a visual and physical inspection that catches shipping damage, tampering, and obvious defects before you invest time in electrical testing.

Packaging and Shipping Assessment

• Document external box condition – Photograph all six sides of each shipping carton before opening. Crushed corners, punctures, or water staining indicate potential internal damage. This documentation is essential for freight claims.
• Verify quantities and serial numbers – Cross-reference received serial numbers against the purchase order and manufacturer packing list. Serial number mismatches between the invoice and the physical unit are a red flag for counterfeit or grey-market hardware.
• Check tamper indicators – New machines from Bitmain and MicroBT include tamper-evident seals on the enclosure. Broken seals on supposedly new units indicate the machine was opened after leaving the factory, which could mean component substitution or unauthorized refurbishment.

Visual Hardware Inspection

• Fan condition – Spin each fan by hand. It should rotate freely with no grinding, clicking, or resistance. Check fan blades for cracks or chips from shipping impact.
• Connector integrity – Inspect all power connectors, data cables, and hash board ribbon cables for bent pins, corrosion, or heat discoloration. Loose or damaged connectors are among the most common causes of intermittent hash board failures.
• Hash board visual check – If accessible, examine the hash boards for bulging capacitors, discolored solder joints, missing components, or signs of liquid exposure. Flux residue around chip packages is normal on new boards but excessive residue or crystallization patterns suggest exposure to humidity during storage or shipping.
• Heat sink attachment – Verify that all heat sinks are firmly attached and have not shifted during transport. A detached heat sink means the thermal interface material bond is broken, and the affected chip will overheat immediately upon startup.

Phase 2: Electrical Benchmarking

After passing physical inspection, each machine undergoes a controlled power-on test to verify electrical performance against manufacturer specifications.

Burn-In Test Setup

Designate a testing area separate from your production floor. This area needs:

• Adequate power capacity for the number of machines being tested simultaneously
• Sufficient cooling (these machines will run at full power for hours)
• Network connectivity to a test mining pool or the manufacturer’s diagnostic server
• A monitoring system to log hashrate, power consumption, temperatures, and error rates per machine

Connect each machine to a properly sized PSU and allow it to run for a minimum of 4-6 hours, ideally 24 hours. Short test runs miss intermittent failures that only manifest after thermal cycling stabilizes.

Key Metrics to Record

• Per-board hashrate – Each hash board should produce within 5% of its rated output. A board delivering 90 TH/s when it should produce 100 TH/s has failing chips that will likely degrade further over time.
• Total unit hashrate – Compare against the manufacturer’s specification. A machine rated at 305 TH/s should sustain at least 290 TH/s (within 5% tolerance) under standard conditions.
• Wall power consumption – Measure actual power draw at the wall, not the software-reported figure. Deviation of more than 10% from the rated wattage indicates electrical problems. A machine drawing significantly more power than rated is likely running with degraded chips that draw excess current.
• Chip temperatures – Log maximum chip temperatures across all boards via the miner’s web interface or firmware dashboard. Temperature variance exceeding 15 degrees between chips on the same board suggests failed thermal interface material or a manufacturing defect.
• Hardware error rate – Track HW errors as a percentage of total shares submitted. A healthy machine produces fewer than 0.5% hardware errors. Error rates above 2% indicate chip-level failures that will worsen over time.
• Fan speeds and RPM – Fans running at maximum RPM while chip temperatures remain elevated suggest blocked airflow or inadequate heat sink contact. Fans that report 0 RPM are either disconnected or failed.

Phase 3: Classification and Disposition

Based on the inspection and benchmarking results, classify each machine into one of four categories:

1. Pass – Deploy to production – Meets all physical inspection criteria and performs within 5% of rated specifications across all metrics.
2. Conditional pass – Minor issues – Physical inspection passed but one metric is marginal (e.g., one board at 93% of rated hashrate). Deploy with a monitoring flag for 30-day observation.
3. Fail – Repairable – Identified defect is fixable (fan replacement, connector re-seat, single board swap). Route to your maintenance team with a specific work order.
4. Fail – Return – Defect exceeds repair threshold or the machine arrived with physical damage indicating shipping or manufacturing fault. Initiate warranty claim or return process immediately with the photographic documentation from Phase 1.

Building a Testing Database

Every machine that passes through your acceptance testing process should have its results recorded in a fleet database. Over time, this data becomes invaluable:

• Baseline performance – When a machine’s hashrate drops six months later, you can compare against its acceptance test results to quantify the degradation and decide whether repair or replacement is more cost-effective. This feeds directly into your resale and depreciation planning.
• Supplier quality tracking – If 15% of machines from one supplier fail acceptance testing versus 3% from another, that data informs your procurement strategy.
• Batch defect identification – Manufacturing defects often cluster in production batches. If three machines with sequential serial numbers all fail with the same board-level issue, you can proactively test the remaining machines from that batch before deploying them.

Modern fleet management platforms like Foreman and Hive OS can automate much of this data collection and trending analysis.

Acceptance Testing for Hosted Mining Customers

If you are purchasing ASICs for deployment at a colocation hosting facility like Rax Mining, acceptance testing is equally important but the logistics differ. In hosted arrangements, the hosting provider often receives hardware shipments on your behalf. Establish in your hosting SLA whether the provider will perform acceptance testing, what test protocol they follow, and how results are communicated to you.

Rax Mining’s hosting customers benefit from our on-site technical team performing incoming inspection and burn-in testing on all hardware before it enters production racks. This ensures that every machine generating revenue for you has been verified against manufacturer specifications.

Start Every Deployment Right

A disciplined acceptance testing protocol takes a few hours per batch of machines and prevents weeks of lost revenue from undetected defects. The cost of testing is negligible compared to the cost of deploying underperforming hardware that quietly reduces your operation’s output.

Whether you are scaling an existing fleet or deploying your first machines, contact our team to discuss hardware procurement, testing protocols, and hosting options that include professional acceptance testing as a standard service.