[ANN] Bitfury is looking for alpha-testers of first chips! FREE MONEY HERE!

[Gomeler]

Is the level-shifter required to transform the voltage coming out of the GPIO pins of the raspberry pi(5v?) down to levels that the bitfury chips operate on(3.3v?)?

[cscape]

Is the level-shifter required to transform the voltage coming out of the GPIO pins of the raspberry pi(5v?) down to levels that the bitfury chips operate on(3.3v?)?

Yes, but the voltages are 3.3 and 1.8.

[Gomeler]

Is the level-shifter required to transform the voltage coming out of the GPIO pins of the raspberry pi(5v?) down to levels that the bitfury chips operate on(3.3v?)?

Yes, but the voltages are 3.3 and 1.8.

Thanks for the confirmation. Maybe one day I'll get around reading up on PCB design and tinker with this.

[klondike_bar]

Is the level-shifter required to transform the voltage coming out of the GPIO pins of the raspberry pi(5v?) down to levels that the bitfury chips operate on(3.3v?)?

Yes, but the voltages are 3.3 and 1.8.

if you are feeling creative, i bet you could grab the 3.3V line directly from the PSU. 1.8V is another challenge though

[zulunation]

Hello everyone,
I'm developing a board for 16 chips and having some problems with power supply.
It is the same like cscape used in his board.
Here is the front view of the power supply:
 


I see that the pin height here is 1.57mm. Is it enough to solder it on the board?
As is understand there must be pads on the board and the ps is soldered into them and pin goes through the whole board and soldered at the other side of the board.
It seems to me that 1.57mm is not sufficient.

[LaserHorse]

Hello everyone,
I'm developing a board for 16 chips and having some problems with power supply.
It is the same like cscape used in his board.
Here is the front view of the power supply:
 
https://dl.dropboxusercontent.com/u/26351037/bitfury_ps_pinheight.JPG

I see that the pin height here is 1.57mm. Is it enough to solder it on the board?
As is understand there must be pads on the board and the ps is soldered into them and pin goes through the whole board and soldered at the other side of the board.
It seems to me that 1.57mm is not sufficient.

These are surface-mount - pins are not intended to go through the board.

In a rush to get my chips hashing, I soldered one of these to a through-hole board.  You need to be careful, but it's possible.

I used 2 pieces of solid core wire wrapped around on each 'pin', routed them under the board and soldered them to massive tracks on the underside.  I considered soldering directly to pads on an upside-down protoboard, but the spacing does not match up, etc.

some notes:
     • Per the datasheet - solder the on/off pin to ground if you won't be using it.
     • The smaller 'pins' on the board easily come loose while soldering extension wire to them
     • Flux is your friend
     • Needless to say - triple-check for bridges *everywhere* before applying power

I've got the board powering 8 chips currently with moderate airflow - temp is good, power is good.

[ssi]

I'm up and running... seeing 6GH on four chips.  Running .77V VDD, and changing the clock doesn't seem to change the average hashrate much.  My clock is currently at ... 0xFF 0x10, and I'm bumping it up slowly.  Haven't seen any hardware errors yet.  Not sure when/if I should try to move up to .9V.

[needbmw]

and changing the clock doesn't seem to change the average hashrate much.  My clock is currently at ... 0xFF 0x10

If using cgminer please note reverse clock bits order, so you have to set 0xff 0x01, 0xff 0x03, ... , 0xff 0x3f, 0xff 0x7f, otherwise your clock will be 8-bits aligned.

[ssi]

and changing the clock doesn't seem to change the average hashrate much.  My clock is currently at ... 0xFF 0x10

If using cgminer please note reverse clock bits order, so you have to set 0xff 0x01, 0xff 0x03, ... , 0xff 0x3f, 0xff 0x7f, otherwise your clock will be 8-bits aligned.

Interesting, thanks.

So, just to make sure I understand you, when I download the legkodymov cgminer fork, it's set like so by default:

Code:

unsigned char osc6[8] = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x7F, 0x00, 0x00 };

So you're saying that one setting faster is:

Code:

unsigned char osc6[8] = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x00, 0x00 };

and one setting faster than that is:

Code:

unsigned char osc6[8] = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x01, 0x00 };

Is that correct?

[ssi]

and changing the clock doesn't seem to change the average hashrate much.  My clock is currently at ... 0xFF 0x10

If using cgminer please note reverse clock bits order, so you have to set 0xff 0x01, 0xff 0x03, ... , 0xff 0x3f, 0xff 0x7f, otherwise your clock will be 8-bits aligned.

Beautiful... getting somewhere now!


 (5s):10.79G (avg):8.436Gh/s | A:205  R:6  HW:0  WU:117.9/m


that's at:

Code:

unsigned char osc6[8] = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0F, 0x00 };

Bottom side of the board is at 29C.

[ssi]

[LaserHorse]

Beautiful... getting somewhere now!


 (5s):10.79G (avg):8.436Gh/s | A:205  R:6  HW:0  WU:117.9/m


that's at:

Code:

unsigned char osc6[8] = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0F, 0x00 };

nice - what's your SPI clock set to?  legkodymov's code defaults to 200KHz

[ssi]

Beautiful... getting somewhere now!


 (5s):10.79G (avg):8.436Gh/s | A:205  R:6  HW:0  WU:117.9/m


that's at:

Code:

unsigned char osc6[8] = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0F, 0x00 };

nice - what's your SPI clock set to?  legkodymov's code defaults to 200KHz

5MHz

[LaserHorse]

wow - guess i need to switch to using external clock 

Do you have the last chip in your chain feeding SPI outs back into the Pi as on the M / H boards?
I've yet to figure out why this is done.

[cscape]

Do you have the last chip in your chain feeding SPI outs back into the Pi as on the M / H boards?
I've yet to figure out why this is done.

It's not necessary, but you could use this to verify the SPI chain is working correctly.

[ssi]

wow - guess i need to switch to using external clock 

Do you have the last chip in your chain feeding SPI outs back into the Pi as on the M / H boards?
I've yet to figure out why this is done.

I'm not using external clock, and no I didn't feed the final SPI outs back anywhere.

[LaserHorse]

It's not necessary, but you could use this to verify the SPI chain is working correctly.

makes sense - thanks for clarifying.

I'm not using external clock, and no I didn't feed the final SPI outs back anywhere.

ah - this is line used to program chips internal oscillator (libbitfury.c line#109):

Code:

unsigned char osc6[8] = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0F, 0x00 };

I'd assumed the Pi was generating external clock for the chips using the above code.
The separate clock connection on the H-board threw me off - guessing c-scape added that to keep the functionality available.

EDIT: reworded for posterity

[buzzdave]

and changing the clock doesn't seem to change the average hashrate much.  My clock is currently at ... 0xFF 0x10

If using cgminer please note reverse clock bits order, so you have to set 0xff 0x01, 0xff 0x03, ... , 0xff 0x3f, 0xff 0x7f, otherwise your clock will be 8-bits aligned.

Beautiful... getting somewhere now!


 (5s):10.79G (avg):8.436Gh/s | A:205  R:6  HW:0  WU:117.9/m


that's at:

Code:

unsigned char osc6[8] = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0F, 0x00 };

Bottom side of the board is at 29C.

Great Job!

[18RATTT]

b-e-a-u-tiful

[ssi]

and changing the clock doesn't seem to change the average hashrate much.  My clock is currently at ... 0xFF 0x10

If using cgminer please note reverse clock bits order, so you have to set 0xff 0x01, 0xff 0x03, ... , 0xff 0x3f, 0xff 0x7f, otherwise your clock will be 8-bits aligned.

Beautiful... getting somewhere now!


 (5s):10.79G (avg):8.436Gh/s | A:205  R:6  HW:0  WU:117.9/m


that's at:

Code:

unsigned char osc6[8] = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0F, 0x00 };

Bottom side of the board is at 29C.

Great Job!

Thanks Dave

It's time for more chips; putting in an order now.