NanoFury Project - Open Source Design

[vs3]

I hope the NF* variants have a unique identifier on USB to make them easy to distinguish on USB.
The nanofury currently nicely identifies itself as:

Code:

iProduct                2 NanoFury NF1 v0.7

Managing variant devices is much easier when some forethought goes into it by the hardware devs thanks

I can send you my code if you like. I modified cgminer 4.2.3 for NFx already.
I get the number of chips by finding "NFx" where x is for chips, i.e. NF1 for 1 chip, NF2 for 2,
NF3 for 3,...NF5 for 5, NF6 for 6...and so on...
The freq of all chips are the same. I have no idea to add an optional to set different freq for each
chip. As the picture I showed that is tested on "NF3".
https://dl.dropboxusercontent.com/u/23682587/DSC_0226.JPG
https://dl.dropboxusercontent.com/u/23682587/DSC_0227.JPG

Later, I will show NF5...

First - CONGRATS! :-) Nicely done!
And that's exactly why those 3 extra test points are there and how they were meant to be used

Oh - and I think that's the first case when I see someone adding a 40mm fan (which was another one of the other design considerations)

Just a word of caution - I see that you're powering the 3rd chip from the same power regulator - be aware that the regulator is rated for 6Amps so if you push the chips a bit too much you may exhaust it. (nothing scary is going to happen - just some serious voltage ripples which will mess with bitfury's internal clock generator and you'll see very low hashing speed and lots of errors)


As for the NFx numbering - that is correct, the "x" was meant to indicate the number of chips. This is all part of the "product string", which is a customizable element.
Luke's bfgminer looks for the text "NanoFury" and that's how it decides to do any further tests with that USB device. There is also an additional test that checks the device (a hardware check). The NFx part is not mandatory for bfgminer, but I've been adding it anyways for consistency.

I've been using the https://github.com/nanofury/NanoFury_Init program for setting the "product string".

The product string can be up to 25 characters long, and I've been adding manufacturer-specific identifiers after the NFx part (e.g. "NanoFury NF2 v0.5 by MLM").

I don't know if any of those strings have been used by cgminer and how it detects the devices (although I could probably take a look at the code and figure it out).

There is one addition that I've asked Luke to add but it is not there yet - I want to add a "default speed" indicator, which will be in the form of "/N" after the NFx part - e.g.
"NanoFury NF2/t" where the letter will represent the default speed, calculated by the formula 30+modifier, where upper-case letters are negative (@=0, A=-1, B=-2, etc) and lowercase are positive (a=+1, b=+2, c=+3, etc), and in this case t=20 so that's default speed 30+20=50 bits.
And the speed designator can be one for all chips, or one for each chip - e.g. if I have 2 chips and if the first should default to 48 bits and the second to 50 the example would be "NanoFury NF2/rt" ("r" is 18+30=48 bits for the first chip, "t" is 20+30=50 bits for the second).
I realize that this has a limitation and will work for individual settings of up to 12 chips, but for anything with that many chips it may just be sufficient to have one default speed for all chips.

Also, there will be one more hardware change specific to the NF6 design - I need an inverse PWRON signal (where ON=0 and OFF=1 or input state) and I need the signal inverted so that I don't have to put an extra transistor (and the signal that is currently in the code is not inverse = ON=1 and OFF=0).
Currently there is no power control and all chips start working and consuming power the moment you plug them in the USB

And that will probably be the only notable difference in v0.2 of the board (plus moving some stuff around).


@simoncc - I know I'm repeating what's already said but could you please share your multi-chip version of the code?
I'll be messing with Luke's code tomorrow and want to try cgminer as well

[simoncc]

I hope the NF* variants have a unique identifier on USB to make them easy to distinguish on USB.
The nanofury currently nicely identifies itself as:

Code:

iProduct                2 NanoFury NF1 v0.7

Managing variant devices is much easier when some forethought goes into it by the hardware devs thanks

I can send you my code if you like. I modified cgminer 4.2.3 for NFx already.
I get the number of chips by finding "NFx" where x is for chips, i.e. NF1 for 1 chip, NF2 for 2,
NF3 for 3,...NF5 for 5, NF6 for 6...and so on...
The freq of all chips are the same. I have no idea to add an optional to set different freq for each
chip. As the picture I showed that is tested on "NF3".
https://dl.dropboxusercontent.com/u/23682587/DSC_0226.JPG
https://dl.dropboxusercontent.com/u/23682587/DSC_0227.JPG

Later, I will show NF5...

First - CONGRATS! :-) Nicely done!
And that's exactly why those 3 extra test points are there and how they were meant to be used

Oh - and I think that's the first case when I see someone adding a 40mm fan (which was another one of the other design considerations)

Just a word of caution - I see that you're powering the 3rd chip from the same power regulator - be aware that the regulator is rated for 6Amps so if you push the chips a bit too much you may exhaust it. (nothing scary is going to happen - just some serious voltage ripples which will mess with bitfury's internal clock generator and you'll see very low hashing speed and lots of errors)


As for the NFx numbering - that is correct, the "x" was meant to indicate the number of chips. This is all part of the "product string", which is a customizable element.
Luke's bfgminer looks for the text "NanoFury" and that's how it decides to do any further tests with that USB device. There is also an additional test that checks the device (a hardware check). The NFx part is not mandatory for bfgminer, but I've been adding it anyways for consistency.

I've been using the https://github.com/nanofury/NanoFury_Init program for setting the "product string".

The product string can be up to 25 characters long, and I've been adding manufacturer-specific identifiers after the NFx part (e.g. "NanoFury NF2 v0.5 by MLM").

I don't know if any of those strings have been used by cgminer and how it detects the devices (although I could probably take a look at the code and figure it out).

There is one addition that I've asked Luke to add but it is not there yet - I want to add a "default speed" indicator, which will be in the form of "/N" after the NFx part - e.g.
"NanoFury NF2/t" where the letter will represent the default speed, calculated by the formula 30+modifier, where upper-case letters are negative (@=0, A=-1, B=-2, etc) and lowercase are positive (a=+1, b=+2, c=+3, etc), and in this case t=20 so that's default speed 30+20=50 bits.
And the speed designator can be one for all chips, or one for each chip - e.g. if I have 2 chips and if the first should default to 48 bits and the second to 50 the example would be "NanoFury NF2/rt" ("r" is 18+30=48 bits for the first chip, "t" is 20+30=50 bits for the second).
I realize that this has a limitation and will work for individual settings of up to 12 chips, but for anything with that many chips it may just be sufficient to have one default speed for all chips.

Also, there will be one more hardware change specific to the NF6 design - I need an inverse PWRON signal (where ON=0 and OFF=1 or input state) and I need the signal inverted so that I don't have to put an extra transistor (and the signal that is currently in the code is not inverse = ON=1 and OFF=0).
Currently there is no power control and all chips start working and consuming power the moment you plug them in the USB

And that will probably be the only notable difference in v0.2 of the board (plus moving some stuff around).


@simoncc - I know I'm repeating what's already said but could you please share your multi-chip version of the code?
I'll be messing with Luke's code tomorrow and want to try cgminer as well

Hi vs3,

1. About freq issue, your ideal is good. I will modify my code to solve this issue.
2. After modifying and finishing test simply, I put my code into public.
3. Regulator on NF2 seems supporting 3 chips and working well. I had made them working for +10hrs last week, and found the heat-sink was very hot. That's why I put a 5V 4-cm FAN on the heat-sink. After that, it's quite cool now.

simoncc

[adib]

Download this tool to modify the product string of MCP2210 :  http://ww1.microchip.com/downloads/en/DeviceDoc/MCP2210Utility_v1.2.1.zip
Read, modify and configure... quite easy. If you use zadig to access MCP2210, you have to remove/uninstall it and change to windows default HID driver such that microchip can access MCP2210 correctly. I will put my source to github here tonight, 'cause that is at home, not in office.
[/quote]

I thought cgminer was suppose to work with the zadig drivers... excuse my ignorange im not very good at software.
As Taungeran said earlier in the thread, mcp2210 is limited by the hid drivers as they are designed for relatively low output of information, limiting the NFx to about 8 chips... as i found out.

[Zeta0S]

Is this the final results?
Dual Chip?

https://www.youtube.com/watch?v=BxyzR_2IYG8

[adib]

Thats BIFURY, same design but communication is done through another chip and through firmware.
NFx design comunicates with the chip aproximately directly...just an usb-spi converter.

NanoFury and RedFury..icefury...etc are not the same.

[vs3]

some more observations with a slightly better hub:

bfgminer version 3.10.0 - Started: [2014-04-07 02:56:32] - [  0 days 00:16:58]
Manage devices Pool management Settings Display options
Connected to stratum.mining.eligius.st diff 32 with stratum as user
Block: ...6f069900 #294632  Diff:6.12G (43.81Ph/s)  Started: [03:01:58]
ST:23  F:0  NB:4  AS:0  BW:[ 58/ 52 B/s]  E:162.41  I:  130uBTC/hr  BS:4.28k
7/21         | 41.73/41.54/43.68Gh/s | A:338 R:3+0(.71%) HW:208/2.0%
-------------------------------------------------------------------------------------------
NFY 0:       |  2.51/ 2.52/ 2.43Gh/s | A: 18 R:1+0(3.3%) HW:  3/.50%
NFY 1:       |  4.50/ 4.50/ 4.48Gh/s | A: 36 R:0+0(none) HW: 12/1.1%
NFY 2:       |  4.26/ 4.26/ 4.40Gh/s | A: 22 R:0+0(none) HW: 10/.95%
NFY 3:       |  4.24/ 4.25/ 4.18Gh/s | A: 43 R:0+0(none) HW:  2/.20%
NFY 4:       |  4.50/ 4.50/ 4.40Gh/s | A: 28 R:0+0(none) HW:  9/.86%
NFY 5:       | 11.12/11.24/12.13Gh/s | A:105 R:2+0(1.7%) HW:101/3.3%
NFY 6:       | 10.26/10.27/11.72Gh/s | A: 88 R:0+0(none) HW: 71/2.5%
-------------------------------------------------------------------------------------------
[2014-04-07 03:14:45] Accepted 02780fff NFY 6e Diff 103/32
[2014-04-07 03:14:48] Accepted 070813ac NFY 5f Diff 36/32
[2014-04-07 03:14:50] Accepted 0605a0b3 NFY 6a Diff 42/32
[2014-04-07 03:14:58] Accepted 019b50b7 NFY 0  Diff 159/32

In another test I tried two different heatsinks - one from the NF2 series (very high performing - probably the best I've seen so far) and another generic one. The generic one had about 8-10C higher temperature, and when pushed very hard (and no fan) the board got to 105-108C and I started seeing a lot of hardware errors.

I've relocated it vertically - again no fan, and that seems to be fine - temp is 76C as of this moment while it does:
 NFY 0:       | 10.37/10.22/10.89Gh/s | A:87 R:0+0(none) HW:75/5.0%

which looks like this:

[simoncc]

I hope the NF* variants have a unique identifier on USB to make them easy to distinguish on USB.
The nanofury currently nicely identifies itself as:

Code:

iProduct                2 NanoFury NF1 v0.7

Managing variant devices is much easier when some forethought goes into it by the hardware devs thanks

I can send you my code if you like. I modified cgminer 4.2.3 for NFx already.
I get the number of chips by finding "NFx" where x is for chips, i.e. NF1 for 1 chip, NF2 for 2,
NF3 for 3,...NF5 for 5, NF6 for 6...and so on...
The freq of all chips are the same. I have no idea to add an optional to set different freq for each
chip. As the picture I showed that is tested on "NF3".
https://dl.dropboxusercontent.com/u/23682587/DSC_0226.JPG
https://dl.dropboxusercontent.com/u/23682587/DSC_0227.JPG

Later, I will show NF5...

First - CONGRATS! :-) Nicely done!
And that's exactly why those 3 extra test points are there and how they were meant to be used

Oh - and I think that's the first case when I see someone adding a 40mm fan (which was another one of the other design considerations)

Just a word of caution - I see that you're powering the 3rd chip from the same power regulator - be aware that the regulator is rated for 6Amps so if you push the chips a bit too much you may exhaust it. (nothing scary is going to happen - just some serious voltage ripples which will mess with bitfury's internal clock generator and you'll see very low hashing speed and lots of errors)


As for the NFx numbering - that is correct, the "x" was meant to indicate the number of chips. This is all part of the "product string", which is a customizable element.
Luke's bfgminer looks for the text "NanoFury" and that's how it decides to do any further tests with that USB device. There is also an additional test that checks the device (a hardware check). The NFx part is not mandatory for bfgminer, but I've been adding it anyways for consistency.

I've been using the https://github.com/nanofury/NanoFury_Init program for setting the "product string".

The product string can be up to 25 characters long, and I've been adding manufacturer-specific identifiers after the NFx part (e.g. "NanoFury NF2 v0.5 by MLM").

I don't know if any of those strings have been used by cgminer and how it detects the devices (although I could probably take a look at the code and figure it out).

There is one addition that I've asked Luke to add but it is not there yet - I want to add a "default speed" indicator, which will be in the form of "/N" after the NFx part - e.g.
"NanoFury NF2/t" where the letter will represent the default speed, calculated by the formula 30+modifier, where upper-case letters are negative (@=0, A=-1, B=-2, etc) and lowercase are positive (a=+1, b=+2, c=+3, etc), and in this case t=20 so that's default speed 30+20=50 bits.
And the speed designator can be one for all chips, or one for each chip - e.g. if I have 2 chips and if the first should default to 48 bits and the second to 50 the example would be "NanoFury NF2/rt" ("r" is 18+30=48 bits for the first chip, "t" is 20+30=50 bits for the second).
I realize that this has a limitation and will work for individual settings of up to 12 chips, but for anything with that many chips it may just be sufficient to have one default speed for all chips.

Also, there will be one more hardware change specific to the NF6 design - I need an inverse PWRON signal (where ON=0 and OFF=1 or input state) and I need the signal inverted so that I don't have to put an extra transistor (and the signal that is currently in the code is not inverse = ON=1 and OFF=0).
Currently there is no power control and all chips start working and consuming power the moment you plug them in the USB

And that will probably be the only notable difference in v0.2 of the board (plus moving some stuff around).


@simoncc - I know I'm repeating what's already said but could you please share your multi-chip version of the code?
I'll be messing with Luke's code tomorrow and want to try cgminer as well

I put my code at this : https://github.com/capricorner/cgminer
About the number of chips and frequency in MCP2210 product string, that was added already. For example, "NanoFury NF2/E v0.5 by MLM"
stands for 2 chips by freq 54. the token in the product string is "NFx/y" or "NFx-y", x is for the number of chips ( 1~99 ), and y is for the number
of frequency ( A ~ F ==> A for 50, B for 51... E for 54, and F for 55 ).
I believe there must be some issues in the code and I will try to fix them ( I found several tonight...)

[Zeta0S]

I hot 6 of that new bitfury miners. Can overclock them up to 6gh/s whit fan and 5gh/s no fan.
Can u guys get more hashing speed out of them? Is this the same chip as the nanofury?

Video of my bitfury miner:
https://www.youtube.com/watch?v=stbhrGpfq1w

[TheRealSteve]

Can u guys get more hashing speed out of them? Is this the same chip as the nanofury?

The chip is the same, but the board is not (there is a dual chip NanoFury, the NanoFury II as well, you can compare the two)
For the device in your video, try this thread instead:
[SUPPORT THREAD] BFx2 Bitfury USB stick miner

[vs3]

Last two panels and a stencil left! $200 shipped!
(I don't have enough from all components though - so no components, even though I'll probably throw in a few of the more important ones)

DYI kits on the way! :-)



I have only 4 panels and one stencil left!
(I wish I knew that DYIs would be so popular and ordered some more panels...)

Several people have asked me for a NF2 DYI kit.
So I ordered some extra boards - they're 8 boards per panel and should be arriving at some point later this week or early next.

I have a few extra panels left. I also have an extra steel laser cut stencil.

Pricing will be as follows:
----------------------------
8-board panel: $80
Stencil : $70
Components pack (all components for 8 boards excluding bitfury chip) : $100
----------------------------
Ground shipping within the US will be free. PM me for international.

Note: heatsink and thermal tape are not included.

TO PLACE AN ORDER:
PM me with what you want and I'll send you a quote with the final amount and payment instructions.
Please put DYI KIT in the message subject line.
For the quote I'll need a shipping address, and if you prefer UPS/FEDEX also a phone number.
Payment options (in the order of preference) will include: BTC (coinbase rate), check, paypal.

I anticipate having everything in my hands and ready for shipping at some time next week. Based on the interest I'll be ordering the missing components on Thursday (with expected delivery on Monday).

Below are images of the panel and actual boards with stencils. Boards on the panel have been specifically laid out for an easy automated SMT machine assembly.


edit: added clarification regarding heatsink and thermal tape

[adib]

Has anyone got Cgminer to work with more than one chip?
I cant figure out how to make the code simoncc provided.
If someone could post some instructions... i would be very happy

[DrZeck]

Has anyone got Cgminer to work with more than one chip?
I cant figure out how to make the code simoncc provided.
If someone could post some instructions... i would be very happy

You can download cgminer from official Con's git, then replace all files related to mcp2210 and bitfury from simoncc git then compile with --enable-bitfury.

I think that should work ... 

[vs3]

Update: ALL DYI KITS HAVE BEEN SOLD.

Last two panels and a stencil left! $200 shipped!
(I don't have enough from all components though - so no components, even though I'll probably throw in a few of the more important ones)

DYI kits on the way!

[simoncc]

One day someone there asked me one question "How to chain NF2s together and access them by one USB interface?"

I had a look at my NF2 board, and found out how to do that. Let me show you the picture:


WARNING HERE:  I HAVE TO SAY ONE THING THAT I DO NOT REALIZE NF2 CHAIN HERE BECAUSE I HAVE ONLY ONE
BF2. SO YOU CAN TAKE THIS FOR REFERENCE ONLY. NO GUARANTEE AGAINST DESTROYING DEVICES...https://dl.dropboxusercontent.com/u/23682587/nf2-chain.jpg

1. unsolder the parts in orange rectanges on 2nd NF2. ( 1st NF2 does not need any modified, only 4 flying wires later...)
2. there are 4 red circle on 1st NF2, and there are 4 blue circle on 2nd NF2. Connect red 1 to blue 1, R2 to B2, R3 to B3,
    and R4 to B4 by flying wires. Note: make sure wires 1 2 3 are the same length.
3. if you want to chain the 3rd NF2, unsolder the parts on 3rd NF2, and connect Red circles on 2nd NF2 to Blue circles on 3rd NF2,
   and so on...
4. if you plug these chained NF2 to the same USB hub, GND plan should be the same. No GND wire should jump each other.
    Yellow circle in the picture is GND point I use.

AGAIN HERE:
WARNING HERE:  I HAVE TO SAY ONE THING THAT I DO NOT REALIZE NF2 CHAIN HERE BECAUSE I HAVE ONLY ONE
BF2. SO YOU CAN TAKE THIS FOR REFERENCE ONLY. NO GUARANTEE AGAINST DESTROYING DEVICES...


You do not populate the parts in oragne rectangles on the rest 7 NF2 boards if you
have DIY kit from vs3... and you can save a little bit of money by this.

[vs3]

First: simoncc - nicely done! :-) And thanks for sharing!



Onto the NF6 story - v0.2 of the design is now confirmed with both gen1 and gen2 chips!


Here is how things look like with Gen1 chips:

Code:

 bfgminer version 3.99.0 - Started: 2014-04-15 02:02:21 -   0 days 00:18:07
 Manage devices Pool management Settings Display options
 Pool 0: ...ning.eligius.st  Diff:32  +Strtm  LU:02:21:55
 Block: ...997c24bf #295953  Diff:6.12G (43.81Ph/s)  Started: 02:04:50
 ST:14  F:0  NB:1  AS:0  BW:365/ 20 B/s  E:25.28  I:70.11uBTC/hr  BS:5.15k
 2/12          | 20.12/19.31/18.80Gh/s | A:127 R:0+0(none) HW:74/1.5%
----------------------------------------------------------------------------------------------
 NFY 0a:       |  1.75/ 1.69/ 1.67Gh/s | A:  9 R:0+0(none) HW: 7/1.6%
 NFY 0b:       |  1.76/ 1.71/ 1.80Gh/s | A: 20 R:0+0(none) HW: 3/.65%
 NFY 0c:       |  1.71/ 1.66/ 1.66Gh/s | A: 10 R:0+0(none) HW: 6/1.4%
 NFY 0d:       |  1.70/ 1.67/ 1.55Gh/s | A:  7 R:0+0(none) HW:11/2.7%
 NFY 0e:       |  1.69/ 1.64/ 1.61Gh/s | A: 11 R:0+0(none) HW: 6/1.5%
 NFY 0f:       |  1.72/ 1.69/ 1.63Gh/s | A: 11 R:0+0(none) HW: 6/1.4%
 NFY 1a:       |  1.63/ 1.59/ 1.40Gh/s | A:  8 R:0+0(none) HW:11/3.0%
 NFY 1b:       |  1.58/ 1.54/ 1.49Gh/s | A: 10 R:0+0(none) HW: 3/.79%
 NFY 1c:       |  1.61/ 1.57/ 1.46Gh/s | A:  9 R:0+0(none) HW: 1/.27%
 NFY 1d:       |  1.57/ 1.53/ 1.62Gh/s | A: 13 R:0+0(none) HW: 3/.73%
 NFY 1e:       |  1.50/ 1.46/ 1.35Gh/s | A: 13 R:0+0(none) HW:10/2.8%
 NFY 1f:       |  1.62/ 1.57/ 1.56Gh/s | A:  6 R:0+0(none) HW: 7/1.7%
----------------------------------------------------------------------------------------------

 2014-04-15 02:22:14 Accepted 07af1852 NFY 0b Diff 33/32
 2014-04-15 02:22:20 Accepted 016516bb NFY 0a Diff 183/32

This is with a crappy power supply (4.6V) and 54 bits, somewhat inferior heatsinks, a tiny bit of airflow, temp 46C and 65C (on the second one heatsink faces downward and that's chips temperature). I haven't checked exact current usage but from indirect measurements it is around 1.3A (which is a bit higher due to the lower voltage, or around 6W total)

Or to summarize it - with Gen1 chips 9GH would be around the norm, and over 10GH with a bit better hub (that provides actually 5V). On a slightly over-voltaged hub (5.25V) it should be able to hit even 12GH and above.


And the screenshot below is with a mix of chips: one NF6 with Gen1 chips and one with Gen2 chips:

Code:

bfgminer version 3.99.0 - Started: 2014-04-15 02:29:03 -   0 days 00:22:28
Manage devices Pool management Settings Display options
Pool 0: ...ning.eligius.st  Diff:16  +Strtm  LU:02:56:07
Block: ...16e4330e #295957  Diff:6.12G (43.81Ph/s)  Started: 02:38:44
ST:14  F:0  NB:2  AS:0  BW:264/ 50 B/s  E:34.41  I:77.92uBTC/hr  BS:2.94k
2/12         | 20.77/20.84/21.59Gh/s | A:439 R:5+0(1.1%) HW:83/1.2%
-------------------------------------------------------------------------------------------------------
NFY 0:       | 10.58/10.68/10.11Gh/s | A:225 R:5+0(2.2%) HW:35/1.1%
NFY 1:       | 10.16/10.17/11.54Gh/s | A:219 R:0+0(none) HW:48/1.3%
-------------------------------------------------------------------------------------------------------
2014-04-15 02:54:21 Accepted 0250a55d NFY 1a Diff 110/16
2014-04-15 02:54:23 Accepted 03a6b93f NFY 1a Diff 70/16
2014-04-15 02:54:27 Accepted 0a40d969 NFY 1b Diff 24/16
2014-04-15 02:54:27 Accepted 0f624add NFY 1e Diff 16/16
2014-04-15 02:54:30 Accepted 088b8b78 NFY 0c Diff 29/16
2014-04-15 02:54:36 Accepted 07b59471 NFY 1a Diff 33/16
2014-04-15 02:54:41 Accepted 0f83ad2c NFY 1e Diff 16/16
2014-04-15 02:54:51 Accepted 07e8cc88 NFY 1e Diff 32/16

Same hub, power usage a bit higher (and voltage down to 4.3V), speed 54 bits for both, a bit better heatsinks (the Gen2 with the best I have around), temperatures: 50C on the gen2 and 73 on the gen1 (which is also facing down so not the most efficient position anyways).

Judging by the low error rate I can probably squeeze out another 0.5-1GH out of each.

And I'm actually rather surprised that I see a stable 11.5+GH with such a low voltage on the Gen2 chips. On a better hub it will certainly exceed 12GH and with a slightly overvoltaged one (5.25V) it will very likely exceed even 15-18GH.

The Gen1 chips in this screenshot are from the older series with datecode 1337.

There are also differences between the various production dates in the Gen1 chips - the first 2 boards have chips with date code 1404 (which were from the last series - I got a few of them forwarded from bitcoinvalet's last group buy) and I am a bit surprised that they perform slightly worse than the older chips that I had around with datecode 1337 (which were leftovers from the NF1 series).

Bitfury Gen2 chips were from the first samples available for purchase from punin (datecode KCTMS):
.

[intron]

Nice, well done:-)

[vs3]

Nice, well done:-)

Thank you, Sir! :-)

And to be honest from the few people that deserve a lot of credit for NF6' existence there are two that stand out:
you - for proving that it is actually doable, and
KNK - for confirming the minute detail that makes it all possible! (which also saved me a dozen chips:)

So - Thank You!

Should our paths meet - drinks are on me!

[intron]

Nice, well done:-)

Thank you, Sir! :-)

And to be honest from the few people that deserve a lot of credit for NF6' existence there are two that stand out:
you - for proving that it is actually doable, and
KNK - for confirming the minute detail that makes it all possible! (which also saved me a dozen chips:)

So - Thank You!

Should our paths meet - drinks are on me!

Praise should go to Valery (bitfury), he
came up with this brilliant 'string' idea:-)

[vs3]

And while we set up the NF6 production line and clarify all details ... I got bored and tweaked the board a bit (removing a few extra resistors and some other minor optimizations) .. and figured - it would be interesting to see how did it evolve :-)

So - here it is - version 1 vs version 2:


then, version 2 vs 3:


or all changes between 1 and 3:

[Cryptixuk]

I take it that you would do a diy kit for the NF6?