So it looks like we've gotten some sort of approval from four of the eight testers. Still waiting to hear anything from MrTeal, CanaryInTheMine, vs3 and ManeBjorn. I'd like to have approval from at least two of them before opening a sales queue. I've gotten word from vs3 and ManeBjorn that sticks have been received, so hopefully something pops up in the next few days. I'm tentatively looking to open sales next Wednesday with at least 600 sticks (more if I hear back from Bitmain about a potential second chip buy). As I've done before on limited-batch opening sales, I'll announce several days beforehand the time (and time zone) and orders will be queued on a first-come first-serve basis.
I'll be focusing on North America, I reckon. I'm supposed to send European customers to the German guy, who has received manufacturing data and tells me he'll be running out a tiny batch to test before doing a full run. Anyone in Europe specifically wanting a GekkoScience-made stick for posterity (like, you know, TheRealSteve) I can probably send a couple, but so's I don't cut in on his turf I'd just as soon stick to the Western Hemisphere.
I think I posted my note to the other thread, so just to repeat it here too - Regarding feedback - I've shared my observations and feedback with sidehack already (in PMs, I'll leave it up to him if he wants to repost them here - that's fine by me). Aside from that this thread already covers 99% of the rest, so I'm not sure if there is any benefit of repeating that. In a nutshell - a very nice design, the kind of solid plug-and-play stuff that you would expect for a USB miner, and of course if anyone wants to overclock it (I guess that includes everyone ) - don't forget to use a powered USB hub and cooling. The only other thing worth mentioning again is cgminer. Version 4.9.0 was the only one that worked consistently for me (on Windows - both 7 and 8 and x86 and x64). For some reason the officially published 4.9.1 and 4.9.2 did not recognize the devices. As for EU - if that's the same German guy (selling NF2s on bitshopper.de) I would highly vouch for him!
|
|
|
Chips are already here. I'm ready to start this thing any time, but my stated preconditions regarding posted public approval from reviewers have not yet been met so now we're waiting on other people (namely MrTeal, vs3, CanaryInTheMine and ManeBjorn) to report in. I know at least three of those guys have received their sticks, and I know at least one of them has run the stick and talked to me about it, and one of them told me a week ago he was starting a review and also had other stuff to send me (none of which has been seen). So with the exception of chips in hand, we're no closer to starting the sales and manufacturing process than we were this time last week.
btw over the last week I've shared my thoughts and feecback with sidehack. philipma1957's review thread already covers 99% of the rest and I don't see much of a benefit to repeat the same. Also, I'm not sure if any of my feedback is worth posting on this thread so I'll leave that to sidehack and I don't mind if he wants to repost it here. In a nutshell - a very nicely put device, good design job and I'm sure it will make a lot of "overclockers" and all other bitcoin enthusiasts happy! One more time - good job sidehack!
|
|
|
I'm also hoping more than just Philipma1957 post information about the sticks in the review thread, since there should be at least six people besides myself who have final-version prototypes in hand. The sooner I get feedback on the design, the sooner I can decide whether to tweak further or open up a sales queue.
And I got today a little package too! Feedback coming soon!
|
|
|
Steve - thanks for covering for me! Here are the actual parts (including DigiKey part numbers) that were used on some of the NanoFury NF1 v0.7 boards: # Designators Value Package "count/board" OEM P/N DigiKey P/N 1 C1 C3 C6 C10 C12 C13 14 C15 C22 C23 C24 C25 CF1.8-1 100nF 0603 13 CL10B104KO8NNNC 1276-1005-2-ND 2 C2 CF1 22uF/6.3V 0805 2 C2012X6S0J226M085AC 445-14469-2-ND 3 C4 C5 C11 C21 22uF/4V 0805 4 AMK212AC6226MG-T 587-3264-6-ND 4 C8 C9 10uF/6.3V 0805 2 CL21F106ZPFNNNE 1276-3012-2-ND 5 C7 1.1nF 0603 1 CL10B102KB8SFNC 1276-1916-2-ND 6 R1 13K 0603 1 RC0603JR-0713KL 311-13KGRTR-ND 7 R2 180 0805 1 RC0805FR-07180RL 311-180CRTR-ND 8 R3 2.4K 0603 1 RC0603FR-072K4L 311-2.40KHRTR-ND 9 R4 R6 R8 R9 2K 0603 4 MCR03EZPEJ202 1056-MCR03J202-CHP 10 R5 2K 0805 1 RMCF0805JT2K00 RMCF0805JT2K00TR-ND 11 R7 R11 R12 1K 0603 3 MCR03EZP5J102 1056-MCR03J102-CHP 12 D1 LED 2PLCC 1 LY T67K-J2M1-26-Z 475-2562-6-ND 13 L1 10R@100MHz 0603 1 "CIB10P100NC(or ACML-0603-060)" 1276-6388-6-ND 14 L2 2.2uH 0630 1 NR6028T2R2N 587-2098-6-ND 15 Q1 BSS138L SOT-23 1 BSS138LT3G BSS138LT1GOSDKR-ND 16 U1 BD9C301FJ HTSOP-8 1 BD9C301FJ-E2 846-1117-6-ND 17 U2 LDO-1.8V SOT-23-5 1 LD39015M18R 497-6977-6-ND 18 U3 LDO-3.3V SOT-23 1 MCP1700T-3302E/TT MCP1700T3302ETTDKR-ND 19 U4 BITFURY QFN48 1 20 U5 MCP2210-I/SS SSOP-2 1 MCP2210-I/SS MCP2210-I/SS-ND 21 X1 12MHz 3SMD* 1 CSTCE12M0G55-R0 490-1197-6-ND 22 CN1 USB-A plug USBA 1 1001-011-01101 1175-1008-ND
And as for the X1 crystal - do not change it. It is used for the USB communication (and will not overclock anything). Clock speed on the BitFury chip is determined by the voltage - and the two resistors you mentioned will change that. (exact formula and details are in the previous posts)
|
|
|
So, general testing update.
But the breakout board, with identical circuitry, worked just fine - except its regulator was more distant than and also orthogonal to the ASIC, which was itself under a large aluminum heatsink.
So I yanked an inductor on a moderately-functional board, wired up some pins to drive it off my test schfifty-three board (the one rigged up for the dual-chip testing with the original breakout boards) an inch or two away (on the other side of a ground plane from the ASIC) and BOOM within a second the chip was too hot to touch, current meter was pegged and I saw about 8GH reported on cgminer.
Yeah - I was lucky that I didn't have too hard time with my power design ... And a good catch on your side! Past that, maybe drill out the inductor pads and mount it on the back of the board, ...
If you can make it work without going to double-side components this will save you a lot on production cost (and not to mention testing headaches). But it is what it is and if you have to - then by all means go for it. There is always the option for another improved version later. I'm really hoping I don't have to do another board design, because that'll shift things back another two weeks or so.
Crossing fingers on that! And let me know if you go past version 7 And actually I think v5 is something like the statistical median for me - first NF1 was 0.6 , NF2 was v5, NF6 was the only lucky one with v3.
|
|
|
Yes, behind. I was hoping to be on TypeZero design by now, and I haven't even gotten working Compac prototypes distributed for testing. Just because I'm not racing other independents doesn't mean it's not a race. I'm also racing new chip generations, and the network as a whole, and my own opinions about how long a thing should take to get done.
Nah, you're good - you're still on v0.3 ... it was version 0.5 for me that was the first that sort of worked ... and 0.6 was the first production one I've been there and I know the pressure and you are definitely ahead of schedule! And it is what it is, and sometimes you have to take a break, step back and take another look, or just give it a day or two break and another solution will most likely appear.
|
|
|
I've succeeded in getting two of four sticks hashing unstably and unreliably but I have seen submitted shares. I can't say for sure what the issue is that gets the chip working haphazardly at about 10% capacity, because the breakout boards were pretty much all-or-nothing. Could be signal integrity issues, I really don't know but I hope that's not the case because if it's a result of PCB layout we got problems. The signal traces are between ground planes - at least they should be if the etch fab didn't get layers out of order. Nicely done! btw one of my first surprises was that the USB 5V was rarely exactly 5V ... and depending on the hub - under load it went all the way down to 3.5V. It looks like you're already beyond that with the bolted-down hub, but do double that just in case. Another thing I had to keep in mind was very short traces to the USB connector - some hubs were very picky about noise there. And thanks for the pics - keep that pr0n coming! Did you hand-solder all components? If you did - you are a HERO! I didn't realize how much of a PITA that was until I started using soldering paste and (cheap) stencils. For example - this took me just about 2-3hrs to complete and was significantly less messy and nicer than my other hand-soldered ones.
|
|
|
by the way this is not exactly a "PC" (in terms of what people usually presume - e.g. "IBM PC"-compatible). Think of it as a stripped-down RaspberryPi 1 (without the HDMI port and SD card) and adding the 2.4GH radio module (for WiFi and BT). Nonetheless - a great undertaking! I quite like it! And already pre-ordered the kernel kit! Thanks for sharing this!
|
|
|
Awesome job Steve! Very entertaining and a lot of fun! Can't wait for the next one!
|
|
|
J = NanoFury 6 We will solve it! Impressive! Most impressive! :-)
|
|
|
B = Nanofury II
Well done!
|
|
|
Here is how we stop it.
1. Open source ASIC chip. 2. Open source low cost miner design. 3. Low cost miners in every hand. 4. P2Pool with those miners.
well - for example all you have to do is buy Bitfury's chip design and open-source it. I have already done #2, and plenty of people did #3. #4 will happen by itself.
|
|
|
* BGA is a terrible idea (...) you need a friggin' X-Ray machine to inspect your solder joints. * The kind of chips Bitmain and Avalon are doing already are the best option (...)
I'd add Bitfury in the above list - the simplicity (in terms of communication and especially packaging) made them stand out! Amen to that! Are you listening chip-makers?
|
|
|
just my 2c on the backplanes idea - stick with The KISS Principle In other words - don't go that route. IMHO it makes the whole design more complicated that what it could be, adds a bunch of extra fail points and really what's the gain? Saving $3 from a converter chip? Which you'll pay in extra connectors cost and more expensive boards and assembly. If you really want to do chains - just add the option to use plain cables. Not that many people will go that route - you may as well just provide a few test points on the boards and those willing to experiment can solder the few wires - not a big deal.
|
|
|
so they could as (everyone else so far) have done before price it to what the market will bare not what it best to make ROI for a home miner a small farm.
yup - that's how pricing usually works. It's always the maximum that the market can bear (which sometimes happens to be the maximum considering the competition).
|
|
|
I suspect it would be somewhat unfair for me to participate ... however here is a little hint: there are different versions of the same device from different manufacturers Have fun and best of luck to all participants!
|
|
|
Companies that still produce publicly accessible (and shipping) ASIC miners of .28 or better. Spondoolies; Bitmain.
I would exclude Bitmain as they've said that they currently have no available chips for sale (information as of 4/26).
|
|
|
I don't think we have any NF1s though- if you have any of those (even under-performing or damaged) we'd gladly send a little BTC.
Sure - Just PM me a shipping address!
|
|
|
@phil We do not have any of those dual chip (NF2) miners, although we do have a couple of the similar six-chip NF6.
I think I have a few NF2s left around - they're considered "defective" as the asics underperformed significantly (and I haven't bother to unsolder them). Aside from that they work sufficiently fine for testing and experiments. If you want PM me your address and I can mail you one.
|
|
|
Ahhh.... it's been a while since I've seen my creations And by the way if you have any questions on them or if I could be of any help with reusing parts of that design - let me know. I already know that the voltage regulator will not be a good fit - it's lower limit is 0.8V (although the dual-chip version used a 6A one which might still be usable). Ripples were quite significant too - 30mV (but that might also be within the tolerable limits). It was a very cheap option though (and which is why I used it) And software is the other major issue - currently the design is stuck with this USB-Serial chip as this is what cgminer supports and that's the only documentation on the BM1384 (yeah! seriously!). The specific USB-serial chip is not really bad but an alternative one could easily save another buck or so from the final price. I guess for the moment that's a reasonable compromise - spend a bit more on components and a lot less on software development. Oh - and almost forgot - note that black heatsink! We paid a bit extra for them but they were well worth the money and in terms of performance they exceeded the competition by a large margin! Really good ones! The downside is that they were 40x40mm so the board was redesigned around them and that's why it is a bit bigger. And kudos to novak and sideshack! Great job!
|
|
|
|