Pages:
Author

Topic: [DIY] - Reward $100 | Antminer S1/S3 Blade on Raspberry Pi - page 30. (Read 82132 times)

legendary
Activity: 1638
Merit: 1005
very curious to see this project's success. would love a controller for multiple blades.  can't wait anymore.  we're running out of time :-)

I tried to find some used control board or heatsinks. it is impossible apparently Tongue

Anyone know what kind of cables i can buy to connect my future boards to the controller ?  I will be getting one for sure !

I'm also trying to figure out an alternative to the existing heatsink... I will probably underclock at the most possible but i will still need some kind of heatsink i'm sure.

Thanks !
tss
hero member
Activity: 742
Merit: 500
very curious to see this project's success. would love a controller for multiple blades.  can't wait anymore.  we're running out of time :-)
hero member
Activity: 840
Merit: 1000
j4bberwock will soon have a controller board ready.. he is waiting for the pcbs to arrive
its first step, next will sofware
contollers are not a so problem

that is why j4bber is sending out 2 controller boards, one to me, and another one so we can get on hacking at cgminer

Yeah, it finally moved from the post office and should be delivered today.
Fingers crossed, knocking on wood.

I have another package that is stuck somewhere at France airport, but hopefully not related to this project.
10/13/2014 12:00:00 AM    France. Arrive at France Airport
It's only been 10 days, and I live 2 hours from this airport...
legendary
Activity: 1358
Merit: 1002
j4bberwock will soon have a controller board ready.. he is waiting for the pcbs to arrive
its first step, next will sofware
contollers are not a so problem

that is why j4bber is sending out 2 controller boards, one to me, and another one so we can get on hacking at cgminer
hero member
Activity: 784
Merit: 504
hi guys! Great theme, cp2102 works fine with a single blade a lot of HW
The board overheating without a heat sink causes it. After lowering the voltage to 0.82V the blade stays cold, but I still screwed it to a piece of the monitor as a heat sink)
The best string that I’ve found is --bitmain-options 115200: 1: 8: 1: 200: 5002 gives the minimum number of errors and gives 38Gh on pool. It’s possible to change frequency with cgminer 4.7.0 with --anu-freq string, it recognizes the blade but it works like a single chip. Frequency greater than 250 dramatically increases the number of errors due to undervolting. At the frequency of 270 it consistently shows 43Gh on the pool. I sometimes managed to achieve speed about 80Gh on the pool with voltage turned to 1.1V.  The main problem is the old version of cgminer, I will try to compile a bimain version and describe the results here
[img]http://s020.radikal.ru/i700/1410/b8/2acae425e183.jpg[/im

put the fans blowing on the back side of the PCB - it wil be a lot more effective than trying to cool the outer side of the chips

Put one fan on each side to make the temperature under control.
Don't forget to add a good heatsink on backside.
newbie
Activity: 13
Merit: 0
j4bberwock will soon have a controller board ready.. he is waiting for the pcbs to arrive
its first step, next will sofware
contollers are not a so problem
member
Activity: 104
Merit: 10
i have an extra s1 yes, thought about getting a controller just to try...seems logical it might work..
hero member
Activity: 518
Merit: 500
why wouldnt a block erupter controller work? curious..
Have you got one and a couple of S1 boards lying around? Try it and let us know .... if you have any other questions along the way, ask here and you'll be answered.
legendary
Activity: 1358
Merit: 1002
j4bberwock will soon have a controller board ready.. he is waiting for the pcbs to arrive
member
Activity: 104
Merit: 10
why wouldnt a block erupter controller work? curious..
newbie
Activity: 17
Merit: 0
its prototype) blade cold, place the fans I checked the temperature dependence of the HW.Heatsink required if the voltage is above 0.9v
legendary
Activity: 2128
Merit: 1005
ASIC Wannabe
hi guys! Great theme, cp2102 works fine with a single blade a lot of HW
The board overheating without a heat sink causes it. After lowering the voltage to 0.82V the blade stays cold, but I still screwed it to a piece of the monitor as a heat sink)
The best string that I’ve found is --bitmain-options 115200: 1: 8: 1: 200: 5002 gives the minimum number of errors and gives 38Gh on pool. It’s possible to change frequency with cgminer 4.7.0 with --anu-freq string, it recognizes the blade but it works like a single chip. Frequency greater than 250 dramatically increases the number of errors due to undervolting. At the frequency of 270 it consistently shows 43Gh on the pool. I sometimes managed to achieve speed about 80Gh on the pool with voltage turned to 1.1V.  The main problem is the old version of cgminer, I will try to compile a bimain version and describe the results here
[img]http://s020.radikal.ru/i700/1410/b8/2acae425e183.jpg[/im

put the fans blowing on the back side of the PCB - it wil be a lot more effective than trying to cool the outer side of the chips
newbie
Activity: 17
Merit: 0
 hi guys! Great theme, cp2102 works fine with a single blade a lot of HW
The board overheating without a heat sink causes it. After lowering the voltage to 0.82V the blade stays cold, but I still screwed it to a piece of the monitor as a heat sink)
The best string that I’ve found is --bitmain-options 115200: 1: 8: 1: 200: 5002 gives the minimum number of errors and gives 38Gh on pool. It’s possible to change frequency with cgminer 4.7.0 with --anu-freq string, it recognizes the blade but it works like a single chip. Frequency greater than 250 dramatically increases the number of errors due to undervolting. At the frequency of 270 it consistently shows 43Gh on the pool. I sometimes managed to achieve speed about 80Gh on the pool with voltage turned to 1.1V.  The main problem is the old version of cgminer, I will try to compile a bimain version and describe the results here
http://s020.radikal.ru/i700/1410/b8/2acae425e183.jpg

legendary
Activity: 1638
Merit: 1005
Soooooo, i took a look at the current PCB and BOM files that Bitmaintech put on github.

For like 100 units, it will cost at least 30$ each. I don't think producing the current board will be economically viable Tongue

I had to check !
newbie
Activity: 13
Merit: 0
Lets keep project alive!

FYI (Who dont know) original controller based on pic32mx250 (ready boards are Penguino or Fubarino Mini) which support 4 uarts. Possible to apply PicKits - but bitmain close source codes =(
PIC32MX795 (ready board http://fubarino.org/sd/index.html) - possible to attach 6uarts
- so needed programmer for this project way

another way - i will work
use USB-UART (based on cp2102 - less 1$ for piece on aliexpress) - attach only one board to the controller...
But in my way needed software ... code open by bitmain
https://codeload.github.com/bitmaintech/cgminer/zip/master
i will work on these... 

check driver-bitmain.h for example Wink

now i am waiting boards and controllers to begin (shipped)
legendary
Activity: 1638
Merit: 1005
The significance of this project is already dead or being dead.
What you guys think?

I Disagree. undervolt them with cheap electricity = win (add some fun too Wink )
legendary
Activity: 1638
Merit: 1005
If Bitmain confirmed that we could use one existing S1 controller for 8 blades. Why don't they share the info to do it ?

The chips used in the controller board have 4 TX/RX lanes, so 4 boards would have been possible, if they were all connected (only 2 are)
At a time, Bitmain were speaking of a new controller that will allow more boards (up to 8?), but they never released anything.

Thanks for the answer ! :-)
newbie
Activity: 1
Merit: 0
What you guys are doing here is great!
I've got an S1 with a dead controller that I would love to get some more life out of (undervolted or not), so I'll gladly contribute $20 to the bounty if this guide is ever completed.
hero member
Activity: 840
Merit: 1000
If Bitmain confirmed that we could use one existing S1 controller for 8 blades. Why don't they share the info to do it ?

The chips used in the controller board have 4 TX/RX lanes, so 4 boards would have been possible, if they were all connected (only 2 are)
At a time, Bitmain were speaking of a new controller that will allow more boards (up to 8?), but they never released anything.
hero member
Activity: 784
Merit: 504
The significance of this project is already dead or being dead.
What you guys think?
Pages:
Jump to: