Hacking a BFL Jalapeno to 32GH and beyond....(???)

[lightfoot]

And I didn't go crazy so I have 13 chips left.. Look for a thread under Computer Hardware if you want them.

https://bitcointalk.org/index.php?topic=395129.new#new

Well, it looks like Danger labs is going for the gold. I just picked up six of these, will distribute as follows:

Jally 1: 6 chip to 8 chips (+2) Looks like the water cooling block may come in handy.
Jally 2: 5 chips to 7 chips (+2)
Jally 3: 5 chip (this is the bad board one) to 7 chips (+2). Might as well set it alight as well.

This will cover my three dead chips; I'll use my last good chip for a friend's jally, giving him a 12gh unit that can be used or sold without super cooling. My reballs I'll put on the danger board if I can, what's the worst that can happen?

I am going so straight to hell here. But it's all for... research.

Right.

Crap, I just realized something: I have one other chip, and one of my slots is bad. So I can only do four. UG. Will request update, if seller does not agree will honor 6 chip request.

PS: I'd ordered two PCI-E to 3 Plug custom power cables when I placed my order. They were not in the box. Logged back into BFL's site and checked my invoice, and it clearly shows I paid $29.98 for the cables and also indicates that they were shipped. I wrote an inquiry to office@butterflylabs.com, but to date have not received a reply.

Same here. At this point it's pointless to get the cable; I'm already at full power here. I guess I can just ask BFL to send me a chip or something instead. Or my money back. :-)

C

[1715269509]

1715269509

[1715269509]

1715269509

[1715269509]

1715269509

[1715269509]

1715269509

[1715269509]

1715269509

[1715269509]

1715269509

[synapse]

Wait, that chip you blew up is a voltage regulator? I thought the six FETs were it, or is that some sort of LM317 for the fans?

And is that big box thing damaged as well?

C

Yes, notice the hole in the chip. The number on it was ST1S10, on mouser it comes up as a voltage reg. This particular jala is the one that took out the same power supply. When I used a good power supply it blew that chip.... doesn't necessarily mean that's the problem though....

The blown caps are on the blocks DC output side, everything else looks fine.  I'm not sure what you mean by big box thing? If you mean the USB controller chip, I ordered a couple. I have 2 jala's that wont register in windows... trust me I researched it on two PC's and 3 different miner programs. Again everything is just to the best of my knowledge, I'm out on a string with some of this stuff.

The 2 mouser chips I got ordered are

ST1S10 (Voltage reg)
FT232HQ (FTDI driver chip for USB)

Yes I'm in the USA unfortunately.

Bugga, I just blew a hole in the exact same chip in the exact same spot. I thought it was new years eve fireworks again.
Power supply tests ok though. the thing powers up with one red led near the power connector. The red leds on the back and front don't light up anymore.

Torn between sending back as an RMA, or having a go at replacing the chip myself.

[lightfoot]

You blew up U15 as well? Hm, interesting; wonder why that is failing.

I think that is the 3.3 volt supply line, if you want to send it over I can take a look at it. I'm sure though they will honor the RMA. Drug, did swapping that chip fix it?

C

[MrTeal]

Remove the chip again, and check the resistance on the 3.3V rail. If it's below about 10 ohms or so, something is shorted and that's what keeps blowing the ST1S10.

Shorts on a main power rail like that are a huge PITA to debug. Depending on what you have available, there's a couple ways to find out what the issue is. The easiest way I know is to hook a lab supply at 3.3V up to 3.3V rail and turn the current limiting down. Grab an IR camera, and turn up the current until you see a hot spot. You can do the same with an IR thermometer, but it's more time consuming. A finger work too, but you need get the problem a lot hotter that way, and if it's something that's a QFN or one of the ASICs you might never notice it get hotter.

Without a current limited lab supply, it becomes more of a crapshoot. It's probably not worth trying to debug yourself if that's the case.

[lightfoot]

Remove the chip again, and check the resistance on the 3.3V rail. If it's below about 10 ohms or so, something is shorted and that's what keeps blowing the ST1S10.

Shorts on a main power rail like that are a huge PITA to debug. Depending on what you have available, there's a couple ways to find out what the issue is. The easiest way I know is to hook a lab supply at 3.3V up to 3.3V rail and turn the current limiting down. Grab an IR camera, and turn up the current until you see a hot spot. You can do the same with an IR thermometer, but it's more time consuming. A finger work too, but you need get the problem a lot hotter that way, and if it's something that's a QFN or one of the ASICs you might never notice it get hotter.

Without a current limited lab supply, it becomes more of a crapshoot. It's probably not worth trying to debug yourself if that's the case.

Yup. However the 3.3 volt supply on a jalapeno is really pretty small. Even the 1 volt supply will just crowbar and die if shorted (verified on the danger board); what blows things up in the FET world is when the FETs short due to overheat on high loads. Then if you feed it from an unlimited 12 volt supply (as opposed to the little supplies BFL sends) then the weakest component will explode.

That's why I have been recommending putting a 7 amp fuse in series with your jally when running it on 300-800 watt supplies. 12 volts at 7 amps will do minor things. 12 volts at 50 amps will do a bit more....

Having bench top power supplies is a godsend; I have a pair of them myself and can feed all three voltages to a board. Oddly enough I am getting a board with blown FETs on it, I'm debating checking the 1 volt rail, clearing the failures, and trying to power it off the danger board's 1 volt supply. A double-headed jalapeno. :-)

It keeps my mind working.

C

[Drug5bitz]

You blew up U15 as well? Hm, interesting; wonder why that is failing.

I think that is the 3.3 volt supply line, if you want to send it over I can take a look at it. I'm sure though they will honor the RMA. Drug, did swapping that chip fix it?

C

I wish I could say yes. I believe when I originally replaced it just powered up for a second. It blew again or something else when a real PSU was used. I think I pulled a pad off when removing it a second time. So I removed the chips and will list on eBay as a dead jallie soon..... Sorry

[arekm]

Remove the chip again, and check the resistance on the 3.3V rail. If it's below about 10 ohms or so, something is shorted and that's what keeps blowing the ST1S10.

Shorts on a main power rail like that are a huge PITA to debug. Depending on what you have available, there's a couple ways to find out what the issue is. The easiest way I know is to hook a lab supply at 3.3V up to 3.3V rail and turn the current limiting down. Grab an IR camera, and turn up the current until you see a hot spot. You can do the same with an IR thermometer, but it's more time consuming. A finger work too, but you need get the problem a lot hotter that way, and if it's something that's a QFN or one of the ASICs you might never notice it get hotter.

Freeze it with spray (or compressed air upside-down) and see where it becomes hot - is another method.

[synapse]

You blew up U15 as well? Hm, interesting; wonder why that is failing.

I think that is the 3.3 volt supply line, if you want to send it over I can take a look at it. I'm sure though they will honor the RMA. Drug, did swapping that chip fix it?

C

Thanks for the offer. I'm in Australia though so probably not worth it for me.
I think u15 blew because I recently changed to an ATX power supply rather than the power brick (all in effort to prevent something like this from happening.)
The sparks came as soon as I un-plugged and re-plugged in the USB cable while trying to get rid of "unresponsive asic" messages in cgminer.
For next time I'm considering wiring in a 30v 6A Poly switch that I have here or maybe a fuse.

Thanks everyone else for the suggestions. I don't have a current limited power supply so this might be a bit too hard to fix myself.
My main issue now is if BFL will try and fix it before sending me a new one. If so they might see that I reflashed to get to 7.8gh/s

[lightfoot]

Hm, yep that's the 4 amp 3.3 volt power supply. Out of curiosity what happens when you plug it in now?

Do you have a volt meter and know how to use it? I'd be really curious to see if the 1 volt power line is up; it feeds off the 12 volt supply line. If that's working and the 3.3 volt short could be cleared you could power the board up with a small bench supply and it might work again. But man that is a serious hack.

Speaking of serious hacks, I have a board coming in with a blown set of 1 volt FETs. I'm going to try powering it off the danger board with jumper wires; technically it might work. Then again the voltage drop might be too much but what the heck?

:-) Keeps me out of trouble.

C

[elaramus]

That's why I have been recommending putting a 7 amp fuse in series with your jally when running it on 300-800 watt supplies. 12 volts at 7 amps will do minor things. 12 volts at 50 amps will do a bit more....

Despite my intentions to do so, I didn't (install a fuse) and thats exactly what happened to the board coming your way.

[lightfoot]

That's why I have been recommending putting a 7 amp fuse in series with your jally when running it on 300-800 watt supplies. 12 volts at 7 amps will do minor things. 12 volts at 50 amps will do a bit more....

Despite my intentions to do so, I didn't (install a fuse) and thats exactly what happened to the board coming your way.

Ah. So what I see when I tested a shorted jally (and I have shorted the danger board a *lot* as of late) is the crow-baring of the cheap-o BFL power supply and not a failure of the 1 volt supply to come up into a short. That is interesting, I wonder how much power it sourced before blowing up. Probably a lot. :-)

Ok, then my protocol of testing a newly balled jally with a BFL supply in all cases is sound. That's good to know.

C

[synapse]

Hm, yep that's the 4 amp 3.3 volt power supply. Out of curiosity what happens when you plug it in now?

Do you have a volt meter and know how to use it? I'd be really curious to see if the 1 volt power line is up; it feeds off the 12 volt supply line. If that's working and the 3.3 volt short could be cleared you could power the board up with a small bench supply and it might work again. But man that is a serious hack.

Speaking of serious hacks, I have a board coming in with a blown set of 1 volt FETs. I'm going to try powering it off the danger board with jumper wires; technically it might work. Then again the voltage drop might be too much but what the heck?

:-) Keeps me out of trouble.

C

I just get the red light near the power connector, none of the other LED's light up at the front or back.

I have a multimeter to test with. I just need to know where to test. I haven't had a chance to look at the board layout and see what is what.
I'ts certainly the most tightly packed board I have ever had to look at.

[lightfoot]

Hm, yep that's the 4 amp 3.3 volt power supply. Out of curiosity what happens when you plug it in now?

Do you have a volt meter and know how to use it? I'd be really curious to see if the 1 volt power line is up; it feeds off the 12 volt supply line. If that's working and the 3.3 volt short could be cleared you could power the board up with a small bench supply and it might work again. But man that is a serious hack.

Speaking of serious hacks, I have a board coming in with a blown set of 1 volt FETs. I'm going to try powering it off the danger board with jumper wires; technically it might work. Then again the voltage drop might be too much but what the heck?

:-) Keeps me out of trouble.

C

I just get the red light near the power connector, none of the other LED's light up at the front or back.

I have a multimeter to test with. I just need to know where to test. I haven't had a chance to look at the board layout and see what is what.
I'ts certainly the most tightly packed board I have ever had to look at.

It's not too bad. Grab the schematics by googling for sc-mainboard-1.0, it's a PDF from BFL. Once opened take a look at the board layout. On the opposite side from the power plug (top side) you will see a square hole, then a space, then five more holes. Right by the JTAG port.

Those are the voltage test points. They are:

ground     1-volt     3.3-volts    5.0-volts     12 volts.

All reference to ground. See what you find with the board plugged in.

C

[synapse]

Thanks C,
I'll test it when I get home tonight.
'spose i'll asses from there and see if I want to try fixing it myself or not.

[synapse]

Hm, yep that's the 4 amp 3.3 volt power supply. Out of curiosity what happens when you plug it in now?

Do you have a volt meter and know how to use it? I'd be really curious to see if the 1 volt power line is up; it feeds off the 12 volt supply line. If that's working and the 3.3 volt short could be cleared you could power the board up with a small bench supply and it might work again. But man that is a serious hack.

Speaking of serious hacks, I have a board coming in with a blown set of 1 volt FETs. I'm going to try powering it off the danger board with jumper wires; technically it might work. Then again the voltage drop might be too much but what the heck?

:-) Keeps me out of trouble.

C

I just get the red light near the power connector, none of the other LED's light up at the front or back.

I have a multimeter to test with. I just need to know where to test. I haven't had a chance to look at the board layout and see what is what.
I'ts certainly the most tightly packed board I have ever had to look at.

It's not too bad. Grab the schematics by googling for sc-mainboard-1.0, it's a PDF from BFL. Once opened take a look at the board layout. On the opposite side from the power plug (top side) you will see a square hole, then a space, then five more holes. Right by the JTAG port.

Those are the voltage test points. They are:

ground     1-volt     3.3-volts    5.0-volts     12 volts.

All reference to ground. See what you find with the board plugged in.

C

Ahh I see, they cut the board so close to those test points the silkscreen of gnd, 1v etc is missing.

OK at the test points I have: 1.01V, 0.0001V, 0V, 13.1V
should I have 5v?

Resistance between ground and:
1v = 41ohm
3.3v = 0.20 ohm
5v = 2.9k ohm
12V = 2k ohm

It's hard to tell if this is the burnt out u15 chip causing the short or something else as well causing this.
If Mr Teal thinks it might be hard to diagnose then it might not be worth removing u15 and RMA'ing instead.

[lightfoot]

OK at the test points I have: 1.01V, 0.0001V, 0V, 13.1V
should I have 5v?

Eh. I think the 5 volt circuit was on some boards and not others. Technically it's only needed for the LCD display, whatever that was. However lack of 3.3 volts will end it.

Resistance between ground and:
1v = 41ohm
3.3v = 0.20 ohm
5v = 2.9k ohm
12V = 2k ohm

Yep, confirmed short on the 3.3 volt line. Given that all of the chips use 3.3 volts it could be anything. I'd say do an RMA on it, but I do wonder what blew up the 3.3 volt line.

C

[synapse]

This one that blew up had the older style chips. no 5v at the 5v test point.
Out of curiosity I checked my other Jally that has two of the new style of chip. That one has 5v present at the 5v test point.

I think i'll just RMA it.
Thankyou for all your help anyway.
I have 5 ASIC chips on the way so i'll pop 3 on the replacement and two on my other one. (this is how i came accross this thread in the first place. I remembered the post with the blown u15 pics)

[lightfoot]

This one that blew up had the older style chips. no 5v at the 5v test point.
Out of curiosity I checked my other Jally that has two of the new style of chip. That one has 5v present at the 5v test point.

Hm. I've seen that too. My weird jally didn't have 5 volts but it seems to be running well now. Maybe they put it in sometimes.

I think i'll just RMA it.

I have 5 ASIC chips on the way so i'll pop 3 on the replacement and two on my other one. (this is how i came accross this thread in the first place. I remembered the post with the blown u15 pics)

That sounds best. I've got 4 chips outside in the mailbox I think; I'll put one of them on my weird jally, taking it to 5 and the other on my other new jally taking it to six on an old-style heat sink. I think that might wind up being a problem, but my water block is not coming till Monday at the earliest.

And I get to look at another blown board. Fun!

Speaking of chip placements, put another one on a 2 chip jally for a friend, no problems. I'm actually waiting 5 minutes with preheat now, until I see the faint wisp of smoke/flux come up, then a count to 90 with the air tool at 450c. The hardest part is placing the chip perfectly on the balls, but I have had a 100% success rate since I started using a loupe to verify all the balls.

C

[lightfoot]

Well, 2 more chips went on the five chip jally taking it to seven. Yes, 28gh. Yes, insane. Thoughts:

With 6 chips on the board I was hashing a good sold 24gh, and was able to keep the system cool with the aluminum heat sink, stock fan pointing down, and a bottom plate. Temps in the 60's. Note I have large heat sinks on the FETs and a fan pointing at the FETs and the FTDI chip.

With 7 chips, forget it. Temps were between 75 and 85 at which my BFGMiner throttles the work. Shut it down, put it on the Copper heat sink plate, fired it up. Temps now between 70 and 75, very hot but still running.

There is little doubt I need more cooling. Might try swapping the AL heat sink for the copper heat pipe one on my old 5 chip jally. 20gh is suddenly a lot slower, and now that I run it open as well, temps are not a problem. This though is really generating some warm.

We do need water cooling here to get to that magical 8 chips. It's funny seeing all the lights but one on the back of the jally burning brightly, that last LED is so lonely being off....

C

[lightfoot]

Was running 75-80 with the bottom sink and top. Shut it down.

Now running at 40c. With a Corsair 100 water cooling block on the bottom.

This um.... solves the heat problem. The mounting is actually quite clever. Will go to 8 tomorrow.

Water is without a doubt the only way to fly.