Topic: [ANN] Bitfury is looking for alpha-testers of first chips! FREE MONEY HERE! - page 24. (Read 176728 times)

The 5 parallel resistors on the back are (counting from corner): 220k, 147k, 187k, 220k, 100k. Did they put silk screen on the back ?

Yeah but let me double check them one by one, thanks for the hint, I will let you know 

Or in the (over)current sense resistors. I assume the resistors are all unmarked ?

Or wrong value. Most likely in compensation loop.

Yeah, 12.000V stable and no ripples, it's very strange, maybe some parts of the TPS are damaged/broken?

@giorgiomassa: did you check the input voltage ? Is it stable ?

Is anybody else getting those crazy spikes out of c-scape's 16 bitfury card power supply? Guess this is why my chain is highly unstable, the length is detected randomly and sometimes even not detected at all...



Any help appreciated!

Thanks

Weird - I've got a 9-chip chain that's maxing out @ 9.5GH as well.  Just did a rework of my DC converter because of this and have tried all the clock settings posted here.

This is why I assumed you were using an external clock when you were running 4 @ 9GH.

SPI clock 500 kHz, core voltage 0.88V, internal slow osc 54 bits.

Update on power circuit.  1/2 day on regulated AC (through UPS & PDU network) with < 5mV jitter @90A maintaining 92% efficiency average for the circuit and 82% from the wall (additional loss is due to the ATX power supply I'm using, which is a Corsair AX1200i).   Running 1/2 day at my home now using residential AC, there is more noise in the output, but still exceptionally stable.  I have gotten 2 samplings thus far of 7mV spikes.  Efficiency from the wall is still at 82%.  Going to finish assembling my bitfury test boards tonight or tomorrow and will start powering real chips.

Our (c-scape & intron) bi•fury board is hashing over 5 GH/sec.

Screenshot from BTC Guild:


Photo of bi•fury prototype board with 2 ASICs, plugged into laptop. A fan is aimed at the board + heatsink, keeping the ASIC temperature at 45 ℃, and the rest of the board (including the regulator) at 40 ℃. Core voltage has been set at 0.84V. 

The extra pushbuttons, jumper, and 3 wire UART are just for testing/debugging.

So I modified cgminer to allow me to run only one chip at a time.

All the chips consume power the whole time, but only one hashes.

They all show >2Gh except for one, which is 1.8GH.  Most of them will maintain >2.5GH.


I have a current meter inline with 12v.  It's reading 1.73A while hashing... ~20.76W.  My power module is supposed to be 79% efficient at .9V, so that'd be 16W delivered.  16W @ .9V is ~18A.  Less than half the 40A rating on the regulator.

seeing about 7-7.3GH on five chips, with it skipping the first chip.  In order to skip, all I did was:


change chip = 0 to chip = 1.


Maybe I should set it up so I can run any single chip, with the full chain intact, and get a hashrate per chip.

How about modifying software to skip first chip ?

So I unbridged between #4 and #5, going back to four chips, and it's back to fast:

 cgminer version 3.3.1 - Started: [2013-08-29 05:42:33]
--------------------------------------------------------------------------------
 (5s):10.56G (avg):9.477Gh/s | A:160  R:0  HW:0  WU:132.4/m
 ST: 2  SS: 0  NB: 1  LW: 251  GF: 0  RF: 0
 Block: 002f2d30baf48c98...  Diff:65.8M  Started: [05:42:33]  Best share: 360
--------------------------------------------------------------------------------
 [P]ool management Settings [D]isplay options [Q]uit
 BITFURY 0: 11.61G/10.63Gh/s | A:176 R:0 HW:0 WU: 148.6/m
--------------------------------------------------------------------------------

 [2013-08-29 05:42:26] Started cgminer 3.3.1
 [2013-08-29 05:42:26] BITFURY chip #1 detected
 [2013-08-29 05:42:26] BITFURY chip #2 detected
 [2013-08-29 05:42:26] BITFURY chip #3 detected
 [2013-08-29 05:42:27] BITFURY chip #4 detected
 [2013-08-29 05:42:30] BITFURY: 4 chips detected!


Next, bridged 4 to 5 and unbridged 5 to six:

cgminer version 3.3.1 - Started: [2013-08-29 05:47:30]
--------------------------------------------------------------------------------
 (5s):9.923G (avg):8.487Gh/s | A:184  R:0  HW:0  WU:118.6/m
 ST: 2  SS: 0  NB: 1  LW: 379  GF: 0  RF: 0
 Block: 0034005b153db46a...  Diff:65.8M  Started: [05:47:30]  Best share: 1.52K
--------------------------------------------------------------------------------
 [P]ool management Settings [D]isplay options [Q]uit
 BITFURY 0: 10.17G/8.956Gh/s | A:192 R:0 HW:0 WU: 125.1/m
--------------------------------------------------------------------------------

 [2013-08-29 05:47:23] Started cgminer 3.3.1
 [2013-08-29 05:47:24] BITFURY chip #1 detected
 [2013-08-29 05:47:24] BITFURY chip #2 detected
 [2013-08-29 05:47:24] BITFURY chip #3 detected
 [2013-08-29 05:47:25] BITFURY chip #4 detected
 [2013-08-29 05:47:25] BITFURY chip #5 detected
 [2013-08-29 05:47:28] BITFURY: 5 chips detected!


slower!

That's the clock setting I've been using, and I've been at 5MHz.

Lemme turn spiclk down and see where that gets me.

For 16 chip chain I use 1 MHz SPI clock, and this clock setting: { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x3F, 0x00 };

What are you using for spiclk speeds and bitfury clock setting?

I've got 40A available; should be plenty!

Something wrong with your setup.
I'm running my longest 90-chips chain at average 237.7Gh/s (today will add 10 more chips to the chain).
Chips require a lot of power to hash faster than 2.5Gh/s per chip (at least 0.88V, 2,5-3A per chip), are you sure your DC/DC is capable of fully power 7 chips chain?