[GUIDE] BitFury Miner Support/Tuning

[goxed]

Guys heads up on pencil mod for overvolting and thus aiding overclocking.
Use pencil mod on R2F, make it around 1.1 - 1.2K, voltage goes from 0.66V to 0.81V. The default value is ~1.3K. It may blow the 30AFuse if you go too high; so make sure you do it only if you have a soldering station handy

After Mod (I have only 3 H cards )

Code:

Bank 1
1: 30.781GH/s
2: 33.315GH/s
3: 30.022GH/s
4: 0GH/s

[darkfriend77]

Can you explain how to do this?

:-)

Maybee some pictures ...

[goxed]

Can you explain how to do this?

:-)

Maybee some pictures ...

a) Get an Ohmmeter or DMM

b) Get an HB or softer Pencil

c) Locate R02F resistor. Donot mod R01F.

d) Make sure R02F measures 1.3KOhm

e) Rub pencil on the surface of R02F 3 - 4 times

f) Measure resistance, if it's 1.2K plug the board in

g) Measure output voltage, and repeat steps e,f if necessary for fine-tuning

h) Hash like a boss

[tom99]

Can you explain how to do this?

:-)

Maybee some pictures ...

a) Get an Ohmmeter or DMM

b) Get an HB or softer Pencil

c) Locate R02F resistor. Donot mod R01F.

d) Make sure R02F measures 1.3KOhm

e) Rub pencil on the surface of R02F 3 - 4 times

f) Measure resistance, if it's 1.2K plug the board in

g) Measure output voltage, and repeat steps e,f if necessary for fine-tuning

h) Hash like a boss

  Maybe you need to post picture of step e) if you could and thank you.

[mr_rulezzz]

Code:

Bank 1
1: 30.781GH/s
2: 33.315GH/s
3: 30.022GH/s
4: 0GH/s

% of HW errors? or ... show your stat.log please

[punin]

Nice mod Please remember to use fans when overvolting & overclocking your unit. If your unit goes "dead" ie. plugged in H-card shows no voltage on top of inductor, this may be due to overcurrent protection of the regulator kicking in. Although I'm generally supportive of hacking, please remember that you're doing so on your own risk. We cannot guarantee long-time functionality of the boards if you mod them.

[darkfriend77]

we need a pictured guide for dummies ... so we can break our boards trying ...

[puffmais]

has someone tried to use a wlan usb stick? how can i do this?

[badalien]

is everyone here from Germany - Hamburg that can make the resistor mod for the H Card, i have two of them ?

[goxed]

we need a pictured guide for dummies ... so we can break our boards trying ...

courtesy c-scape

[-Redacted-]

Dear Lord that thing is really small on a version-1 M-board (not the mod being discussed - the pencil mod is done on each H-board).  

My eyes aren't what they used to be, but are R01F and R02F REALLY size 0603 on the m-board??  If they are, how is it even possible to manually place 0402 parts

[darkfriend77]

i got those pictures sent from a user ...

to guide through the process of the pencil mod ....

***be aware that you loose warranty of the h-board in doing this ... and you can brick your board!!!***

details:

- temp +10-15°
- 1.6 times higher electricity needed
- up to 50% increased hashrate

Step 1: unplug your h-board ... and locate the R02F Resistor



Step 2: use ur multimeter on the resistor and verify that it shows 1.3k Ohm



Step 3: use a HB (medium grade) pencil (http://en.wikipedia.org/wiki/Pencil) and rub 2-3 times



Step 4: use ur multimeter on the resistor and check how much it was lowered in resistance ...as long as your higher then 1.2k restart step 3 & step 4 ...
goal resistance is 1.1k - 1.2k, if you get under 1.1k just swip a little away with your finger ...

Step 5: plug in again and start




have fun ...

[Isokivi]

Any word on the error rate, I managed to get to 30-33Ghs for short periods and huge error rates, with 0 (zero) actual gains.

[goxed]

Any word on the error rate, I managed to get to 30-33Ghs for short periods and huge error rates, with 0 (zero) actual gains.

How did you look for the error rates? can it be calculated from .stat.log?

[Isokivi]

How did you look for the error rates? can it be calculated from .stat.log?

open bitfury.ip/stat.json for constantly updating counts per board.
or ssh in and

Code:

less run/shm/.stat.log

This is a 5 min log that is rewritten over and over again

[-Redacted-]

or leave an ssh window open and  run this:

while (true) do tail -3 /run/shm/.stat.log; sleep 305; done


( not run/smh )

[Isokivi]

or leave an ssh window open and  run this:

while (true) do tail -3 /run/shm/.stat.log; sleep 305; done


( not run/smh )

corrected my post, thanks ...I keep making that same typo over and over again

[spiccioli]

or leave an ssh window open and  run this:

while (true) do tail -3 /run/shm/.stat.log; sleep 305; done


( not run/smh )

or

Code:

watch -n 305 tail -3 /run/shm/.stat.log

spiccioli

[-Redacted-]

Code:

watch -n 305 tail -3 /run/shm/.stat.log

Mine should run on any version of Linux or UNIX released since 1975, to include V2R5, Solaris, BSD, HPUX, OS/390, z/OS, Mac OS, and more.  
Yours, not so much....  

[darkfriend77]

Anybody knows what "speed->0" at the end of the .stat.log line actually mean?

2       AIfDSo  55      0.243   2.019   17      126     0       0       191     [0:1]   726     0 1 2 2 1 0 0 0 2 0 3 0 4 1 1 0         9 8 7 7 8 9 9 9 7 9 6 8 5 8 8 9    speed->0

10      AIfDSo  55      0.272   1.924   19      110     0       0       182     [0:9]   722     0 2 1 2 2 2 0 2 0 1 1 1 2 1 1 1         8 6 7 6 6 6 8 7 8 7 7 7 6 7 7 7    speed->0

#2 and #10 chips are not doing what they suppose to.  I'm not sure if it is assembly or my configuration.

EDIT:

After a while the rates drop completely:

2       AiFDso  55      0.014   0.550   1       35      0       0       52      [0:1]   723     0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0         2 2 2 2 3 3 3 3 2 2 1 2 2 2 2 2
10      AiFDso  55      0.072   0.528   5       34      0       0       50      [0:9]   715     1 1 0 0 0 1 0 1 0 1 0 0 0 0 0 0         2 2 3 2 2 1 2 1 2 1 2 2 3 3 3 3

thats the tuning mod that was used for this cycle ...

each cycle it will use a autotuning mod depending on the errors u got ....

as long you got no errors it will speed up the clock +1 ...

Code:

13. Autoclock
    - shut down = Chip will be shutdown and unavailable (Aifdso)
    - speed->0 = If still more errors than correct nonces -> shut down
    - speed up = If you have 0 errors
    - speed down = If you have >5 errors clock down