Topic: Hacking Bitmain Antminers (S7 & S9) because man a lot of these break...... (Read 2187 times)

the boards i have repaired light up when they are connected to the controller and start hashing. when a miner doesnt detect the board i look for the one without the red led on and thats the one that failed...in my case at least. i could be wrong but i think the light only comes on when the board is being checked for connection to the controller....when the chips are being checked and tuned....and finally it stays solid when the miner is actually mining away. hope this helps.

Question : on a working Hashboard, when powering it up with only 12v rails and not conected to a controller,  is the red Tf led supposed to stay solid red?
 Ca someone who poseses god miners and defect boards test this?
This ,I guess, would be a quick and easy method to see wich board is defect on your machine when one board is going defect...
(And give me a hint when a repair was succeful or not)
 A solid red Tf led could indicate wheter clock and data lines are broken between the first and last chip on the chain and could prevent people from seeing "magic smoke" when trying to test boards...


Regards,
Mihai

Hy all,
Fairly new to this forum so don't take my words as reliable.
I recently bought a dead S9 hashboard to play around with it. (I do not own any complete miners, but a verry curios person I am )
It is a autotune version from March 2017.
Some notes to sumarize what you all have written in the previous posts: powerwize- there is one string of 21 chips in series with 3chips in paralel for each step (63 chips total); comunicationwise- 63 chips daisychained.
 If one chip goes bad and pulls any clock or data line to ground, the board is dead. Same situation if a chip interupts any af these lines.
Nice thing is that Bitmain has included testpads for each chip, so troubleshooting is fairly easy.
O this board, powering it up by itself, does not do any good. I get 0.38v at the buck converter output. (I am thinking that by missing the 3.3v line the PIC is not powered on, so no output).
Removing the heatsinks off the chips is easy if using hot air, so is scaping the leftover thermal adhesive from the chip by using a
hot soldering tip.

(Correct me here please) Measuring for short on the powerline steps with resistance setting on my meter is not helpfull because the chips have a verry low resistance, under 1 ohms each... I also noticed some resistor paralelled with each step of chips noted R500 wich would make measuring for shorted chips a bit harder.

These boards are missing some back heatsinks near the intake fan(12 to be exact), so overheating in this area is a posible reason of failure. (The 2 changed chips on this board were located precisely in the middle of the heatsinkless area).


I think I was lucky enough to buy a board that someone has allready attempted to repair, so the guy has replaced 2 chips. Unfortunately for him the resoldering process was not good, therefore some pins were not soldered to their pads.
After resoldering the 2chips, all measures tell me that the board is fixed. I am now sad that I removed all the heatsinks glued to the chips, as for further tests, I will have to clean all chips and reglue the heatsinks.
I have also an idea of repositioning the back heatsinks in a chess like pattern, to see if this improves cooling at the begining of the board...
If only I had a controller to test it... Does anyone located within EU have a spare controller for lending it to me?
Regards,
Mihai

Just in case anyone is wondering, yes, it is possible to reflow a whole board as a last ditch attempt for boards where the source of the fault is extremely tricksy.

 I don't have a proper oven which can cope with the maaaaaasive thermal mass of all of the heatsinks, so this puts the board *way* outside of the reflow parameters (it takes about 15 minutes to get up to reflow).

... Take off the electrolytics and the power connectors, clean board (I use isopropyl dip), apply thinned down flux to the topside only, ensure that you are heating on a solid surface.

On a completely unrelated note, the black thermal adhesive copes remarkably well with heat. The white stuff... Not so much. Completely degrades and a slight touch will knock off the heatsink. So, I now have a completely bare of heatsinks board and am desperately trying to think of a way to properly heatsink it. I may just end up oil bathing the damn thing (it's a 54 chip board, so there are no non-SMD electrolytics).

If you leave the connectors on, they will survive (foil wrapped) but look unhealthily yellowed. Subtext: Don't do this, there are *far* better ways to spend your money and time.

Meantime been busy here working on things. Here's an interesting picture: The S9 in question is dead, and oddly enough does not respond. Checking clock signals with the scope shows clock to the end of the board but not on the other side. So I pulled the back sinks and checked. Sure enough the three test points on the bottom of the board were ok, the top ones were dead.

Pulled the heat sinks, cleared off the crud, and we see this on the third chip, bottom side:



If you look carefully on the right side of the chip you see the solder doesn't look right. What happened is either a cold solder joint from the glue factory (Bitmain) or more likely the chip overheated and melted the solder out. Either way no contact, not going to work.

Off to reflow-world....

They are called molex mini fit jr

They are a couple bucks each from mouser or digikey, or maybe $1.00 more to upgrade to the gold finish.

The ones you linked will work fine

First, thank you for this thread it's invaluable! I recently powered down my S7 for routine maintenance and cleaning and noticed that my power connectors were fused to the white PCB headers (very similar to your photo) on 2 out of 3 boards.  It was hashing fine when I shut it down and probably would have kept running if I wasn't so paranoid. I was using an old X11 board (I think v4 or v6) and there seems to be varying opinions on whether it was a surge or a if the opposite wires became exposed and "arced" (I am very new to comprehending electricity)?

I found a local guy who does soldering and he's agreed to replace the 3 headers on each board for $70 USD total ($35/board) if I provided the components.  I have found and ordered a bunch from this site but I am uncertain if they will definitely work. https://www.moddiy.com/products/6%252dPin-Graphics-Card-PCIe-Male-Header-Connector-%252d-90%25-Angled-%252d-Black.html Could you please verify that this part is an acceptable replacement.  Also about those nickel ones, would you mind sharing where you ordered from?  And do you think the price quote I received is a fair one?  (I will likely refer him to this post for your step by step instructions) If I continue down this path I may perhaps need his services again in the future.  And tbh after reading this post I'm starting to think it may be time to try my hand at soldering.

I'll sell you a chip for $100.

Problem is, it's about 0.4mm pitch blind mount and not likely to be seated successfully without a split vision system.

Also, fails are often more than one chip.  Haven't read the most recent parts of this thread, but the fails I've seen cascaded over multiple tiers of chips.

I'm not trying to talk you out of it, but I don't want complaints that my chip is bad just because you can't place it.

Or, if you're sure it's only one chip, have the whole top tier pulled off and use one of those.  If you don't adjust the pic you'll be running at 50mV overvolt for the remaining tiers..  not that big a deal

I've also done business with them:  I sent them 5 bad older fixed-frequency boards.  They managed to fix 2 of them (for a reasonable fee, but it took 2+ month) and sent the other 3 back a bit worse-for-the-wear without charge.

Currently they have (2) newer boards and a controller that I screwed up loading firmware on (somehow got out of sync while doing 8 at a time and managed to get the reflashing process to start twice on a single board).  They have had them since December 12th.  Communication feels spotty - I ask weekly how its going and unfortunately don't always get a reply.

These last 2 boards were actually only a few weeks old, so under warranty, but Bitmain insisted I send the entire miner for each - meaning I would have (4) otherwise fully functional hash boards offline for a couple of months.  Instead I opted to just pay the repair fee to BItmainWarranty.

MyRig (previously known as BitmainWarranty) is legit, I have had few paid repairs done at their facility and you can also ask philipma1957 who has also done business with them.

Just bitmain, after getting scammed by syan and almost by a few others, I find it hard to trust anyone. I took a chance with lightfoot and he is a straight up guy, but has no chips to help me get my board going.

MyRig is a separate company and not a part of Bitmain.

Did you ask them also or just Bitmain only?

Bitmain offered no inside USA help, even after I offered to pay for the repair, they still wanted me to ship the whole unit back to China

bitmain has an option to send the board to a repair place in ca for in warranty repairs. even if you have to send ur whole miner in its far less than $100 if ur inside the us. as for people being happily willing to pay it...im not so sure about that one lol. bitmain warranty will repair a board for $200 or so shipped. they will charge $200 even if the chip is provided. i can remove chips from the boards if needed but i rather not price gouge the hell outta people for $100 per chip. its not right and the markup is insane especially since theres no way to guarantee that the chips removed will function properly.

I think a good option for you is to contact MyRig (previously known as BitmainWarranty).
They do out-of-warranty paid repair service for Antminers.

Contact them and figure out your issue and the repair service process.

https://myrig.com/contacts/

Bitmain was requiring me to send the whole miner to China and other people claim it takes at least 8 weeks to get it back,  plus the chip is bad because of my own fault and really shouldn't be under warranty.
Like I said before, I would be more than happy to pay you $100 for a working chip if you want to scrap your board so I don't have to. Its not gouging when people are willing to pay it happily.

lol u think a working board broken down is worth $6200?? i guess technically $6100 since one chip is bad but either way have u lost ur damn mind man??

1 the miner is only 2 months old so repairs are covered under warranty

2 ur hopin to price gouge worse than bitmain does and u somehow think thats ok to do lol. r u payin the nice guy that did all the work $5k of that $6100?? i highly doubt it.

ne one that thinks 1 miner with 3 boards is worth $18k+ is long past the point of return. jesus!

I have a 2 month old S9 that lost a chip, lightfoot has it and posted a pic of it above, since I cannot get a replacement chip, I am considering paying him to remove the 62 working chips so I can sell them for $100ea. Would there be any interest in this? I know if someone was offering to sell me a working chip for $100, I would jump on it.

I can give you a link for a Taobao seller advertising fake antminer chips.  No joke - he is openly selling counterfeits that other sellers will in turn resell to us.  Seeing that kind of turned me off to looking for a source other than bitmain.

Shuts down the power supply? Is it an S7 or 9?

C