The third board in this S7 was also a '21' / 10.26V board. So on these three boards we have: Pot Value -> Vchain 16 -> 10.38V 21 -> 10.26V {{:ip.bitcointalk.org.jpg?nolink&200|}} Doing a bit of maths on the values measured equates to about 0.024V / step. The Pot has 127 positions this would give a full scale range of about 7.75V to 10.75V. On reset the Pot is set to mid position which equates to about 9.2V and this ties up with the measured start up voltage. This is original board without fix. You need to connect potentiometer between U2--R17 connection and GND which can be found on C76. http://pantin.cz/20160209_155344.jpg Firstly, use silicone or any other suitable glue to glue potentiometer to the board. After it dries out, you can solder its pins to R17 and C76. {{:s7_pot_soldered_working.jpg?nolink&200 |}} http://www.pantin.cz/20160401_095409.jpg I've been fiddling about a bit with the VRM and stuff on the board. The LM27402 is a quite nice chip, goes all the way down to .6v output - and the WebBench tools are absolutely awesome. So, theoretically, if the top side of the board has too many faulty ASICs, you could jumper the heatsinks to "cut them out" of the chain. Also looking with envious eyes at the NEC Tokin caps as used in the PS3 - insanely low ESR. So I started checking things. The voltage on a good board at the choke with the board plugged into a PS and no controller is about 8.5 volts. So there is a stock setting. One the two bad boards, the voltage is a solid 0.0 volts. Checking resistance on a disconnected board I see about 18.4k at R17-U2 and ground on a good and bad board. Thus I think the variable resistor is ok. Checking resistance on the choke to ground shows a resistance cold of about 30 ohms, so the chips are good. Checking resistance on the +12 to ground I see the 1k and up reading as the caps charge, so that's good. Checking the FETs I see the gates are not shorted to ground, source and drain go to ground normally. It's *possible* the low side FETs are shorted, can't tell yet (If they were, the controller chip would shut down the high side FET when it saw the cut-through current go way up). Interesting. The 27402 buck controller is pretty dirt-simple, feeding a 60a low side FET and a pair of 60a high side FETs. Given the voltage differential is minimal they can get away with this using 60a FETs on the low side with a fast switch cycle of ground. Clever. Still I wonder why it's not coming up at all... It's the 25mhz crystal that is just barely under one of the heat sinks north of the coil. That I believe is what provides the master xclock to the chips; they put in a whole bunch of pads for secondary clocks but I think they drive each chip's clock from the previous one in a line. No clock=no chip running. There's usually 3 things each miner needs: Power Clock Signal Get those three and working chips and you should be underway in some capacity. ===== help questions ===== Hi, any repair guides for S7 boards showing the #48 asics? If im not wrong we have only the 54 chip version of the board. Thank you There is essentially not a lot you can do without QFN rework facilities and understanding the board. If it's a chip fallen off situation (desoldered), you may be able to get away with a DIY fix, but even then it's not easy. Likely faults in high-temp environments on the 54 chip boards from what I've heard are: Cap failure - which is not impossible to fix, the heatsinks on the back come off fairly easily. Failure of the tiny boost converter which does the I/O voltages for the last (couple?) of sets of chips. Hard to fix unless you know what you're doing. Failure of an ASIC (solder joint issue). Possible to flux and reflow, but cleaning without softening the adhesive holding all the rest of the heatsinks on is not fun. Failure of an actual ASIC chip. Find the chip and replace. Not going to happen if you don't already do fine-pitch SMD work. There is no PIC or voltage control on a 54 chip board. I haven't seen a 54 chip board with black adhesive, but I don't have a large sample size. You can get some diagnostic information by reading the voltages between each "set of 3" heatsinks - there should be around .666 volts per chip (some small amounts of variance is normal). However, even then, there's not a lot you can do without SMD rework. Check the kernel log - I would expect to see timeouts because it can't see any of the chain - hence why it's defaulting to 48 or 30 chips. temp failure: That said, if you want to freeze a miner *then* check the voltage that could be a neat data point. I've been split on the cold temp not starting thing; either it's a temp sensor on the DC-DC converter flagging low and keeping the osc from firing, or it's a chip that loses connection with the board when cold. If there is no voltage on a board when cold but voltage when warm we have identified the problem that kills R4's and S7's and probably S9's. ==== chips unsoldered: ==== I am looking at a couple of S7 boards with a few dead chips. Many upstream chips look good, but I'm having trouble communicating with them to verify. Does anyone know if the BM1385 CRC5 is the same as the BM1384 CRC5? Meaning 84 00 00 11 is a complete getstatus command.. I was considering pulling the whole tier of chips and undervolting ala this thread to compensate, but I *believe* that the ribbon cable's TX is non-level shifted, while the CI at the input side of each tier IS level shifted. Meaning it won't talk unless I negatively level shift the ribbon's TX, which I'm not thrilled about doing. I am currently considering pulling one tier + one chip of the next tier to access a non-level-shifted CI and strapping in a diode or a flipped-over 1385 to take up the extra current, but that's major surgery. Best to make sure I'm sending the right command before pulling 4 chips and getting my splice on, heh. chipset of s7: bm1385 datasheet: http://www.pincho.org/libros/libros/subidas/BM1385_Datasheet_v2.0.pdf ===== averias documentales ===== {{:antminer_s7_board_averia_d24.jpg?nolink&200|}} diodo d24: smd 0402 SL ===== Intentado averiguar el orden de encadenamiento de los asics ===== voltajes mosfet: {{:detalle_voltajes_medidos_en_s7_mosfets.png?nolink&400|}} {{:voltajes_en_una_s7_v3_sin_disipadores.jpg?nolink&600|}} Measure voltage between inductor and GND. If the voltage is lower than 9V, you can be sure PIC is not working. If the voltage is higher and less than 10.5V, PIC is malfunctioning, partially working and you need to remove it or remove U2. Correct voltage of string: 10.5V Partially working voltage: 9-10.5V PIC not working voltage : less than 9V (usually 8.7V) despues de conectar 3.3 en el conector de 18 pines el voltaje de 8.32 sata a 10.56 Parece que hay tres chips en serie, y luego 15 grupos en paralelo bitcointalk-sidehack: thats about exactly what you should be seeing, so it means the power circuit is functional. Next thing would be to poke reset, TX, RX and clock lines at the 18-pin and see if those signals make it all the way to the last chip in the chain. This should be done while the board is powered up, of course. Signals in Antminers go uphill, so the first chip would be on the lowest-volt node and the last is at the highest. I'm not sure offhand where those are but it shouldn't be too hard to tell. === averiguar === digital pot present? MCP4017 pic: pic12f1572 (u3) (pic present, check firmware) donde esta el sensor de temperatura? que modelo es el regulador 4vk4 y su datasheet que modelo es U2 ? versiones: 162-chip, it's straight off the PSU with no regulation. The later 135-chip batches are regulated. driver de linux: https://github.com/hashrabbit/bitmain-spi comando para probar desde beaglebone: /usr/bin/cgminer --bitmain-dev /dev/bitmain-asic --bitmai n-options 115200:32:8:5:200:0782:0725 --bitmain-checkn2diff --bitmain-hwerror -- version-file /usr/bin/compile_time --queue 8192 --api-listen --default-config /c onfig/cgminer.conf -T configuracion de pruebas: cgminer.conf { "pools" : [ { "url" : "stratum.antpool.com:3333", "user" : "pincho", "pass" : "123" }, { "url" : "stratum.antpool.com:443", "user" : "pincho", "pass" : "123" }, { "url" : "stratum+tcp://stratum.antpool.com:3333", "user" : "pincho", "pass" : "123" } ] , "api-listen" : true, "api-network" : true, "api-allow" : "W:0/0", "bitmain-use-vil" : true, "bitmain-freq" : "700", "bitmain-voltage" : "0706" } firmware: https://github.com/bitmaintech/Antminer_firmware https://github.com/ckolivas/cgminer Buy a Cyclone IV dev board, and a Raspberry Pi if you want the no frills basics. The code for sha256, scrypt, and X11 are all freely available. It’s slow, and won’t make money, but it teaches how to build the cores, and get them hashing. The next step is tearing down an antminer