Antminer S15 Maintenance & Repair Guide

Introduction: Bitmain’s First 7nm Miner

The Bitmain Antminer S15 holds a unique place in Bitcoin mining history. Released in December 2018, it was the first production Antminer to use 7nm ASIC chips — the BM1391BE. While the S9 defined an era with its 16nm BM1387 chips and the S17/S19 generations would go on to dominate industrial mining, the S15 was the bridge. It proved that Bitmain could execute a full node shrink ahead of most competitors, and it shipped real hashrate to real miners at a time when the bear market was crushing everyone who wasn’t running the most efficient hardware possible.

At 28 TH/s in high-performance mode and 1596W power draw, the S15 delivered an efficiency of 57 J/TH — a massive leap from the S9’s ~98 J/TH. It also introduced an energy-saving mode at 17 TH/s and ~900W, pushing efficiency down to 50 J/TH. For home miners running on metered power, that mode was a game-changer.

The S15 also introduced the APW8 power supply, a voltage-regulated PSU that communicates with the control board — a departure from the dumb-rail APW7 that powered the S9. This tighter integration between PSU and control board meant better voltage regulation and higher efficiency, but it also meant new failure modes that the S9 generation never encountered.

Today, the S15 is no longer a profitable miner at most electricity rates. But it still appears regularly in repair shops, on secondhand markets, and in home-heating setups where its ~1600W of thermal output serves double duty. If you are running one, maintaining one, or repairing one, this guide covers everything you need to know — from routine cleaning to board-level diagnostics.

Technical Specifications

Before you crack the case on an S15, you need to understand what you are working with. The S15’s internal architecture differs significantly from both its predecessor (S9) and its successors (S17/S19). Here are the complete specifications.

S15 in Context: Generational Comparison

Understanding where the S15 sits in Bitmain’s lineup helps you contextualize its design decisions and failure modes.

The S15 was Bitmain’s proving ground for 7nm technology. The BM1391BE chip it carries is the direct ancestor of the BM1397 used in the S17 and S17 Pro. Many of the architectural concepts — voltage-regulated PSU, tighter thermal management, multi-domain hashboard design — debuted on the S15 before being refined for the S17 generation.

Before You Begin

Safety Warnings

If you are not confident working with high-current electronics, board-level soldering, or ASIC diagnostics, there is no shame in sending the unit to a professional. That is literally what we do. See the When to Call a Professional section at the end of this guide.

Understanding S15 Architecture

Before you diagnose or repair anything, you need a mental model of how the S15 is built. This section walks through the internal architecture so you can reason about failures instead of just pattern-matching symptoms.

Hashboard Structure

Each S15 hashboard contains 72 BM1391BE chips arranged in 12 voltage domains. Each voltage domain contains 6 chips wired in series. The 12 domains are themselves connected in series, creating a voltage chain across the entire board.

Key architectural details:

• Voltage per domain: approximately 1.53V per domain
• Total voltage across all domains: approximately 18.36V (first to last domain)
• Clock signal (CLK): A 25MHz crystal oscillator (Y1) generates the master clock. The signal passes in series from chip #1 through chip #72. Standby and computing voltage: 0.9V ±0.1V
• TX signal (CI/CO): Transmits from the IO connector (pin 7) through chip #1 to chip #72. Voltage: 0V (unplugged), 1.8V (computing)
• RX signal (RI/RO): Returns from chip #72 back to chip #1, then to the control board via IO pin 8. Voltage: 0V (unplugged), 1.8V (computing)
• RST signal: Reset line from IO pin 3, passes through chip #1 to #72. Voltage: 0V (standby), 1.8V (computing)
• Boost (B) signal: Forward from chip #1 to #72. Voltage: 0V (standby), approximately 0V impulse during computing

Boost Circuit

The S15 hashboard includes a 23V boost circuit that steps up the input voltage from 19.2V to 23V for the DC-DC converter chain. The circuit works as follows:

1. Switching power supply IC (U7) generates a switching signal
2. Energy storage inductor (L4) charges and discharges through boost rectifier diode (D4)
3. Capacitors C73–C75 smooth the output
4. Electrolytic capacitor EC26 filters the final 23V output
5. Voltage regulator U170 steps down to 1.8V, then U171 provides 0.8V

This boost circuit is a common failure point. If the inductor L4, diode D4, or the switching IC U7 fails, the entire hashboard loses power and will not be detected by the control board.

IO Connector Pinout

The S15 hashboard IO port is a 2×9 pin header (18 pins total, 2.0mm pitch, 90-degree right-angle mount). Understanding this pinout is essential for diagnosing communication failures between the control board and hashboards.

APW8 Power Supply Architecture

The APW8 is not a dumb power supply. Unlike the APW3++/APW7 used with the S9, the APW8 is a voltage-regulated PSU that requires communication with the control board to output its main voltage rail. Key characteristics:

• AC Input: 200–240V AC (does NOT support 110V)
• Main DC Output: 16.32V–20.04V, adjustable, max 95A
• Standby Output: 12V at 5A (always on when AC is connected)
• Max Power: 1850W
• Behavior without control board: APW8 outputs only 12V standby. The 19.2V main rail will not engage without the control board requesting it

Routine Maintenance

Preventive maintenance is the difference between a miner that runs for years and one that dies in months. The S15 is a reasonably robust machine, but 7nm chips running at full tilt generate significant heat, and heat is the enemy of longevity. Here is your maintenance schedule.

Recommended Maintenance Schedule

Visual Inspection

Start every maintenance session with a careful visual inspection before powering down. While the miner is still running:

1. Check airflow: Hold your hand near the exhaust side. You should feel strong, consistent hot air. Weak or uneven airflow suggests a fan issue or internal dust blockage.
2. Listen for abnormal sounds: A healthy S15 produces a consistent 76 dB hum. Grinding, clicking, or rattling sounds indicate fan bearing wear.
3. Check LEDs: Green LED solid = normal operation. Red LED = fault condition. No LEDs = power or control board issue.
4. Check the dashboard: Log into the web interface at the miner’s IP address. Verify all 4 hashboards are detected, chip counts are full (72 chips per board), and temperatures are within 40°C–85°C.

After powering down and waiting 10 minutes for cooling:

1. Inspect the enclosure: Look for dust accumulation at the air inlet grille
2. Check all cable connections: Verify flat ribbon cables from hashboards to control board are seated firmly. Check PSU power cables for signs of heat damage or discoloration
3. Inspect heatsinks: The S15 has small individual heatsinks on the front and back of each chip. Look for any that have shifted, detached, or show gaps in the thermal adhesive
4. Check the PCB: Look for any discoloration (brown/yellow burn marks), bulging capacitors, or corroded components

Dust Cleaning

Dust is the silent killer of ASIC miners. It insulates heatsinks (reducing thermal dissipation), clogs fan blades (reducing airflow), and creates conductive pathways on PCBs (causing shorts). Clean your S15 thoroughly every 3 months at minimum — more frequently if you are running in a dusty environment.

Cleaning procedure:

1. Power down completely. Disconnect the power cord from the wall outlet. Wait 10 minutes for the miner to cool.
2. Move to a well-ventilated area. You are about to displace a lot of dust. Do this outside or in a garage, not above your other miners.
3. Remove the top cover. Remove the screws securing the enclosure cover and lift it off.
4. Blow out the fans. Using compressed air at a 45-degree angle, blow through each fan from both sides. Hold the fan blades stationary with a finger while blowing — letting the compressed air spin the fan can damage the bearings. Blow for 2–3 seconds per gap, repeat 2–3 times per fan.
5. Clean the heatsinks. Blow compressed air across all heatsink fins on every hashboard. Work from one end of the board to the other in a consistent direction. Focus on the narrow gaps between fins where dust compacts most densely.
6. Clean the control board. Gently blow any dust off the control board, Ethernet port, and connector areas.
7. Brush stubborn deposits. Use the anti-static brush to gently dislodge any dust that compressed air cannot remove. Pay particular attention to the area around heatsink bases and the edges of chips.
8. Inspect after cleaning. With the dust cleared, this is the best time to look for damaged components, shifted heatsinks, or burned traces.
9. Reassemble and test. Replace the cover, reconnect cables, and power on. Verify all 4 hashboards come online and temperatures are in range.

Thermal Paste & Heatsink Maintenance

The S15 uses individual heatsinks bonded to each chip with thermal adhesive. Over time, this adhesive degrades, dries out, or loses its bond — especially in environments with thermal cycling (the miner heats up during the day and cools at night). When thermal coupling degrades, chip temperatures rise, leading to throttling and eventually chip failure.

Signs that thermal paste/adhesive needs replacement:

• Individual chip temperatures significantly higher than neighbors (check the dashboard per-chip view)
• Heatsinks that are loose, wobble, or have visibly shifted
• Dried, cracked, or missing thermal compound visible at heatsink edges
• Miner throttling despite clean heatsinks and good airflow

Heatsink removal and re-application:

1. Carefully pry off the old heatsink using a plastic spudger. Never use metal tools directly on the chip surface. If the adhesive is stubborn, warm it gently with a heat gun at 100°C to soften it.
2. Clean old thermal material from both the chip surface and heatsink base using 99% IPA and a lint-free wipe. The surfaces should be clean and shiny.
3. Apply new thermal compound. Use black heat-conducting adhesive (3461 or equivalent) for permanent bonding, or high-quality thermal paste (Arctic MX-4, Noctua NT-H1) for a non-permanent application that allows easier future removal.
4. Seat the heatsink firmly and apply even pressure for 10–15 seconds. Ensure the heatsink is centered on the chip with no overhang that could contact adjacent components.
5. Allow adhesive to cure per manufacturer specifications before powering on (typically 4–8 hours for thermal adhesive).

Fan Maintenance

The S15 uses two fans — one intake (front) and one exhaust (rear). Both are critical for airflow through the hashboard heatsink stack. A single failed fan will cause thermal shutdown.

Fan health checks:

• RPM monitoring: Check fan RPMs through the miner web interface. Both fans should spin at similar speeds. A fan running significantly slower than its partner is failing.
• Listen for bearing noise: Grinding or clicking sounds indicate bearing wear. Replace before complete failure.
• Check for blade wobble: With the miner powered off, gently spin each fan by hand. It should spin freely and smoothly with no visible wobble or resistance.
• Connector inspection: Verify the fan connectors are firmly seated on the control board. Loose connectors can cause intermittent fan faults.

Fan replacement procedure:

1. Power off the miner and disconnect from power
2. Remove the enclosure cover
3. Disconnect the fan cable from the control board
4. Remove the 4 screws securing the fan to the enclosure/bracket
5. Slide out the old fan and slide in the replacement
6. Secure with the 4 screws — snug, not overtightened
7. Reconnect the fan cable to the control board (match the original connector position — front fan to front header, rear to rear)
8. Replace the enclosure cover, power on, and verify the new fan reports correct RPM in the dashboard

Diagnostics & Troubleshooting

When an S15 is not hashing correctly, you need to systematically narrow down the fault. Start with the web interface, move to SSH diagnostics, then progress to physical inspection and measurement. Never skip steps — a simple cable reseat fixes more miners than any soldering iron.

LED Status Indicators

The S15 control board has two LED indicators (green and red) that provide immediate visual feedback on the miner’s state.

Web Interface Diagnostics

Access the miner’s web interface by entering its IP address into a browser. Default login: root / root.

On the Miner Status page, check:

• Hashboard count: All 4 boards should appear. Missing boards = connection or hardware issue.
• Chip count per board: Each board should report 72 chips. Fewer chips = dead chip(s) in the chain.
• Real-time hashrate: Should be near 28 TH/s (high-perf) or 17 TH/s (energy-saving). Significantly lower = dead chips, throttling, or hashboard fault.
• Chip temperatures: Normal range is 40°C–85°C. Any chip above 95°C is overheating. Any chip reading 0°C is dead or disconnected.
• Fan RPMs: Both fans should report speed. Zero RPM on either fan = fan failure or connector issue.
• Pool status: Should show “Alive” for your configured pool. “Dead” or “Rejected” = network or configuration issue.

SSH Diagnostic Commands

For deeper diagnostics, SSH into the miner. The S15 runs a Linux-based firmware with several diagnostic tools available via the command line.

# SSH into the S15 (default credentials: root / root)
ssh root@MINER_IP

# View the kernel log for hardware-level messages
dmesg

# View the miner-specific log (varies by firmware version)
cat /var/log/messages

# Check the miner process status
ps | grep bmminer

# View real-time miner output (press Ctrl+C to exit)
tail -f /var/log/messages

# Check detected hashboards and chip counts
cat /var/log/messages | grep "chain"

# Check for ASIC chip errors
cat /var/log/messages | grep -i "error|fail|timeout"

# View hashboard temperatures
cat /var/log/messages | grep "temp"

# Check fan status
cat /var/log/messages | grep "fan"

# View network configuration
ifconfig eth0

# Check if the miner can reach your pool
ping -c 4 stratum.pool.address

# View system uptime and load
uptime

# Check available memory (low memory can cause instability)
free

# From any computer on your network — use curl to pull diagnostics via CGI
# Replace MINER_IP with your S15's IP address

# Get the kernel log
curl -u root:root http://MINER_IP/cgi-bin/get_kernel_log.cgi

# Get miner status (hashrate, temps, chips)
curl -u root:root http://MINER_IP/cgi-bin/get_miner_status.cgi

# Run network diagnostics
curl -u root:root http://MINER_IP/cgi-bin/network_diag.cgi

# Get system info
curl -u root:root http://MINER_IP/cgi-bin/get_system_info.cgi

Common Error Codes & Messages

These are the most frequent error messages you will encounter in the S15 kernel log and web interface. Each one points to a specific failure category.

Hashboard Testing with Multimeter

When a hashboard is not detected or producing low hashrate, you need to test it at the board level. The S15 hashboard has 5 test points labelled TP1 through TP5 that correspond to different voltage domains along the chip chain.

Resistance test (power OFF, hashboard removed):

1. Set your multimeter to resistance mode (ohms)
2. Measure resistance between each test point (TP1–TP5) and ground
3. Expected resistance: approximately 10Ω per test point
4. Significantly higher resistance indicates open circuit (broken trace or dead chip)
5. Near-zero resistance indicates short circuit (failed chip or shorted capacitor)

Voltage test (power ON with test jig):

1. Connect the hashboard to a test fixture with APW8 power supply
2. Measure voltage at each test point while the board is initializing
3. Check signal voltages at the BM1391 chip pins:
   • CLK: 0.9V ±0.1V (both standby and computing)
   • TX (CO): 1.8V during computing, 0V unplugged
   • RX (RI): 1.8V during computing, 0V unplugged
   • RST: 1.8V during computing, 0V standby
4. Voltage difference between adjacent domains should not exceed 0.05V. Larger differences indicate a faulty chip in that domain

Common Repairs

This section covers the most frequent S15 repairs — from simple cable reseats to board-level chip replacement. Repairs are ordered from easiest to most difficult. Always start at the top of this list and work your way down.

Hashboard Not Detected

One or more hashboards missing from the miner status page is the single most common S15 issue. Work through these causes in order:

1. Reseat the flat ribbon cable. Power off, unplug the flat cable connecting the hashboard to the control board, and firmly re-insert it. These cables work loose over time from vibration. This fixes the problem roughly 30% of the time.
2. Replace the flat cable. Flat ribbon cables degrade. Pins corrode, conductors break inside the cable from repeated flexing. Try a known-good cable.
3. Check the IO connector on the hashboard. Inspect the 18-pin header for bent, broken, or corroded pins. A single bent pin on the TX/RX/RST lines will prevent the entire board from being detected.
4. Check Pin 13 (PLUG0). This is the board detection pin. If this line is open (broken trace or corroded pad), the control board will not even attempt to initialize the hashboard.
5. Test with a different hashboard slot. Move the suspect hashboard’s cable to a different slot on the control board. If it comes up on the new slot, the original control board slot is damaged.
6. Inspect the hashboard power connection. The hashboard receives power through heavy-gauge leads from the PSU bus bar. Verify these connections are tight and not discolored from heat damage.
7. Check the boost circuit. If the 23V boost circuit (U7, L4, D4) has failed, the hashboard cannot power its chips. Measure the output of the boost circuit with a multimeter — you should see approximately 23V.

Missing Chips (Low Chip Count)

When the miner detects a hashboard but reports fewer than 72 chips, one or more BM1391 chips have failed or lost communication. This results in reduced hashrate proportional to the number of missing chips.

Diagnosis:

1. Identify which chips are missing from the miner status page. They are usually reported by position number.
2. The S15 chip chain is serial — signals pass from chip #1 to #72. A dead chip in the middle can mask all chips after it in the chain. Start by checking the first missing chip in sequence.
3. Perform a visual inspection of the suspect chip area. Look for:
   • Shifted or detached heatsink
   • Discoloration or burn marks on the PCB near the chip
   • Cracked solder joints visible at the chip edges
4. Measure the voltage domain containing the suspect chip. A voltage significantly different from adjacent domains confirms the chip’s domain has failed.

Repair options (in order of invasiveness):

1. Reflow the chip. Apply flux around the chip’s perimeter, heat with a hot air station at ~350°C, and gently press the chip down. This re-melts cold or cracked solder joints without removing the chip. Success rate: ~40%.
2. Replace the chip. Remove the failed BM1391 with hot air, clean the pads, reball or stencil a new chip, and solder it down. This requires proper BGA rework equipment and experience.
3. Check surrounding passives. A failed 0402 resistor or capacitor near the chip can cause it to fail detection. Measure each passive component near the suspect chip with a multimeter.

Voltage Domain Imbalance

When the voltage across the 12 domains is not evenly distributed, you have a voltage domain imbalance. This indicates that one or more chips within a domain are consuming abnormal current (if the voltage is low) or not conducting at all (if the voltage is high).

Normal domain voltage: ~1.53V per domain, total ~18.36V across all 12 domains.

Diagnosis:

1. Measure voltage across each domain using the test points on the hashboard
2. Any domain with voltage more than 0.05V different from its neighbors needs investigation
3. A domain with higher-than-normal voltage has a chip that is open (not conducting) — current is being forced through fewer chips
4. A domain with lower-than-normal voltage has a chip with excessive leakage current or a shorted component

Repair: Identify and replace or reflow the faulty chip within the affected domain. Check adjacent passive components.

Heatsink Displacement

The S15’s individual chip heatsinks are bonded with thermal adhesive. During shipping, operation vibration, or thermal cycling, heatsinks can shift or detach entirely. A displaced heatsink causes the underlying chip to overheat, leading to throttling, hash errors, or permanent chip death.

Fix:

1. Power off and cool the miner
2. Remove the displaced heatsink
3. Clean old adhesive from both surfaces with IPA
4. Apply fresh thermal adhesive (black 3461 heat-conducting glue)
5. Re-seat the heatsink, ensuring centered placement
6. Allow adhesive to cure before powering on

Power Supply Issues

The APW8 introduces unique failure modes compared to the simpler APW3++/APW7.

APW8 won’t output main voltage rail:

• Verify AC input is 200–240V. The APW8 does not support 110V input. Running it on 110V will produce zero output or unstable voltage.
• Verify the control board is connected and communicating. The APW8 requires the control board to request the main voltage rail.
• Check the 12V standby rail — if this is present but the main rail is not, the PSU electronics are likely fine and the issue is communication with the control board.
• Inspect the PSU output cables for damaged connectors or heat damage.

APW8 making clicking/buzzing sounds:

• Clicking at power-on followed by shutdown = overcurrent protection tripping. This usually means a hashboard has a short circuit that is overloading the PSU.
• Remove all hashboards. If the PSU starts normally with only the control board connected, add hashboards back one at a time to identify the shorted board.

Voltage fluctuation or instability:

• Check AC input with a multimeter at the wall. Unstable mains voltage (brownouts, generator power) causes downstream instability in the APW8.
• Inspect the output capacitors inside the APW8 (if you are comfortable opening a PSU). Bulging or leaking electrolytic caps are a common failure point.

Network & Control Board Issues

Cannot find the miner’s IP address:

1. Press the IP Reporter button on the control board for 1 second. Listen for an audio announcement of the IP address (if your firmware supports this feature).
2. Check your router’s DHCP client list for the miner’s MAC address.
3. Use a network scanner tool (Angry IP Scanner, nmap) to find the device on your subnet.
4. Try connecting with a static IP by temporarily connecting the miner directly to your computer via Ethernet.

Miner boots but no hashrate:

1. Check pool configuration. A typo in the stratum URL or worker credentials will result in the miner running but not submitting shares.
2. Verify DNS resolution. If the miner cannot resolve your pool’s hostname, it cannot connect. Try using the pool’s IP address directly.
3. Check the Ethernet cable. Replace it with a known-good Cat5e cable.
4. Inspect the Ethernet jack on the control board for bent pins or corrosion.

Control board dead (no LEDs, no network):

1. Verify the control board is receiving 12V from the APW8. Check with a multimeter at the control board power connector.
2. Inspect the control board for visual damage — burnt components, cracked solder joints, corrosion.
3. Try a known-good control board. If the miner works with a replacement control board, the original is dead.

Firmware & Software

Firmware on the S15 controls everything — chip initialization, frequency tuning, voltage regulation, fan curves, pool communication, and the web interface. Running outdated or corrupted firmware is a surprisingly common cause of performance issues and instability.

Firmware Updates

Updating via the web interface:

1. Download the latest S15 firmware from Bitmain’s support site or a trusted third-party source (BraiinsOS for advanced users)
2. Log into the miner’s web interface
3. Navigate to System > Upgrade
4. Select the firmware file and click Flash image
5. Wait for the process to complete. Do NOT power off during flashing. Interrupting a firmware write will brick the control board
6. The miner will reboot automatically. Allow 5 minutes for hashboards to initialize

SD Card Recovery

If the firmware is corrupted and the miner cannot boot (no web interface, red LED on boot, boot loop), you can reflash via MicroSD card.

1. Download the correct firmware image for the S15
2. Format a MicroSD card (<16GB) as FAT32
3. Write the firmware image to the card following Bitmain’s flashing instructions
4. Power off the miner
5. Depending on the control board revision:
   • Shift the JP4 jumper to the flash position, OR
   • Short the AB pads on the control board to enable SD card boot
6. Insert the MicroSD card into the control board slot
7. Power on. The firmware will flash automatically (LED may blink rapidly during this process)
8. After flashing completes, power off, remove the SD card, reset the JP4 jumper/AB pads to normal position, and power on

Factory Reset

A factory reset clears all custom configuration (pools, frequencies, network settings) and returns the miner to its default state. This can resolve issues caused by misconfiguration or corrupted settings.

Method 1: Web interface

1. Log into the web interface (root / root)
2. Navigate to System > Factory Reset
3. Click OK to confirm
4. Wait for the miner to reboot (several minutes)
5. Reconfigure pool settings and network after reboot

Method 2: IP Reporter button

1. With the miner powered on, press and hold the IP Reporter button for 10 seconds
2. Release the button. The miner will reboot and reset to factory defaults
3. The miner will revert to DHCP. Check your router for its new IP address

Configuration Best Practices

• Always set 3 pool URLs — primary, secondary, and tertiary. If your primary pool goes down, the miner fails over automatically instead of sitting idle
• Use energy-saving mode if your electricity cost exceeds $0.08/kWh. The efficiency gain from 57 J/TH to 50 J/TH can make the difference between profitable and unprofitable operation
• Set a static IP address for the miner to prevent IP changes after router reboots that can make the miner unreachable
• Change the default password — leaving the miner on root/root is an open invitation for anyone on your network to reconfigure or hijack it
• Monitor regularly — set up monitoring alerts (via your pool’s dashboard, BTCTool, or a custom script) to be notified immediately if hashrate drops

Advanced Board-Level Repair

This section is for technicians with board-level rework experience. If you do not have a hot air rework station, BGA stencils, and at least a year of electronics repair experience, skip to the When to Call a Professional section.

Signal Tracing Methodology

When a hashboard has missing chips or is not detected, and cable/connector issues have been ruled out, the next step is signal tracing. The S15’s serial chip chain means that a fault at any point in the chain blocks all signals after it.

Tracing procedure:

1. Start with CLK. The 25MHz clock from crystal Y1 propagates from chip #1 to #72 in series. Using an oscilloscope, verify the clock signal at each chip along the chain. The point where the clock disappears is your fault location.
2. Check TX (CO). The TX signal propagates forward from chip #1 to #72. Verify the signal passes through each chip.
3. Check RX (RI). The RX signal propagates in reverse — from chip #72 back to chip #1. Follow this signal backward from the last chip.
4. Check RST and B signals along the forward path.
5. Measure VDD, VDD_0V8, VDD_1V8, and VDD_2V5 at the suspect chip. All should match the expected voltages for that chip’s domain. A missing or abnormal voltage on any rail points to the specific failed component.

BM1391 Chip Replacement

1. Remove the heatsink from the failed chip. Heat gently with a heat gun at ~100°C to soften the thermal adhesive, then pry with a plastic spudger.
2. Apply flux around the chip perimeter.
3. Heat with hot air at 350°C–380°C until solder melts. Gently lift the chip with tweezers.
4. Clean the pads on the PCB using desoldering wick and IPA. Inspect for lifted pads or damaged traces.
5. Prepare the replacement chip. If using a BGA reballing stencil, align the stencil and apply 0.4mm solder balls with paste. Reflow on a preheater to create clean, uniform solder bumps.
6. Place the new chip on the cleaned pads. Ensure proper orientation (the chip has an alignment mark).
7. Reflow with hot air. Apply flux, heat evenly at 350°C until the solder wets and the chip self-aligns.
8. Inspect solder joints visually and with a multimeter (check for shorts between adjacent pins).
9. Apply thermal gel to the chip surface and re-attach the heatsink with thermal adhesive.
10. Test the board using the test fixture. Run at least two test cycles with cooling time between them.

Post-Repair Validation

After any board-level repair, validation is critical. A chip may pass initial power-on but fail under sustained load.

1. First test: Immediately after repair, connect to test fixture and run full chip detection and hash test. All 72 chips should report in.
2. Cooling period: Let the board cool completely (15–20 minutes).
3. Second test: Run the full test again. Thermal cycling between hot and cold can reveal marginal solder joints that passed the first test.
4. Extended burn-in: Install the hashboard in the miner and run it for 24 hours. Monitor for hash errors, temperature anomalies, or chip drops during this burn-in period.
5. Document the repair. Record the failed component, its position, the replacement part used, and the root cause. This data is invaluable for identifying batch defects or recurring issues.

S15-Specific Failure Modes

The S15 has several failure modes that are unique to its design or significantly more prevalent than in other Antminer generations. Understanding these helps you prioritize your diagnostic approach.

BM1391 Chip Degradation

The BM1391BE was Bitmain’s first mass-produced 7nm chip. Like many first-generation node shrinks, it shows higher degradation rates than its successors. S15 miners that have been running for several years often exhibit gradual chip death — losing 1–3 chips per hashboard over time. Each lost chip reduces hashrate proportionally.

Signs: Gradual hashrate decline over weeks/months. Chip count slowly decreasing on the dashboard. No single catastrophic event.

Mitigation: There is no way to prevent chip degradation on already-manufactured chips. Ensure optimal thermal management to slow the process. Replace dead chips as they fail, or accept the reduced hashrate. If more than 10–15% of chips on a board have failed, the board becomes uneconomical to repair.

Thermal Adhesive Failure

The S15’s individual-chip heatsink design with thermal adhesive is a known weakness. The adhesive used in factory production degrades faster than the thermal interface materials used on later models. Heatsink detachment cascades — one overheating chip raises the temperature of its neighbors, accelerating adhesive degradation on adjacent heatsinks.

Prevention: During routine maintenance, check heatsink adhesion by gently pressing each heatsink. Any that rock or shift need to be re-bonded immediately, before they detach during operation.

APW8 Failure Patterns

The APW8 PSU is less robust than the later APW9/APW9+. Common APW8 failures include:

• Capacitor degradation: Internal electrolytic capacitors dry out, especially in hot environments. Symptoms: voltage ripple, random miner restarts, clicking sounds.
• Fan failure inside the PSU: The APW8 has its own internal cooling fan. When it fails, the PSU overheats and enters protection shutdown.
• Connector heat damage: The output connectors carry high current. Poor connections at the hashboard power bus cause resistive heating, which damages the connector and the cable, creating a feedback loop of increasing resistance and heat.

Flat Ribbon Cable Fragility

The flat ribbon cables connecting S15 hashboards to the control board are more fragile than those used in later models. They are prone to conductor breaks from repeated bending, and their connectors corrode more easily in humid environments. Keep spare cables on hand — they are cheap insurance against a common failure.

Frequently Asked Questions

When to Call a Professional

There is a clear line between DIY maintenance and professional repair. Here is when to stop and send the miner to someone who does this for a living:

• Multiple hashboards dead. If 2 or more of the 4 hashboards are not detected and cable reseats have not fixed it, you likely have board-level failures that require professional diagnostics.
• Chip replacement needed. BGA rework on 7nm chips is not a kitchen-table job. You need a preheater, a calibrated hot air station, BGA stencils, and experience. One slip damages the PCB pads and turns a $50 repair into a $200 one.
• PSU making abnormal sounds. If the APW8 is clicking, buzzing, or emitting a burnt smell, do NOT open it. PSUs contain high-voltage capacitors that can deliver a fatal shock even when unplugged. Send it for repair or replace it.
• Control board dead. If the control board shows no signs of life (no LEDs, no network) after verifying power supply, it likely has a failed component that requires micro-soldering.
• Firmware bricked. If SD card recovery has failed and the miner will not boot, the control board’s NAND flash may need to be reprogrammed with specialized equipment.
• You are not confident. This is the most important one. There is no shame in admitting a repair is beyond your current skill level. A professional repair costs far less than a destroyed hashboard.

Maintenance Workflow Summary

Here is the complete diagnostic and maintenance workflow distilled into a decision tree. Follow it top-to-bottom whenever your S15 needs attention.

1. Check the dashboard. All 4 hashboards detected? Full chip counts? Temperatures in range? Fan RPMs normal? Pool connected? If yes → the miner is healthy. Continue routine maintenance.
2. LED is red or no LEDs? Check power connections first. Verify AC voltage at the wall. Verify APW8 is producing 12V standby. If no standby voltage → PSU is dead. Replace the APW8.
3. Hashboard missing? Reseat the flat cable. Try a different cable. Try a different control board slot. Inspect IO connector pins. Check boost circuit voltage (23V). If all pass → hashboard has a board-level fault.
4. Low chip count? Identify the first missing chip. Visual inspect the area. Check domain voltage. Reflow the chip. If reflow fails → replace the chip (or send for professional repair).
5. High temperature? Check fans (RPM and physical spin). Clean dust. Check heatsink adhesion. Reduce ambient temperature. Switch to energy-saving mode. If all pass → thermal interface has degraded; re-paste heatsinks.
6. No network? Check Ethernet cable. Check router. Try a static IP. Scan your subnet. If no response → control board network interface may be dead.
7. Unstable hashrate? Check pool connection stability. Check for temperature fluctuations. Update firmware. If the issue persists, a hashboard may have marginal chips that hash intermittently → test each board individually.

Final Notes

The Antminer S15 was a milestone machine. It brought 7nm efficiency to Bitcoin mining before most of the world even knew what a nanometer was. It is not the fastest or the most efficient miner on the market today, but for home miners with low electricity costs — especially those of us in Canada where winter is long and heating bills are real — it still has a place.

Maintain it properly, keep those heatsinks bonded, keep the dust out, and it will keep hashing. And when a board-level repair is beyond your tools or your time, remember that D-Central has been fixing these machines since before most people knew Bitcoin mining was something you could do at home.

Stay sovereign. Keep hashing. Decentralize the hash.

Questions? Reach us at 1-855-753-9997 or contact us online. We are based in Laval, Quebec, and we ship parts and repaired miners across North America.

Interactive Hashboard Schematic

Explore the ANTMINER S15 hashboard layout below. Toggle layers to isolate voltage domains, signal chains, test points, key components, and thermal zones. Hover over any region for quick specs — click for detailed diagnostics, failure modes, and repair guidance.