Klondike - 16 chip ASIC Open Source Board - Preliminary

[tiros]

I'm going to release the firmware and driver code once it's working. I was thinking of all that stuff to protect it but to be honest that's just is a hassle. I'm going to just trust that people will choose to support the vendors that kick back a small fee. Or even better maybe they'll mine a few hours for me by their own choice. I'll put the mine-to address out there when it's time. I'm getting a lot of help/support from forum members here and I think it's better to respect that by not having locked code at all.

Bravo for the effort and for the spirit. World surely need more people like you are.
24 hrs mining time donations would be fair thing to propose.

[FFMG]

I'm going to release the firmware and driver code once it's working. I was thinking of all that stuff to protect it but to be honest that's just is a hassle. I'm going to just trust that people will choose to support the vendors that kick back a small fee. Or even better maybe they'll mine a few hours for me by their own choice. I'll put the mine-to address out there when it's time. I'm getting a lot of help/support from forum members here and I think it's better to respect that by not having locked code at all.

Thanks for the effort and sharing, if/when I get my miner I will be sure to mine some coins your way!

FFMG

[marto74]

Complete miner
8x K16 stacked on 4 200x100 heatsinks with fans and enclosure

[marto74]

I have a question to BKKcoins
How long can be a wire betwen two daisychain conectors?

[BkkCoins]

I have a question to BKKcoins
How long can be a wire betwen two daisychain conectors?

You shouldn't have problems with up to a meter. I'd suggest keeping them as short as possible to aid in having more devices on the same bus. I've seen reports on the web of using I2C between 25 LED blinky boards running 1 ft. apart but I wouldn't count on that being possible.

I will be making up some "klego ties" that are small pcb connectors that can join boards together. I'll also make a vertical one that joins "upwards" but I'm not sure about a standard height for that yet. The "klego ties" will be usable passively but also have a spot for a repeater chip to extend beyond what one bus can handle.

There are fairly cheap buffer chips as well and NXP has an app note out showing the P82B96 allows using Cat5E twisted pair for 50M even at higher speeds than I use here. So if later there is demand for long I2C lines that could be provided for.

http://www.nxp.com/documents/application_note/AN10658.pdf

[Biomech]

I'm going to release the firmware and driver code once it's working. I was thinking of all that stuff to protect it but to be honest that's just is a hassle. I'm going to just trust that people will choose to support the vendors that kick back a small fee. Or even better maybe they'll mine a few hours for me by their own choice. I'll put the mine-to address out there when it's time. I'm getting a lot of help/support from forum members here and I think it's better to respect that by not having locked code at all.

Bravo for the effort and for the spirit. World surely need more people like you are.
24 hrs mining time donations would be fair thing to propose.

I'm down with that. Even if I go with a different product, as I think this should be rewarded. And a bit to the cgminer crew as well, since I intend to use their miner.

[fasmax]

@BkkCoins
I was just looking at the Avalon schematic and noticed an inductor connected to each ASIC on the analog VCC input(lead 7). Do you think this inductor is necessary or have you filtered this voltage in some other way.
Thanks

[BkkCoins]

@BkkCoins
I was just looking at the Avalon schematic and noticed an inductor connected to each ASIC on the analog VCC input(lead 7). Do you think this inductor is necessary or have you filtered this voltage in some other way.
Thanks

You mean the M1 / MB part? I took that to be a jumper/test point of some sort because it has no value and doesn't appear on the board. I'm pretty sure an inductor is not needed there.

[BkkCoins]

I received a USB PID assignment from MicroChip just now. So looks like Klondike boards will have their own legit unique VID/PID. Technically, this is only good for the first 10,000 boards produced, though I wonder how that can even be counted in this case.

That was pretty fast and easy - and didn't cost $2000 to get. Yay, MicroChip Tech.

[LaserHorse]

I received a USB PID assignment from MicroChip just now. So looks like Klondike boards will have their own legit unique VID/PID. Technically, this is only good for the first 10,000 boards produced, though I wonder how that can even be counted in this case.

That was pretty fast and easy - and didn't cost $2000 to get. Yay, MicroChip Tech.

I love this - very professional.

[chaoztc]

the nxp i2c buffers are cool - did i2c at 300m cat5 some years ago

[fasmax]

@BkkCoins
I was just looking at the Avalon schematic and noticed an inductor connected to each ASIC on the analog VCC input(lead 7). Do you think this inductor is necessary or have you filtered this voltage in some other way.
Thanks

You mean the M1 / MB part? I took that to be a jumper/test point of some sort because it has no value and doesn't appear on the board. I'm pretty sure an inductor is not needed there.

That's the part.I think it is a ferrite bead low resistance at DC high resistance for AC. I just looked at a photo of an Avalon board and it looks like there is a part installed for what is marked MB. It looks like a capacitor but it is a inductor here is a link  http://www.mouser.com/ProductDetail/Murata-Electronics/BLM21BD102SN1D/?qs=SsukgLyLLEeY%252bx0h1h0Bsw== hope I did that link right

[BkkCoins]

That's the part.I think it is a ferrite bead low resistance at DC high resistance for AC. I just looked at a photo of an Avalon board and it looks like there is a part installed for what is marked MB. It looks like a capacitor but it is a inductor here is a link  http://www.mouser.com/ProductDetail/Murata-Electronics/BLM21BD102SN1D/?qs=SsukgLyLLEeY%252bx0h1h0Bsw== hope I did that link right

If that's the case then I've missed it and we'll get a chance to see how it works without on the test board.

How did you find that part#? You can't tell from the board or schematic, or is it a guess/example. It can't be 0805 size as shown on mouser. 1000 Ohms seems a bit high for being in the supply path.

[fasmax]

That is just an example part. I don't know what Avalon used.

[fasmax]

Oh is .4 ohm at DC but 1000 ohm at 100MHZ.

[BkkCoins]

That is just an example part. I don't know what Avalon used.

Oh is .4 ohm at DC but 1000 ohm at 100MHZ.

Oh ya. That's make more sense.

[fasmax]

Found some more info on GitHub in the commits tab.
"finished hash unit BOM"
It lists MB as magnetic beads quantity (11) 30 ohm at 100MHZ does not specify DC resistance.

[BkkCoins]

Found some more info on GitHub in the commits tab.
"finished hash unit BOM"
It lists MB as magnetic beads quantity (11) 30 ohm at 100MHZ does not specify DC resistance.

That doesn't seem like much of a roll off. Most of the traces for AVDD on the K16 board are long enough that I can cut and patch one in if need be but I suspect it will be ok. No doubt the intent is to keep high freq. PLL noise from passing via the supply between the ASICs. I wonder how unstable it is and if it was put in out of necessity or just for extra precaution.

[marto74]

Full enclousure 8 x K16 with PSU and space for linux router

[erk]

Interesting to look at the pics of the KNCminer boards in this thread:

https://bitcointalk.org/index.php?topic=214285.0

A lot less parts count than the Avalon or BFL boards. I believe it's the  6 GH/s device called Mars.