Topic: BFL board project COINTAMINATION - EU facility - ORDERS OPENED for Chili - page 32. (Read 73072 times)

hi


Glad the forum is fixed


Apparently the boards work but the incredibly crappy low grade sample chips were mostly dead.

out of two boards 1 of the chips works on one, and 8 chips work on the other.

As far as I know the board design and assembly we did here in USA worked fine, it is just some of those sample chips are very very low quality.


I would expect a statement from Lucko today about how to order from him - for assembly at his facility in Slovenia.

I am waiting for him now on irc to see how things are moving forward.

-jj

The forum is back and I need my news fix. Did Form get the board assembled and hashing? Have you received the batch 1 chips yet? Did you get notice about delivery of batch 2? I was considering bombarding you guys with a bunch of questions but I'll leave it at that...

OK I hope that pictures did the job... Thanks form...

Now if anyone would like to vote for Gigahash-man can do that on BFL forums

https://forums.butterflylabs.com/dbtgallery.php?do=view_image&id=1376&gal=usergallery

I think it is a nice piece of art. It would be batter if it would not existed but this is the price of progress...

Now I might comment on a BFGMiner picture for all who are unfamiliar with it. You see 3 hashing speeds. First one is 5 second average. Next one is weighted all time average and the last one is work utility but not in shares per minute but in GH per second. So it shows what pool shows.

This are also some additional data.
Little single is completely stock. 8 chips A and B grade...
Coinado... 5 chips running 16 chips firmware... Realy bad chips. Only one is D grade all other are less then that. Running one job per chip. Old 1.2.5 based firmware.
Cointamination 1. One chip working. Below D gared. Running one job per chip. New 1.2.9 based firmware.
Cointamination 2. 8 chips working. From A to below grade D chips. Running one job per board. New 1.2.9 based firmware.

On Cointamination we also find new problem that cut hashing speed to 15 to 16 GH. Half-dead chips can make interference in system bus but form solved this problem with coding. So nice job... But still some added HW errors and less speed. But hahsing speed is close to theoretical max(second number).

Now again if you are not professional with hi end euipment don't mount chips on yourself. You will get yourself Gigahash-man and even with some smart coding from form less speed.

EDIT: new 1.2.9 also test engine 0 and enables it if it works. But we did not yet find a chips that would work with it...

Agent form is reporting back from shadows :-)
I would really appreciate if you want to donate. Here are some chars and numbers: 1DeLV4jBk1wFzjEooc53efVTRGztaSSv4V

It was a long time in front of the soldering iron, in front and even in several reflow-ovens, fighting with hotair-guns and heating-plates against the COINTAMINATION boards...
That all was done to get an idea if the board-design is working or not, before we hired the professional assembly fab. Unfortunately the manual solder-tries didnt work out so well - but it was enough to see that the design is actually working. Many chips died caused by too much heat. I needed some deaths to figure out how to treat them right. That was only possible by the sample chips some people donated. THANK YOU VERY MUCH for that possibility!

We will have no big news in the next 1-2 days (waiting for JJ's fab-assembled board to arrive), so meanwhile i'll post some hardware-porn pics.

http://my.root4u.de/COINTAMINATION-BETA/TN_gigahash-man.JPG
I raised a monument for all of the lost soldiers.



http://my.root4u.de/COINTAMINATION-BETA/TN_CIMG4596.JPGhttp://my.root4u.de/COINTAMINATION-BETA/TN_CIMG4603.JPG
This is a test to mount a stock cpu-cooler - As you see, a adapter-plate is needed to reach all chip clusters.



http://my.root4u.de/COINTAMINATION-BETA/TN_CIMG4609.JPG
But it seems to work with this easier construction, too.
(This has still to be verified with the professional assembled board with 16 chips which is on the way from JJ to me.)


http://my.root4u.de/COINTAMINATION-BETA/TN_CIMG4607.JPGhttp://my.root4u.de/COINTAMINATION-BETA/TN_Image2.PNG
This is my last handsolder-try with 16-chips. Actually only half of them are working.

All i can say now is: The boards itself are working fine and very very stable. No issues so far. If you look at the bfgminer-screenie you'll see the much lower temperature which is caused by non-overloaded power-regulators, as they are on BFL-boards. (Most heat on BFL-boards comes from MOSFETs, not the ASICs!)

I will come back here if i got the boards from JJ and tell some news then. They won't arrive tomorrow (legal holiday here), but certainly on friday.
Stay tuned and happy hashing (if you can)...

GB batch 1. If prototype works yes. I don't have 16 chips(only 7 samples) so I can't test... But we can make some agreement to host boards if needed...

No idea about size, depends on a radiator. It is about 200W per board so that is the guideline...

Flow would probably be the factor how many you can chain... So again this is the question...

Custom cooler is in the price of a board if you replace it with WB there will probably be some difference but I need to see cost... We are still at prototype stage... Probably you will see custom cooler pictures tomorrow(will send one to form ASAP to test)... WB will be available after that but no idea how soon at the moment. Need to test for leakage. If there will be time I will do my best to send one to form to test it with 16 chips...

But if you use WB you need to add some fans(not hi power really low speed) to get some air flow over power modules... I was looking into a way to cool them with WB but I can't do it at low cost...

Appreciate the honesty JJ.  Form & Luke-JR - You're up!

+1.  Good call to sent it to Luke-Jr. 

Thanks for your update.

Good Morning,

sorry I never came back on IRC last night. You can count on me to be here every day for a period of time but right now there is no possible way I be online all day long. I am in the process of setting up a 6 th/s datacenter  way up in northern NY near the Canadian border. (cheap cooling and electricity!)

So heres what I did because I know people want to see the board hashing and I will be completely honest I don't really understand how to setup this board properly, there is a bit of a language/time zone barrier between form, lucko and myself its possible I didnt understand the instructions properly and its possible I just did something wrong they are much more technically inclined then myself, and this is their design after all.

so as I stated yesterday I have two fully assembled 16 chip production boards, I had three but I killed one of them because I loaded the firmware too early.

At any rate here are the problems:

1)I don't want to kill any more boards.
2)People want to see video of this hashing before they place an order which I understand.

So here are my solutions:
I send two boards out:

One went to Luke-JR, he will have it today or tomorrow at the latest, and this will allow him to test with latest bfgminer and also post video of board hashing so everyone can see proof that the board works
which we already know it does, but I understand people want to see for themselves.

The other board I sent to fedex express directly to Form in Germany he should get it by early next week at the latest.

Now as far as contacting me it would be better to email me at [email protected] - the limitations of the PM system on this forum are cumbersome and I do not have time for that.

Right now I have only accepted one order for six boards, his ship date is coming up very soon and to be honest I really want to see how this first customer order turns out before I accept any others.

Because I am not sure If I am going to be setting the boards up after production or if someone else will do this role, but either way it will get done and we will make sure your boards are set up, running properly and shipped out to you ASAP, we really appreciate customer trust and loyalty and we will take care of you. It will be awesome to see these boards finally made with real high grade chips.

If anyone has any problems with this they can certainly place an order with Lucko, even if you are from the US if you have your chips already and are in a hurry to get them made your welcome to place an order with him, he has an assembly facility in Slovenia and I am sure they can make these boards just as well. My main concern here is that everyone gets a good product and in a timely fashion, and I don't want to be in a situation where customers are unhappy.

My role on this project was to get the PCBs made and the production Boards professionally Assembled, which I have done. I thought I could handle setting them up but as I stated before I ruined one board in the process and dont want to make that mistake again so I am sending the other two production boards to people who already know how to set them up and there is no learning curve involved here.

As soon as Luke-JR/Form receive their boards you will see them in action and we can move forward.

If you have any specific questions you need answered right away please contact Lucko I cannot guarantee I will be able to respond right away but I will make sure I respond to emails everyday so if you have a question that needs answering right away I would contact him.

I have to run, lucko - form I will see you on IRC in a little bit

-jj

form is hashing with hand solder board so yes board is hashing. It should be 16 chip but handsoldering didn't work as good as hoped so it is only 8 chip and that why we are waiting JJ for 16 chips pictures. It is not a board problem it is soldering problem. I asked form to make some pictures to come you down. Sample chips are not the best possible but he is getting 26 to 27 GH out...

We are unfortunately not sure if JJ is hashing. He told us he will be right back yesterday but he wasn't...

Are you hashing with any of the chips/engines?  You should be able to issue some query commands to the controller even if chips are not on the board or are dead. No?  Maybe you did not flash it right.  If it is firmware like BFL's, that is how it should work. i.e. the chip map will be empty.

Any idea when form will get your board?  I just hope it is just a software issue.  If it is a hardware problem, the project will be pretty much kaput as most people will just order from Mr. Teal ($600 without cooling elements) and start hashing before KNC gates open wide.  

Hi Lucko

Will this waterblock be available in time for group buy batch 1?

and what size radiator will it need?

and how many will we be able to chain together?

final question, will it have a cost impact?

Its a HELICOPTER

It needs a power brick that size to keep it on the desk.

Nice to see pictures of the finished product.
Congratulations to you all on the arrival of the new baby.

OK that was a near miss... Noting critical since you have enough samples but still you lost some hashing power Anyway I (EU) also have a board with 7 chips(that is all the samples I have left) getting ready for production. It should be done by the end of the weak or start of the next. I'm waiting for solder stencil to get ready. After that we can go on any day with some warning... It takes up to 5 days.

So I will do my best not to fall a sleep so we can setup the bored(you need to learn that anyway) so the one send to form will be just in case we have problems today... We will make this puppy hashing...

BTW look at the pictures send to you how to put on cooling elements... They are put that way to get more air over power modules... But with this fan it doesn't make much of a difference...

af_newbie,

we really appreciated your donation and it helped us a lot to get this developed, you should get whatever discount that was offered to you for making a donation. If you were a part of the group buy from Lucko you get what $5 per chip deducted from your price? If no discount was offered PM me and we can work out a deal to reimburse you for your donation.

now onto the board itself, I have some pictures of it with cooling system on, we certainly wont have any problems with heat dissipation! this fan is a monster.

I think it would be wise to have it incorporated into mining software or firmware to control the fan speed, but thats something to take care of down the line.

I'm not hashing yet because I am having a little problems getting the trimpots right and I dont want to blow any chips - each one has to be set to 1 volt (according to form), so I just have them all turned way down right now, I have to buy a better multi-meter today because the one I have is not cutting it, and I want to get these trim pots set exactly right so I dont blow the chips.

anyways heres the pics, I gotta run to radio shack.









ill be around all day playing with this thing. if you need me send email or Pm I will respond shortly.

-JJ


p.s. after just reading some mail from lucko, I realize now that I did probably kill the first board because I loaded firmware without heatsinks, and thats why I was having problems yesterday, ive got two boards left.

so I think what im going to do is keep testing this board, and send the last untouched but fully assembled board to form along with cooling parts so he can get it working himself, he is much more technically inclined than I am with this type of thing, looks like im going to fedex this morning

that way he can get it working and then give me some better instructions on how to set the thing up, its actually much more complicated than I thought it was, these trimpots have no stopping point so you really dont know when your all the way down or all the way up for that matter, without the correct voltage measuring equipment basically all you can do is turn the damn thing counter clock wise 20-30 times and hope you turned it down all the way These boards work for sure and the last thing I want to do is destroy another one.

form, lucko I am on IRC now we need to talk about this, I really need your help to set these up ,I dont want to destroy another good board for no reason
thank god they are just sample chips on here

PM... I would guess there was some misunderstanding...

Lucko,

What is the price for the assembly (for the donors like me)?  JJ, told me $599 with the "discount".  Not sure what that means.

I still would like to assemble 3 boards as I have 48 unsold chips.  But I think I'm getting shafted, being chip donor and all.

Thanks,
af_newbie

No they can't be chained... I'm waiting for JJ to agree on all prices... But I would guess that it would cost about the same as fully assembled with chips since the price in my assembly is the same. It takes just 10 to 20 second less for pick and place machine... So I don't think there will be a difference. But it will be a difference if you will not put chips on right. So I would suggest not to do that. But you can if you would like to do it.

How much is the board with all of components installed and no chips?
Can boards be chained?

It also confusing me why he always forgetting this...

You all need to get the story straight here.  This is what is confusing the shit out of people.