Ir directamente al contenido

Reg CRC Errors on Antminer S19 & S21: Causes, Diagnosis, and Repair

Antminer hashboard ASIC chip diagnosis with thermal camera for reg CRC error repair
A reg CRC error is one of the most frustrating faults on an Antminer hashboard — but it's almost always a hardware problem with a fixable root cause. Here's how to diagnose and repair CRC errors on S19 and S21 platforms.

Few error messages are as universally frustrating — or as widely misunderstood — as the reg CRC error on an Antminer hashboard. It shows up in kernel logs across the entire S19 and S21 lineup, it causes hashrate drops and HW error spikes, and in many repair shops it gets mislabeled as "random chip failure" when the real root cause is something much more specific.

This guide breaks down what reg CRC errors actually are, why they happen on Antminer platforms, and the step-by-step diagnostic and repair procedures that restore affected hashboards to full operation.

What a Reg CRC Error Actually Is

CRC stands for Cyclic Redundancy Check — a low-level data integrity mechanism used across digital communication protocols. On an Antminer hashboard, the firmware continuously performs CRC checks on the data returned from each ASIC chip in the chain. When the data returned doesn't match the expected checksum, the miner logs a "reg CRC error" against that chip or domain.

The critical thing to understand: a reg CRC error is not the disease. It's a symptom. It tells you that something is interfering with clean data communication between the control board and one or more chips. The job of the repair technician is to find what.

Data from one ASIC repair specialist with over 13,000 hashboard repairs completed is unambiguous on this point: reg CRC errors are almost always hardware-related. Firmware reflashes, reboots, and pool configuration changes may occasionally mask the symptom, but they don't fix the underlying fault. Component-level intervention is what resolves the problem long-term.

The Four Root Causes of Reg CRC Errors

1. Voltage Domain Instability

Antminer hashboards organize their ASIC chips into voltage domains — groups of chips that share a regulated power rail. When the regulator for a domain fails, ages out, or experiences increased resistance in its power delivery path, the chips in that domain receive unstable voltage. Chips running on out-of-spec voltage behave erratically, and erratic behavior produces CRC errors.

Common culprits inside a voltage domain: failing buck converter ICs, aged electrolytic capacitors that have lost capacitance, and bad solder joints on power traces that have developed resistance from thermal cycling.

2. PLL Synchronization Loss

Each ASIC chip contains a Phase-Locked Loop (PLL) that synchronizes the chip's internal clock with the hashboard's clock signal. When a chip's PLL loses lock — either from internal chip degradation or from harsh operating conditions like prolonged overheating — the chip runs out of sync with the rest of the chain. Out-of-sync data produces phase-shifted returns that fail the CRC check.

PLL loss is particularly common on boards that have experienced repeated thermal shutdowns or that run in facilities with poor cooling. It's one of the most reliable indicators of a chip nearing end-of-life.

3. Failed or Dying ASIC Chips

ASIC chips don't always fail cleanly. A partially failed chip may still respond to commands but return incorrect or garbled data. On a healthy chip, the per-chip hashrate in test logs should read around 5000. A chip returning ~1092 or other drastically reduced values is almost certainly the source of a CRC error and needs replacement.

Chip degradation is accelerated by thermal cycling, voltage overshoot events, and prolonged operation above rated junction temperatures.

4. Bad Solder Joints

This is the most underestimated cause. ASIC chips on Antminer hashboards are BGA-mounted — their solder balls sit under the chip package, invisible from above. A solder joint that looks perfect from a visual inspection may actually be cracked, cold-soldered, or not connected at all. Repair technicians frequently report cases where pins appeared properly soldered from the side, only to discover upon chip removal that those pins had never formed a connection in the first place.

Bad solder joints typically manifest after thermal cycling stress: the board hashes fine when cool but develops CRC errors once the chips heat up and micro-movements break marginal connections.

Diagnosing a CRC Error: The Efficient Path

Working through CRC errors manually with a multimeter and oscilloscope is possible but slow. The efficient approach uses a dedicated ASIC tester to isolate the fault in minutes.

Step 1 — Identify the affected board and chip position. Read the kernel log to determine which hashboard is throwing the error and whether it's localized to a specific chip number or spans multiple chips. If the error is localized to a narrow range, that range usually corresponds to a single voltage domain — which points you toward either a chip failure or a power delivery fault in that domain.

Step 2 — Bench-test the suspect board. Remove the hashboard from the miner and connect it to a diagnostic tester. A K9 ASIC Multifunctional Tester supports 70+ mining machine models — including the full S19 and S21 lineups — and will scan the board chip-by-chip, reporting the exact position of any faulty chip along with diagnostic data on voltage and hashrate per chip. This converts hours of guesswork into minutes of targeted diagnosis.

Step 3 — Verify the failed component before replacement. For S21-platform boards, a dedicated BM1368 Series ASIC Chip Tester can verify whether the chip itself is bad before you remove it. For S19 XP and S19K Pro boards, the equivalent is the BM1366 Series ASIC Chip Tester. These testers also let you verify replacement chips before installation — critical if you're sourcing chips from the secondhand market.

The S19 vs. S21 Repair Context

Different S19 and S21 variants use different ASIC chips, and sourcing the correct replacement matters:

  • Antminer S19 / S19 Pro / S19j: BM1397 series chips (TSMC FinFET, ~30 J/T efficiency)
  • Antminer S19j Pro: BM1362 series (notably the BM1362AC)
  • Antminer S19 XP / S19K Pro: BM1366 series (BM1366BS, BM1366BP)
  • Antminer S21 / S21 Pro / S21 XP / T21 / S21 Hydro: BM1368 series (BM1368AA, BM1368PA, BM1368PB, BM1368PM, BM1368PV)

Beyond the chip family difference, the S21 platform introduces a meaningful architectural change from the S19: the BM1368 eliminates the MOS and PIC circuits present on earlier Antminer chips, and adds 11 operational amplifiers from the second voltage domain forward for signal summation. In practice, this means CRC diagnosis on an S21 board sometimes requires checking the op-amp circuits at the 2-domain handover position — a diagnostic step that doesn't apply to S19 boards.

Repair Workflow: From Error Log to Working Board

Once diagnostics have isolated the fault to a specific chip or power component, the repair workflow follows a consistent pattern:

1. Reflow before replacement. For marginal solder joint issues, reflowing the suspect chip is the cheapest fix. Apply flux around the chip, heat each solder joint to a dissolved state with a hot air station, and allow the tin to re-flow into clean contact with the pad. If the error clears after reflow, the issue was solder-related and the chip itself is still good.

2. Chip replacement if reflow fails. If the CRC error persists after reflow, the chip itself is faulty and needs removal. Use a hot air station to lift the chip, clean the pads, and install a new chip using a tin-planting stencil to ensure consistent solder ball formation. Always verify the replacement chip with a dedicated chip tester before soldering — bad new chips are a real risk when sourcing from the secondhand market.

3. Voltage domain component replacement. If diagnosis points to a power delivery fault rather than a chip fault, replace the suspect buck converter, capacitors, or other power components in the affected domain. Re-test the board under load before returning it to service.

4. Full board re-test. After any repair, bench-test the board through a full hashrate cycle. CRC errors that only appear under thermal load will not show up in a cold test — run the board long enough for all chips to reach normal operating temperature before declaring the repair complete.

When Repair Isn't Viable

Not every CRC-erroring board is worth repairing. A board with multiple failed chips across different voltage domains, extensive corrosion, or damage to the base PCB is usually beyond economic repair — the parts and labor cost approaches or exceeds the value of a replacement board. In those cases, harvesting salvageable components (good chips, heatsinks, connectors) and replacing the board entirely is the right call.

For everything else, the combination of the right diagnostic tools and quality replacement chips turns what looks like a dead hashboard into a day's work and a restored machine.

Get the Right Parts and Tools

We stock BM1368, BM1366, BM1397, and BM1362 series ASIC chips, dedicated chip testers for each platform, universal hashboard testers, and the full range of tools needed for CRC error diagnosis and repair — with factory-direct pricing from Shenzhen and DDP shipping to the USA.

Browse ASIC Hashboard Test Fixtures & Chip Testers

Need help identifying the right chip or tester for your specific Antminer model? Our technical team can match the correct parts to your hardware. Reach us at contact@lys-sz.com or via WhatsApp.

Regresar al blog
Te podría gustar