Topic: Efudd Z-Series Fuddware 2.3 -Z11/Z11e/Z11j/Z9/Mini - page 31. (Read 45506 times)

@ZergeZ

A lil bit off topic, could you post some images of that immersive cooling and possibly share how much did it cost you to install it.

Cheers.

Ends in 3 or 8, +7. Otherwise +6

Absolutle so!.  I also have several batch1 devices  working in liquid. Temperature +45-48 degrees.  They all can work at 775MHz - but hashrate become lower. After 750Mhz, voltage on hash chips dosn't change - 0.85v, and this voltage not enough for  higher than 750Mhz freq.

My batch 1's run at 762 anything higher it loses hashrate but does not error so I know they can handle higher. I ran it as high as 800 but 1 board dropped within a few minutes. You also have to follow the clocking pattern it seems to go in intervals of 6 with the exception of a few where it jumps 7

It will vary by machine. You will have to see how yours behaves at a given frequency.

i have batch 1 running on 750Mhz, are u suggesting that goign above has no sense as the hash will drop due to not enough power?

That is just shell expansion and filters on the inputs stream instead of using things like sed and awk.

Just presentation. 100% sure. “Final answer” as some tv show used to say.

I find it odd if you look at how they are pairing the variables.... You have speed paired with fan then then temp and pool. Kind of odd pairs just to pass some variables  to the ui

Read only presentation variables for the UI.

The reason is clear. Lack of power supply. For frequencies above 750, it is necessary to increase the power supply to at least 0.9V with appropriate cooling of the chips and PWM drivers. At a frequency of 750, the chips work and do not fall off, but due to internal errors, the hashrate drops

I have the same issue I could reach a higher freq but the hashrate starts to drop the batch 1's I can run at 762 and the batch 4's at 737 theZ9 full size I can only run at 656

Jason

What do you get of this code here... it appears to be double checking fan and temps versus hashrate etc.... It is in the file sbin minerlcd there is more this is just a snippet of it. I assuming the word temp is the temp in c or f and tmp is what it is a tmp file or folder. Even though there is no lcd this routine still runs for the API I believe. Is it possible you can gain control over the fans at this point

    if [ "${ant_tmp}" == "Socket connect failed: Connection refused" ]; then
       ant_ghs=0
       ant_fan=0
       ant_temp=0
       ant_poolurl=no
       ant_pooluser=no
    else
       ant_ghs=${ant_tmp#*GHSavg=}
       ant_ghs=${ant_ghs%%,fan=*}
       
       ant_fan=${ant_tmp#*fan=}
       ant_fan=${ant_fan%%,temp=*}
       
       ant_temp=${ant_tmp#*temp=}
       ant_temp=${ant_temp%%,pool=*}
       
       ant_poolurl=${ant_tmp#*pool=}
       ant_poolurl=${ant_poolurl%%,user=*}
       ant_var=${ant_poolurl:0:7}
      if [ "${ant_var}" == "http://" ]; then
         ant_poolurl=${ant_poolurl:7}
       fi
      ant_pooluser=${ant_tmp#*user=}
       ant_pooluser=${ant_pooluser%%|*}
    fi

      if false; then
    echo "IP:${ant_ip}"                 > /tmp/lcd.data
    echo "${ant_ghs}"              >> /tmp/lcd.data
    echo "${ant_fan} ${ant_temp}"  >> /tmp/lcd.data
    echo "Pl:${ant_poolurl}"         >> /tmp/lcd.data
    echo "Ur:${ant_pooluser}"        >> /tmp/lcd.data
   fi

Fan simulator solution gives nothing. I am using now fan simulator because I have no fans - I am cooling the asics by liquid. The speed of fan  is 30600rpm. But as the frequency rises, the voltage on the chips does not change (for 3-4 batch) or changes from 0.8 to 0.85v when the frequency changes from 500 to 750 MHz(for 102 batch). If I set the frequency higher than 750, the voltage remains constant at 0.85V and the hashrate does not change, but even drops.

Anyone beta testing with HiveOS?

Pic microcontroller aren't very complicated most pic chips can be unlocked if you have the right equipment which I do but at the moment I don't want to rip one off a board just to dump the code. As far as the PWM  many of its routines off the fan rpm.

I don't fully get that part and Efudd has a better chance of getting past it right now then anyone else out there at the moment. 

One workaround could be to use a fan simulator telling the unit the fan rpm is over 5880 and this will allow for the unit to say ok we can give it some more power so at this point we need to figure out the code sent to the pic saying give more juice. Sometimes the fix could be a simple hardware mod by adding a resistor allowing a little more juice to get to the chip.

I know that. But the crux of the matter is that the PWM stabilizer, which in turn is programmed by 16f1704, controls the voltage and current protection. It is necessary to understand how this microcontroller works together with the PWM stabilizer and change its program.

In other words if you cant get good control of the ASIC power thru cgminer then a hardware mod could possibly be done to the hashboards to gain the extra power needed to clock them up higher

I didn't understand nothing in you answer!  Epxplain me detail -  what you wanted to say

And I still don't have a problem paying you for a version. I asked you if you were giving beta testers a free license or not so your suspicions are wrong. I only fixed a fan issue on the batch 4 minis which you stated you were not interested is messing with at this time due they were only available in china. I don't get it but whatever.

Beta testers -- I've pulled the Z9 mini image as there is an issue which will prevent it from mining properly. I will provide an updated version soon.

Thank you,

Jason