As long as you are getting a fan error you should ignore everything else as the miner might work just fine when the fan is replaced, don't attempt to flash a different firmware until you change fan [4], if the problem persists you can go ahead with the recovery firmware and flash any firmware you like, avoid the most recent as it would lock your miner.

Do you mean board 2 works fine on its own but not when plugged in with either board 1 or 3? this line is a bit confusing, but i'd guess based on the system status on the image you uploaded is that you should flash different firmware, all your hash boards showing 63 Asics which is a good sign that either the firmware/control board or a PSU is causing the issue.

Try to flash this multi-option version (Antminer-S9i-xilinx-201907302056-autofreq-user-Update2UBI-NF-sig.tar.gz) from the official website and then try the different settings, start with -2th and see if you get any different results on your second hashboard.

Please use code tag to post the kernel log, it looks so messy and hard to read.

A quick report:


Your chain 2 which is the to the most right seems dead, your miner should be hashing just fine with 2 boards at 2/3 the hashrate which would be at about 25-27th, but in the kernel log I see something strange here:


These temps are not normal by any means, is it freezing cold where the miner is running? also what firmware does it use? please post a few screenshots of the status and main page.

It works now but it will eventually fail as long as it's mining on Nicehash you should expect to face all kind of issues ahead, including but not limited to your funds being "hacked",  and you don't really make any extra mining on Nicehash, in fact, the last 7 days Nicehash made 3% less profit in BTC against mining directly to a PPS pool, the on-going difficulty changing and all the mess they create will go to as far as bricking your gear and many of us here have personal experience with such issues.

The only time I would consider nicehash is if there is a massive temporary pump on the rates there, and that will be for a quick run and I'd leave, but mining 24/7 on Nicehash is a terrible idea regardless of how you look it at.

philipma1957, a fellow bitcointalk member, a very nice lad.


It won't consume power, maybe a tiny amount, but it will slow down the boot, the miner will keep trying to read asics for a few seconds, so just unplug the power cord and leave it there.


Given that the value of each hash board is slightly less than 33% (You have a control board and a PSU) that's a decent deal, the seller was really more than decent to give you such a refund.

No, they are slightly different in size, not the overall size but the heatsink size on both directions are different, shorter on one side and taller on the other, so you can't fit 3 S9 hash boards in an S9i case, it's easy to spot a fake S9i but not a fake S9j, hope that makes sense.

I am 100% sure they are S9s with the J sticker, or S9j miner with S9 boards inside, at $350 per S9j they could just sell that in China at a much higher profit, so be careful when prices are too good to be true, always double-check with grace, scott and the other reliable resllers.

Not quite accurate, it's more like this:

DO NOT BUY S9k and S9SE.
avoid S9 unless it's pretty cheap.
If you trust the seller, go for S9j.
if you don't trust the seller, go for S9i.

The difference between I and J is really nothing at all, so all in all, it doesn't matter, but many sellers will put the S9j sticker on the S9 and sell it as S9j which they can't do with S9i, so unless you know the origin of the miner or truly trust the seller, S9i will be the best to buy.


Fair? in my standards? I can't tell, I have bought them as cheap as $40-50, it's too hard to stomach the fact that they go for about $500 today, on the 28th of last month I bought a few for a client at $450 each excluding shipping, BTC price was at about 47k, now we are up to 54-56k so my guess is that they are back to $500+ range in China.

I believe China is the best place to get your miners fixed, check Zeusbtc they have repair centers there, you could also send it to Bitmain although someone mentioned they told him they can't ship the gear back and they offered him some coupons instead, and a buyer for those coupons lol.


Yes pretty normal, it will work sometimes and won't on the others, eventually, it will just die.

In the kernel log before you flashed the junk piece of firmware your 3 hash boards were showing 63 Asics which is great


You are asking for troubles when using Nicehash as your pool, and you are asking for even more troubles by using their firmware, get rid of that firmware and test your miner on a proper pool, use F2pool or Viabtc, flash Bitmain firmware (avoid the 2019-2020 versions) and repost your results.

It's not displayed on the miner status because it's powred off by miner.


In plain English, this is a dead hashboard, the S9 you bought could be 4 years old and has not rested expected for the shipping duration to your address, it's perfectly normal for it to lose a board or two, and 90% of the time it starts with the middle board exactly like how your miner did, the middle board doesn't have the best air-flow due to the fan motor, that's why some people like phill use a spacer to allow more cooling.

You could try the basic PSU / ribbon cables switching but chances are, it's a dead hashboard just leave it there and mine with the other 2 boards.

I sent you merit for doing this before posting, most folks won't do any basic troubleshooting and expect to find all the answers online.

Things to be done:


Check the screws on both the PSU and the hashboard rails.


That will just be a waste of compressed air.

if all fails, try the freezer trick, you might get lucky.

This is normal nothing is weird about the result, it's a bit confusing but let me try to explain.

Board 0 is bad
Board 2 is good

The control board connects to hash boards this way

0 control board > 0 hash board >Bad
1 control board > 1 hash board >NA
2 contorl board > 2 hash board >Good

Say you switchboard 2 and 0, the reading on the contorl board which is what you see in the kernel will be

0 control board > 2 hash board >Good
1 control board > 1 hash board >NA
2 contorl board > 0 hash board >Bad

Notice that despite the fact the reference on the control board has changed, the control board numbering remains the same, board 0 is still bad, if however, you get this

0 control board > 2 hash board >Bad
1 control board > 1 hash board >NA
2 contorl board > 0 hash board >Good

That's good news, it means both boards 0 and 2 are fine and something is wrong with the wires or the PSU busbars, and that is very, very unlikely to be the case, so good luck.

While I don't like Slushpool at all, I don't think it's the cause of your problem, it's likely just a terrible coincide.


The S17e uses 135 Asics on each hashboarrd, your board 0 (usually the most left when facing the miner's front-side) shows 0 asics, and chain 1 which is the middle board is showing only 134 (so it's missing a single asic), this problem is very common with all the 17 series, the heatsinks solder they use on those is pretty bad and eventually, chips start dying.

There is a good chance that your board 1 will come back to normal after a couple of reboots, chances are slim for board 0, some times people get lucky with custom firmware so you might check if any trusted devs have the S17e on their list.

And please, use the code tag when posting Kernel logs.

Maybe, but unlikely, the miner has temp sensors that stop it from mining when temps go past a certain degree, if you have not disabled that then it's unlikely that you toasted them.


Yes you could measure the voltage and resistance of each individual chip using a multimeter, or buy a special fixture tool, but your problem seems like a bad control board, I believe that if you Sdcard the miner using recovery files and then flash 2017-2018 firmware your miner might just come back to life.


If it's a bad heatsink or two, yes, you just need the adhesive and a heat gun, but this isn't a common fault with the S9s, usually, when an S9 hashboard is dead it's so because of a toasted chip or a few of them, which means you will need to order some chips and obviously a fixture tool which doesn't make much sense since this is just a single S9.

It all depends on what you are trying to achieve and obviously your power rate, if the cost of electricity is too high then eco will be a winner in most cases, even with the lower hashrate you get since eco improves the gear's efficiency, if you think the difficulty is going to spike and you want to maximize your gains now then you might want to overclock the gear and go with the +/performace settings, everyone has a different sweet spot for tuning their miners if it was up to me, I'd just keep the factory mode.

You won't get anywhere, trust me, we have gears running just fine for 4 years and we have gears that failed after running just for a few weeks, some for days, some for months, there are no official statements from the manufacturers but for an example Jihan wu the former CEO admitted that the 17 series had high failure rate, he did not the deny the claim he was replying to which states >30% failure, and that was just a couple months after they hit the market.

But then, the next batches "supposedly" improved, and since the 30% came in after a short period of time, it increased for most others, if you ask Phill he will tell you his failure rate on the 17 series gears from Bitmain is pretty low if you ask me, it's well beyond 50%-60 by now IIRC, both of us are telling the truth but which piece of info are you going to use?

if you look around the support section, you'll find people reporting dead miners after a few weeks of purchasing them, some will say they bought this off Alibaba which has probably mined for way too long already, if you ask the Avalon fans they will tell you a very low failure rate, if you ask someone who happened to buy T17es from Malaysia batches their failure rate could be close to 100% by now, you will get completely different results and non of it will be a fair representation of the overall market.

If you trust the seller always go for the S9J version, if you don't, then ask for the S9I version, avoid the original S9 unless it's cheaper by say 20-30%, and of course, avoid both S9k and S9SE like you avoid putting your wet fingers inside the electricity box.

I was going to respond to your PM but then I saw this post, sorry for the delay. The short version of the story is you have two dead boards and only one works fine, did you just buy the miner used? also, keep in mind sometimes if the PSU is bad it could result in the same error, but in most cases, it's just a bad hashboard that's beyond a simple fix.

Are the rest of the miners you have giving you the same error? I suggest you post about them 1 by 1, don't create a new topic but 1 post per miner in this topic, wait for members to respond, troubleshoot, and move on to the next one, this way you keep it simple and easy for us to follow and won't spam the support board.

It's inner Mongolia for the record, and you are right about this, the difference overall for this won't be noticeable, one Chinese miner I know told me they would be moving to Sichuan anyway since the rainy season is just a few weeks head, around April is where the hashrate swap happens in China, expect to see some weird ups and down within the same epoch but overall, nothing will happen.

My guess is that the Chinese look after one another's business, they gave miners enough time to reallocate, and asked them to shutdown just at the beginning of the moon soon, this means miners in Inner Mongolia have until September to worry about what to do with their gears, with the super cheap hydropower there, all gears would ROI by then and the only losers will be those buying the gears at the peak of the cycle.