ANTMINER S4 Discussion and Support Thread

[nexus99]

The 400 coupon seems to make these one of the better buys. Competitive with the SP20.

[1714104205]

1714104205

[1714104205]

1714104205

[pak13]

I'm assuming there is no issue with stacking S4s? Or would running them vertical (PSU intake side up) be preferable?

They have rack mount ears--they are designed to be mounted in a DC horizontally.

What I meant was stacking them directly on top of each other, without a cabinet supporting the weight of each unit.

They're not that heavy, although I wouldn't go more than 4 in a stack. You could probably go more but I wouldn't.

If you do follow this dogie employee's blah blah advice, remember to strap them as one. They will vibrate or worst case, if you live in an earth quake region, they could potentially fall, kill or create a fire.

[pak13]

The 400 coupon seems to make these one of the better buys. Competitive with the SP20.

SP20 far cheaper long term. And from what we've seen, more reliable.

The new low power chips are less than a month away. This is the same fire sale as with the S2's before the newer models come out.

[Cassey]

How so on the SP20?  Lower output, similar power requirements....

[pak13]

How so on the SP20?  Lower output, similar power requirements....

Power is always the primary concern for cost over time.

Real world numbers ... 15% more power = 15% less profit or 15% longer to get a positive ROI assuming initial costs are equal.
1200w for 1.7TH 0.7
+0.8 for S3 S4

Ask the question, why it's difficult to get hold of a spondoolies but easy to buy a bitmaintech. Why are we still stuck on batch 11 for the s3's. Why are they giving such great deals at the moment. End of life products. (within the month most likely)

[Cassey]

How so on the SP20?  Lower output, similar power requirements....

Power is always the primary concern for cost over time.

Real world numbers ... 15% more power = 15% less profit or 15% longer to get a positive ROI assuming initial costs are equal.
1200w for 1.7TH 0.7
+0.8 for S3 S4

Ask the question, why it's difficult to get hold of a spondoolies but easy to buy a bitmaintech. Why are we still stuck on batch 11 for the s3's. Why are they giving such great deals at the moment. End of life products. (within the month most likely)

1200W for 1.7Th = 706 W/TH for SP20
1400W (worst case) for 2.0Th - 700W/TH for SP4

Where are you getting 15% difference?

[pikachuy]

Please post all of these coupons and others similar in the Marketplace. That is what it's for.

I don't like having a post alert and come and check to see it was only coupon offerings/for sale...

[notlist3d]

Thanks for coupon the first one after is now up and running

And shipping in 2 days!

[Cyper_BLC]

Hi everyone,
i have got a strange problem with my Antminer S4 batch2, i listed founded problems:
1-my S4's has different connections with FPGA controller, one of them is top left hash board connected to J7, other s4 same board is connected to J6. this is very strange or poor assembly.
2-this J7 connector is going to the first hash board i think, but other  hash boards red led is off except this J7 connected board led blink. 5 minutes waiting then this board getting hot and another five minutes all boards leds are on. then all of them is hot but THERE IS NO ETHERNET CONNECTION.  
3-i worried power comp. and i changed 0720 to 0700 then i measured all hash boards voltages on 1500uF/2,5V capacitor.. wouuwww THERE WAS A PROBLEM! because we can never change 1 hash board's (j7 connected) voltage. this is allways 0750V. than i read confidential pdf of CSD95372B and TPS53641 this is i2c controlled buck converter and this is controlled by pic12f1822 on every hash board. So this 0,750 Volt i measured is over voltage protection limit.
i am an electronic engineer by the way.

I ASK TO BITMAINTECH,
WHY OUR HASH BOARDS HASHING SOMETHING STANDALONE (or why warming up) AND WHY 3 HASH BOARDS VOLTAGE CONFIGURABLE BUT 1 IS NOT ?

thanks, any ideas ?

Dogie said to me "paste pictures to forum" then i did.
Now any ideas for this? when i did sniff controller->buck i2c communication, i will make a seperate or online live voltage changer ):
software bugs are minor but we have got a major hardware bugs.

[dmwardjr]

Hi everyone,
i have got a strange problem with my Antminer S4 batch2, i listed founded problems:
1-my S4's has different connections with FPGA controller, one of them is top left hash board connected to J7, other s4 same board is connected to J6. this is very strange or poor assembly.
2-this J7 connector is going to the first hash board i think, but other  hash boards red led is off except this J7 connected board led blink. 5 minutes waiting then this board getting hot and another five minutes all boards leds are on. then all of them is hot but THERE IS NO ETHERNET CONNECTION.  
3-i worried power comp. and i changed 0720 to 0700 then i measured all hash boards voltages on 1500uF/2,5V capacitor.. wouuwww THERE WAS A PROBLEM! because we can never change 1 hash board's (j7 connected) voltage. this is allways 0750V. than i read confidential pdf of CSD95372B and TPS53641 this is i2c controlled buck converter and this is controlled by pic12f1822 on every hash board. So this 0,750 Volt i measured is over voltage protection limit.
i am an electronic engineer by the way.

I ASK TO BITMAINTECH,
WHY OUR HASH BOARDS HASHING SOMETHING STANDALONE (or why warming up) AND WHY 3 HASH BOARDS VOLTAGE CONFIGURABLE BUT 1 IS NOT ?

thanks, any ideas ?

Dogie said to me "paste pictures to forum" then i did.
Now any ideas for this? when i did sniff controller->buck i2c communication, i will make a seperate or online live voltage changer ):
software bugs are minor but we have got a major hardware bugs.

Any symptoms prior to this other than ethernet going out?  Such as 1 or 2 boards much warmer Celsius than the others, etc...?

[Cyper_BLC]

Others perfecly normal only J7 connected hash board is unconfigurable, if you want to proof of that disconnect this board power cord and you will see chain 1 is gone and look at your power wattage. or try to configure voltage and repower again. i have 2 S4 and all of them have same problem.
and i found best configuration and best efficiency (0.52 but not stable) => 187.5 Mhz at 0.685 Volt (you can add this value like that = 0685)
this is the best...

[dmwardjr]

Others perfecly normal only J7 connected hash board is unconfigurable, if you want to proof of that disconnect this board power cord and you will see chain 1 is gone and look at your power wattage. or try to configure voltage and repower again. i have 2 S4 and all of them have same problem.
and i found best configuration and best efficiency = 187.5 Mhz at 0.685 Volt (you can add this value like that = 0685)
this is the best...

Wow!

So, you recommend all of us with batch 2 to switch to 187.5 Mhz at 0.685 Volt?

Will doing this lower my hash rate? 

If so, by how much will it lower the hash rate?

[Cyper_BLC]

of course that is lower than 2Thash but also eat lower energy and also lower Temperature and lower noise
like 1800Ghash 1060W

i think that voltage adjust problem is master reason of psu's blowing because developers unexpected situation like that.
they put a 12V 120A top limit psu and they expect 1200w power usage but not, some of hash boards still working warning limits...

[dmwardjr]

of course that is lower than 2Thash but also eat lower energy and also lower Temperature and lower noise
like 1800Ghash 1060W

I was just about to ask you for the actual hash rate and wattage.

Thank you...

[dmwardjr]

Others perfecly normal only J7 connected hash board is unconfigurable, if you want to proof of that disconnect this board power cord and you will see chain 1 is gone and look at your power wattage. or try to configure voltage and repower again. i have 2 S4 and all of them have same problem.
and i found best configuration and best efficiency (0.52 but not stable) => 187.5 Mhz at 0.685 Volt (you can add this value like that = 0685)
this is the best...

Curious:

Does this also mean board J7 will remain at 0.725 Volts while the others are at 0.685 Volts?  Cause I think you said J7 was un-configurable?

This would mean we can expect this one board to probably be considerably hotter than the other boards in miner status.

[Cyper_BLC]

Others perfecly normal only J7 connected hash board is unconfigurable, if you want to proof of that disconnect this board power cord and you will see chain 1 is gone and look at your power wattage. or try to configure voltage and repower again. i have 2 S4 and all of them have same problem.
and i found best configuration and best efficiency (0.52 but not stable) => 187.5 Mhz at 0.685 Volt (you can add this value like that = 0685)
this is the best...

Curious:

Does this also mean board J7 will remain at 0.725 Volts while the others are at 0.685 Volts?  Cause I think you said J7 was un-configurable?

This would mean we can expect this one board to probably be considerably hotter than the other boards.

Yes thats right.  i can measure stable results, only after i found a master problem and solve it

[dmwardjr]

Others perfecly normal only J7 connected hash board is unconfigurable, if you want to proof of that disconnect this board power cord and you will see chain 1 is gone and look at your power wattage. or try to configure voltage and repower again. i have 2 S4 and all of them have same problem.
and i found best configuration and best efficiency (0.52 but not stable) => 187.5 Mhz at 0.685 Volt (you can add this value like that = 0685)
this is the best...

Curious:

Does this also mean board J7 will remain at 0.725 Volts while the others are at 0.685 Volts?  Cause I think you said J7 was un-configurable?

This would mean we can expect this one board to probably be considerably hotter than the other boards.

Yes thats right.  i can measure stable results, only after i found a master problem and solve it

And the way to solve it [for now] is to reduce to 187.5 Mhz at 0.685 Volt?  Just making sure you did not do something else as well?

[Cyper_BLC]

Others perfecly normal only J7 connected hash board is unconfigurable, if you want to proof of that disconnect this board power cord and you will see chain 1 is gone and look at your power wattage. or try to configure voltage and repower again. i have 2 S4 and all of them have same problem.
and i found best configuration and best efficiency (0.52 but not stable) => 187.5 Mhz at 0.685 Volt (you can add this value like that = 0685)
this is the best...

Curious:

Does this also mean board J7 will remain at 0.725 Volts while the others are at 0.685 Volts?  Cause I think you said J7 was un-configurable?

This would mean we can expect this one board to probably be considerably hotter than the other boards.

Yes thats right.  i can measure stable results, only after i found a master problem and solve it

And the way to solve it [for now] is to reduce to 187.5 Mhz at 0.685 Volt?  Just making sure you did not do something else as well?

No... that is not a solution. this is my best working parameters. not a solution........

[dmwardjr]

Others perfecly normal only J7 connected hash board is unconfigurable, if you want to proof of that disconnect this board power cord and you will see chain 1 is gone and look at your power wattage. or try to configure voltage and repower again. i have 2 S4 and all of them have same problem.
and i found best configuration and best efficiency (0.52 but not stable) => 187.5 Mhz at 0.685 Volt (you can add this value like that = 0685)
this is the best...

Curious:

Does this also mean board J7 will remain at 0.725 Volts while the others are at 0.685 Volts?  Cause I think you said J7 was un-configurable?

This would mean we can expect this one board to probably be considerably hotter than the other boards.

Yes thats right.  i can measure stable results, only after i found a master problem and solve it

And the way to solve it [for now] is to reduce to 187.5 Mhz at 0.685 Volt?  Just making sure you did not do something else as well?

No... that is not a solution. this is my best working parameters. not a solution........

That's what I thought...

I was just making sure.

Thanks again for your patience; time; and hard work.

[dmwardjr]

One more question:

In your opinion, is there any harm with leaving it at the default configuration as is?