2056 is loop 1 failure + pg1 (power issue) If it was a PSU issue normally you would see the PG1 error but not the loop 1 error. Looking at the unit from the side the PSU plugs into that is referencing the board on the left. You can try to put it on its own RPi but a board failure code plus a power failure code is usually a fried board.

This thread can be closed, I was able to find the manufacturers spec sheet for the fans.


These units have been set to the defaults, I almost never force manual fan speeds (learned my lesson on the A6's). I figured out the issue with my replacement fans. I had to swap the tach and the control wires in the 4 pin fan plug and leave the control wire disconnected  to get them running. I now have the S7 running on aftermarket delta fans albeit with no ability to control the rpm anymore. With the weather warming up here the fans are running at max pretty much all the time so losing the ability to control fan speed is a non issue at this point. Thanks to everyone who replied!

Reading this hurts my soul. Windows 10 is garbage, almost as bad as this Windows 8 POS I am running at work :/

That link doesnt provide any specs for the fan. Searching for B01D88TURO just brings up this forum thread and one or two others.

Fans die, there is no specific reason for it. They are just cheapo chinese fans and the bearings seize up and they stop working.

Same with the second link. Spending $15 + shipping for a $5 fan makes no business sense or I would have ordered them from bitmain already.

Thank you all for the replies.

I have tried Minera but it doesnt seem to have the features I need as of yet. I had an S7 fan die and found no warnings or notices of any kind with Minera. In the settings all I can find is scheduled restarts, nothing about active monitoring of the units.

cryptoGlance seems to not be scalable in the way I need. I cant imagine trying to monitor thousands of miners each with their own little widget open.

AwesomeMiner is just not a financially viable solution for the amount of units I have to monitor.

All seem to be quality polished products and if I was just running a couple miners at home they would both be effective solutions.

The search continues!

Does anyone know the actual fan specs for the S7? These cheapo fans have no labeling of any type to look up the info. I have tried multiple replacement fans and none of them run properly. The only other fan I have found that I can replace them with are from some old S4s I have that were just gathering dust. I have had to replace 5 S7 fans in the last 6 weeks and would really like to get some replacements ordered so I dont have to disassemble functional units for them.

Thanks in advance!

When they say largest mine are they talking about total land usage? 10 petahash is nowhere near the largest bitcoin mine. I can think of 3-4 locations off the top of my head that are bigger than that.

As of now there is no coding in the software to make it work. You must use the two inner data ports as well as the inner fan port or the controller will not recognize anything. I think they were planning a larger multi S5 style unit controlling more boards but it never came to fruition.

I have the Minera image downloaded and on an SD card that I plan on testing but from the screenshots and interface it looks like it has too much going on for what I need. I also couldnt find any information on people using it to monitor a large number of machines, or even a cap on the number of machines it can monitor before resources become an issue. Is there a default IP that it is on? I do not have a monitor available with HDMI so I need to be able to connect to it remotely from the start and when I booted the system on the RPi it did not capture an IP from my DHCP server.

Hello all,

Can anyone make any recommendations on server farm monitoring software for antminers? The main ones I have found on this site so far seem to be geared more towards the home miner monitoring a small number of miners. They have too many bells and whistles that are unnecessary on a more industrial scale. I am looking for something to do basic fault monitoring on a large number of units (1000+). Is the unit running? Is it overheating? Has there been a hardware failure? I dont need altcoin mining, pool switching, profit calculators, or a fancy cluttered UI. Avalon has something very similar to what I am looking for that they call AMS (Avalon Monitoring System). It shows number of expected vs running units, has a color coded temperature diagram to zero in on heating issues quickly, and shows errors based on the RPi error codes. With all the homebrew creations out there I am hoping someone has a few suggestions for me.

Thanks in advance!

Running two S7s @ 1400w each means you are running 2800w on a 2880-2980w PSU. You really dont want to push power supplies to nearly 100% of their rated power, server class hardware or not.

This would be a very bad assumption. China really doesnt have cheap power at all. The VAST majority of power produced in China is coal based. That means it costs a good $0.07/kwh just to produce the power on average when you factor in the retrieval, transportation, and processing of the coal to make power.

517 is a combination of three error codes

code 1 - idle
code 4 - loop 0 fail
code 512 - no fan


The fan error means that the unit cannot sense the control line the fan (this wire is how the unit adjusts the fan speed). The fan will just be spinning at maximum but the unit will not hash in this state. I would recommend completely powering down the unit, unplug the fan, then wrap a small piece of tape around the connector before plugging it back in. The plugs are not always snug and just the slightest movement from vibrations can send the unit into the 512 state.

The loop 0 error is saying that one of your hashing boards is in a failure state. This can be a false reading due to the fan error. Once the fan has been taken care of you should power the unit up and and see if the error code is gone. Some other issues can cause a false positive with the board failure codes. If you get an error code 4 after doing this it means the board on the right side of the unit (looking at the unit from the PCIE power plug side) has failed. If you are still under warranty I would call immediately to get a replacement.

I am not sure if anyone ever responded to this so I will go ahead. Just to be safe I would verify the 4 pin cable is good and change the micro usb cable but it sure looks like you are seeing a continuous restart loop of the AUC dongle. You will almost assuredly need to replace it. As far as power requirements go the Rpi should have at least a 1.5 amp power supply (I would recommend a minimum of 2amp personally). A 1 amp power supply will not support the AUC dongle and the RPi at the same time. The dongle will sit idle with just a single green LED on and the Pi itself will crash and only have its red power LEDs on. My impression is it is corrupt firmware on the dongle causing the boot loop. Avalon does has firmware for the AUC on its canaan creative site but I do not know how you would actually reflash the dongle itself or if that will even fix the problem.

 The first temperature reading is not ambient intake temp, it is the temperature of the control board itself. The machine will shut down and flash red at 45c. It will restart the controller and continue hashing once the temps have gone back down.

I would like to add this to help some people out.

If you have an unresponsive Avalon unit (E.G. it will not boot up to the flashing amber light) you can run the MM update and it will actually update the unit still as long as it is on the chain. This is not documented anywhere but I know from personal experience that it will 'save' a seemingly 'dead' controller board.


Also as far as updating MM goes....they wouldnt put an update out if it didnt actually fix something.



New MM firmware released for AvalonMiner 6.0, Hardware model: MM-60, Version: 601512-57532250

Firmware Details:
    http://downloads.canaan-creative.com/software/avalon6/mm/2015-12-28/

ChangeLog:
* Fix:Don't touch fan by miner when mm is too hot
* Fix:Update idle process with toohot detect, stratum detect, and invalid mcu protect
* Fix:Fix coretest failed cause by incorrect pll setting
* Update:Update API_TIMEOUT_CONST
* Add:Support adjust freq by CGMiner when mm is working