I think you have your answer: Hate to say it but S9's are disposable.

Toss on ebay or junk it. I haven't done the ROI on power vs. difficulty but I don't know if an S9 can mine profitably anymore. So the best answer may be to not mine.

Sad, but the fate of many a miner....

You got to read the whole thread :-) In a nutshell you download the .7z file from my dropbox link, then decompress it with 7zip to an iso file. Then you use a program like w32diskimager to write the ISO file onto an 8gb or bigger memory card, plug it into a bbb, and you are off to the races.

The BB is faster than the Pi, and yes does fix the annoying amount of stale shares submitted. If you like it send me some litecoin.

Meantime I still have to fix your board.....

Yeah, ground fault. You know, it costs ten cents for a damn optoisolator pair, buffer the SCL signal in and out, and this sort of stuff doesn't happen. But no, everyone wants to go straight TTL and save a dime....

Ahem. Anyway, if it's under warranty send it back to bitmain, but they may laugh at you except for the fact that you did use their power supply. I could fix one of the boards, maybe the ground looped one, but that's up to you. Really the best thing to do may be to put it on Ebay with a $50 opening bid and let someone pick it up. That will give you the true worth right there.

Thanks for posting all the pics and such, it's an interesting thread because of them.

Edit: Holy cow, these things go for $200 on Ebay? Serious price collapse there...

Sometimes. One thing I have noticed is that Bitmain uses cheap plated brass plugs, they do not conduct as well as nickel ones but they cost a few pennies less.

And while loose connectors can be a problem I think the bigger problem is voltage sag: If you check the voltage at the miner pins and it's less than 12v then you have voltage drop between the supply and the miner. Aside from being resistance (warming the wires) it results in the miner pulling more current to maintain output. P=I*E, so as voltage (E) goes down, current goes up which stresses the pins even more.

Thus the connector warms up, resistance goes up, and the connector burns.  This happens a *lot* on Neptunes running with 500 watt Corsair supplies. Annoying. Then if you have 3 plugs in parallel, one burns, goes high resistance and the current load is distributed across the other two. With the usual results.

Side question is did the ribbon cable burnt: I can see one of the ribbon cable plugs looks melted, if the grounds opened on the final plug before the hots it's possible for the miner to use the ribbon cable's ground as a return, burning that and possible traces on the board.

Hm. Not too bad; if the boards themselves are not damaged then the connectors can be swapped. The trick is to use preheat under the board, not too much but enough to then use an iron or hot air to pull the plugs. Then clean up, and use new plugs. PM me if you need a hand.

What kind of power supply were you using? Also did it burn up any of the ribbon cables?

At this point I would consider this a dead pool. Sad, I had some shelved shares so to speak but you would have better support on a P2pool node.

But thanks for the ride WK. You ran a good pool for a long time, Butcoin and such just changed around all of us.

If he's just pulling the choke he can get away with a good iron and some chip-quik (the cheap bastard solution to SMD rework). That pulls down the melting temp of the solder to something that is low enough you can pull with an iron. Handy crud but can't easily get under a TO type case on a transistor.

I'd say check the resistance on the FETs by checking the resistance from +12 on the connector to the transistor side of the choke. If it's low resistance then one of the FETs has shorted. Then you get your preheater (God in a box), warm up the board to 150c or so, put some Kesterel liquid flux on the FET, pull it with 380c hot air, then clean everyting up, tin with good old rosin core tin/lead solder, then flux and put the new FETs on.

Result, happy board.

---

I'm guessing the PIC chip is the waveform generator/general purpose processor, which then drives the 3420 as a dual POWER Fet driver, which then drives the gates for the um... FETs. You need power to drive the gates because they are running at a high frequency, and each off needs power to be done quickly in order to keep both FETs from being open.

Still think the high side FET shorted closed, which would mean the +12 supply would have gone straight to the chips, which would have blown the weakest component being the choke. Did you have a big honking power supply plugged to the board, smaller supplies would probably have crowbared first.

Interesting stuff, keeps my brain working.

Edit: If my posts are not technical enough, pls let me know and I'll fiddle with something else.

Agreed, that's a simple choke/inductor. The much bigger question is why did it burn out. Hm....

Get some close ups of the transistors next to it please, and check to see if they are shorted source to drain (or across the gate). I'm willing to bet a banana that the high side fet shorted, putting +12 across the string of chips which overloaded the choke.

There's another pair of transistors and choke on the right, so I wonder if there are two power planes, or if these two run 180 degrees out of phase to smooth the load on the one big inductor. Slightly different type of design (doesn't need two low side FET's, interesting) I'd be interested in knowing what kind of transistors/fets they used.

Hm. Can you post a pic of the chip and the board? Wonder why it burned.....

Do they know about segwit transactions? :-)

The 6510 was in the C64. I remember doing a fair bit of assembly and FORTH on the Vic-20, it could do a good bit with what you had. First computer was a TRS80 Model 1 Level 1. I still remember the game "Termites" followed by the game "Introduction to programming".

Wonder about first computers for people here.

Better than 6502. I think the whale that is running the market was a VIC20 owner.

Still, we're all going to hell.

Ok, so with no titans connected it displays spimux and never the IP address, correct?

If so the controller has a problem. Send it in and I'll look at it.

Are you comfortable with taking a cube apart and removing a board, or if you're in the US you can send over a whole cube and cable and I'll check it out.

1) 10k entry is pretty cheap overall. I'd prefer a Patek Phillipe myself or a Breuget over a Hublot.
2) They want a picture of my passport to pre-register? That's got to be the fucking ballsiest thing I have seen all week.

Fuck this. Go to Vegas, cash out 50-80k and buy a really nice Patek repeater. Do not go to the Cartier store in Wynn unless you're looking to pick up a 1.3m platinum watch....

I really should not check this stuff after waking from weird dreams. 3.79tps, not 379 tps.....

https://www.blockchain.com/btc/unconfirmed-transactions

Hm. Are we holding at almost 400 transactions a second?

Remember when 3 tps would crash the network?

Ok, hm. Try this:

Connect one cube

Power up

Go to 50mhz on each die, -.0366 volts

See if it hashes at all.

This will tell you if the heat sink is loose; the cubes can run at 50 for a bit without much of a sink. If so you need to repaste the dies and ensure the sink is on snug enough that it can't be wiggled around on the chip.

Let me know. If that doesn't work you can try to reflash the controller or send the controller and one cube in and I'll take a look.

Bitmex is the hot leggy redhead. She will give you the ride of your life, blow through every account you have, then trash your life by making you wrap your own car around a tree.

Worth every minute of it.

My father once told me that slavery is the most stupid concept ever and to never own slaves.

Reason: Why own an asset when you can lease it....

Well if you're in the army then yes, termination does take on another meeting. But remember the words of Bob Dobbs:

The worst thing that happens if they fire you
Is you get a better job....