Author

Topic: ANTMINER S3+ Discussion and Support Thread - page 322. (Read 710164 times)

sr. member
Activity: 384
Merit: 256
Who is having batch5  with ups and  has ups update their status? Any problem so far

I use UPS and have incoming batch #5 miners that have been updated today. Is there something specific you're looking for? Where are you located? (I'm in the US)

My previous order was for 50 batch 4 units and they arrived earlier than advertised. I had a call from customs to confirm my SSN/EIN, but that didn't seem to delay anything. I never get calls from customs unless the order on the invoice is over ~$10k.

My batch 5 miners will arrive Wednesday. We sell in the US and Canada if anyone's interested. We've already fulfilled over 100 AntMiner S3s to date with 100% satisfaction rate.
member
Activity: 65
Merit: 10
Who is having batch5  with ups and  has ups update their status? Any problem so far

Looks like mine shipped this morning via UPS and will be here in TX on Wednesday afternoon
hero member
Activity: 784
Merit: 504
Can someone explain all this talk on thermal grease? I've read a lot of this thread, but it isn't entirely clear to me.

From what I understand, you are taking off the heat sinks, adding thermal grease, and then putting them back. Why do you have to do this? It seems it makes it cooler - does that make them hash faster? Which heat sinks do you replace? Is this a problem only with Batch 1-3 or something, or would you recommend this for all of them?

A lot of questions, I will try to answer them all with in my capacity.
If you have watch this thread you may already know why all are doing the heat sink hack.
In most of the miners various users found that the thermal paste is not applied correctly.
Some miners don't have enough thermal paste, others have a lot of them leaking to the pins of the chips.
That's why a lot of hash rate issues in S3, bitmaintech even advised to do it.

Chip throttled to lower clock frequency when they are overheated. When the thermal paste is not properly applied
this will lead to overheating due to lower heat conductance and the result is lower hash rate.
So removing the old thermal paste and applying a thermal grease of minimum conductivity of 3 will help the heat
to dissipate properly which will further lead to higher/rated hashrate.

The smaller heatsink attached to the outer side or front side or the chip side is what we are discussing here.
Since the opposite side is a flat surface chances of wrong application on that side is comparatively low.

We are not replacing heatsink but the thermal paste.
If you can apply to all miners its good. But generally fix those who are not working properly.
Keep an eye on temperatures, fans speed, hashrate.
hero member
Activity: 857
Merit: 1000
Anger is a gift.
Can someone explain all this talk on thermal grease? I've read a lot of this thread, but it isn't entirely clear to me.

From what I understand, you are taking off the heat sinks, adding thermal grease, and then putting them back. Why do you have to do this? It seems it makes it cooler - does that make them hash faster? Which heat sinks do you replace? Is this a problem only with Batch 1-3 or something, or would you recommend this for all of them?

Taking off the heat sink and removing the absurd amount of paste on the chips. Why? I had issues with units not starting and having X's on the chips. Replacing the paste will make them hash better ( a few GH/s)  and maybe a little cooler (1-5C). You are not replacing the heat sink itself, just the paste between it and the top of the chips.
newbie
Activity: 19
Merit: 0
Can someone explain all this talk on thermal grease? I've read a lot of this thread, but it isn't entirely clear to me.

From what I understand, you are taking off the heat sinks, adding thermal grease, and then putting them back. Why do you have to do this? It seems it makes it cooler - does that make them hash faster? Which heat sinks do you replace? Is this a problem only with Batch 1-3 or something, or would you recommend this for all of them?
newbie
Activity: 15
Merit: 0
Couple of points for those who might be interested...

Connecting the blades up to non-specific controllers...

Not easy to do. The blades are driven by a specific microcontroller (PIC of some sort) which is connected via USB to the AR9331. All of this is integrated onto the mainboard, and can't be removed. It also limits the number of blades you can connect to two, as the MCU in question doesn't have any connectors for more and likely lacks the resources to support them anyway.

As for the controller, it is already pretty much maxed out and would probably not be able to handle more even with the room to connect them. If you could get the MCU without the AR9331 board and also programmed with the appropriate firmware, you might be able to get it running on a generic PC or other more powerful host. You would however need access to the driver to compile it for a different architecture (not impossible, but difficult).

People having issues with controller boards...

Why not do as I do and post extracts from system logs here? Those who have worked with linux, and specifically OpenWRT might be able to help out, or at least give more specific diagnosis of issues...

Not feasible for me as my controller board's Ethernet interface is completely dead - so it's not even possible to get into it with the default IP address to configure anything let alone see the logs. 

Other people have comparatively minor issues and at least get some hashing out of their units.  Mine is completely dead in the water going on a week now, extending to at least another week.

I have 3 S1 units, and thought about swapping the controller from them, but they're still profitable and I don't want to muck anything up that is actually still running.

BITMAIN - PLEASE JUST SEND REPLACEMENT BOARDS TO THOSE WHO NEED THEM AND HAVE ESTABLISHED THAT IS THEIR PROBLEM.  WAITING 2+ WEEKS FOR A SOLUTION KILLS OUR PROFITABILITY.

But apparently the only thing that seems to matter to them is their own profitability.
sr. member
Activity: 392
Merit: 250
Who is having batch5  with ups and  has ups update their status? Any problem so far
legendary
Activity: 1235
Merit: 1202
I've now tried this...

Blade 1 -> Cable1 -> PORT2 in CP
Blade 2 -> Cable2 -> PORT1 in CP

The result is swapped, now Chain1 shows all dashes with one "o" and Chain2 is working fine. My guess is the blade is gone?

Good to see the results. So now, we found who is the culprit in the couple.
The Blade1 is damaged. And I'm almost sure that its the heat sink - thermal paste problem.
Connecting the damaged blade like this and running the miner may harm the chip in your damaged blade.
So, the option is to unplug the damaged blades power supply from PSU and the cable from control panel.
Then run the single blade or shut down the S3 completely to save your chips in the non-working blade.

Same time unscrew the small lengthy heatsink which is attached to the front side of the blade(Chips)
and check the condition of thermal paste application. If its not good clean it and apply some new thermal paste.
If its good contact bitmain and tell them you did all these thinks, may link this post. Bitmain will probably do RMA your miner.

Already looked at the heatsink, had it off, paste looked like it was applied evenly and heatsink seemed fastened properly. I reattached it now but still no nice Sad

Just my luck eh?

Thanks again for the help, you've been brilliant!

Hope you have the images of the heatsink and blade when opened.
Upload it here to have a look. Also as a last resort, do the below and if not working, get ready for RMA.
Backup the current firmware and update with the latest available from bitmaintech website.


Didn't take any pics when it was apart sadly (I didn't think of it to be honest) Good idea about the firmware, I'll probably give it a go tomorrow but I'm not liking my chances. I'll wait to hear back from Bitmain and see what they have to say also Smiley
hero member
Activity: 784
Merit: 504
I've now tried this...

Blade 1 -> Cable1 -> PORT2 in CP
Blade 2 -> Cable2 -> PORT1 in CP

The result is swapped, now Chain1 shows all dashes with one "o" and Chain2 is working fine. My guess is the blade is gone?

Good to see the results. So now, we found who is the culprit in the couple.
The Blade1 is damaged. And I'm almost sure that its the heat sink - thermal paste problem.
Connecting the damaged blade like this and running the miner may harm the chip in your damaged blade.
So, the option is to unplug the damaged blades power supply from PSU and the cable from control panel.
Then run the single blade or shut down the S3 completely to save your chips in the non-working blade.

Same time unscrew the small lengthy heatsink which is attached to the front side of the blade(Chips)
and check the condition of thermal paste application. If its not good clean it and apply some new thermal paste.
If its good contact bitmain and tell them you did all these thinks, may link this post. Bitmain will probably do RMA your miner.

Already looked at the heatsink, had it off, paste looked like it was applied evenly and heatsink seemed fastened properly. I reattached it now but still no nice Sad

Just my luck eh?

Thanks again for the help, you've been brilliant!

Hope you have the images of the heatsink and blade when opened.
Upload it here to have a look. Also as a last resort, do the below and if not working, get ready for RMA.
Backup the current firmware and update with the latest available from bitmaintech website.
hero member
Activity: 784
Merit: 504
Observation.
In one of the Batch 4 machines I got, there is a machine with a July 28 firmware date within the Kernel Log.

Anyone noticed that?  Is that an experimental firmware?

The cooling fans report constant 7000-9000 RPM range.  That doesn't look normal.  Recommendations?

Got my B4 S3's today.  Both of them have also the July 28 firmware ("Linux version 3.10.12 (xxl@ubuntu) (gcc version 4.6.4 (OpenWrt/Linaro GCC 4.6-2013.05 r38031) ) #140 Mon Jul 21 17:27:04 CST 2014").
Both of them run with very high fan speeds, much higher than the B1 units

3780 3960
4680 10560

The B1 units, at freq=225 run 1740 1620  and 1680 1500 rpm

The Bitmain support page does not mention this firmware.  Any clues why the high fan speed?
I'm now getting fan speeds of ~14000, it 'sounds' like a software bug more that a super-sonic fan.... (the fans on the new B4 units sound similar to the older B1 units with the July 21 firmware) - BITMAIN what is happening with this July 28 firmware on the Batch 4 Antminer S3s?

Cheers

Batch 4 Antminer S3 seems more stable than the Batch 1 to 3.
I think the manufacturer purposefully did this. Else there is no reason to have a higher fan speed.
Don't forget to take a backup of your firmware before doing any update.
since, the current so called different firmware is not available in bitmaintech website.
Any one here with B4 S3-es having lower fan speeds below 2000? Also post your ambient and clockspeeds.

Hey, I forgot to ask you some thing, which is your power supply and how you are connecting your miners.
Also tell how much load or miners connected to your Power supply.
legendary
Activity: 1235
Merit: 1202
I've now tried this...

Blade 1 -> Cable1 -> PORT2 in CP
Blade 2 -> Cable2 -> PORT1 in CP

The result is swapped, now Chain1 shows all dashes with one "o" and Chain2 is working fine. My guess is the blade is gone?

Good to see the results. So now, we found who is the culprit in the couple.
The Blade1 is damaged. And I'm almost sure that its the heat sink - thermal paste problem.
Connecting the damaged blade like this and running the miner may harm the chip in your damaged blade.
So, the option is to unplug the damaged blades power supply from PSU and the cable from control panel.
Then run the single blade or shut down the S3 completely to save your chips in the non-working blade.

Same time unscrew the small lengthy heatsink which is attached to the front side of the blade(Chips)
and check the condition of thermal paste application. If its not good clean it and apply some new thermal paste.
If its good contact bitmain and tell them you did all these thinks, may link this post. Bitmain will probably do RMA your miner.

Already looked at the heatsink, had it off, paste looked like it was applied evenly and heatsink seemed fastened properly. I reattached it now but still no nice Sad

Just my luck eh?

Thanks again for the help, you've been brilliant!
sr. member
Activity: 252
Merit: 250
Couple of points for those who might be interested...

Connecting the blades up to non-specific controllers...

Not easy to do. The blades are driven by a specific microcontroller (PIC of some sort) which is connected via USB to the AR9331. All of this is integrated onto the mainboard, and can't be removed. It also limits the number of blades you can connect to two, as the MCU in question doesn't have any connectors for more and likely lacks the resources to support them anyway.

As for the controller, it is already pretty much maxed out and would probably not be able to handle more even with the room to connect them. If you could get the MCU without the AR9331 board and also programmed with the appropriate firmware, you might be able to get it running on a generic PC or other more powerful host. You would however need access to the driver to compile it for a different architecture (not impossible, but difficult).

People having issues with controller boards...

Why not do as I do and post extracts from system logs here? Those who have worked with linux, and specifically OpenWRT might be able to help out, or at least give more specific diagnosis of issues...
hero member
Activity: 784
Merit: 504
Observation.
In one of the Batch 4 machines I got, there is a machine with a July 28 firmware date within the Kernel Log.

Anyone noticed that?  Is that an experimental firmware?

The cooling fans report constant 7000-9000 RPM range.  That doesn't look normal.  Recommendations?

Got my B4 S3's today.  Both of them have also the July 28 firmware ("Linux version 3.10.12 (xxl@ubuntu) (gcc version 4.6.4 (OpenWrt/Linaro GCC 4.6-2013.05 r38031) ) #140 Mon Jul 21 17:27:04 CST 2014").
Both of them run with very high fan speeds, much higher than the B1 units

3780 3960
4680 10560

The B1 units, at freq=225 run 1740 1620  and 1680 1500 rpm

The Bitmain support page does not mention this firmware.  Any clues why the high fan speed?
I'm now getting fan speeds of ~14000, it 'sounds' like a software bug more that a super-sonic fan.... (the fans on the new B4 units sound similar to the older B1 units with the July 21 firmware) - BITMAIN what is happening with this July 28 firmware on the Batch 4 Antminer S3s?

Cheers

Batch 4 Antminer S3 seems more stable than the Batch 1 to 3.
I think the manufacturer purposefully did this. Else there is no reason to have a higher fan speed.
Don't forget to take a backup of your firmware before doing any update.
since, the current so called different firmware is not available in bitmaintech website.
Any one here with B4 S3-es having lower fan speeds below 2000? Also post your ambient and clockspeeds.
hero member
Activity: 784
Merit: 504
I've now tried this...

Blade 1 -> Cable1 -> PORT2 in CP
Blade 2 -> Cable2 -> PORT1 in CP

The result is swapped, now Chain1 shows all dashes with one "o" and Chain2 is working fine. My guess is the blade is gone?

Good to see the results. So now, we found who is the culprit in the couple.
The Blade1 is damaged. And I'm almost sure that its the heat sink - thermal paste problem.
Connecting the damaged blade like this and running the miner may harm the chip in your damaged blade.
So, the option is to unplug the damaged blades power supply from PSU and the cable from control panel.
Then run the single blade or shut down the S3 completely to save your chips in the non-working blade.

Same time unscrew the small lengthy heatsink which is attached to the front side of the blade(Chips)
and check the condition of thermal paste application. If its not good clean it and apply some new thermal paste.
If its good contact bitmain and tell them you did all these thinks, may link this post. Bitmain will probably do RMA your miner.
hero member
Activity: 616
Merit: 500
Observation.
In one of the Batch 4 machines I got, there is a machine with a July 28 firmware date within the Kernel Log.

Anyone noticed that?  Is that an experimental firmware?

The cooling fans report constant 7000-9000 RPM range.  That doesn't look normal.  Recommendations?

Got my B4 S3's today.  Both of them have also the July 28 firmware ("Linux version 3.10.12 (xxl@ubuntu) (gcc version 4.6.4 (OpenWrt/Linaro GCC 4.6-2013.05 r38031) ) #140 Mon Jul 21 17:27:04 CST 2014").
Both of them run with very high fan speeds, much higher than the B1 units

3780 3960
4680 10560

The B1 units, at freq=225 run 1740 1620  and 1680 1500 rpm

The Bitmain support page does not mention this firmware.  Any clues why the high fan speed?
I'm now getting fan speeds of ~14000, it 'sounds' like a software bug more that a super-sonic fan.... (the fans on the new B4 units sound similar to the older B1 units with the July 21 firmware) - BITMAIN what is happening with this July 28 firmware on the Batch 4 Antminer S3s?

Cheers
legendary
Activity: 1235
Merit: 1202
I've now tried this...

Blade 1 -> Cable1 -> PORT2 in CP
Blade 2 -> Cable2 -> PORT1 in CP

The result is swapped, now Chain1 shows all dashes with one "o" and Chain2 is working fine. My guess is the blade is gone?
legendary
Activity: 1235
Merit: 1202
Just got my two S3's today! Setting up one however and ran into a problem, one chain is hashing away fine with oooooooo oooooooo but the other comes back with -------- -------- -------- -------o

What the hell does this mean?!? It worked ok on first run, restarted the machine and it's been doing it since. Any ideas?

Have you tried swapping PCI power connectors? Hope you have done it.
Then if not working, Open your faulty antminer S3 and swap the cables from blades to control panel.
You can also swap the control board of another working S3 for testing as a second option.
If now the other blades are working its a control panel or cable problem.

Now, open the front side (ASIC Chip side)heatsink and check for thermal paste lease or absence.
If its so, just buy some good thermal grease and apply, the blade will be ok.
Else its a clear case of RMA. Contact bitmaintech as soon as possible via email.
I assumed you are using a proper power supply.

Ok, so as I said before tried swapping PCI-E plugs to no effect...

Opened up the unit, swapped cables, and still the same problem.... Errors still occur on the same chain so it's either the chain itself that is dead or the control board itself. I'm not really keen to take out the control board out of the working unit just to test while it's running so well. I think I'll wait to hear back from support before I go further. Any ideas what the go is here? Do they send me a replacement part or do I need to send the unit back?
We can try to solve it if possible.
Tell me clearly the result of the below you already did.
Blade 1 -> Cable1 -> PORT1 in CP
Blade 2 -> Cable2 -> PORT2 in CP
You need to do this [1]
Blade 1 -> Cable1 -> PORT2 in CP
Blade 2 -> Cable2 -> PORT1 in CP
Further steps after the result of this. If possible attach images of Miner Status page before and after this mod.

If you are ordering a new unit they may send the replacement if its control panel with that.
For blade replacement I think they will only send if you are sending yours first.
If a return courier facility is there that should be utilized.
You should discuss with bitmain about various options if you are not able to fix this here.
As said earlier, time is money. So insists them to send first, you may give them some money as escrow for that.
Hope they will carry the shipping charge, I have no experience in sending for RMA.

Ok, so....

Blade 1 -> Cable1 -> PORT1 in CP
Blade 2 -> Cable2 -> PORT2 in CP

This results in Chain1 working fine and Chain2 with all dash's but the final one which is all dashes and one "o"

I switched the cables and put them in the same ports and blades with no change. Am trying Blade 1 -> Cable1 -> PORT2 in CP and Blade 2 -> Cable2 -> PORT1 in CP now Smiley
hero member
Activity: 784
Merit: 504
Just got my two S3's today! Setting up one however and ran into a problem, one chain is hashing away fine with oooooooo oooooooo but the other comes back with -------- -------- -------- -------o

What the hell does this mean?!? It worked ok on first run, restarted the machine and it's been doing it since. Any ideas?

Have you tried swapping PCI power connectors? Hope you have done it.
Then if not working, Open your faulty antminer S3 and swap the cables from blades to control panel.
You can also swap the control board of another working S3 for testing as a second option.
If now the other blades are working its a control panel or cable problem.

Now, open the front side (ASIC Chip side)heatsink and check for thermal paste lease or absence.
If its so, just buy some good thermal grease and apply, the blade will be ok.
Else its a clear case of RMA. Contact bitmaintech as soon as possible via email.
I assumed you are using a proper power supply.

Ok, so as I said before tried swapping PCI-E plugs to no effect...

Opened up the unit, swapped cables, and still the same problem.... Errors still occur on the same chain so it's either the chain itself that is dead or the control board itself. I'm not really keen to take out the control board out of the working unit just to test while it's running so well. I think I'll wait to hear back from support before I go further. Any ideas what the go is here? Do they send me a replacement part or do I need to send the unit back?
We can try to solve it if possible.
Tell me clearly the result of the below you already did.
Blade 1 -> Cable1 -> PORT1 in CP
Blade 2 -> Cable2 -> PORT2 in CP
You need to do this [1]
Blade 1 -> Cable1 -> PORT2 in CP
Blade 2 -> Cable2 -> PORT1 in CP
Further steps after the result of this. If possible attach images of Miner Status page before and after this mod.

If you are ordering a new unit they may send the replacement if its control panel with that.
For blade replacement I think they will only send if you are sending yours first.
If a return courier facility is there that should be utilized.
You should discuss with bitmain about various options if you are not able to fix this here.
As said earlier, time is money. So insists them to send first, you may give them some money as escrow for that.
Hope they will carry the shipping charge, I have no experience in sending for RMA.
legendary
Activity: 1235
Merit: 1202
Just got my two S3's today! Setting up one however and ran into a problem, one chain is hashing away fine with oooooooo oooooooo but the other comes back with -------- -------- -------- -------o

What the hell does this mean?!? It worked ok on first run, restarted the machine and it's been doing it since. Any ideas?

Have you tried swapping PCI power connectors? Hope you have done it.
Then if not working, Open your faulty antminer S3 and swap the cables from blades to control panel.
You can also swap the control board of another working S3 for testing as a second option.
If now the other blades are working its a control panel or cable problem.

Now, open the front side (ASIC Chip side)heatsink and check for thermal paste lease or absence.
If its so, just buy some good thermal grease and apply, the blade will be ok.
Else its a clear case of RMA. Contact bitmaintech as soon as possible via email.
I assumed you are using a proper power supply.

Ok, so as I said before tried swapping PCI-E plugs to no effect...

Opened up the unit, swapped cables, and still the same problem.... Errors still occur on the same chain so it's either the chain itself that is dead or the control board itself. I'm not really keen to take out the control board out of the working unit just to test while it's running so well. I think I'll wait to hear back from support before I go further. Any ideas what the go is here? Do they send me a replacement part or do I need to send the unit back?
hero member
Activity: 616
Merit: 500
Observation.
In one of the Batch 4 machines I got, there is a machine with a July 28 firmware date within the Kernel Log.

Anyone noticed that?  Is that an experimental firmware?

The cooling fans report constant 7000-9000 RPM range.  That doesn't look normal.  Recommendations?

Got my B4 S3's today.  Both of them have also the July 28 firmware ("Linux version 3.10.12 (xxl@ubuntu) (gcc version 4.6.4 (OpenWrt/Linaro GCC 4.6-2013.05 r38031) ) #140 Mon Jul 21 17:27:04 CST 2014").
Both of them run with very high fan speeds, much higher than the B1 units

3780 3960
4680 10560

The B1 units, at freq=225 run 1740 1620  and 1680 1500 rpm

The Bitmain support page does not mention this firmware.  Any clues why the high fan speed?
Jump to: