bcp19
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 05, 2014, 01:33:21 PM |
|
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.
Nice cooling! Are you going to include pics of this monster?
|
I do not suffer fools gladly... "Captain! We're surrounded!" I embrace my inner Kool-Aid.
|
|
|
lightfoot (OP)
Legendary
Offline
Activity: 3150
Merit: 2257
I fix broken miners. And make holes in teeth :-)
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 05, 2014, 03:08:14 PM |
|
Nice cooling! Are you going to include pics of this monster?
Need to wait for the sun to come up for a better picture, but you can see all seven lights a blinking.... ![](https://ip.bitcointalk.org/?u=http%3A%2F%2Ffarm6.staticflickr.com%2F5548%2F11776589723_35816de28c_z.jpg&t=665&c=UYkeTnpBUjBKcA) Thinking about going to 8 this morning: The issue is the bottom of the board under the FETs is still hot. I mounted the block so that it uses the default BFL holes and the fan as the other side against the AL heat sink (brilliant!) however this requires a piece of electrical tape in there to insulate those stupid VIAs. That's cutting down on the block's cooling ability. But still I am concerned about the FETs going to 8. This might be as far as I want to go right now; if I blow the unit I lose 28gh of power. Maybe I'll try boosting the air cooled one. BFL people: Any thoughts on how much more those FETs can handle? Hm.... A thought: I wonder if I could tie two jallies together, one with only 4 chips, the other with 8 to balance the 1 volt loads. Why not? Maybe I'll do that: Focus on getting the danger jally and the melted jally working, then try tying their power grids together. C
|
|
|
|
helipotte
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 05, 2014, 06:11:27 PM |
|
Lightfoot, I just got 4 jalapenos and I have gotten the hackit-to-go-faster bug. Already changed out the 7k resistor with a larger one to get more voltage to the ASICs. This reduced the errors by 1% and increased the speed by 500Mhs. Thinking about putting maybe two more chips in each unit. Do you know where the early type (no laser etch on top) chips can be obtained? ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) To answer you question in the above post the schematic indicates 75-80A on the main 1V supply. I would guess in their totally stock trim (180Mhz), these chips use maybe 8-10A each? Running at little single speeds (250Mhz+), maybe 12-15A. If you install 8 chips, run them at 200Mhz, this should yield 24Ghs if All 15 engines are enabled. I do not know the current usage on the chips vs speed. Just guessing here. I figure the longboard in the single/little single is different than the jala board. Otherwise BFL would just plop a 8 chip jala board in the little single and call it a day. Maybe the VRM on the longboard has a higher current rating? Seems like I read somewhere that the longboard can do 100A@1Vdc.
|
|
|
|
lightfoot (OP)
Legendary
Offline
Activity: 3150
Merit: 2257
I fix broken miners. And make holes in teeth :-)
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 05, 2014, 09:33:23 PM |
|
Well..... I've had a productive day.
First I added a second chip to my friend's jally, bringing it to 16gh. However this chip was a reballed unit; I used one of my old wrecked chips to learn on, figured out how to use the reballing tool and .050 screen (that's it) with .045 balls (bit smaller means less chance of shorting the damn thing), then took a chip I screwed up, reballed it, and put it on.
Worked perfectly.
Then I took another unit up to 6 chips, no problems, then took my air cooled jalapeno (the one with the cooler master fan) to 7 chips. Note: water cooling kicks more ass than God. I put a small piece of electrical tape over the vias, and just mounted the IBM X46 heat sink to the whole bottom of the board, and put the cooler master on top. It's running at 70c with all that sinking. The water cooled unit with *one* fan running on the radiator is at 45c.
Moral: Screw air cooled heat sinks. However this one is working, both are hashing at 27-28gh, and I'm done for the day. I still have two more chips I can reball, not sure where I will put them. I need more jallies. (no no no no no). :-)
So there you go: 6 chips can be handled by the stock fan and custom power supply. 4 chips is the max for the stock PS. Seven chips is for the men.
C
|
|
|
|
lightfoot (OP)
Legendary
Offline
Activity: 3150
Merit: 2257
I fix broken miners. And make holes in teeth :-)
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 05, 2014, 09:39:50 PM |
|
Lightfoot, I just got 4 jalapenos and I have gotten the hackit-to-go-faster bug. Already changed out the 7k resistor with a larger one to get more voltage to the ASICs. This reduced the errors by 1% and increased the speed by 500Mhs. Thinking about putting maybe two more chips in each unit. Do you know where the early type (no laser etch on top) chips can be obtained? ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) Hm. If you want to reball them, I have at least three. .02btc each, unknown if they work (two probably do). maybe 8-10A each? Running at little single speeds (250Mhz+), maybe 12-15A. If you install 8 chips, run them at 200Mhz, this should yield 24Ghs if
All 15 engines are enabled. I do not know the current usage on the chips vs speed. Just guessing here. I figure the longboard in the single/little single is different
They seem to pull about 12-13 watts each, which at 1 volt is 12-13a. The original chip specs were supposed to be 8-10 watts each, that's where they ran into trouble. 8 10 amp units is 80a, 8 13a units is 104a. Big difference for power and heat dissipation. than the jala board. Otherwise BFL would just plop a 8 chip jala board in the little single and call it a day. Maybe the VRM on the longboard has a higher current rating? Seems like I read somewhere that the longboard can do 100A@1Vdc.
The longboard has more space, better cooling, and FETs that can work together to share the load. Even with that they limit to 60gh, that's fair enough. But running the little single with 7 chips requires literally water cooling to handle the load. I'd recommend not going past 5 chips to be honest. Six is do-able if you get heat sinks for the FETs. Seven rocks. C
|
|
|
|
Drug5bitz
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 05, 2014, 11:32:31 PM |
|
Could you please clarify 0.05 screen and 0.045 balls? You don't mean 0.5mm screen and 0.45 balls? If what you said originally is correct I can seem to find the supplies on eBay. Congrats on the water cooling!!! Kinda thinking, I didn't go this way but on eBay they have water blocks for like 10-15$ Hook um all up with a pump and a big enough rad and that's pretty tits. Just an idea as I know those self contained units arn't cheap on their own. Check out the water cooling they did with the 60 Ghs models.
FYI I have 8 12v rails 25A, 2 jalapenos a piece. My custom 16 gauge wires are a little warm on the 2 jalapenos 1x6 chips 1x4 chips. The 18 gauge wires from the power supply are actually warm-hot to the touch. The insulation isn't melting yet but I think I need to pull the weak link out. 6 chips still running magical under 60c on air..........
|
If you would like to donate to my jalapeno mods, or just buy me a b33r it's all appreciated.
BTC Address 1DX24XAojH2qjAgFzbME81o9BD3yDjfGLR
|
|
|
lightfoot (OP)
Legendary
Offline
Activity: 3150
Merit: 2257
I fix broken miners. And make holes in teeth :-)
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 06, 2014, 01:18:16 AM |
|
Could you please clarify 0.05 screen and 0.045 balls? You don't mean 0.5mm screen and 0.45 balls? If what you said originally is correct I can seem to find the supplies on eBay. Congrats on the water cooling!!! Kinda thinking, I didn't go this way but on eBay they have water blocks for like 10-15$ Hook um all up with a pump and a big enough rad and that's pretty tits. Just an idea as I know those self contained units arn't cheap on their own. Check out the water cooling they did with the 60 Ghs models.
FYI I have 8 12v rails 25A, 2 jalapenos a piece. My custom 16 gauge wires are a little warm on the 2 jalapenos 1x6 chips 1x4 chips. The 18 gauge wires from the power supply are actually warm-hot to the touch. The insulation isn't melting yet but I think I need to pull the weak link out. 6 chips still running magical under 60c on air..........
Yep, added an extra zero. .5 screen, .45 balls. Or maybe they are .5 balls but I think they're a bit smaller. Works well enough. Don't use too much heat to melt them on; I really cooked a few off on my second reballing. Oh well, clean and try again. Double up the wires coming from your power supply; I tripled the 18 gauge wires together and they seem to hold. But the Corsair has 16 gauge wire by default and that thing runs all sorts of stuff. It's got an 8 chip chili, 7 chip air cooled, and 5 chip jalapeno. I think I'll move the water cooled 7 chip in there as well, or maybe another 5 chip unit. It's a very good power supply. Wish I could get the danger jally up and going, seems to just fast-flash. Might be a firmware problem, I'll check that next. C
|
|
|
|
lightfoot (OP)
Legendary
Offline
Activity: 3150
Merit: 2257
I fix broken miners. And make holes in teeth :-)
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 06, 2014, 04:06:40 AM |
|
This is a Jalapeno running 7 chips, 28gh mining speed. Air cooled, 70c temp: ![](https://ip.bitcointalk.org/?u=http%3A%2F%2Ffarm4.staticflickr.com%2F3812%2F11792059936_725f176830_z.jpg&t=665&c=NtHxC9QCx3eXWg) This is a Jalapeno running 7 chips, 28gh mining speed, water cooled, 40c temp. ![](https://ip.bitcointalk.org/?u=http%3A%2F%2Ffarm3.staticflickr.com%2F2867%2F11791290435_9ccfa84bcc_z.jpg&t=665&c=4MJuszKrEyo2_Q) ![](https://ip.bitcointalk.org/?u=http%3A%2F%2Ffarm6.staticflickr.com%2F5550%2F11777145636_8239db482a_z.jpg&t=665&c=cRuLsqMnUXhjzg) ![](https://ip.bitcointalk.org/?u=http%3A%2F%2Ffarm8.staticflickr.com%2F7310%2F11776782214_1b51a47e2d_z.jpg&t=665&c=04AMbfyxonN9rA) I think if there's a Bitcoin conference I'd be happy to give a talk on hacking your bitcoin miner to full power. :-) Actually maybe I should write something up and submit it to Defcon this year. It's fun, and they love this sort of stuff....
|
|
|
|
freddyfarnsworth
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 06, 2014, 04:25:23 AM |
|
You are having to much fun ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) Great pictures, can you give dimensions of PCB of Jally boards, I saw 1/2 inch uncovered on the air cooled one right by where you said the mosfets were getting heat from the back plate > I have a few 40ft storage bins of old computers and NOS servers and parts, I never throw anything away, and at the time the stuff was so hard to find (best of the best) NOS I hate to sell any of it. It is irreplaceable no longer on Bay or elsewhere, ALR I collected for years. I think I have some 30+ HUGE Copper xeon Intel sinks, may work for Jallys, need to know the size of the bottom plate (PCB). About 2 lbs or so each 1/2 inch pure copper. (Old Skool).
|
BTC: 1F1X9dN2PRortYaDkq89YJDbQ72i3F5N3h MEOW: KAbvy9jrrajvN5WLo7RWBsYqYfJKyN9WLf DOGE: DAyKSrTiVeRZaReTu1Cyf5Je6qPdKTuKKE
|
|
|
Pt0x
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 06, 2014, 04:38:48 AM |
|
Nice hack, congrats!
|
BTC: 17sz6AoYVpwXjaStmnVCsGTufUhvrAMhTw
|
|
|
synapse
Member
![*](https://bitcointalk.org/Themes/custom1/images/star.gif)
Offline
Activity: 79
Merit: 10
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 06, 2014, 08:12:33 AM |
|
BADASS ![Shocked](https://bitcointalk.org/Smileys/default/shocked.gif) Just out of curiosity, do both new and old style asic chips go on with the same technique? do you have to vary it between the two?
|
Part-time Computer Systems Engineering student - Full time Service Assurance (faults) for a large Telco.
|
|
|
lightfoot (OP)
Legendary
Offline
Activity: 3150
Merit: 2257
I fix broken miners. And make holes in teeth :-)
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 06, 2014, 06:23:35 PM |
|
BADASS ![Shocked](https://bitcointalk.org/Smileys/default/shocked.gif) Just out of curiosity, do both new and old style asic chips go on with the same technique? do you have to vary it between the two? Thank you! These things are definitely doing well, and once I get this other jally running (maybe) I will see about taking the two of them to 8 chips. Which will basically give me a single/60. :-) As for the chips: I have had issues with the old style chips, specifically I was shorting one of them because heating from the top caused the solder holding the chip to the board to go molten. That's a big problem; it implies that one needs to use a reflow oven instead of a rework gun since the chip and carrier do not conduct the heat to the pads. However I am also finding out that leaving the board at 375f on the pre-heat board is enough to make the components almost "jiggly" so that even a small amount of heat can blow a component off the board. It might be possible to preheat the board to hell, then use something like 375c air to melt the balls. Maybe.
|
|
|
|
lightfoot (OP)
Legendary
Offline
Activity: 3150
Merit: 2257
I fix broken miners. And make holes in teeth :-)
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 06, 2014, 06:26:11 PM |
|
You are having to much fun ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) Great pictures, can you give dimensions of PCB of Jally boards, I saw 1/2 inch uncovered on the air cooled one right by where you said the mosfets were getting heat from the back plate > I have a few 40ft storage bins of old computers and NOS servers and parts, I never throw anything away, and at the time the stuff was so hard to find (best of the best) NOS I hate to sell any of it. It is irreplaceable no longer on Bay or elsewhere, ALR I collected for years. I think I have some 30+ HUGE Copper xeon Intel sinks, may work for Jallys, need to know the size of the bottom plate (PCB). About 2 lbs or so each 1/2 inch pure copper. (Old Skool). Sure I'll measure the sides tonight. Oddly enough the holes in the corner fit the corsair PC mounting bracket with just the smallest amount of filing, so it's pretty close to a standard. The key is to cool the back of the board, the chips can deal with a heat sink. I've got some old hardware myself, mostly pdp11's. For awhile there I kept the mit-ai.arpa system safe and sound before sending it over to Paul. Still have a KS10 out in the shed, I fire it up every once in awhile and think about what might have been....
|
|
|
|
freddyfarnsworth
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 06, 2014, 09:15:42 PM |
|
Those are very old ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) ALR was PPro hex and quads era. Big Tower cases. ""Finishing up our PDP-11 tour are two more LSI-11 (microcomputer) based PDP-11's. One is a very special machine to my friend Hannes. It is the H11, a joint effort of Digital Equipment and Heathkit."" Heathkit, love the things... ""KS10[edit] The KS10 design was crippled to be a Model A even though most of the necessary data paths needed to support the Model B architecture were present. This was no doubt intended to segment the market, but it greatly shortened the KS10's product life."" Must have been for a small business model, or ? On Soldering, this tut is over a hour, many may find the whole page very informative. It did not go into BGA tho, but after the movie, some links did, not as good as this one, really eye opening for me. I was always trying to heat with no solder flooding to remove, ha ! flood it and a little heat it comes right off, then clean up. He make it look so easy http://projectklondike.org/how-to-buildIf I find any good sinks sizewise, I will pm and send you one my dime, thanks for the info.
|
BTC: 1F1X9dN2PRortYaDkq89YJDbQ72i3F5N3h MEOW: KAbvy9jrrajvN5WLo7RWBsYqYfJKyN9WLf DOGE: DAyKSrTiVeRZaReTu1Cyf5Je6qPdKTuKKE
|
|
|
lightfoot (OP)
Legendary
Offline
Activity: 3150
Merit: 2257
I fix broken miners. And make holes in teeth :-)
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 06, 2014, 10:49:47 PM |
|
An H11. The old 11/03 in a 4 slot chassis. Yep, good little system; I ran a 4 user BBS using MU-BASIC back in the day.
Interesting note: I may have hit my first problem: My 7 chip air cooled jally was reporting 300gh speed, 26gh accepted. Slightly odd from the POV of BFGminer but when I unplugged it and plugged in again one chip didn't come up. Shut down, started again, back to 26gh.
I might be banging into a limit here. Without watercooling; maybe I should buy a small block for this one, or that one with only 5 chips....
C
|
|
|
|
Brassguy
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 06, 2014, 11:53:09 PM |
|
An H11. The old 11/03 in a 4 slot chassis. Yep, good little system; I ran a 4 user BBS using MU-BASIC back in the day.
Interesting note: I may have hit my first problem: My 7 chip air cooled jally was reporting 300gh speed, 26gh accepted. Slightly odd from the POV of BFGminer but when I unplugged it and plugged in again one chip didn't come up. Shut down, started again, back to 26gh.
I might be banging into a limit here. Without watercooling; maybe I should buy a small block for this one, or that one with only 5 chips....
C
Probably not... I had something similar on one of my 6 chip jallys.... one chip had poor contact to the board. I had the corresponding led flashing, and the hashrate on cgminer was showing like 125gh... in my case the chip was disabled after power cycling. A reflow fixed the error and has been hashing steady for 9 days now at ~24gh.
|
15xNxXy2PfFv3rz8rnfkV6L7WQiwuYax2K
|
|
|
lightfoot (OP)
Legendary
Offline
Activity: 3150
Merit: 2257
I fix broken miners. And make holes in teeth :-)
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 07, 2014, 03:03:35 AM |
|
Probably not... I had something similar on one of my 6 chip jallys.... one chip had poor contact to the board. I had the corresponding led flashing, and the hashrate on cgminer was showing like 125gh... in my case the chip was disabled after power cycling. A reflow fixed the error and has been hashing steady for 9 days now at ~24gh.
It's possible: I checked my notes and that jally had the back plate on it when i went from 6 to 7 chips. I use heat sink compound to put on the back plates, and in that case I forgot to remove it. So it got less pre-heat from the heating table, and it did take longer to make the chip "drop" into place. If it happens again I'll rework it, thanks to your diagram it's easy to find the right chip. However it's at a happy 27gh for now so I'll let it be. Thank you!
|
|
|
|
lightfoot (OP)
Legendary
Offline
Activity: 3150
Merit: 2257
I fix broken miners. And make holes in teeth :-)
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 07, 2014, 04:42:56 AM |
|
Probably not... I had something similar on one of my 6 chip jallys.... one chip had poor contact to the board. I had the corresponding led flashing, and the hashrate on cgminer was showing like 125gh... in my case the chip was disabled after power cycling. A reflow fixed the error and has been hashing steady for 9 days now at ~24gh.
Well darn, chip light 6 is unhappy. Where is that really nice map of all the chips again; I drew my own but I can't find it now. C
|
|
|
|
lightfoot (OP)
Legendary
Offline
Activity: 3150
Merit: 2257
I fix broken miners. And make holes in teeth :-)
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 07, 2014, 05:53:40 AM |
|
Update: Found my chip map. #6 is the chip top right with the board pointing up, power plug bottom left.
And like I'm going to sleep on this :-) Pulled the heat sink, sure enough chip 6 was the one that didn't float well enough. Due to having the back plate on when I soldered it. Moral: Bottom pre-heat is incredibly important. Cleaned the heat sink crap, put some liquid flux under it, 375 preheat then 450c for 90 seconds. Chip dropped as it should, is same plane as the others. Successful boot with the little power supply, remounted sinks, hashing happily at 27gh.
Thanks for the help. I'll always remember to ensure that plate is off...
C
|
|
|
|
Brassguy
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 07, 2014, 05:26:51 PM |
|
Update: Found my chip map. #6 is the chip top right with the board pointing up, power plug bottom left.
And like I'm going to sleep on this :-) Pulled the heat sink, sure enough chip 6 was the one that didn't float well enough. Due to having the back plate on when I soldered it. Moral: Bottom pre-heat is incredibly important. Cleaned the heat sink crap, put some liquid flux under it, 375 preheat then 450c for 90 seconds. Chip dropped as it should, is same plane as the others. Successful boot with the little power supply, remounted sinks, hashing happily at 27gh.
Thanks for the help. I'll always remember to ensure that plate is off...
C
True story! I had the bottom plate on mine as well... didn't want to have to clean and regrease it... too lazy to do it right the first time, so ended up doing it twice! I'm using an IR station, but I concur... bottom heat is critical to a good chip installation.
|
15xNxXy2PfFv3rz8rnfkV6L7WQiwuYax2K
|
|
|
|