ANTMINER S3+ Discussion and Support Thread

[philipma1957]

bitmain has gained greatly in value from yesterday up to today.  the sp30 is a bust.  no one knows had big of a bust yet but it is a bust.  the ceo the the company admitted so on the sp30 thread.

 so will it hash at 4th and pull 2400 watts.

 it is not going to do the promised 6th.

we also do not know what the efficiency will drop to.  .6 watts  .7 watts?

 this means the s-3 has jumped up in value.  How big a jump depends on how bad the sp30 turns out to be.

As a s-3 owner this will also help with the diff.

This is a very big deal and has serious consequences for the home miner.

He forgot to say the CEO also said that this would not impact delivery dates.

So if its slightly less than 6TH and slightly more than 0.5watts the overall effect will not be that different to everyone else than if they had delivered on spec.

 

you are correct  if it is 5.5th and .55 watts and ships on time   not much of an effect.  It still makes the s-3 look more valuable.

The questions are :

6th = 4th or 5th

more importantly does the .46-.5 watt move to .6-.7 watts?

[bobsav2121]

<snip>

From the datasheet:


Note the line for natural convection, and how current begins to drop off as ambient temperature rises.

So opening them, putting a box fan behind them in a AC room, should help them alot being stable, and better OC performance if possible. right .?

Your chart is meaningless unless we know the devices current requirement.

The chip is spec'd at 9.33W @ .75V - you can do the math on that amiright?

You need to know what the miner is drawing on the device. what the device is rated for is only half the question and not the answer.

[jimrome]

The chip is spec'd at 9.33W @ 7.5 0.75V - you can do the math on that amiright?

thats about 12.5A per chip with 2 chips per regulator - only about 25A draw or so which is well within the limits of the TPS53355. Either bitmain got some crappy/knockoff regulators or the problem is something else.

in the S1 the TPS53355 was able to push 35A without issues in most miners.

Without issues? I guess that depends on how lucky you were with the batch of regulators used for your particular S1s. This part was also used in the bitfury "25GH" boards and offered a ton of room for overclocking - provided that some airflow was provided.

Thing is, the part may have outperformed the specification in the past, or in certain batches, but that's no guarantee of future performance - which is why they provide min/typ./max info in the first place. Assuming that a critical part is going to perform above spec is either ballsy or retarded depending on your POV.

Yes, 25A is under the 30A "max" spec'd by TI, but to get the whole picture you need to consider thermal derating which I mentioned in an earlier post. I have some thermal pics at home, I'll upload them if there's any interest.

[TracerX]

The chip is spec'd at 9.33W @ 7.5 0.75V - you can do the math on that amiright?

thats about 12.5A per chip with 2 chips per regulator - only about 25A draw or so which is well within the limits of the TPS53355. Either bitmain got some crappy/knockoff regulators or the problem is something else.

in the S1 the TPS53355 was able to push 35A without issues in most miners.

Without issues? I guess that depends on how lucky you were with the batch of regulators used for your particular S1s. This part was also used in the bitfury "25GH" boards and offered a ton of room for overclocking - provided that some airflow was provided.

Thing is, the part may have outperformed the specification in the past, or in certain batches, but that's no guarantee of future performance - which is why they provide min/typ./max info in the first place. Assuming that a critical part is going to perform above spec is either ballsy or retarded depending on your POV.

Yes, 25A is under the 30A "max" spec'd by TI, but to get the whole picture you need to consider thermal derating which I mentioned in an earlier post. I have some thermal pics at home, I'll upload them if there's any interest.

Please upload your thermals, I'm interested to see them.

[dsheaffer]

Anyone have a clue as to when the next batch of S3s is going on sale?

[bitfor8888]

Got mine yesterday.  One work fine at 449G per pool stat., but the other one is not working  . 

Half of the ASIC was xxxxx.  I tried to change psu, cable, shutdown the one working and switch to this one (in case my new RM1000 is not working or not have capacity), reset it, change ethernet cable.  I even change the place 20 km apart!  Nothing work. 

I did show it got all 3 workers alive and the first one even get work (but discard them all).  Pool stat. said it was touched in 11 hours ago but not lately.  I double check the tailoring is the same (DHCP etc.) and hence it might not be mine.

(I did open the case see whether there is any thermal paste.  Nothing.  Hence not for that reason.)

Have to get back to Bitman for maintenance.  A bit disappointed.  6 S1 since April is all fine and hence may have too high expectation on its working.

[bytechanger]

I noticed the intake fan on one of my units does not work.  I wonder if I should open the unit up and ensure the cable is properly attached...

Does that void warranty? 

[halcy0n3]

There are no stickers on the screws to break (you know the ones that are placed there essentially to prove that you undid the screw)

And there are no electrical connections to the case to sense that you opened it.

I cant imagine that it could be proved you opened the case, supposing that youre not allowed to.

That said, i know from past experience that when there are problems bitmain support will direct you to re-seat electrical connections. Cant imagine that it would void anything, its just the case.

Im confident the case is there to look cool and improve airflow

go nuts

[bit_NINj4]

wow, interesting, this is exactly what I hit in 100d on S1 (on average). they must have some formula to predict this.
14% in 100 days is 60% a year. Don't tell me you hit this on the stock market on anything else.

I would say BitMain is well aware of our margins, I am tracking everything in a spreadsheet (IRS) for S1, S2, S3..  they all seem to be on that mark, S1's early appearance really gave it the edge. My first S1 I paid 1.9BTC ($1554.96) ran it for 180 days and then sold it on ebay, net was $634.61 / $1554.96 = 41% Net profit. My first S2 is @ day 129 and is currently $1012.95 in the black, other 2- S2's are 64 days in and @ 49% ROI. SP10's are 63 days in and @ 37% ROI ;(

Just like the market you have to hedge your bets, keep your eye on the future offerings, and get IT while you can! Basically I have watched this grow from a $1500 hobby into a 30K small business. Bitmain is helping me keep that going by NOT raping every nickel of value out of their offerings.

THIS

[64dimensions]

Here is a my take on the DC-DC regulator issue:

This part can cool off in 3 ways: conduction, convection and radiation. A back-of-the-envelope calculation shows that you can drop radiative cooling.

I think the new S3 design has a (slight?) flaw in the cooling of the DC-DC regulator part. With the cover on, the cooling of this part depends on airflow.

The problem is that the closer to the exhaust, the warmer the air is flowing over the regulator. So from back to front the regulator chips are running hotter. You could could use heat sinks to try to bandaid the problem but you probably could only place them on the last few chips because you don't want to block the air flow. So the issue is are the back chips running too hot?

I think the ultimate solution is removing the top cover, but what does that do to everything else and the warranty?

[macsheadroom]

Got my late batch 2 miners today, All seems well except these "discarded".

On the S1 I have after 24 hours, 23000 accepted and 6000 discarded.

On the S3 I have after 6 hours, 6000 accepted and 75000 discarded.

Software is 3.10.12 on both, Why like this?

 

[mrpark]

Here is a my take on the DC-DC regulator issue:

This part can cool off in 3 ways: conduction, convection and radiation. A back-of-the-envelope calculation shows that you can drop radiative cooling.

I think the new S3 design has a (slight?) flaw in the cooling of the DC-DC regulator part. With the cover on, the cooling of this part depends on airflow.

The problem is that the closer to the exhaust, the warmer the air is flowing over the regulator. So from back to front the regulator chips are running hotter. You could could use heat sinks to try to bandaid the problem but you probably could only place them on the last few chips because you don't want to block the air flow. So the issue is are the back chips running too hot?

I think the ultimate solution is removing the top cover, but what does that do to everything else and the warranty?

Can you post a photo of the chips in question. Removing the case would not be a good idea unless you had airflow directly hitting the chips. The tunnel helps move the heat across the chips.

[midnightcruze]

Got my late batch 2 miners today, All seems well except these "discarded".

On the S1 I have after 24 hours, 23000 accepted and 6000 discarded.

On the S3 I have after 6 hours, 6000 accepted and 75000 discarded.

Software is 3.10.12 on both, Why like this?

 

I am seeing the same thing.  I wonder if they just have the two values reversed?  According to the pool I am getting share credit at the expected rate.... so I'm not sure what the high discard means if it's real?

[real1st]

Hi All,

Bought 2 S3.  Arrived today.

1.  First one, turns on, immediately non-stop ringing.  Never turns off. 
2.  Second one, turns on, no ringing, but the temperature of the first blade says 2C while the second is 44c.

They are both being tested with the SAME Corsair 1000W PSU.  The ringing is the worst part, but the temperature thing also concerns me.

Anyone know about this?  I am not talking about the intermittent beep that happens, it happens on the second unit, but the first unit is constant ringing, non-stop.

Thoughts?

real1st

[macsheadroom]

Got my late batch 2 miners today, All seems well except these "discarded".

On the S1 I have after 24 hours, 23000 accepted and 6000 discarded.

On the S3 I have after 6 hours, 6000 accepted and 75000 discarded.

Software is 3.10.12 on both, Why like this?

 

I am seeing the same thing.  I wonder if they just have the two values reversed?  According to the pool I am getting share credit at the expected rate.... so I'm not sure what the high discard means if it's real?

Yea pool side were all good, I think the numbers are transposed or some goofy thing like that.

[nanook7]

Quick Question.

Any one else having their S3 report only one working blade.

I've tried
-different power supply
-multiple reboots
-different pci-e plugs
-Currently power all 4 pci-e plugs with a 1600 watt supply

I've filed a ticket with Bitmain but nothing yet.

Anyone have any suggestions short of opening the box and checking connections.

http://i819.photobucket.com/albums/zz114/oneilljf/s3_bad_blade_zps6ef30a19.jpg

[Brassguy]

Quick Question.

Any one else having their S3 report only one working blade.

I've tried
-different power supply
-multiple reboots
-different pci-e plugs
-Currently power all 4 pci-e plugs with a 1600 watt supply

I've filed a ticket with Bitmain but nothing yet.

Anyone have any suggestions short of opening the box and checking connections.

Bitmain will tell you to open up the case and reseat the plugs from control board to hashing board. 99% of the time that will fix your problem.

[Biodom]

Re firmware flashing: it gives the checksum before flashing, but what I supposed to compare it to?
also, "image size 7.5mb (8.1 mb available)"-what is this supposed to mean?

[faetos]

Re firmware flashing: it gives the checksum before flashing, but what I supposed to compare it to?
also, "image size 7.5mb (8.1 mb available)"-what is this supposed to mean?

The firmware hash should have a corresponding hash on their site. The two should match ensuring integrity. The 8.1mb available is the space available on the system disk. I ran a 'df -h' on the files system and this is what it showed 29M available in /tmp.

I've also been trying to track down how to turn off the system bell (beep) and haven't found anything. It's loud and annoying. I'll let you know when I find out how to disable it completely through software.

[Biodom]

Re firmware flashing: it gives the checksum before flashing, but what I supposed to compare it to?
also, "image size 7.5mb (8.1 mb available)"-what is this supposed to mean?

The firmware hash should have a corresponding hash on their site. The two should match ensuring integrity. The 8.1mb available is the space available on the system disk. I ran a 'df -h' on the files system and this is what it showed 29M available in /tmp.

I've also been trying to track down how to turn off the system bell (beep) and haven't found anything. It's loud and annoying. I'll let you know when I find out how to disable it completely through software.

thanks, but why size discrepancy-when I download, it shows as a 8.1 mb file, but once I upload to s3 , then it says that it is 7.5mb.
besides, i don't see the checksum posted anywhere on their site.