Welcome to this Antminer T19 maintenance guide, provided by D-Central, your reliable source for critical ASIC repair resources. We have put a lot of effort into creating this guide and would appreciate your support. If you find this guide helpful, please consider giving us a shoutout, a review, a share, a subscription, or a tip. Bitcoin mining is our passion, and we are thrilled to share our knowledge with the world. This guide provides a comprehensive overview of the steps involved in maintaining an Antminer T19, including visual inspection, testing with a test fixture, troubleshooting faulty chips, and re-soldering them. Additionally, we outline other important matters that require attention when conducting maintenance on the Antminer T19. By carefully following these steps and taking preventive measures, you can ensure that your miner runs at peak performance for extended periods.
Preparation and Maintenance Guidelines
It’s essential to take the time to properly prepare and maintain components before, during, and after installation. This includes applying thermal gel for better heat transfer, forming air ducts for better airflow, connecting power supplies in the correct sequence, fixing chips to prevent overheating, and ensuring test fixtures meet production requirements. Additionally, these guidelines should also include instructions on cleaning components with approved solvents such as isopropyl alcohol or distilled water, as well as how to store components away from extreme temperatures and humidity levels safely. Finally, regularly scheduled maintenance checks should be carried out every few months or at least annually to guarantee the proper functioning of all parts within the system.
Preparation Requirements for Repair Platform, Tools, and Equipment
I. Platform Requirements
To perform maintenance work, an anti-static maintenance workbench is required. It should be grounded, and an anti-static wristband and grounding are necessary.
II. Equipment Requirements
Constant temperature soldering iron (350°C-380°C) with a pointed tip for soldering small patches such as chip resistors and capacitors.
Hot air guns and BGA rework stations are used for chip / BGA disassembly and welding.
Multimeter with welded steel pins and heat-shrinkable sleeves for easy measurement. The recommended model is the Fluke 15b+ multimeter.
Oscilloscope. The recommended model is UTD2102CEX+. A network cable is required for an internet connection and a stable network.
III. Test Tool Requirements
ARC Kit
ARC Antminer Hashboard Tester
Lab PSU 10-30V / 1-15A
Bitmain Kit
APW12 power supply: AP12_12V-15V_V1.2 and power adapter cable. It is recommended to use thick copper wire for the positive and negative poles of the power supply to connect the power supply and the power board and only limited to PT1 and maintenance test use.
Use the test fixture of the V2.3 control board (test fixture material number ZJ0001000001). The positive and negative poles of the test jig need to be installed with discharge resistors. Using a cement resistance of 20 ohms and 100W or more is recommended.
IV. Maintenance Auxiliary Materials/Tools Requirements
Mechanic lead-free circuit board cleaner cleans up the flux residue after maintenance.
Thermally conductive gel is used to apply to the chip surface after repair.
Ball-planting steel mesh, desoldering wick, and solder balls (the recommended ball diameter is 0.4mm).
When replacing a new chip, it is necessary to tin the chip pins and then solder them to the hash board. Apply thermally conductive gel evenly on the chip’s surface, then lock the heatsink.
Serial port code scanner.
Serial port adapter board RS232 to TTL adapter board 3.3V.
Self-made short-circuit probe (use the pins for wiring and welding and heat the shrinkable sleeve to prevent short-circuit between the probe and the small heatsink).
V. Common Maintenance Spare Material Requirements
0402 resistor (0R, 10K, 4.7K,)
0402 capacitor (0.1uF, 1uF)
Maintenance Requirements
When replacing a chip, pay attention to the operation method. After replacing any component, check that the PCB board has no obvious deformation. Check the replacement and surrounding parts for missing, open, and short circuits.
Maintenance personnel must have electronic knowledge, at least one year of maintenance experience, and proficiency in BGA/QFN/LGA packaging and welding technology.
After repair, the hashboard must be tested more than twice, and all tests must pass.
Check the tools to ensure that the test fixture can work typically. Determine the parameters of the maintenance station test software, the version of the test jig, and other related parameters.
To test repairing and replacing the chip, test the chip first and then do the functional test after it passes. The functional test must ensure that the small heatsink is welded correctly, the large heat sink is installed, and the thermal adhesive gel is applied evenly. Two hash boards should be placed simultaneously to form an air duct when using the chassis to dissipate heat. For single-sided testing in production, the air duct must also be formed.
When measuring the signal, use fans to dissipate heat and ensure the fans are at full speed.
When powering on the hashboard, connect the negative copper cord of the power supply first, then the positive copper cord of the power supply, and finally, insert the signal cable.
When disassembling, reverse the order of installation. First, remove the signal cable, then pull the positive copper cord of the power supply, and finally, remove the negative copper cord. If you do not follow this order, it may cause damage to U1 and U2.
Before testing the pattern, the repaired hashboard must cool down before testing; otherwise, it will lead to testing NG.
Pre-tin the chip pins with solder paste to replace a new chip and then solder them to the PCB for repair.
Overview of Antminer T19 Components
T19 Hashboard Structure
The T19 hash board comprises 76 BM1398 chips, divided into 38 groups of 2 ICs each. The BM1398 chip used in the T19 hash board operates at a voltage of 0.36-0.38V. The 32-38 groups are powered by 19V DCDC, and the 38th domain’s VDD13.64V provides an output of 1.8V to the LDO from the DCDC of the 31st domain, which then provides 1.8V to all other domains. As the voltage passes through each domain, it reduces by 0.36V until the 1.8V domain provides 0.8V output through the LDO.
T19 Hashboard Boost Circuit
The boost circuit of the T19 hash board is powered by a power supply that ranges from 14V to 19V.
T19 Chip Signal Direction
The CLK (XIN) signal is generated by the Y1 25M crystal oscillator. It flows from chip No. 01 to chip No. 76. During operation, the voltage is 1.8V (measured by an oscilloscope), and the multimeter measures about 0.7-1.2V.
The TX (CI, CO) signal flows from the IO port 7 pin (3.3V) to the level conversion IC U2 and then transmits from chip No. 01 to chip No. 76. When the IO line is not inserted, the voltage is 0V, and during operation, the voltage is 1.8V (measured by an oscilloscope).
The RX (RI, RO) signal flows from chip No. 76 to chip No. 01, through U1, back to the 8th pin of the signal cable terminal, and then back to the control board. When the IO signal is not inserted, the voltage is 0.3V; during operation, the voltage is 1.8V (measured by a multimeter).
The BO (BI, BO) signal flows from chip No. 01 to chip No. 76. The multimeter measures 0V.
The RST signal flows from the IO port 3 and then transmits from chip No. 01 to chip No. 76. During standby, if no IO signal is inserted, it is 0V, and during operation, it is 1.8V (measured by a multimeter).
Antminer T19 Structure
The miner comprises three hash boards, one control board, an APW12 power supply, and four cooling fans.
Identifying Common Issues with Hashboards and Troubleshooting Procedures
Phenomenon 1: The detection chip on a single hash board test shows 0
Check the power output
Check the power supply input to the hash board.
Measure the voltage output of the power supply using a multimeter.
Ensure that the voltage output matches the required voltage level for the hash board.
If the voltage is too low or non-existent, check the power cable connection and power supply unit.
If the voltage is too high, check the power supply unit for any faults or damage.
If the power supply is functional and the voltage output is correct, move on to the next troubleshooting step.
Check the voltage domain voltage output
The voltage of each domain is about 0.36V, and the 14V power supply generally has a domain voltage. Test the output of the power supply terminal of the hash board and check whether the MOS is short-circuited (measure the resistance between pins 1, 4, and 8). If 14V is powered, but there is no domain voltage, continue troubleshooting.
Identify the domains and their respective voltage levels.
Measure the voltage output of each domain using a multimeter.
Ensure that the voltage output matches the required voltage level for each domain.
If the voltage is too low or non-existent, check the circuit for faults, such as short circuits or damaged components.
If the voltage is too high, check the circuit for any faults or damaged components that may be causing the voltage to be increased.
If the voltage output is incorrect, continue troubleshooting to identify and fix any underlying issues that may be causing the problem.
Check the PIC circuit
Measure whether the second pin of U3 has output. If there is no 3.3V output, check whether the connection status of the test fixture cable and the hash board is good, and re-burn the PIC record. When the hash board is powered, the 14V voltage first passes through the U10 circuit and outputs a voltage of about 21V to supply power to the load circuit of the R41 and R42 resistors. Pin 11 of PIC U3 will output a voltage of about 3V to supply power to pin 1 of Q4 (the voltage of pin 1 of Q4 to pin 2, measured by a multimeter). After pin 1 of Q4 receives the voltage, it will turn on pin two and pin three and pull down the voltage of pin 1 of Q5 so that pins 2 and 3 of Q5 will be turned on. Finally, the 4-pin batteries of Q2, Q3, Q6, and Q7 are pulled up to 21V and turned on, and then the hash board has the domain voltage.
Check the boost circuit output
Identify the C55 component on the hash board.
Measure the voltage output of C55 using a multimeter.
Ensure that the voltage output is around 19V.
If the voltage is too low or non-existent, check the boost circuit for faults, such as short circuits or damaged components.
Check the power supply unit to ensure it provides the required voltage input to the boost circuit.
If the voltage output is still not correct, continue troubleshooting to identify and fix any underlying issues that may be causing the problem.
Check each group of LDO 1.8V or PLL 0.8V output
Identify the groups of LDO 1.8V or PLL 0.8V on the hash board.
Measure the voltage output of each group using a multimeter.
Ensure that the voltage output matches the required voltage level for each group.
If the voltage is too low or non-existent, check the circuit for faults, such as short circuits or damaged components.
If the voltage is too high, check the circuit for any faults or damaged components that may be causing the voltage to be increased.
If the voltage output is incorrect, continue troubleshooting to identify and fix any underlying issues that may be causing the problem.
Check the signal output of the chip (CLK/CI/RI/BO/RST)
If the chip’s signal pins’ output voltage is normal, and the chip’s detection is still incomplete, short-circuit RO and 1.8V, and then check the detection of the chips. If 74 chips can be detected after short-circuiting, then 74 chips can be short-circuited until a certain chip is short-circuited, and the fixture also reports this chip. This indicates a problem with the next chip.
Identify the signal pins of each chip on the hash board (CLK/CI/RI/BO/RST).
Measure the voltage output of each signal pin using a multimeter.
Refer to the signal direction description for the voltage range, and ensure that the voltage output falls within the correct range.
If the measurement encounters a large deviation of the voltage value, compare it with the measurement value of the adjacent group.
If the voltage output is not correct, check the circuit for any faults, such as short-circuits or damaged components.
If there are any issues with the signal output, troubleshoot and fix the underlying issues that may be causing the problem.
Phenomenon 2: The single-board detection chip is incomplete
To troubleshoot a chip error, it’s essential to check the relevant signals (CLK/CI/RI/BO/RST) both before and after the error reporting position. If any signal is abnormal, it can help to locate the bad position by identifying which integrated circuit (IC) is responsible for the problematic signal. Then, the signal direction and voltage value range can be examined and maintained.
However, simply checking the chip’s output voltage of the signal pins is not always sufficient. For instance, if 75 chips are detected, and there is still an issue, a short-circuit test can be performed by connecting RO and 1.8V. If only 74 chips can be detected after short-circuiting, it suggests that one of the chips is the problem. The process can be repeated by short-circuiting 74 chips until a particular chip is identified. The test fixture will report the problematic chip, and it can be determined that the next chip has an issue.
In summary, when troubleshooting a chip error, it’s crucial to check relevant signals, locate the problematic IC, and maintain the signal direction and voltage range. Additionally, short-circuit testing can help identify which chip is causing the problem if a simple voltage check is insufficient.
Phenomenon 3: Single board Pattern NG, that is, the reply nonce data is incomplete (PT2 station)
When encountering a pattern NG error, it indicates that the characteristics of the chip are different from those of other chips, and the solution is to replace the chip. Based on log information, the replacement rule is to replace the chip with the lowest response rate in each domain.
For example, in one of the test logs, it can be seen that the recovery rate of asic[36][37][43][75] is low. Since 36 and 37 are in the same domain, the chip with the lowest nonce in that domain should be replaced. Additionally, both chips 43 and 75 should be replaced as well. It’s important to note that domain numbers and ASICs start from 0.
In the case of a PATTERN-NG error, the screen displays a constant refresh with the message BTC_check_register: reg_value_buf buffer is full! This issue indicates at least one defective chip on the PCBA. To repair it, a 50M single-board PT2 can be used to test the firmware and perform a single-board test.
During testing, when the LOG displays chip detection of 76 chips and starts to refresh the screen, a short-circuit pin is used to short-circuit one of the chips. If the short-circuited chip is not the issue, the next chip can be tested by short-circuiting its RO input and 1.8V for 15 seconds. If the screen stops refreshing after 15 seconds, it proves that the tested chip is not the problem. However, if the screen continues to refresh after short-circuiting the 6th chip, it indicates a problem with that particular chip. The issue could be due to incorrect soldering or a defective chip.
Phenomenon 4: Abnormal temperature reading during the test (PT2 station)
To troubleshoot temperature-related issues, there are a few steps to follow. First, check whether the matching resistors are the four temperature sensors: U4, R28R30, U6, R31R33, U7, R34R36, U8, R37R39. It’s important to ensure that the matching resistors are not abnormally welded. These components are typically located on the front of the PCB, and it’s crucial to ensure that the temperature-sensing 3.3V power supply is functioning correctly.
Next, it’s essential to inspect the welding quality of the heat sink on the opposite side of the chip connected to the temperature sensor. If the heat sink is not welded correctly, it can significantly impact the temperature difference and cause temperature-related issues.
In summary, to troubleshoot temperature-related problems, it’s crucial to check the quality of the matching resistors and ensure that they are correctly welded. Additionally, it’s important to inspect the heat sink’s welding quality, which is connected to the temperature sensor on the opposite side of the chip.
Troubleshooting Common Miner Failures
Preliminary test of the whole machine
Referring to the test process document, the general problems that arise are assembly process problems and control board process problems. Common problems include the inability to detect the IP, abnormal number of detected fans, and abnormal detected chain. If any issues arise during the test, they should be repaired according to the monitoring interface and the test LOG prompts. The maintenance methods for the initial test and the aging test of the whole machine are the same.
Aging Testing of the Miner
The aging test should be repaired according to the monitored interface test. For instance:
Abnormal fan display: In such cases, we need to check whether the fan works normally, whether the connection with the control board is normal, and whether the control board is abnormal.
Less chain: This means that three boards are missing one board. In most cases, there is a problem with the connection between the hash and control boards. Check the cable to see if there is an open circuit. If the connection is OK, you can test the board to PT2 to see if it can be tested. If it can be tested, it can basically be determined that it is the control board. If the test fails, use the repair method of PT2 maintenance.
Abnormal temperature: Generally, the temperature is high. The maximum PCB temperature set by our monitoring system cannot exceed 90℃. The fan will alarm and it will not work normally. Generally, the ambient temperature is too high, and the abnormal operation of the fan will also cause abnormal temperature.
Insufficient number of chips: If the number of chips is insufficient, you can refer to PT2 for testing and repair.
After running for a period of time, there is no hashrate, and the connection of the mining pool is interrupted, check the network.
If the miner still loses hashrate, reduce the frequency and other conditions remain unchanged. Let the miner mine to see if it will lose hashrate and whether the hash board will hit X. If it still hits X in losing hashrate, then remove the heat sink of the hashboard for mining and wait for the hashrate to drop. Measure whether the domain voltage is normal. Generally, the domain voltage will be abnormal in the problematic domain. Then measure the RI signal to see if the RI signal is broken. If the RI signal is missing, basically, the chip is short-circuited or damaged after being tinned.
Other Considerations and Maintenance Flow Chart
To conduct a routine inspection of a hash board, there are several steps to follow. First, visually inspect the board for signs of PCB deformation and scorching. If there are any such signs, they must be addressed first. Additionally, check for any parts with burn marks, missing parts, or impact offset.
Next, perform a visual inspection after the routine inspection to ensure there are no issues. Test the impedance of each voltage domain to detect any short circuits or open circuits. If found, it must be addressed before proceeding. Also, ensure that the voltage of each domain is approximately 0.36v.
Once the routine inspection is complete, the chip can be inspected with a test fixture. Determine the location of the faulty chip based on the test results of the fixture. Check the chip test points (CO/NRST/RO/XIN/BI) and ensure that VDD0V8 and VDD1V8 have equal voltage.
Next, transmit the RX signal in the reverse direction (No. 76 to No. 1 chip), and several signals such as CLK, CO, BO, and RST in the forward direction (No. 1 to No. 76 chip) according to the signal flow. This can help locate the abnormal fault point through the power supply sequence.
When the faulty chip is located, it needs to be re-soldered. Apply flux (preferably no-clean flux) around the chip and heat the solder joints of the chip pins to a dissolved state. This promotes the chip pins and pads to re-run and close the tin to achieve re-soldering. If the fault persists after re-soldering, the chip can be replaced directly.
After repairing the hash board, it’s essential to perform more than two passes to judge it as a good product when testing the test jig. After the first replacement of the parts, wait for the hash board to cool down, use the test fixture to test the pass, and put it aside before cooling. Wait a few minutes for the hash board to cool down before running a second test.
In summary, to inspect and repair a hash board, it’s crucial to follow a specific process that includes visual inspections, impedance tests, test fixture inspections, and re-soldering. Additionally, multiple passes are required to judge the repaired board as a good product.
Disclaimer: The information provided on this blog is for informational purposes only and should not be taken as any form of advice.
Ready to become an ASIC expert?
The D-Central team is a well-established and recognized authority in matters of ASIC repair. Our Training Course is tailored to equip individuals with the skills they need to tackle even the toughest repair jobs with confidence. The consulting services we offer provide our clients with expert guidance and support, allowing them to get the most out of their repair sessions. By bringing their own equipment, participants can learn to fix up to five broken devices during each session, making the trip pay for itself. Don’t let technical details hold you back.
