Bitcoin Forum

https://i.imgur.com/jFUuY.png

https://i.imgur.com/jiRyf.png

Directly from BFL.

Looks like it should do the job :)

Very nice!  I hope the ones I get will be in this new rev 3 batch.  Perhaps it won't need the fan on the bottom anymore, then.

Looks like a nice update, with the heat pipe. I like it!

Nice :)

If you ordered in the last week or two, I believe you'll be getting rev 3.  I ordered 30 units myself.

A bit strange that heatsink fins are facing different ways...

Agreed. I don't see any direct path for good airflow.

Looks like they have mounting holes.  I wonder if they would sell me a "naked" boards.  No power supply, no case, no heatsink, no fans.

Perv.

Not worried about the reward halving, huh?

They'll pay for themselves before that happens.

Pardon the newb question (this is the newb forum after all :D ) but where can I get my claws on a few of these, and how much?  I've been drooling on the X6500R3 but seems they're not yet for sale anywhere.

Anyone know what CAD software they rendered that board in? It looks rather primitive. Makes me curious what tool they are using for electrical design.

Solidworks?

What do you expect? It's not supposed to look like it's rendered in Maya!

I´m also tempted to place another order, but I first want to receive something before ordering more.

According to the bitcoin profitability calculator on bitcoinx.com it will take 280 days for one single to pay for itself ($600).  (Assuming a modest $0.11/KHW).

So, since the block reward is going to halve in December, I'd say you're cutting it pretty close for them to pay for themselves by the time the reward halves.  Plus you just ordered them and we know they're having trouble processing all the orders so who knows how long it will be till you actually have all 30 of them in your hands. 

So, in conclusion, if it takes them a month or more to ship them, then they actually won't pay for themselves by the time the reward halves.  Obviously not really a big deal since as you pointed out they will be mostly paid off by then. But still something to think about, especially for people who haven't placed orders yet like myself.  If I order today, it will likely be a month or more till I get the items.

I expect delivery in about a month, yes, that leaves ~8 months... and power cost is 0, since they will most likely be housed in my DC.  But even factoring in power, depending on what BTC prices do, it may be paid off sooner (or later if the fall is dramatic)  I based my calculations on $4/btc.

Hmm I see.  Is there a limit to how many you can put on 1 PC or are you planning to do all 30 on 1 PC with a bunch of USB hubs?

127 per USB port.   I imagine you'd run out of power points before your PC ran out of USB ports :)

This is the number from the usb-specification, and even if somebody said, it will not require any CPU-Power from the PC and a "healthy" usb-port is enogh, I would not be too sure. Depends all on the amount of data which is transfered and on the latencies. I really don´t know what data is transferred, but if that will turn out as a problem, you could use an usb3-host-port on your PC, together with a huge usb3-hub. That will give you some air with all the usb2-devices. But with just 10 singles you will not encounter any problem. Just my 2 BTC-cent on this...

Anyone have a general idea on how long the wait is, currently, between ordering and shipping?

Go here and draw your own conclusions:

https://bitcointalk.org/index.php?topic=67887.msg791080#msg791080 (https://bitcointalk.org/index.php?topic=67887.msg791080#msg791080)

No one can tell you what the current wait is, not even BFL. People that have ordered in mid-December have started getting theirs (some have, some haven't). People who ordered in early January still haven't received anything. It isn't yet clear whether BFL is now catching up, keeping up, or getting further behind. As for how long will people who order now have to wait, your guess is as good as mine.

Probably on the slow boat from china now.

Ordering now is probably after rev 3, unless BFL steps us their production size big time. Which they certainly should, considering that most people's reservations in ordering the singles is how long it would take until it gets into their hands.

I have had my orders in and it is looking like I may be getting the rev3s??  yeah!!     ;)8)

BFL really does need to get themselves together though and get me my unitz.  Can't wait for the end of the month.......

I'm surprised they are still keeping their 4-6 week ship guarantee. I wonder what they're backing it with...

If they don't meet the 4-6 weeks you can cancel and get your money back ... guaranteed.  There is nothing else "backing" it.  Not a single person has received units in <6 weeks.  Some early January orders just shipped.

Awesome. Seems like they're making some good progress. I hope this indicates confidence in their design such that it will accelerate the order fulfillment.

I'm not seeing where they are claiming 4 - 6 weeks on the website anymore.  Beyond that, I don't recall them ever guaranteeing 4 - 6 weeks, just an estimate... though I may be wrong about that.

I believe the 4-6 week statement was on the order form no?

Actually I don't see it anywhere now. Hmmm

You can see it on the order page for the Singles. They use the word 'guaranteed'.

http://www.butterflylabs.com/pre-order-form/ (http://www.butterflylabs.com/pre-order-form/)

Look at the 'Order Terms' section on the right side of the page, above the photo of the old Rev 1 Single PCB they are no longer producing.

When you get the confirmation page after placing an order (through Paypal, not sure if other methods get the same page) there is a block of text:

Ah got it, I missed that... was scanning for 4 - 6 instead of four to six heh... oops!

Your not alone Inaba, same here.  Doh!    :-[

I assume this is a special processing unit built for Bitcoin mining? You just plug it into a USB port and you are good to go?

???
I suggest you do some more reading before posting.

Maybe newbies jail should be extended from 5 to 10 posts...

Actually, he is quite correct. It is a FPGA(google that term), built specifically for SHA256-hashing, or Bitcoin mining in this case.

Of course he's "quite correct".

Are we in the mode of stating the obvious now?!
That's _much_ more annoying than simply stating "subscribe".

You realize that this is the newbie board right? I reserve the right to act like an idiot.

Then you haven't see the worst questions yet IMHO. He is quite reasonable in his questions already, given that this is the newbies' thread.

 ;D ;D ;D
You just brightened up my day right now, seriously.
Yes, you totally have the full right to ask your question here.
Absolutely right. I should have taken a look at the top of the page and notice the board.
I'll go and get a cup of coffee now, seems I need it.

John's right too, it could be worse...

Bitcoins are not actually coins.

Oh and subscribed.

Has there been any change to the power supply unit in Rev3 singles?  Has anyone come up with a way to power multiple units vs using the included PSU for each single?  If you had 10, 20 or 30+ units, you would need to have 10, 20 or 30+ outlets.

Thanks in advance,

Tossil

Power bars are cheap. No, the PSU of the Rev 3 Singles is identical to the Rev 2.

Alternately you could use a standard ATX power supply. Each of its 12V lines could power one Single. For example, you could take a PSU that had 4 6-pin (or 8-pin) PCIe connectors and use it to power 12 Singles (each Single takes 12V @ 6A, or 72W. Each PCIe connector has 3 12V lines).

A good 850W or higher supply would work well. You'd have to wire up the barrel connectors (5.5mm x 2.5mm x 13mm) and ground the 'power good' line to get the PSU to turn on without a motherboard, but other than that you'd be good to go. Use at least 16ga wire (not 18ga) to reduce power loss in the cable runs; 6A is a lot of current to run in small wires for more than a few feet.

Do you think that this table here is bogus then?

http://www.rbeelectronics.com/wtable.htm

They seem to say that for 12 amps on 12v circuit you could do 18AWG up to 10'.  That just doesn't seem right.   ???

That table isn't bogus, but is oversimplified. One thing to keep in mind is that the 'length' is actually the return trip length; an electrical circuit is always a loop round-trip, never one-way. If you want a 6' cable going from your PC to your Single, its 'length' is actually 12' (6' to the Single, 6' back to the PC). The other thing to keep in mind is that cable has a resistance, and that resistance depends on (a) the wire gauge and (b) the wire length.

So for that 12' run of 18ga wire carrying 6A, you will lose 0.95V at the load end. That gives you 11V at the Single, not 12V. That's fine in and of itself, but it represents a power loss of 0.95V x 6A or almost 6W just in the wire.

For fun let's take the same run but use 16ga wire instead of 18ga. For 16ga wire, the voltage drop will be only 0.6V giving you a loss of 0.6V x 6A or only 3.6W. It is still significant, but clearly much better than the 18ga case.

So if you want to lose ~6W per Single by using 18ga wire (assuming a 6' cable), no one is stopping you. Use the handy calculator at the bottom of the following link if you want to play with various scenarios:

http://www.powerstream.com/Wire_Size.htm (http://www.powerstream.com/Wire_Size.htm)

I think the thing to take away from all this is (a) use the heaviest gauge wire that's practical (I'd say 16ga is the smallest you want to go, regardless of how short your run is), and (b) make the run as short as possible (i.e. if your Single is going to be 3' from your PC, don't make a 6' cable for it).

That is good to know.  I had been to the powerstream site before but the max amps for chassis wiring vs. power transmission listings were a little confusing.

This also brought a question to mind.  The wall warts that are supplied with said units really are not giving the BFL a solid 12v then are they?  I understand that these devices have a range of input voltage that it can take but I just assumed that the supplied materials were much more suited to the power design.  The wall warts are usually around 5ft and maybe 18AWG so the BFL really doesn't see a 12v load during normal usage correct?  How will this affect the power transmission wire releasing all this wasted wattage as heat?

I guess I am just trying to understand the details of DC power transmission.  I'll send you a tip Epoch,  Thanks.

Thanks for generous tip, Cablez. Completely unnecessary, but certainly appreciated.

Regarding the BFL wall warts: I don't have one to dissect, but I've seen detailed photos. The 12V line from the power brick is no more than 2.5'-3' long, and the cable itself looks pretty beefy ... I'd guess it is heavier than 18 gauge. So there probably isn't much loss in that short cable.

BFL has stated elsewhere that the Singles themselves take 70W DC but, due to the inefficiency of the supplied wall wart, it pulls 85W at the wall. That's an efficiency of 82%. If you were making your own multi-drop supply, the ATX 80+ Gold has efficiencies approaching 90%. So at most you'd be saving 8% power, or 7-8W per Single. I'm not sure the expense of a $100 ATX power supply is worth it. Say it will power 12 Singles (740W DC, 825W at the wall). Your power savings over 12 stock wall warts would be about 12x8W or 96W. Daily that's 24x96=2.3kWh. Assuming the North American average of $0.11/kWh, that's $0.25/day. Or $92 savings per year.

So it seems that in the long term (1+ years) it could pay off. But that's assuming you are powering 12 Singles. If you are powering less than that, your savings would be proportionally less and your ATX PSU would take longer to pay back. The ATX PSU does have the advantage of having less clutter (you only need 1 power plug total, not 1 per Single).

Not to mention that most miners probably have 1 or 2 spare or underutilized PSUs anyway, if they are looking at FPGAs, so that cost most likely becomes moot.  :D

For me, when I see setups like xiangfu's Icarus haven, I look at all the cables and go batty.  :-\    I would much rather utilize my ATX PSUs and keep it as clean as I can.

Also, don't fret about the tip,  Bitcoin is meant to be spent.  ;D

I will be running my miners on solar. The plan is to run them pure DC, not dc to inveter to psu back to DC.

Hmmm, my calculation on $4 per BTC and 0 electricity puts it at 282 days ROI?

How are you doing yours? Maybe I oversimplified mine?
1. Network Ghash/s    11242
2. value per coin            $4.00
3. coins per reward    50
4. rewards / day            144
5. reward / Mhash/s    0.00256 ((4*3*2)/(1*1000))
6. elect rate $ / kWh    $0.00
7. rate                        832 Mhash
------------------------------------
 $2.13 per day = 282 days which also assumes the network rate does not increase. I am really thinking I should revise all my projections now with all the FPGAs coming online.

I'm not questioning your numbers, just checking that my reasoning is not drastically off somehow and trying to become enlightened before my CFO (wife) takes a look at my business plan (crazy idea to buy a minirig or 2).

I have ordered one Single for myself to UK, but with the mini rig now out and using second generations chips, I am considering canceling the order and getting the mini rig instead.