Show Posts
|
Pages: « 1 2 [3] 4 »
|
So is there any economic benefit for you to use this rig, instead just using S9? 29 TH @ 2800 WATTS 39 TH @ 3800 WATTS
How much did it approximately cost to buy the equipment and make it work?
Here's the main advantages of using such system: 1. The heat is being driven to exterior very easy. You can install a heat exchanger and benefit from the hot water. 2. Reduced cooling costs, it can be cooled with running cold water or with externally mounted radiators 3. The allows overclocking IC’s with 40%, obtaining more hashing power with less units. 4. Reduced service costs, the water blocks are easy to disassembly. 5. Increases electronic components life span, running between 35 and 50 degrees in comparison with 58 to 95 degrees with air cooling. 6. Eliminates the necessity of air filtration. For the costs, for 6 boards, the conversion kit is $2000, this includes the following: 1. 6 x waterblocks kits (12 waterblocks in total, as it's one waterblock for each side 2. 36 Thermal pads used between the board and waterblock 3. Thermal putty for the CPU's 4. 12 KW water radiator, capable of running at outside temperatures of over 45 Degrees Celsius and still keep the machine at optimal temps 5. Water pump 3500 liters per hour 6. Watter fittings and water distribution system 7. 2 x 2400 W PSU's The noise will be under 40 Db, it was noisy because of the PSU fans, but I changed those to bigger fans and solved this issue, I will upload another video soon explaining all the functionality and safety features.
|
|
|
Alex is using the s9 waterblocks you can find from alibaba if memory is correct.
No, I've designed my own blocks, the alibaba ones are not fitting properly on the board.
That's neat. Is the s9 different than the s9i? s9i already mines at 14.5 per bitmain.
This will work on any S9 machine with a board versions higher than v4 . The older versions don't overclock very well, but they can be run at smaller frequency. The chips are the same but the power module is more stable on the board, that's why it allows better OC.
|
|
|
Interesting work. Did you use the stock controller?
so 40th and about ? watts
Thanks! 3800 W for power usage and yes, it's stock controller but I added three more ports to the board.
So these are basically 6 boards from 2x S9j miners, right?
They are antminer s9i boards, it will work is s9j as well.
No, they are not s9 hydro boards, I am waiting for that to see what's inside.
|
|
|
Just saw your waterblock testing video!
Looks like you've made a lot of progress very quickly.
Have you figured out how to test the I2C chain efficiently to figure out which chips are faulty?
Hi, thanks! We did some work on the i2c which is talking with the PIC to enable the voltage for the IC's, we made it enable the board so we can communicate with the BM's, but we stopped there as we managed to modify the original firmware to obtain the things we were after, like modifying frequency and board voltage so we can overclock the board properly. The controller talk via UART with the BM's. There are 2 ways to talk with the IC's.. One option which looks a bit simpler is shown on this page: http://mightydevices.com/?p=615 . The second way to talk is the Bitmain way which is quite complex due to the lack of docs and the not very clean code they written for the firmware. Given the time we could put together a tool to test the IC's individually and come with a good or bad report, but we are focusing our efforts on something else right now. In the future we might look into this as well. BTW, for others who did not see the waterblock video, check it out https://www.youtube.com/watch?v=tj-qNBYHM0wIt turned out better than I was hoping, it keeps the local temp of the board to 41 degreese celsius just with a water radiator and a fan. I plan on installing an active water chiller and attempt to overclock and get the most of these boards as I think they can be run at at least 8TH/s per board in normal conditions by using these dual water block system. The best part is that these water blocks can be installed on any antminer s9 board with not too much skill required. If you are interested to discuss more, you can join this discord channel: https://discord.gg/5Eq5hUB
|
|
|
Hi, I just saw this thread. Are the boards ready for sale yet? I have a lot of BM1387 chips, would I be able to use them with this board to make an antminer s9? Please contact me: nikhilooga@gmail.comHi, yes, the board is working properly at 4.5 TH/s. I replied you by e-mail, will continue the conversation there. Alex
|
|
|
Hi guys,
My name is Alex, I have recently finished my work on developing a board similar to S9 boards. It's hashing with 4.5 TH/s with no issues and we are now working on improving the system.
We begin working on our own control board for raspberry pi for now, in the future we might be able to write the custom firmware on the original board as well.
I would like to know if there are people interested in such things to get more control over the miners.
Here's some of the features we are thinking to implement in the first version:
1. Ability to control the voltage of the board (for advanced users) 2. Ability to set frequency of the IC's 3. Use the official CGMiner build 4. Control multiple antminers from a single interface, giving you the option to switch pools to all the miners at once or monitor them on a single page.
Let me know if there are other features that you think the original interface lacks.
I appreciate your time for reading this topic!
Have a great day!
Alex
|
|
|
Salut, Recent am terminat un proiect cu o placa similara cu cea de antminer s9 de 4.5TH/s . Acum lucrez la un sistem de racire water block pentru a experimenta cu overclocking-ul chip-urilor. Ele ruleaza la 0.45V acum si modele mai vechi mergeau pana la 0.8V, plus frecventa merge schimbata. Aici puteti vedea un video pe care l-am facut: https://www.youtube.com/watch?v=a70ns3xpKUkAs vrea sa stiu daca sunt impatimiti Romani (programatori in principiu) care ar fi interesati sa dezvolte un controller bazat pe CGMiner pentru a inlocui controller-ul de antminer s9. Astept un PM in caz de sunt oameni interesati.
|
|
|
Hi guys, First of all nice job with the R808, it actually works. The downside is that I could get it just up to 70GH/s but I didn't buy it to mine is more for studying. I have made a mining board using 63 x BM1387, it's a replacement board for antminer s9 . I would like to know if any of the developpers which worked on this project would be interested to get in touch and work with me on a controller for this board which is going to use the original cgminer code instead of using an antminer board. I think there's huge potential to overclock the board with proper cooling as right now the chips are running at 0.45 Volts, which is almost half of what they would support IMHO. If you plan to create R808 with BM1387 chips, let me know, I can share some of the info on how the chips connect in exchange of teaching me how to control it . I wish you the best!
|
|
|
Hey guys, Here's the proof I promissed. It took some time as I was getting the board working outside the aluminum case and when I was putting it inside it was showing 0 Asics found. I had to find out that the uncovered vias used for prototyping are not very happy touching the aluminum case. I used some electrical tape to insulate the margins and it works fine. Glad it didn't burn anything. In the manufacturing version all the vias will be covered with solder mask so this issue will disappear. Here's a video showing how it works: https://www.youtube.com/watch?v=a70ns3xpKUkIn case you have any ideas on overclocking this unit, what needs to be changed, voltage, frequency or both. I plan to experiment on this after the waterblock is completed for both sides, to make sure I get to use this board a little longer. Have a great day!
|
|
|
Hey guys, I can confirm that the design is proven by the second prototype board.
The issues I had with the first board were fixed and I will make a video with the working board on Monday.
The board is hashing with 4.5 TH/s . I will now continue the work on the waterblock and install it on the board then begin tweaking settings to see what kind of overclock can be made.
|
|
|
Hi,
We should have the second board prototype in about a week. Will inspect, test and come with a video showing the pefromance.
One thing I want to let you know is that's not going to be cheaper than what you get from Bitmain, the reason is that the IC's BM1387 are much more expensive than what Bitmain is paying to produce them, 3-4 times more expensive actually.
So expect one board to be $1000+ for low quantity order and a slight discount if ordering a big batch.
Will keep you posted.
Alex
|
|
|
pm me, please.
thx
zwilla
Hi, You need to enable the "allow newbie to contact" I also enabled it. You can get in touch on skype as well, Alexander1560
|
|
|
Are your boards ready for sale? What is your best suggestion to improve s7 to s9? We have many S7 which is not profitable now.
Hi, the boards are not ready for sale yet. I am waiting for the second prototype board to be assembled, in about 2 weeks should be ready. I have not had the chance to look into S7 so far, but we are trying to improve the cooling of S9 to be more efficient in terms of power consumption. If you would be interested to donate an S7 machine so we can have a crack at it, we could try different cooling systems and see if there's any chance of overclocking if it would run cooler in liquied or with a waterblock. Today I am going to a factory to check the waterblock I have designed for the S9 machine, I will do some tests as soon as I have the system working. You can get in touch on skype Alexander1560 is my id if you are willing to send an S7 machine. I can look into improving it and then I can send it back or send you a free cooling system for the machine in exchange. Alex
|
|
|
Hello, can we replace bm1385 with bm1387 in s7 to have s9? Hi Amir, The IC's are very different, even the footprint is dissimilar between the chips. There's no way this could be done.
|
|
|
Hey guys. Quick update on the project. I have found multiple small issues which I have fixed on the prototype board. I managed to get it connecting with the original controller board and we are also working on making a rasp pi controller version which will allow controlling more than 3 boards on the same machine. I will make the changes on the design and update the parts list and will order a fully assembled board this time, as it took me 20 hours to manually place the parts (1465 parts) on the board and a lot more to inspect and test everything. It will take another 2-3 weeks to get the new prototype board automatically assembled to remove the error of bad soldering or applying uneven heat to the chips. I have made a workflow for servicing these boards using a DIY reflow oven with my arduino controller and will test this further on the future boards. The main problem I faced so far was that I was using chips from other boards as I did not wanted to distroy any new chips, and some of those were bad from the begining and as you know, the chain will stop working if a chips is bad, and there's no easy way to know which chip doesn't work. I will have to develop a testing gear to know which chip is bad or good. If there's someone with more knowledge about IC's, drop me a message, I would love to have a chat about how to debug these things. Lot's of things to take care of, but I am optimistic that this will work and there's soo much room for improvement on these and will begin working on that after the first batch. We already planed some changes to the cooling system, looking into using water blocks instead of air cooling, see some design pictures attached. https://ibb.co/fYCN2chttps://ibb.co/kSBFNchttps://ibb.co/foSBvxThe truth is that I intended finishing this project without asking for money from the comunity, but it took me already 2 months to make the design and I think that it will take another 30-40 days for a finished product including the extra features for efficiency. I am thinking to setup a crowdfunding campaign to help pushing through with this and enable me to get the first batch of 50 boards made. I will keep you updated.
|
|
|
|