This manual demonstrates how to disassemble and fix the Bitmain Antminer S19 Series hash board. Certain electronic knowledge is required of the maintenance operators.
Care must be taken when tearing down an Antminer S19 and making the most of your hash board testers. The supporting fixture of the hash board tester should allow the hash board to dissipate heat and facilitate signal measurement. Update the hash board tester control board FPGA for the first time using the SD card swiping teardown program for the Antminer S19 series.
The file should be extracted, copied to an SD card, and then put in the card slot on the hash board tester. For roughly a minute, turn the device on. The update is complete when the control board indicator double flashes three times (if it is not updated, it may result in a particular chip failing the test if it is not updated).
Table of Contents
Platform/tool/equipment Preparation Requirements
1. Platform requirements:
Grounding, an anti-static wristband, and a rubber blanket repair workbench are required.
2. Equipment requirements:
Heat gun, BGA rework station, multimeter with soldering steel pin and heat-shrinkable T bush, oscilloscope (Agilent recommended), network cable (requirements: Internet connection, stable network) are used for chip/BGA assembly and soldering. Small patches, such as chip resistors and capacitors, are soldered using a constant temperature soldering iron (350°C – 380°C) and pointed soldering iron tip.
3. Requirements for test tools:
Power adapter cable (homemade: connect the power supply at the positive and negative poles with thick copper wires, and the hash board with the APW12 power supply, APW12_12V-15V_V1.2). The hash board should be made of 4AWG copper wires that are 60 cm or shorter in length. Use the V2.2010 control board’s test fixture. Discharge resistors must be installed on the test fixture’s positive and negative poles. It is recommended to use a cement resistor of 25 ohms and more than 100W.
4. Maintenance auxiliary materials/tool requirements:
Thermal gel (specification: Fujipoly SPG-30B) is used to spread the surface of the chip after repair; tin-planting steel mesh, solder absorption wire, and solder ball (the ball diameter is recommended to be 0.4mm); when replacing a new chip, you need to tin the chip pins and then solder to the hash board, and then locate it.
5. Common maintenance spare material requirements:
0402 resistance (0R, 51R, 10K, 4.7K,); 0402 capacitor (0.1uf, 1uf)
Antminer S19 Teardown and Issues That Require Care
1. When changing the chip, pay close attention to the procedure. The PCB board shouldn’t be visibly deformed after any accessories have been replaced. For open circuit and short circuit issues, examine the replacement parts as well as the nearby components.
2. The maintenance technicians must be experts in BGA/QFN/LGA package soldering technology, have a certain level of electronic knowledge, and more than a year of maintenance experience.
3. The Antminer S19 hash board needs to pass more than two tests after being repaired in order to be approved!
4. Check whether the tools and hash board testers can work normally, determine the maintenance station to test software parameters, test fixture versions, etc.
5. The chip must first be tested in the test for repairing and replacing the chip, and only after passing should the function test be carried out.
The function test must ensure that the small heat sink is welded qualified. Thermal gel must be evenly applied to the chip’s surface before the large heat sink is installed, and the cooling fan must be running at full speed. Two hash boards should be placed side by side to create an air duct when using the chassis as a heat sink.
6. Use four fans to help with heat dissipation while measuring the signal, and make sure the fans are running at full speed.
7. The user must connect the signal cable, the positive copper wire, and the negative copper wire of the power supply before turning on the hash board. The order of installation must be reversed when removing. Remove the signal cable first, then the power supply’s positive copper wire, and finally the power supply’s negative copper wire. It is very simple to damage R8, R9, U1, and U2 if the user doesn’t follow this order (not all chips can be found). The repaired hash board needs to cool down before testing the pattern; otherwise, PNG testing may result.
8. Printing pins and soldering paste are necessary to ensure that a new chip is pre-soldered and then soldered to the PCBA for repair when replacing an old one.
Hash Board Tester Making and Matters Needing Attention
The supporting fixture of the hash board tester should allow the hash board to dissipate heat and make signal measurement easier.
1. For the first time, update the hash board tester control board FPGA using the 19 series hash board tester SD card swiping program. Unzip the file and copy it to the SD card, then place the card in the hash board tester card slot. Power on the device for about a minute, then wait for the control board indicator to double flash three times, signaling the end of the update (if it is not updated, it may result in a particular chip failing during the test).
2. Make the test SD card according to the requirements, and directly unzip the compressed package of the single-sided heat sink inspection chip to make the SD card;
3. The test SD card will be made according to the requirements, and the double-sided heat sink 8-times patter test needs to make an SD card, as shown in the figure below;
S19 Hash Board’s Operational Structure:
The BM1398 chip used in the S19 hash board has an operating voltage of 0.36V, and for the 38th, 37th, 36th, 35th, 34th, 33rd, and 32nd groups (a total of 7 groups), the LDO is powered by the 19V output from the boost circuit U9 and outputs 1.8V. There are 76 BM1398 chips on the hash board, which are organized into 38 groups (domains) that each contain two ICs.
The voltage of each domain that has retreated is reduced by 0.36V, and the 30th group and the first group are supplied by VDD13.64V through the LDO to produce 1.8V. Figure 4-1 illustrates how the 1.8V output of this domain uses LDO to supply the entire 0.8V;
Circuit for An S19 Hash Board Boost and S19 Chip Signal Trend:
The boost circuit, as previously shown, increases the power supply’s output voltage from 14V to 19V.
1) TX (CI, CO) signal flow direction, from IO port 7 pin (3.3V) into IC U2 through level conversion, and then transmitted from chip 01 to chip 76; the voltage is 0V when the IO signal is not inserted, and the voltage is 1.8V during operation;
2) CLK (XIN) signal flow direction, generated by Y1 25M oscillator, transmitting from chip 01 to chip 76; voltage of 0.7V-1.3V; and 2)
3) When the IO signal is not inserted, the voltage is 0.3V, and the voltage will be 1.8V during computation;
2) RX (RI, RO) signal flow direction, from chip 76 to chip 01, return to the signal cable terminal pin 8 through U1, and then return to the control board;
4) The multimeter measurement value for the BO (BI, BO) signal flow direction, from chip 01 to chip 76, is 0V;
5) If no IO signal is inserted and the equipment is in standby, the voltage is 0V, and when computing, the voltage is 1.8V; the RST signal flow is from pin 3 of the IO port to chip 01;
A Complete Miner’s Architecture
As seen above, the key components of the entire miner are the three hash boards, one control board, the APW12 power supply, and the four cooling fans.
Common issues and hash board troubleshooting techniques
Single board test detection chip is 0 (PT1/PT2 stations), phenomenon
Examine the area below that is circled to determine the power output in the first step.
The following action is to examine the voltage output in the voltage domain.
The voltage in each voltage domain is roughly 0.36V. Most of the time, domain voltage is present when there is a 14V power source. The output of the hash board’s power supply terminal and whether the MOS is shorted out (measure the resistance between pins 1, 4, and 8) are given priority. Continue checking whether 14V has power supply but no domain voltage.
The third step entails inspecting the PIC circuit.
Continue troubleshooting if there is output and the voltage on U3’s second pin is about 3.3V. If not, check that the hash board tester cable is properly connected, that the hash board is in good shape, and that the PIC needs to be programmed.
Steps for Programming a PIC in a Antminer S19 Teardown
1. A hash board is used to program a PIC.
Get the programming device here: PICkit3. Pin 1 of the PICkit3 cable, along with pins 1, 2, 3, and 4, must be connected to J3 on the PCB.
2. Computer programming tools
Start MPLABIPE, then select a device. Select the PIC16F1704’s power supply mode by clicking power, and then click operate.
First, choose file to locate the.HEX file to be programmed; next, choose connect to establish a regular connection;
The final step is to select “program” and then “verify” to trigger a verification to show that the programming was successful.
3. Check that C55 in the boost circuit output test shown in Figure 4-9 is capable of measuring a voltage of 19V.
4. check the output of each group of LDO 1.8V or PLL 0.8V
5. Check the chip signal output for CLK, CI, RI, BO, and RST.
Refer to the voltage value range that the signal trend describes. If there is a significant voltage value deviation during the measurement, it is possible to compare the measured value of the following group with the measurement.
PS: the Chip Will Report 0 If the Hash Board is Not Powered Or Switched Off in Accordance With the Test Sequence, Leading to the Burnout of R8, R9, U1, and U2;
- Verify that the welding of U5 is normal when the EEPROM NG is presented on the LCD screen of the hash board tester;
- If the test read temperature is abnormal and the PIC sensor NG is displayed on the LCD screen of the hash board tester, then troubleshoot as follows:
A) Verify the normality of the welding of PINs 2, 3, and U3 as well as the welding of the four resistors of R24–R27;
B) Verify the normal operation of the temperature-sensitive 3.3V power supply as well as the four temperature senses U4, R28–R30, U6, R31–R33, U7, R34–R36, U8, R37–R39, and the matching resistance welding, which are all placed on the back of the PCB and are all illustrated in Figure 5-14.
3. The little heat sink and the heat-sensitive chip should both be checked for welding quality. The chip will have trouble dissipating heat, and the massive heat sink material’s deformation will have an effect on the temperature differential.
4. Observation: The PT1/PT2 stations’ single board detection chip is incomplete.
a) If (0), measure the total voltage of the measurement domain and the boost circuit is normal, short-circuit the RO test point and the 1V8 test point between the first and second chips using the short-circuit probe, and then use the software to locate the chip. If 0 chip is still there at this point after looking at the serial port log,
One of the Following Scenarios Will Occur:
a-1) Utilizing a multimeter, determine whether the voltages at the 1V8 and 0V8 test points are 1.8V or 0.8V. The majority of these problems are caused by short circuits of 0.8V and 1.8V patch filter capacitors (measure the resistance of the patch filter capacitors related to the front and back of the PCBA). If not, it means that either the two ASIC chips in this domain are improperly soldered or the 1.8V or 0.8V LDO circuit is abnormal.
A-3) Check the resistance of R8 or R9 with a multimeter to see if it is within 10 ohms and that the reading is stable. Should it not be necessary, swap out these two resistors.
a-4) Inspect the first chip’s pins to see if they are properly soldered (it was found in repair that the pins are tinned observing from the side, but the pins are not stained with tin at all when the chip is removed).
b) When only one chip can be found in step a), the prior circuit and the first chip are probably in good shape. Similar steps should be taken to inspect the following chips. As an example, short-circuit the 1V8 test point between chips 38 and 39 and the RO test point.
Inspect the U1 and U2 Circuits for Anomalies Like Resistance Welding, Etc.
The first 38 chips are fine if the log can only find those 38. If the log can only locate 0 chips, then examine the 1V8 first; if it is normal, then there is a problem with the chip after 38. Dichotomous investigation should be continued until the issue chip is located.
Assuming the Nth chip has a problem, N-1 chips can be located when the 1V8 and RO between the Nth and N-1st chips are short circuited, but the Nth chip itself cannot be located when the 1V8 and RO between the Nth and N+1th chips are short circuited.
c) LCD display ASIC75: (Reporting 75) This indicates that while the hash board can identify 76 chips at 115200 baud rate, it only finds 75 chips at 12M baud rate, with one chip being undetectable at this speed;
Antminer S19 Teardown Repair Strategy:
Through the short-circuit probe, short-circuit the 1V8 test point and the RO test point situated between chips 38 and 39 using the dichotomy approach. There is no problem with the first 38 chips if the log can find the remaining 38 chips. The visual inspection is not problematic if the log only uncovers 46 chips after short-circuiting 47 chips because the 47th chip is not visible. Usually, the 47th chip needs to be changed;
d) LCD display ASICNG:
In the Following Two Scenarios, a Certain Chip is Fixed:
The term “test time” describes the amount of time that passes between pressing the start test button and seeing the ASICNG: (X) result on the LCD. In the first case, the test duration is essentially the same as that of the good board (normally, the value of X does not change from test to test).
Users should focus on these 6 resistors as they are probably to blame for the atypical resistance welding of the Xth chip’s front and back CLK, CI, and BO. The low likelihood is caused by X-1, X, and X+1, which means that among the three chips, the following pins experience aberrant chip welding conditions:
The following information is typically found in the log at this point (the red number may not be 13 depending on which seat the hash board tester is connected to); during the test, assume that the domain voltage of all the fields in front of the abnormal position is nearly less than 0.3V and the domain voltage of the back fields. d-2) The second scenario is where the test time is nearly twice as long as the effective board. Usually affecting the 1.8V, 0.8V, RXT, and CLK pins, this problem is caused by subpar soldering on the chip. It is suggested that you directly read the domain voltage to find the troublesome domain. The abnormal position can also be found using the 1V8 and RO short-circuit approach described in section a);
- Single-board Pattern NG phenomenon, which indicates insufficient response nonce data (PT2 station).
PatternNG is the result of the significant distinction between the chip’s properties and those of other chips. Because it has been determined that the chip die needs to be replaced, proceed accordingly.
The Log Information Indicates the Following Replacement Rules:
If no damage is done to the chip’s appearance, simply replace the chip with the lowest response rate for each domain. Four chips asic has a low response rate, as can be seen in one of the test logs, which is displayed in the accompanying image. Replace the one with the lower nonce in 36 as 36 and 37 are in the same domain.
Make sure the chip test went well and that the PT2 function test serial port was not interrupted (long-distance running).
Running the PT2 test as a repair method while keeping an eye on the serial port print log. When the serial port starts to run for an extended period of time, short-circuit RO&1.8V using a short-circuit probe. The short circuit begins with the first chip.
If the serial port stops functioning permanently after the short circuit, the first chip is fine. Find the chip that is still experiencing a long-term operating failure using this technique after a certain chip has been short-circuited. Typically, a particular chip damage is to blame, so just replace it;
- Even though the PT1 chip test is satisfactory, a specific chip consistently reports as NG on the PT2 function test.
Check the appearance, test the chip capacitor or resistance up front; typically, a chip capacitor, damaged resistor, or anomalous resistance is the cause;
The Following Issues Are Brought on by a Control Board Problem.
The entire miner is not working.
1) Check to see if the voltages are normal at the various voltage output points. It is possible to first disconnect U8 in case 3.3V is short-circuited. If the CPU is still short-circuited after testing, it can be unplugged. If there are any additional voltage anomalies, swap out the corresponding converter IC.
2) Check to see if the voltage is normal on the DDR/state CPUs before welding.
3) Attempt an SD card upgrade for the flash software;
The next two procedures must be taken if the control board card recovery miner is to start normally:
a) The power must be shut off and restarted when the card recovery is successful, and the green LED indicator will remain lit;
b) The time it takes to turn on OTP is 30 seconds. After that, turn it on again.
c) OTP (One Time Programmable), a memory type for MCUs, stands for one-time programmable, meaning that once a program has been loaded into an integrated circuit (IC), it cannot be altered or cleared again;
(1) The control board won’t be able to access the OTP function if there is a sudden power outage during OTP or if the time is less than 30s. To fix the problem that the control board cannot start (not networked) (main control IC FBGA of control board), the user must modify the U1. U1 has been replaced and is no longer functional with the 19 series.
(2) The control board U1 cannot be utilized on other series of devices while the OTP function is activated;
- The IP cannot be found by the miner in its entirety.
The IP cannot be found, probably because of aberrant operation. Refer to the first point when troubleshooting.
Check for appearance and welding problems with the CPU, network port, and network transformer T1.
- There is no way to enhance the miner as a whole.
Check the CPU, network port, and network transformer T1 for cosmetic and welding flaws.
- The miner has fewer links or is unable to read the hash board as a whole.
A. Check the cable connection’s condition.
B. Examine the control board parts that match the chain.
C. Check the pins of the plug-in interface for resistance and the wave soldering of the pins.
Antminer S19 Teardown Failure Mode of the Entire Miner
- The entire mining test
Common occurrences include chain behavior, odd fan counts, and the inability to identify IP. If there is a test issue, adhere to the monitoring interface and test LOG maintenance recommendations.
1) There is a need to look into an abnormal fan display. In order to proceed, we must first ascertain whether the fan is operating normally and whether the connection to the control board is normal or abnormal.
2) Less link: This phrase alludes to the fact that one board out of the three is missing. The connection between the hash board and the control board is typically flawed. Examine the cable to check for open circuits. the connection is stable. The user can test the single board PT2 to see if it can pass the test. In essence, it is known that the control board is the problem’s root cause if the test is successful. If the test is unsuccessful, fix it using the PT2 repair procedure.
- Unusual temperature: Usually, this is the result of a high temperature. A 90 degree limit is set by our monitoring system for the PCB. The miner will issue a warning and cease normal operation if the temperature rises above 90 degrees. Typically, the miner’s inability to function properly is due to the high ambient temperature and unusual fan operation. Awkward temperatures will also result from aberrant fan functioning.
- If, after working for some time, there is no hash rate and the mining pool connection is lost, examine the network;
Check the Standard Miner’s State After Antminer S19 Teardown
5) The hash rate of one hash board is low: you can use the Putty program to connect to the IP address to see if the domain functioning voltage of this board and the NONCE return are appropriate in this situation. As instructed by the Putty LOG prompt, you can fix it.
6) Putty application: the command tail-f /tmp/nonce.log-NONCE print the voltage print command with tail -f /tmp/adc.log-domain
These are the precise operations:
1. Click OPEN in Putty after entering the problematic miner’s IP address.
Enter the user name, password, and test command to verify the status of the voltage domain and the NONCE response. If the domain voltage and NONCE are abnormal, the user can perform measurements and maintenance using the printed abnormal chip.
Other Issues That Require Consideration During Antminer S19 Teardown
Routine inspection First, visually inspect the hash board that needs to be fixed to look for scorching or PCB distortion. After the visual inspection is successful, it is possible to determine whether there is a short circuit or an open circuit by testing the impedance of each voltage domain. If the response is affirmative, the matter must be processed right away, regardless of whether the parts have glaring burn marks, are offset, are missing, etc. If found, it must be taken care of right away. Make sure that each domain has a voltage of about 0.36V as well.
The chip can be tested using a hash board tester after the routine test is successful (the general routine test’s short-circuit test is necessary to prevent the chip or other materials from being burned when the power is turned on), and the positioning can be decided based on the test results.
In accordance with the results of the hash board tester detection, verify the chip test points (CO/NRST/RO/XIN/BI) and voltages such VDD0V8 and VDD1V8 starting from the vicinity of the problematic chip.
In the power supply sequence of the Antminer S19 Teardown, the anomalous failure location can be found, and the signal flow shows that all signals—aside from the RX signal, which is transmitted backward from chip No. 76 to chip Chip No. 1 transmits data in the forward direction. 1 to chip No. 76.
The chip needs to be rewelded after finding the flawed chip. The technique involves adding flux (ideally no-clean flux) around the chip and heating the chip pin solder junctions to a dissolved state to cause the chip pins and pads to re-run in and collect the tin, creating the effect of re-tinning. In the event that the problem still exists after re-soldering, simply replace the chip.
The restored hash board can be considered a quality item if it passes the hash board tester test more than twice. When installing new parts, allow the hash board to cool before testing it with the tester. If it passes the test, you should then put the hash board aside to cool down. Repeat the test once more after the hash board has had a chance to cool off.