You checked the miner’s status online and found that your BITMAIN S19Jpro hash rate is significantly reduced. The hash rate of an Antminer S19jpro should be approximately 100 Th/s but you only get 70 Th/s. After a more thorough investigation, you observe that one of your three hash boards is not hashing. You checked the kernel log and found 125 ASICs error which indicates that the reason your hash board isn’t hashing is because only 125 ASICs can be found out of 126.
Now you know why your S19jpro isn’t working but you have no idea what caused the problem. There are several reasons why you might encounter a 125 ASICs error (Also known as “Some ASICs”). The three main reasons are: 1) There is a problem with the boost circuit, 2) You have a bad switching power supply and, 3) Your temperature sensor may have failed or may have cold solder joints. To solve a problem like this, one must be familiar with S19Jpro hash board working principles. Normally, the problem is evident in the error code, i.e., 125 ASICs error. However, sometimes the hash board might have other problems that cause this error. Moreover, 125 ASIC error does not necessarily mean that chips #125 or #126 need to be replaced. A detailed diagnostic of the voltages and signals must be done to fix your hash board.
We recently worked on a hash board that shows 125 ASICs error in the kernel log. Although, the Universal Test device, a tool for diagnosing the hash board, confirmed the 125 ASICs error, nonetheless, we did measured all relevant voltages on board. First, we confirmed that the board is getting powered and all MOSFET are switching correctly. Then we measured 15V core voltage supplied to the hash board, the boost circuit which convert the 15V to 20V in other to power the last seven domains, and the voltage domains. One common fault in the S19Jpro is the LDO and the PLL, so we measured them, and the voltage outputs shows the correct output voltages of 1.2V and 0.8V. After confirming that my LDOs and PLLs are working, we proceeded to measure the signals. We found that the return signal, RI, is flowing from chip 126 to chip #1 as it’s supposed to, but the transmitting signal, which flows from chip #1 to chip #126, is not flowing through chip #125. The image shows the chip #125 that was taken off the S19Jpro hash board. Upon a visual inspection, the chip looks burnt and damaged. In addition, we observed that one of the temperature sensors has a solder ball, which might indicate the root of the problem. After replacing the chip with a new one, the hash board was tested, and 126 ASIC chips was confirmed. To prevent the problem from occurring again, we reflowed all temperature sensors on the board.