At 3:47 a.m. on a Tuesday in April 2026, a dashboard alert lit up in the operations office at the MillionMiner Missouri facility. Rack 14, position 07. One of Jason’s two Antminer S21 XP units had dropped from 270 TH/s to 178 TH/s. The web interface showed a single line of text: chain 2 missing. Three hashboards in that miner. One had just stopped responding. Jason, the owner, was in Tulsa, asleep.
By 6:00 a.m., our on-site technician had pulled the unit, confirmed that hashboard 2 showed no voltage across domains 4 through 12, and moved the machine to the repair bench. By 9:00 a.m., Jason had an email with three photographs, a diagnostic summary, and a recommendation. The board had a failed BM1368 chip in domain 7. Repair was possible. He approved it. By Friday, the machine was back in the rack, hashing at rated capacity.
That story has a happy ending because the unit was hosted. For the thousands of miners running units in basements, garages, and small shops across the US and Europe, the story plays out differently. The error appears. The hashrate drops. And the owner sits there asking a question that nobody on the internet answers clearly: what do I actually do next?
Every guide you will find on hashboard repair falls into one of two camps. On one side sit deeply technical manuals that walk you through voltage-domain diagnosis, BM1368 chip architecture, and PT1 test procedures. Excellent detail, but written for technicians who have already decided they are repairing the board. On the other side are repair vendors pitching their service, helpful once you know what you want, but useless while you are still deciding.
This article sits in the gap. It is written for the owner staring at a dashboard alert, trying to understand what just broke, what the four realistic options are, what each one actually costs and takes, and which one is right for their specific unit. We run hosting operations for more than 1,200 customers across facilities in Nebraska, Missouri, and Mississippi, and our repair bench sees every failure mode the Antminer line can produce. What follows is the full decision framework, the full Bitmain repair process, and the cost math for each generation at April 2026 prices.
What just happened: decoding the error on your dashboard
Before any repair decision, you need to know what actually failed. The Antminer web interface gives you three pieces of information that together identify the failure mode in under a minute: the chain status line, the chip count per chain, and the domain voltage readings. Knowing how to read them saves you from misdiagnosing a cable fault as a hashboard failure, or a firmware glitch as a chip failure.
The chain status line Every Antminer has three hashboards, labeled chain 0, chain 1, and chain 2 (or chain 1, 2, 3 on some older web interfaces). The dashboard shows the status of each chain. If all three show the expected hashrate, the unit is healthy. If one shows zero or "missing", that chain has a serious problem. If two or three show reduced hashrate with error flags, the problem is almost certainly not the hashboards themselves. Two hashboards rarely fail at the same time. When multiple chains show problems simultaneously, suspect the control board, the PSU, or the data cables between the control board and the hashboards.
The chip count per chain Log into the miner at http://[miner-ip] with root / root and navigate to the Miner Status page. Each chain should report its expected chip count in full. The S19 shows 114 chips per chain (342 total). The S21 XP shows 108 chips per chain (324 total). If you see a chain reporting 0, 1/3, or 2/3 of the expected chip count, that chain has a chip detection failure. ASIC 0 in the log means no chips are being detected, typically a signal chain problem. ASICNG means a specific chip failed during the self-test.
Advanced firmware (including Bitmain stock firmware with debug mode enabled) shows domain voltage readings for each hashboard. A healthy S21 XP shows approximately 1.2 V across all 12 domains. A domain reading 0 V, or significantly below 1.2 V, points to a voltage regulator (LDO) or boost-circuit failure in that specific domain. Across the S21 generation, domains 11 and 12, which draw power from the MP2019 boost ICs, are the most common failure points.
THE 60-SECOND TRIAGE One chain missing, two chains healthy: suspect a hashboard fault. Proceed to the repair decision tree. Multiple chains showing errors: check PSU output voltage, reseat data cables, and try a firmware reflash before assuming hashboard failure. A miner that boots, shows hashrate briefly, and then crashes: thermal protection kicked in. Fix the cooling problem first, not the hashboard.
Your four options: the repair decision tree
Once you have identified a hashboard failure, you have four realistic paths forward. Each comes with a different cost, timeline, and risk profile. Before we walk through each one in detail, here is the overview.
By 6:00 a.m., our on-site technician had pulled the unit, confirmed that hashboard 2 showed no voltage across domains 4 through 12, and moved the machine to the repair bench. By 9:00 a.m., Jason had an email with three photographs, a diagnostic summary, and a recommendation. The board had a failed BM1368 chip in domain 7. Repair was possible. He approved it. By Friday, the machine was back in the rack, hashing at rated capacity.
That story has a happy ending because the unit was hosted. For the thousands of miners running units in basements, garages, and small shops across the US and Europe, the story plays out differently. The error appears. The hashrate drops. And the owner sits there asking a question that nobody on the internet answers clearly: what do I actually do next?
Every guide you will find on hashboard repair falls into one of two camps. On one side sit deeply technical manuals that walk you through voltage-domain diagnosis, BM1368 chip architecture, and PT1 test procedures. Excellent detail, but written for technicians who have already decided they are repairing the board. On the other side are repair vendors pitching their service, helpful once you know what you want, but useless while you are still deciding.
This article sits in the gap. It is written for the owner staring at a dashboard alert, trying to understand what just broke, what the four realistic options are, what each one actually costs and takes, and which one is right for their specific unit. We run hosting operations for more than 1,200 customers across facilities in Nebraska, Missouri, and Mississippi, and our repair bench sees every failure mode the Antminer line can produce. What follows is the full decision framework, the full Bitmain repair process, and the cost math for each generation at April 2026 prices.
What just happened: decoding the error on your dashboard
Before any repair decision, you need to know what actually failed. The Antminer web interface gives you three pieces of information that together identify the failure mode in under a minute: the chain status line, the chip count per chain, and the domain voltage readings. Knowing how to read them saves you from misdiagnosing a cable fault as a hashboard failure, or a firmware glitch as a chip failure.
The chain status line Every Antminer has three hashboards, labeled chain 0, chain 1, and chain 2 (or chain 1, 2, 3 on some older web interfaces). The dashboard shows the status of each chain. If all three show the expected hashrate, the unit is healthy. If one shows zero or "missing", that chain has a serious problem. If two or three show reduced hashrate with error flags, the problem is almost certainly not the hashboards themselves. Two hashboards rarely fail at the same time. When multiple chains show problems simultaneously, suspect the control board, the PSU, or the data cables between the control board and the hashboards.
The chip count per chain Log into the miner at http://[miner-ip] with root / root and navigate to the Miner Status page. Each chain should report its expected chip count in full. The S19 shows 114 chips per chain (342 total). The S21 XP shows 108 chips per chain (324 total). If you see a chain reporting 0, 1/3, or 2/3 of the expected chip count, that chain has a chip detection failure. ASIC 0 in the log means no chips are being detected, typically a signal chain problem. ASICNG means a specific chip failed during the self-test.
Advanced firmware (including Bitmain stock firmware with debug mode enabled) shows domain voltage readings for each hashboard. A healthy S21 XP shows approximately 1.2 V across all 12 domains. A domain reading 0 V, or significantly below 1.2 V, points to a voltage regulator (LDO) or boost-circuit failure in that specific domain. Across the S21 generation, domains 11 and 12, which draw power from the MP2019 boost ICs, are the most common failure points.
THE 60-SECOND TRIAGE One chain missing, two chains healthy: suspect a hashboard fault. Proceed to the repair decision tree. Multiple chains showing errors: check PSU output voltage, reseat data cables, and try a firmware reflash before assuming hashboard failure. A miner that boots, shows hashrate briefly, and then crashes: thermal protection kicked in. Fix the cooling problem first, not the hashboard.
Your four options: the repair decision tree
Once you have identified a hashboard failure, you have four realistic paths forward. Each comes with a different cost, timeline, and risk profile. Before we walk through each one in detail, here is the overview.
The choice depends on three things: whether your unit is still under Bitmain warranty, whether you have the skills and tools for in-house repair, and whether the economic math justifies saving the board at all. An out-of-warranty S9 is almost never worth repairing in April 2026. A warranty-covered S21 Hydro almost always is. Between those two extremes, most buyers land in Option 2 (third-party repair) because it combines speed, cost, and flexibility better than any other path.
Option 1: Bitmain warranty RMA (the official path) If your miner is still under warranty and you have not done anything to void the coverage, the official Bitmain RMA is the cheapest path. You pay nothing for the repair itself. You do pay for shipping in both directions, and you wait.
Warranty coverage by model
Bitmain warranty periods vary by product line. According to Bitmain’s official product warranty documentation, the standard warranty on most Antminer models is 180 days (6 months) from the date of official dispatch. The S19 and S21 series carry a longer 365-day (12-month) warranty. New PSUs carry their own 365-day warranty. The warranty period starts from the date Bitmain ships the unit, not the date you receive or activate it, so factor in transit time when calculating your remaining coverage window.
Check your exact coverage before doing anything else. Bitmain maintains an online warranty lookup tool where you enter your serial number and get back the ship date and remaining warranty days.
The eight-step RMA process
1. Confirm coverage. Find the miner’s serial number on the label near the LAN port. Enter it at service.bitmain.com/support/warranty to verify the ship date and remaining warranty days.
2. Gather your documentation. You need the original proof of purchase, the serial number, a written description of the fault (when it started, what the miner does or does not do, any error messages), and ideally short video showing the problem.
3. Submit the ticket. Open a repair ticket at support.bitmain.com. Attach your documentation. Describe the fault precisely.
4. Wait for assessment. Bitmain support responds within 3 to 7 business days with one of three outcomes: a remote firmware fix, a request to ship the unit back, or a direct replacement authorization.
5. Receive your RMA number. If physical return is required, Bitmain issues an RMA number. Do not ship anything before you have this number in writing. Unauthorized returns are refused at the receiving facility.
6. Pack and ship. Use the original foam inserts if you still have them. If you are shipping hashboards only, pack each one in antistatic packaging. Write the RMA number clearly on the outside of the package. Ship via a tracked courier.
7.Wait, the repair timeline is where RMA loses most of its appeal. Industry warranty guidance puts Bitmain's typical repair turnaround at 4 to 10 weeks, including international transit both ways. From the US, expect 6 to 8 weeks; from Europe, closer to 4 to 6.
8. Receive the repaired unit. Bitmain sends the repaired hashboard or full unit back. A repair under warranty restores the balance of your original warranty period but does not extend it. A board repaired in month 10 of a 12-month warranty has 2 months of coverage left, not a fresh 12.
What voids Bitmain warranty coverage
This is where most RMA attempts fail before they start. Bitmain defines unauthorized modification broadly, and opening a ticket for a unit that has been modified is an automatic rejection. The most common warranty voiders are:
- Custom firmware. Installing Braiins OS+, LuxOS, VNish, or any other non-stock firmware voids the entire unit’s warranty. This is the single most common reason home miners lose coverage. If you are running custom firmware for underclocking or autotuning, you have already traded away your RMA path.
- Unauthorized repair. Any repair or modification performed outside Bitmain or its authorized repair centers voids coverage. This includes fan replacements, thermal paste re-application, or replacing a blown capacitor. If you opened the case and touched anything, document nothing and say nothing.
- Power damage. Failures caused by power surges, brownouts, incorrect voltage, or use with non-APW power supplies are not covered. Bitmain’s engineers can usually identify power damage from burn patterns on the PSU and hashboards.
- Normal wear. Fan failures, thermal paste degradation, and PSU capacitor aging are classified as normal wear and explicitly excluded. Fans are, ironically, the single most common failure point on Antminers and one of the few things not covered.
If any of these apply to your unit, skip Option 1 entirely. Move to Option 2.
Option 2: Third-party repair shop (the practical path)
For most home miners and small operations, third-party repair is the right answer. Turnaround is measured in days instead of weeks. Costs are transparent and usually modest. The tradeoff is that you are paying for service that would have been free under warranty, and you lose Bitmain coverage on anything you send to a third party for repair.
The third-party repair landscape
The third-party Antminer repair market splits into three rough tiers. US-based independent shops offer fast mail-in repair, often with 2 to 3 business days of processing from receipt. Chinese shops run lower prices but longer international shipping windows both ways. Others sit in the middle, pairing service with deep technical documentation. Across all of them the trade-off is the same: you are shipping a dead board and waiting, with no income while it is gone. If that downtime is the real cost, hosted mining with on-site maintenance sidesteps it entirely.
Typical pricing (April 2026)
Pricing varies by generation and by shop, but the range is reasonably consistent across US-based providers. Older generations command lower prices because the chips are cheaper and more available. Newer generations with 5nm silicon cost more because the chips are harder to source and the soldering is more specialized.
- S9, S9j, S9j Pro: $25 to $45 per hashboard, though at April 2026 BTC prices most S9-class boards are not economically worth repairing.
- S17, S17+, T17: $45 to $85 per hashboard. Also borderline economic at current hashprice.
- S19, S19 Pro, S19j Pro, S19a: $45 to $129 per hashboard. This is the sweet spot where third-party repair makes clear economic sense.
- S19K Pro, S19 XP: $79 to $129 per hashboard. Same logic as S19, with slightly higher costs due to BM1366 chip scarcity.
- S21, T21: $199 per hashboard at most US shops. Almost always worth repairing.
- S21 Pro, S21+, S21 XP, S21 Hydro: $199 to $350 per hashboard depending on shop and complexity. Always worth repairing if damage is not burn-related.
What a standard third-party repair includes
Most US-based repair shops follow a similar workflow. The advertised repair cost typically covers: board diagnostics, EEPROM flashing if needed, replacement of up to three failed ASIC chips, component-level repair of voltage regulators and boost circuits, bench testing under load before return shipment. Repairs that require more than three chip replacements, or repairs to boards with burn damage, are either quoted separately or refused outright.
Turnaround is typically 2 to 3 business days from when the shop receives your board. Most shops offer a 30-day service warranty on the repair itself. If the same fault recurs within 30 days, they will re-repair at no additional cost. Longer service warranties (90 days, 6 months) exist but are usually at a premium.
What shops will not fix
- Burn damage. Cracked or charred PCB traces, widespread component burning, or visible PCB delamination. Most shops return these boards at the customer’s shipping cost, or dispose of them if the customer does not want to pay return.
- Liquid damage. Boards exposed to hydro coolant leaks, immersion fluid contamination, or moisture from a humid basement. The corrosion patterns are obvious and the damage is rarely worth the repair cost.
- Physical impact damage. Dropped boards, bent PCBs, or boards that arrived with shipping damage. These are sometimes fixable but the risk of additional latent failures makes most shops refuse.
Inside the repair shop: what actually happens to your board
Understanding what a technician does with your hashboard helps you read repair quotes, ask the right questions, and recognize a shop doing good work versus one cutting corners. The process looks similar whether you send the board to Bitmain, to a third-party shop, or do it yourself.
Option 1: Bitmain warranty RMA (the official path) If your miner is still under warranty and you have not done anything to void the coverage, the official Bitmain RMA is the cheapest path. You pay nothing for the repair itself. You do pay for shipping in both directions, and you wait.
Warranty coverage by model
Bitmain warranty periods vary by product line. According to Bitmain’s official product warranty documentation, the standard warranty on most Antminer models is 180 days (6 months) from the date of official dispatch. The S19 and S21 series carry a longer 365-day (12-month) warranty. New PSUs carry their own 365-day warranty. The warranty period starts from the date Bitmain ships the unit, not the date you receive or activate it, so factor in transit time when calculating your remaining coverage window.
Check your exact coverage before doing anything else. Bitmain maintains an online warranty lookup tool where you enter your serial number and get back the ship date and remaining warranty days.
The eight-step RMA process
1. Confirm coverage. Find the miner’s serial number on the label near the LAN port. Enter it at service.bitmain.com/support/warranty to verify the ship date and remaining warranty days.
2. Gather your documentation. You need the original proof of purchase, the serial number, a written description of the fault (when it started, what the miner does or does not do, any error messages), and ideally short video showing the problem.
3. Submit the ticket. Open a repair ticket at support.bitmain.com. Attach your documentation. Describe the fault precisely.
4. Wait for assessment. Bitmain support responds within 3 to 7 business days with one of three outcomes: a remote firmware fix, a request to ship the unit back, or a direct replacement authorization.
5. Receive your RMA number. If physical return is required, Bitmain issues an RMA number. Do not ship anything before you have this number in writing. Unauthorized returns are refused at the receiving facility.
6. Pack and ship. Use the original foam inserts if you still have them. If you are shipping hashboards only, pack each one in antistatic packaging. Write the RMA number clearly on the outside of the package. Ship via a tracked courier.
7.Wait, the repair timeline is where RMA loses most of its appeal. Industry warranty guidance puts Bitmain's typical repair turnaround at 4 to 10 weeks, including international transit both ways. From the US, expect 6 to 8 weeks; from Europe, closer to 4 to 6.
8. Receive the repaired unit. Bitmain sends the repaired hashboard or full unit back. A repair under warranty restores the balance of your original warranty period but does not extend it. A board repaired in month 10 of a 12-month warranty has 2 months of coverage left, not a fresh 12.
What voids Bitmain warranty coverage
This is where most RMA attempts fail before they start. Bitmain defines unauthorized modification broadly, and opening a ticket for a unit that has been modified is an automatic rejection. The most common warranty voiders are:
- Custom firmware. Installing Braiins OS+, LuxOS, VNish, or any other non-stock firmware voids the entire unit’s warranty. This is the single most common reason home miners lose coverage. If you are running custom firmware for underclocking or autotuning, you have already traded away your RMA path.
- Unauthorized repair. Any repair or modification performed outside Bitmain or its authorized repair centers voids coverage. This includes fan replacements, thermal paste re-application, or replacing a blown capacitor. If you opened the case and touched anything, document nothing and say nothing.
- Power damage. Failures caused by power surges, brownouts, incorrect voltage, or use with non-APW power supplies are not covered. Bitmain’s engineers can usually identify power damage from burn patterns on the PSU and hashboards.
- Normal wear. Fan failures, thermal paste degradation, and PSU capacitor aging are classified as normal wear and explicitly excluded. Fans are, ironically, the single most common failure point on Antminers and one of the few things not covered.
If any of these apply to your unit, skip Option 1 entirely. Move to Option 2.
Option 2: Third-party repair shop (the practical path)
For most home miners and small operations, third-party repair is the right answer. Turnaround is measured in days instead of weeks. Costs are transparent and usually modest. The tradeoff is that you are paying for service that would have been free under warranty, and you lose Bitmain coverage on anything you send to a third party for repair.
The third-party repair landscape
The third-party Antminer repair market splits into three rough tiers. US-based independent shops offer fast mail-in repair, often with 2 to 3 business days of processing from receipt. Chinese shops run lower prices but longer international shipping windows both ways. Others sit in the middle, pairing service with deep technical documentation. Across all of them the trade-off is the same: you are shipping a dead board and waiting, with no income while it is gone. If that downtime is the real cost, hosted mining with on-site maintenance sidesteps it entirely.
Typical pricing (April 2026)
Pricing varies by generation and by shop, but the range is reasonably consistent across US-based providers. Older generations command lower prices because the chips are cheaper and more available. Newer generations with 5nm silicon cost more because the chips are harder to source and the soldering is more specialized.
- S9, S9j, S9j Pro: $25 to $45 per hashboard, though at April 2026 BTC prices most S9-class boards are not economically worth repairing.
- S17, S17+, T17: $45 to $85 per hashboard. Also borderline economic at current hashprice.
- S19, S19 Pro, S19j Pro, S19a: $45 to $129 per hashboard. This is the sweet spot where third-party repair makes clear economic sense.
- S19K Pro, S19 XP: $79 to $129 per hashboard. Same logic as S19, with slightly higher costs due to BM1366 chip scarcity.
- S21, T21: $199 per hashboard at most US shops. Almost always worth repairing.
- S21 Pro, S21+, S21 XP, S21 Hydro: $199 to $350 per hashboard depending on shop and complexity. Always worth repairing if damage is not burn-related.
What a standard third-party repair includes
Most US-based repair shops follow a similar workflow. The advertised repair cost typically covers: board diagnostics, EEPROM flashing if needed, replacement of up to three failed ASIC chips, component-level repair of voltage regulators and boost circuits, bench testing under load before return shipment. Repairs that require more than three chip replacements, or repairs to boards with burn damage, are either quoted separately or refused outright.
Turnaround is typically 2 to 3 business days from when the shop receives your board. Most shops offer a 30-day service warranty on the repair itself. If the same fault recurs within 30 days, they will re-repair at no additional cost. Longer service warranties (90 days, 6 months) exist but are usually at a premium.
What shops will not fix
- Burn damage. Cracked or charred PCB traces, widespread component burning, or visible PCB delamination. Most shops return these boards at the customer’s shipping cost, or dispose of them if the customer does not want to pay return.
- Liquid damage. Boards exposed to hydro coolant leaks, immersion fluid contamination, or moisture from a humid basement. The corrosion patterns are obvious and the damage is rarely worth the repair cost.
- Physical impact damage. Dropped boards, bent PCBs, or boards that arrived with shipping damage. These are sometimes fixable but the risk of additional latent failures makes most shops refuse.
Inside the repair shop: what actually happens to your board
Understanding what a technician does with your hashboard helps you read repair quotes, ask the right questions, and recognize a shop doing good work versus one cutting corners. The process looks similar whether you send the board to Bitmain, to a third-party shop, or do it yourself.
Step 1: Visual inspection
Every shop starts with a visual inspection under magnification. The technician looks for obvious damage: cracked traces, burned components, offset parts from shipping impact, missing components, PCB deformation from heat stress. Any of these changes the repair path. A board with burn marks near the boost circuit is usually scrap. A board with a single blown capacitor and no other damage is a 10-minute fix.
Step 2: Impedance and voltage domain testing
Before applying power to the board, the technician checks impedance across each voltage domain with a multimeter. A reading that indicates a short circuit tells them immediately where the fault is. On a healthy S21 hashboard each working domain reads approximately 1.2 V, with domains 11 and 12 boosted from the MP2019 circuit. A domain reading zero indicates a short; a domain reading significantly low indicates a damaged LDO. Both are standard repairs.
Step 3: Test fixture diagnosis (PT1 and PT2)
If the voltage and impedance checks pass, the board goes on the test fixture. The fixture connects the board to a diagnostic control board running Bitmain test firmware. Two test modes are standard: PT1 (chip detection, checking that all 108 or 114 chips respond) and PT2 (function test, running each chip at rated hashrate to check for performance faults). The test fixture log tells the technician exactly which chips failed and in which domain.
Step 4: Signal chain tracing
When the test fixture identifies a failed chip, the technician traces the signal chain around it with an oscilloscope. Every BM1368 chip passes a CLK signal (from the 25 MHz crystal oscillator Y1), a TX signal (to the next chip), a BO signal (bypass), and a RST signal (reset) through a specific sequence from chip BM1 to chip BM108. The oscilloscope shows whether the signal is arriving at the chip, whether the chip is processing it correctly, and whether the output is reaching the next chip. If the input is present and the output is missing, the chip is failed. If the input itself is missing, the upstream chip or a signal-path resistor is the problem.
Step 5: Chip replacement
BM1368, BM1398, BM1366, and BM1362 chips are BGA-packaged silicon. Replacing one requires a BGA rework station: heated plate below, hot air gun above, precise temperature control, and a steady hand. The technician heats the failed chip until the solder balls underneath reflow, lifts the chip with vacuum tweezers, cleans the pads, applies fresh solder paste (or places a pre-tinned replacement chip), and reflows the new chip into position. The whole process takes 10 to 20 minutes per chip for an experienced technician. The failure rate for this step is real: even experienced shops ruin boards during chip replacement occasionally.
Step 6: EEPROM reflashing
Every Antminer hashboard has a small PIC16F1704 chip that stores calibration data in EEPROM. This data tells the control board what voltage, frequency, and clock profile this specific board runs at, based on factory binning. If the EEPROM gets corrupted during repair (or was already corrupted before the repair), the board will be unreadable by the control board. Reprogramming requires a PICkit3 programming tool and the factory HEX file for that specific board variant. The HEX files are leaked, traded, and sometimes outright purchased within the repair community. A shop that advertises "EEPROM flashing" as part of their standard service has this capability; one that does not will simply return the board as "unrepairable" if the EEPROM is bad.
Step 7: Final load testing
A reputable shop does not call a repair complete until the board has been tested under load on a PT2 fixture and passed twice consecutively. Both runs must show full chip count, stable voltage domains, and rated hashrate. Only then is the board cleaned of flux residue, reassembled if necessary, and packaged for return shipping. Zeus Mining’s published guidelines require "two consecutive OK results" before a board clears repair, and that standard is industry-typical.
Option 3: In-house repair (when it is worth it and when it is not)
In-house repair is the option most home miners dream about and most farm operators eventually settle into. It trades time and tool investment for per-repair cost savings. Whether it makes economic sense depends entirely on how many miners you run.
The tool investment
Board-level ASIC repair is not a job for a soldering iron and luck. The minimum viable toolkit costs $2,000 to $4,000 and includes:
- Temperature-controlled BGA rework station (hot air plus heated preheat plate)
- Precision digital multimeter (Fluke 15B or equivalent)Oscilloscope for signal chain tracing (Hantek or Agilent entry-level)
- Hashboard test fixture matched to your specific miner generation
- PICkit3 programmer plus factory HEX files for your boards
- Antistatic workstation with grounded wrist strap
- BGA solder paste, flux, no-clean cleaning alcohol, fine-tip tweezers - APW3 or APW9 power supply rated for hashboard testing voltage and current
- Replacement chips sourced from verified suppliers (BM1368, BM1398, BM1366 depending on your fleet)
On top of the tools, you need the skills. D-Central puts it bluntly in their S19 Pro repair guide: board-level ASIC repair is a specialized trade that requires years of practice, thousands of dollars in tools, and a deep understanding of Bitmain’s proprietary designs. BGA soldering on modern ASIC silicon is specifically difficult because the solder balls are small, the PCB has many layers, and static discharge can damage chips you cannot even feel.
The economic breakeven
A single Antminer fleet of 3 or 4 home-deployed units will never justify in-house repair. The tool investment alone exceeds what you would spend on third-party repair over the useful life of the fleet. The breakeven sits at roughly 20 miners or more in continuous operation, assuming a failure rate of 1 board per 10 miners per year at steady-state. At that fleet size, you are looking at 6 to 8 hashboard repairs per year, and the tool investment amortizes over three years of operation.
Below 20 miners, third-party repair is cheaper, faster, and less risky. Above 20 miners, in-house repair becomes a genuine cost advantage, especially if you can also reduce downtime by fixing boards within hours instead of days.
The fix rate reality
First-attempt fix rates for home repair are lower than most people expect. D-Central estimates 60% for experienced hobbyists, 70% for professional shops, and close to 90% for Bitmain’s own repair centers. The failure rate is where home repair gets expensive: a botched chip replacement can destroy a board that would otherwise have been worth saving, turning a $129 third-party repair into a $600 replacement board. The math for beginners is brutal. Fail on two boards while learning, and you have spent more than a shop would have charged you for dozens of repairs.
Option 4: Replace or scrap (the hard math)
Not every failed hashboard is worth fixing. Older generation hardware, burn-damaged boards, and units whose economic life has run out all belong in this category. Making the call honestly saves you more money than any other decision on this page.
The replacement option
Bitmain sells replacement hashboards for current-generation machines directly and through authorized distributors. Prices range from $400 for S19 Pro boards to $1,800 for S21 Hydro boards. You are effectively paying for a new, factory-tested hashboard with a fresh warranty period. For a unit that has lost one hashboard but has two healthy ones, replacement is sometimes the right answer: you keep the control board, PSU, chassis, and two of the three hashboards, and you get rated hashrate back immediately.
Used hashboards are available on eBay and in the grey market for 50% to 70% of new prices. Quality varies wildly. A used S19 Pro board from a reputable seller with a burn-in test log is a reasonable buy. A used board from a generic seller with no test history is a coin flip. Assume 40% of grey-market used boards will need some level of repair within 6 months.
The scrap decision
At April 2026 hashprice of $33.25 per PH per day, the economic floor for hashboard repair is roughly 14 to 17 J/TH efficiency at $0.08/kWh hosted. Anything worse than that loses money continuously even when running. An S9 (100 J/TH) has no path to profitability at any electricity rate. An S17 (42 J/TH) barely works at $0.03/kWh. An S19 standard (29.5 J/TH) is on the edge at $0.07/kWh. Before repairing any older board, run the math: at your electricity rate and current hashprice, does this unit produce positive daily margin after the repair cost amortizes over 12 months?
Our full profitability analysis in Is Bitcoin Mining Profitable in 2026? runs five real scenarios at April 2026 prices. Use it to verify whether your specific miner still earns money before you spend anything on repair.
The cost math: repair vs replace vs RMA by generation
Here is the full comparison across six Antminer generations, updated for April 2026 pricing, hashprice, and realistic RMA eligibility. The verdict column is based on a 12-month earnings horizon at $0.08/kWh hosted. Adjust up or down based on your electricity rate and BTC price conviction.
Every shop starts with a visual inspection under magnification. The technician looks for obvious damage: cracked traces, burned components, offset parts from shipping impact, missing components, PCB deformation from heat stress. Any of these changes the repair path. A board with burn marks near the boost circuit is usually scrap. A board with a single blown capacitor and no other damage is a 10-minute fix.
Step 2: Impedance and voltage domain testing
Before applying power to the board, the technician checks impedance across each voltage domain with a multimeter. A reading that indicates a short circuit tells them immediately where the fault is. On a healthy S21 hashboard each working domain reads approximately 1.2 V, with domains 11 and 12 boosted from the MP2019 circuit. A domain reading zero indicates a short; a domain reading significantly low indicates a damaged LDO. Both are standard repairs.
Step 3: Test fixture diagnosis (PT1 and PT2)
If the voltage and impedance checks pass, the board goes on the test fixture. The fixture connects the board to a diagnostic control board running Bitmain test firmware. Two test modes are standard: PT1 (chip detection, checking that all 108 or 114 chips respond) and PT2 (function test, running each chip at rated hashrate to check for performance faults). The test fixture log tells the technician exactly which chips failed and in which domain.
Step 4: Signal chain tracing
When the test fixture identifies a failed chip, the technician traces the signal chain around it with an oscilloscope. Every BM1368 chip passes a CLK signal (from the 25 MHz crystal oscillator Y1), a TX signal (to the next chip), a BO signal (bypass), and a RST signal (reset) through a specific sequence from chip BM1 to chip BM108. The oscilloscope shows whether the signal is arriving at the chip, whether the chip is processing it correctly, and whether the output is reaching the next chip. If the input is present and the output is missing, the chip is failed. If the input itself is missing, the upstream chip or a signal-path resistor is the problem.
Step 5: Chip replacement
BM1368, BM1398, BM1366, and BM1362 chips are BGA-packaged silicon. Replacing one requires a BGA rework station: heated plate below, hot air gun above, precise temperature control, and a steady hand. The technician heats the failed chip until the solder balls underneath reflow, lifts the chip with vacuum tweezers, cleans the pads, applies fresh solder paste (or places a pre-tinned replacement chip), and reflows the new chip into position. The whole process takes 10 to 20 minutes per chip for an experienced technician. The failure rate for this step is real: even experienced shops ruin boards during chip replacement occasionally.
Step 6: EEPROM reflashing
Every Antminer hashboard has a small PIC16F1704 chip that stores calibration data in EEPROM. This data tells the control board what voltage, frequency, and clock profile this specific board runs at, based on factory binning. If the EEPROM gets corrupted during repair (or was already corrupted before the repair), the board will be unreadable by the control board. Reprogramming requires a PICkit3 programming tool and the factory HEX file for that specific board variant. The HEX files are leaked, traded, and sometimes outright purchased within the repair community. A shop that advertises "EEPROM flashing" as part of their standard service has this capability; one that does not will simply return the board as "unrepairable" if the EEPROM is bad.
Step 7: Final load testing
A reputable shop does not call a repair complete until the board has been tested under load on a PT2 fixture and passed twice consecutively. Both runs must show full chip count, stable voltage domains, and rated hashrate. Only then is the board cleaned of flux residue, reassembled if necessary, and packaged for return shipping. Zeus Mining’s published guidelines require "two consecutive OK results" before a board clears repair, and that standard is industry-typical.
Option 3: In-house repair (when it is worth it and when it is not)
In-house repair is the option most home miners dream about and most farm operators eventually settle into. It trades time and tool investment for per-repair cost savings. Whether it makes economic sense depends entirely on how many miners you run.
The tool investment
Board-level ASIC repair is not a job for a soldering iron and luck. The minimum viable toolkit costs $2,000 to $4,000 and includes:
- Temperature-controlled BGA rework station (hot air plus heated preheat plate)
- Precision digital multimeter (Fluke 15B or equivalent)Oscilloscope for signal chain tracing (Hantek or Agilent entry-level)
- Hashboard test fixture matched to your specific miner generation
- PICkit3 programmer plus factory HEX files for your boards
- Antistatic workstation with grounded wrist strap
- BGA solder paste, flux, no-clean cleaning alcohol, fine-tip tweezers - APW3 or APW9 power supply rated for hashboard testing voltage and current
- Replacement chips sourced from verified suppliers (BM1368, BM1398, BM1366 depending on your fleet)
On top of the tools, you need the skills. D-Central puts it bluntly in their S19 Pro repair guide: board-level ASIC repair is a specialized trade that requires years of practice, thousands of dollars in tools, and a deep understanding of Bitmain’s proprietary designs. BGA soldering on modern ASIC silicon is specifically difficult because the solder balls are small, the PCB has many layers, and static discharge can damage chips you cannot even feel.
The economic breakeven
A single Antminer fleet of 3 or 4 home-deployed units will never justify in-house repair. The tool investment alone exceeds what you would spend on third-party repair over the useful life of the fleet. The breakeven sits at roughly 20 miners or more in continuous operation, assuming a failure rate of 1 board per 10 miners per year at steady-state. At that fleet size, you are looking at 6 to 8 hashboard repairs per year, and the tool investment amortizes over three years of operation.
Below 20 miners, third-party repair is cheaper, faster, and less risky. Above 20 miners, in-house repair becomes a genuine cost advantage, especially if you can also reduce downtime by fixing boards within hours instead of days.
The fix rate reality
First-attempt fix rates for home repair are lower than most people expect. D-Central estimates 60% for experienced hobbyists, 70% for professional shops, and close to 90% for Bitmain’s own repair centers. The failure rate is where home repair gets expensive: a botched chip replacement can destroy a board that would otherwise have been worth saving, turning a $129 third-party repair into a $600 replacement board. The math for beginners is brutal. Fail on two boards while learning, and you have spent more than a shop would have charged you for dozens of repairs.
Option 4: Replace or scrap (the hard math)
Not every failed hashboard is worth fixing. Older generation hardware, burn-damaged boards, and units whose economic life has run out all belong in this category. Making the call honestly saves you more money than any other decision on this page.
The replacement option
Bitmain sells replacement hashboards for current-generation machines directly and through authorized distributors. Prices range from $400 for S19 Pro boards to $1,800 for S21 Hydro boards. You are effectively paying for a new, factory-tested hashboard with a fresh warranty period. For a unit that has lost one hashboard but has two healthy ones, replacement is sometimes the right answer: you keep the control board, PSU, chassis, and two of the three hashboards, and you get rated hashrate back immediately.
Used hashboards are available on eBay and in the grey market for 50% to 70% of new prices. Quality varies wildly. A used S19 Pro board from a reputable seller with a burn-in test log is a reasonable buy. A used board from a generic seller with no test history is a coin flip. Assume 40% of grey-market used boards will need some level of repair within 6 months.
The scrap decision
At April 2026 hashprice of $33.25 per PH per day, the economic floor for hashboard repair is roughly 14 to 17 J/TH efficiency at $0.08/kWh hosted. Anything worse than that loses money continuously even when running. An S9 (100 J/TH) has no path to profitability at any electricity rate. An S17 (42 J/TH) barely works at $0.03/kWh. An S19 standard (29.5 J/TH) is on the edge at $0.07/kWh. Before repairing any older board, run the math: at your electricity rate and current hashprice, does this unit produce positive daily margin after the repair cost amortizes over 12 months?
Our full profitability analysis in Is Bitcoin Mining Profitable in 2026? runs five real scenarios at April 2026 prices. Use it to verify whether your specific miner still earns money before you spend anything on repair.
The cost math: repair vs replace vs RMA by generation
Here is the full comparison across six Antminer generations, updated for April 2026 pricing, hashprice, and realistic RMA eligibility. The verdict column is based on a 12-month earnings horizon at $0.08/kWh hosted. Adjust up or down based on your electricity rate and BTC price conviction.
Three patterns stand out. First, the S9 and S17 generations are economically dead for repair at current hashprice; the only reason to fix one is if you specifically need a cheap Scrypt-style heater that hashes something. Second, the S19 family is the sweet spot for third-party repair because chips are cheap, labor is commoditized, and earnings are still meaningful. Third, any S21-class unit is almost always worth saving, either by RMA if eligible or by third-party repair if not, because the per-unit earnings are high enough to justify aggressive preservation.
Then there is the uncomfortable case in the middle. An S19j Pro with two failed chips, 14 months out of warranty, with an owner running it on $0.14/kWh residential electricity. The $129 third-party repair is technically achievable. The board will hash again. But at $0.14/kWh the unit loses roughly $0.80 per day. The repair buys you 12 months of continuous negative cash flow. Scrap it. Or move it to hosting.
Prevention: what actually reduces hashboard failure risk
Once you have been through this cycle once, you start thinking about how to avoid the next one. Most hashboard failures are not random. They cluster around three root causes: thermal stress, power instability, and environmental contamination. Five monthly practices significantly reduce your failure rate.
- Dust management. Dust accumulation is the leading cause of gradual thermal degradation. Air-cooled Antminers pull aggressive airflow through their chassis and trap dust in the heatsinks. Monthly compressed-air cleaning of intakes and heatsinks extends hashboard life measurably. For a 10-miner home setup, this is 15 minutes per month.
- Temperature monitoring. Bitmain’s documented safe operating range is 5°C to 40°C ambient, with chip junction temperatures up to 85°C. Sustained operation above these thresholds accelerates chip degradation substantially. Monitor chip and board temperatures from the web interface monthly, and investigate any unit where any single hashboard runs more than 5°C hotter than its siblings.
- Power conditioning. Brownouts and voltage sags are silent killers. A surge protector is a minimum; a UPS rated for the full miner load is better; a properly sized, properly grounded circuit is best. Most residential miners fail the last point. If you are running miners at home, have an electrician verify that the circuit you are using is correctly rated and grounded.
- Humidity control. Humidity above 65% relative creates condensation risk inside miners when they power cycle. Basements, garages, and unheated spaces in temperate climates are the worst offenders. A simple dehumidifier in the miner room pays for itself in avoided board failures.
- Firmware discipline. Custom firmware can push boards harder than stock firmware, and that extra performance comes at the cost of silicon life. If you run Braiins OS+ or LuxOS for profitability reasons, accept that your boards will have shorter useful lives. For warranty-period units, stock firmware only.
The honest answer: when hosting removes this problem entirely
The whole decision framework you just read exists because you own the hardware and you are responsible for its uptime. That responsibility is not always the right one to carry. Our ASIC miner hosting service includes free on-site repairs for every unit under our management. If your hashboard fails at our Nebraska or Missouri facility, our technician diagnoses it, fixes it if it is fixable, and returns the miner to service within 48 hours in most cases. The repair bench sees every failure mode described in this article, every week, and our fix rate sits at 87% for S19-class hardware and 91% for S21-class.
Hosting is not always the right answer. If you are running two miners as a hedge and hobby, the economics favor home deployment and occasional third-party repair. If you are running ten or more miners and cannot afford a 6-week RMA window when something breaks, hosting starts winning on pure math. At 30,000 miners hosted across our four US facilities, our repair workflow is part of the service, not a bolt-on.
A brief caveat on credentials: our current Trustpilot rating is 4.3 out of 5 across 109 reviews. The negative reviews in that set are worth reading, not skipping. They describe the realistic friction points of running mining operations at scale, and they describe how we responded. We do not hide them. We fix the specific problems they flagged, and we keep going. Honest trust signals matter more than inflated ones.
Frequently asked questions
How much does it cost to repair an Antminer hashboard in 2026?
Third-party repair pricing in April 2026 runs $45 to $129 per hashboard for S19-class hardware, $199 per hashboard for S21 and T21, and $199 to $350 per hashboard for S21 Pro, S21 XP, and S21 Hydro. Older generations (S9, S17) are cheaper to repair but rarely economic to save. Prices include up to three chip replacements, EEPROM flashing if needed, and bench testing before return.
How long does a Bitmain RMA repair take?
Typical Bitmain RMA turnaround is 4 to 10 weeks including international shipping both ways, per D-Central and industry norm. From the US, expect 6 to 8 weeks. From Europe, 4 to 6. Actual repair time at Bitmain’s facility is usually 1 to 2 weeks; the rest is shipping and queue.
What is the Antminer warranty period?
Bitmain standard warranty is 180 days (6 months) for most Antminer models from the official dispatch date. The S19 series and S21 series carry a longer 365-day (12-month) warranty. New PSUs also carry a 365-day warranty. The warranty starts from the ship date, not the arrival or activation date.
Does running Braiins OS+ or LuxOS void my Antminer warranty?
Yes. Any non-stock firmware, including Braiins OS+, LuxOS, VNish, or other aftermarket firmware, voids the entire unit’s Bitmain warranty. This is the most common way home miners lose warranty coverage. If you run custom firmware for autotuning or underclocking, accept that RMA is no longer an option and third-party repair is your only path for covered periods.
Can I repair an Antminer hashboard myself?
Technically yes. Practically, it depends on your tools and skills. Minimum viable toolkit is $2,000 to $4,000 (BGA rework station, oscilloscope, multimeter, test fixture, PICkit3 programmer, antistatic setup). Skills required include BGA soldering proficiency on modern silicon, signal chain diagnosis with an oscilloscope, and patience. First-attempt fix rates for self-taught hobbyists sit around 60%. Economic breakeven is roughly 20 miners in continuous operation. Below that, third-party repair is cheaper and lower risk.
What does "chain 2 missing" mean on my Antminer?
It means the control board cannot detect hashboard 2 at all. The three most common causes are a failed hashboard, a loose or damaged data cable between the control board and the hashboard, or a PSU voltage fault preventing that board from powering up. Start by reseating all data cables and PSU connections. If that does not resolve the error, the hashboard itself is the most likely culprit and requires repair or replacement.
Is it cheaper to repair a hashboard or buy a new one?
For S19-class and newer hardware, repair is almost always cheaper than replacement. An S19 Pro hashboard repair at $79 to $129 beats a $400 to $600 replacement. An S21 XP repair at $199 to $350 beats an $1,800 replacement. The exceptions are boards with burn damage (not economically repairable) and older generations like S9 and S17 where the repair cost exceeds the residual value of the whole unit.
What does a third-party Antminer repair service warranty cover?
Most US-based third-party shops offer a 30-day service warranty on the repair itself. If the same fault recurs within 30 days, they re-repair at no additional cost. Some shops offer longer warranties (90 days, 6 months) at a premium. The service warranty covers only the specific repair performed; unrelated future failures are separate repairs. MYRIG, for example, offers up to 30 days on all Bitmain Antminer repairs and parts replacements per their published terms.
Can I ship my hashboard alone for repair, or do I need to ship the whole miner?
Most third-party shops accept hashboards shipped alone in antistatic packaging, which saves shipping cost dramatically. Bitmain RMA typically requires the full unit unless their support ticket explicitly authorizes hashboard-only return. Read the specific shop’s shipping policy before packaging. A hashboard-only shipment needs foam inserts to prevent bend damage and antistatic bagging to prevent ESD during transit.
When should I scrap an Antminer instead of repairing it?
Scrap when three conditions are all true. First, the unit is older than the S17 generation (efficiency worse than 30 J/TH). Second, your electricity rate is residential ($0.10/kWh or higher). Third, the repair cost plus 12 months of projected operating cost exceeds the 12-month earnings projection at current hashprice. If all three apply, the repair is throwing good money after bad. Strip the unit for parts (PSU, chassis, fans are reusable) and recycle the boards.
The bigger picture
Jason’s S21 XP was back in the rack by Friday afternoon. The repair cost was absorbed by his hosting contract. He received a five-line status email and went back to not thinking about his miners. That is the outcome hosting is designed to produce. The miner broke. Someone else fixed it. The hashrate recovered. Life continued.
For most of the miners reading this article, that is not the outcome available today. You own the unit. You see the alert. You have to decide what to do. The framework in this article is the same one our ops team uses when a board hits the bench. Check warranty eligibility first. If RMA is viable and you can tolerate a 6 to 10 week window, take it. If not, a third-party shop will fix most S19-class and S21-class boards in under a week for $45 to $350. If you are running a fleet of 20 or more miners, in-house repair eventually pays for itself. And if the board is burn-damaged, or the unit is economically dead, scrap it without guilt.
For the deeper economic picture behind any of these decisions, our profitability analysis with five real scenarios runs the numbers at April 2026 hashprice across every current-generation SKU. For the foundational primer on how any of this works, our guide to what Bitcoin mining actually is covers the protocol mechanics. For the strategic case on why any of this matters in the current geopolitical environment, the hedge against uncertainty piece zooms out further. And if your repair question applies to a Z15 Equihash unit or an X-series RandomX unit instead of a SHA-256 miner, the Zcash Z15 guide and Monero X-series guide cover the equivalent ground for those architectures.
A failed hashboard is the single most common serious hardware event in the life of a mining operation. Nobody enjoys it. The only thing worse than having one fail is not knowing what to do when it does. Now you do.
Then there is the uncomfortable case in the middle. An S19j Pro with two failed chips, 14 months out of warranty, with an owner running it on $0.14/kWh residential electricity. The $129 third-party repair is technically achievable. The board will hash again. But at $0.14/kWh the unit loses roughly $0.80 per day. The repair buys you 12 months of continuous negative cash flow. Scrap it. Or move it to hosting.
Prevention: what actually reduces hashboard failure risk
Once you have been through this cycle once, you start thinking about how to avoid the next one. Most hashboard failures are not random. They cluster around three root causes: thermal stress, power instability, and environmental contamination. Five monthly practices significantly reduce your failure rate.
- Dust management. Dust accumulation is the leading cause of gradual thermal degradation. Air-cooled Antminers pull aggressive airflow through their chassis and trap dust in the heatsinks. Monthly compressed-air cleaning of intakes and heatsinks extends hashboard life measurably. For a 10-miner home setup, this is 15 minutes per month.
- Temperature monitoring. Bitmain’s documented safe operating range is 5°C to 40°C ambient, with chip junction temperatures up to 85°C. Sustained operation above these thresholds accelerates chip degradation substantially. Monitor chip and board temperatures from the web interface monthly, and investigate any unit where any single hashboard runs more than 5°C hotter than its siblings.
- Power conditioning. Brownouts and voltage sags are silent killers. A surge protector is a minimum; a UPS rated for the full miner load is better; a properly sized, properly grounded circuit is best. Most residential miners fail the last point. If you are running miners at home, have an electrician verify that the circuit you are using is correctly rated and grounded.
- Humidity control. Humidity above 65% relative creates condensation risk inside miners when they power cycle. Basements, garages, and unheated spaces in temperate climates are the worst offenders. A simple dehumidifier in the miner room pays for itself in avoided board failures.
- Firmware discipline. Custom firmware can push boards harder than stock firmware, and that extra performance comes at the cost of silicon life. If you run Braiins OS+ or LuxOS for profitability reasons, accept that your boards will have shorter useful lives. For warranty-period units, stock firmware only.
The honest answer: when hosting removes this problem entirely
The whole decision framework you just read exists because you own the hardware and you are responsible for its uptime. That responsibility is not always the right one to carry. Our ASIC miner hosting service includes free on-site repairs for every unit under our management. If your hashboard fails at our Nebraska or Missouri facility, our technician diagnoses it, fixes it if it is fixable, and returns the miner to service within 48 hours in most cases. The repair bench sees every failure mode described in this article, every week, and our fix rate sits at 87% for S19-class hardware and 91% for S21-class.
Hosting is not always the right answer. If you are running two miners as a hedge and hobby, the economics favor home deployment and occasional third-party repair. If you are running ten or more miners and cannot afford a 6-week RMA window when something breaks, hosting starts winning on pure math. At 30,000 miners hosted across our four US facilities, our repair workflow is part of the service, not a bolt-on.
A brief caveat on credentials: our current Trustpilot rating is 4.3 out of 5 across 109 reviews. The negative reviews in that set are worth reading, not skipping. They describe the realistic friction points of running mining operations at scale, and they describe how we responded. We do not hide them. We fix the specific problems they flagged, and we keep going. Honest trust signals matter more than inflated ones.
Frequently asked questions
How much does it cost to repair an Antminer hashboard in 2026?
Third-party repair pricing in April 2026 runs $45 to $129 per hashboard for S19-class hardware, $199 per hashboard for S21 and T21, and $199 to $350 per hashboard for S21 Pro, S21 XP, and S21 Hydro. Older generations (S9, S17) are cheaper to repair but rarely economic to save. Prices include up to three chip replacements, EEPROM flashing if needed, and bench testing before return.
How long does a Bitmain RMA repair take?
Typical Bitmain RMA turnaround is 4 to 10 weeks including international shipping both ways, per D-Central and industry norm. From the US, expect 6 to 8 weeks. From Europe, 4 to 6. Actual repair time at Bitmain’s facility is usually 1 to 2 weeks; the rest is shipping and queue.
What is the Antminer warranty period?
Bitmain standard warranty is 180 days (6 months) for most Antminer models from the official dispatch date. The S19 series and S21 series carry a longer 365-day (12-month) warranty. New PSUs also carry a 365-day warranty. The warranty starts from the ship date, not the arrival or activation date.
Does running Braiins OS+ or LuxOS void my Antminer warranty?
Yes. Any non-stock firmware, including Braiins OS+, LuxOS, VNish, or other aftermarket firmware, voids the entire unit’s Bitmain warranty. This is the most common way home miners lose warranty coverage. If you run custom firmware for autotuning or underclocking, accept that RMA is no longer an option and third-party repair is your only path for covered periods.
Can I repair an Antminer hashboard myself?
Technically yes. Practically, it depends on your tools and skills. Minimum viable toolkit is $2,000 to $4,000 (BGA rework station, oscilloscope, multimeter, test fixture, PICkit3 programmer, antistatic setup). Skills required include BGA soldering proficiency on modern silicon, signal chain diagnosis with an oscilloscope, and patience. First-attempt fix rates for self-taught hobbyists sit around 60%. Economic breakeven is roughly 20 miners in continuous operation. Below that, third-party repair is cheaper and lower risk.
What does "chain 2 missing" mean on my Antminer?
It means the control board cannot detect hashboard 2 at all. The three most common causes are a failed hashboard, a loose or damaged data cable between the control board and the hashboard, or a PSU voltage fault preventing that board from powering up. Start by reseating all data cables and PSU connections. If that does not resolve the error, the hashboard itself is the most likely culprit and requires repair or replacement.
Is it cheaper to repair a hashboard or buy a new one?
For S19-class and newer hardware, repair is almost always cheaper than replacement. An S19 Pro hashboard repair at $79 to $129 beats a $400 to $600 replacement. An S21 XP repair at $199 to $350 beats an $1,800 replacement. The exceptions are boards with burn damage (not economically repairable) and older generations like S9 and S17 where the repair cost exceeds the residual value of the whole unit.
What does a third-party Antminer repair service warranty cover?
Most US-based third-party shops offer a 30-day service warranty on the repair itself. If the same fault recurs within 30 days, they re-repair at no additional cost. Some shops offer longer warranties (90 days, 6 months) at a premium. The service warranty covers only the specific repair performed; unrelated future failures are separate repairs. MYRIG, for example, offers up to 30 days on all Bitmain Antminer repairs and parts replacements per their published terms.
Can I ship my hashboard alone for repair, or do I need to ship the whole miner?
Most third-party shops accept hashboards shipped alone in antistatic packaging, which saves shipping cost dramatically. Bitmain RMA typically requires the full unit unless their support ticket explicitly authorizes hashboard-only return. Read the specific shop’s shipping policy before packaging. A hashboard-only shipment needs foam inserts to prevent bend damage and antistatic bagging to prevent ESD during transit.
When should I scrap an Antminer instead of repairing it?
Scrap when three conditions are all true. First, the unit is older than the S17 generation (efficiency worse than 30 J/TH). Second, your electricity rate is residential ($0.10/kWh or higher). Third, the repair cost plus 12 months of projected operating cost exceeds the 12-month earnings projection at current hashprice. If all three apply, the repair is throwing good money after bad. Strip the unit for parts (PSU, chassis, fans are reusable) and recycle the boards.
The bigger picture
Jason’s S21 XP was back in the rack by Friday afternoon. The repair cost was absorbed by his hosting contract. He received a five-line status email and went back to not thinking about his miners. That is the outcome hosting is designed to produce. The miner broke. Someone else fixed it. The hashrate recovered. Life continued.
For most of the miners reading this article, that is not the outcome available today. You own the unit. You see the alert. You have to decide what to do. The framework in this article is the same one our ops team uses when a board hits the bench. Check warranty eligibility first. If RMA is viable and you can tolerate a 6 to 10 week window, take it. If not, a third-party shop will fix most S19-class and S21-class boards in under a week for $45 to $350. If you are running a fleet of 20 or more miners, in-house repair eventually pays for itself. And if the board is burn-damaged, or the unit is economically dead, scrap it without guilt.
For the deeper economic picture behind any of these decisions, our profitability analysis with five real scenarios runs the numbers at April 2026 hashprice across every current-generation SKU. For the foundational primer on how any of this works, our guide to what Bitcoin mining actually is covers the protocol mechanics. For the strategic case on why any of this matters in the current geopolitical environment, the hedge against uncertainty piece zooms out further. And if your repair question applies to a Z15 Equihash unit or an X-series RandomX unit instead of a SHA-256 miner, the Zcash Z15 guide and Monero X-series guide cover the equivalent ground for those architectures.
A failed hashboard is the single most common serious hardware event in the life of a mining operation. Nobody enjoys it. The only thing worse than having one fail is not knowing what to do when it does. Now you do.
Ready to Start Mining?
Free worldwide DDP shipping. Professional hosting from $0.055/kWh.
