cooling 12 miners s5

[nevillebartos]

Hey guys, new to mining (aside for some investigation I did using my GPU).  I picked up 12 S5s to continue my preliminary investigations into BTC.  Before we get caught up on the economics, the game here is to collect first hand data and go from there down the track.  12 miners because that's the dollar amount I was willing to risk on this exercise, and will hopefully give me sufficient data.  If it's of any interest my current kw/h goes for about US$0.10.  I'd like to keep them on site at this stage, which basically means I have what should really be a comms cabinet to work with.

I calculated and vented for 10 air changes per minute (unobstructed).  This is nowhere near enough, I can run 6 PSUs and 2 miners mining staying within acceptable temps (30c inside w/24c ambients).  Setup is 2 miners and 1 PSU per rack, the chimney effect kicks the goolies out of the top 6 if I power them all up.  I'm interested in solutions but some info on the heat I'm trying to remove would be useful.

I can relatively easily jack up the airflow, but I'm not convinced anything short of a wall of fans is going to do job.  I'm familiar with water cooling and could potentially run a loop through the neighboring room and from there vent to outside, but I'm not sure how much rad I'm looking at.  I could run through a water chiller, but I suspect they PC types although overkill for overclocking may be too small for this job.  Finally there's the aircon route, but I don't know the BTU's I'm looking for, and I suspect it'll kill the margins anyway.

My current intake is passive, I'd have a crack at firing air in from that and possibly something else firing upwards, just don't think it's gonna cut it.  So what do we do here guys?? Got 10 S5s sitting on the floor looking very sorry for themselves 

[1715012346]

1715012346

[1715012346]

1715012346

[1715012346]

1715012346

[notlist3d]

Hey guys, new to mining (aside for some investigation I did using my GPU).  I picked up 12 S5s to continue my preliminary investigations into BTC.  Before we get caught up on the economics, the game here is to collect first hand data and go from there down the track.  12 miners because that's the dollar amount I was willing to risk on this exercise, and will hopefully give me sufficient data.  If it's of any interest my current kw/h goes for about US$0.10.  I'd like to keep them on site at this stage, which basically means I have what should really be a comms cabinet to work with.

I calculated and vented for 10 air changes per minute (unobstructed).  This is nowhere near enough, I can run 6 PSUs and 2 miners mining staying within acceptable temps (30c inside w/24c ambients).  Setup is 2 miners and 1 PSU per rack, the chimney effect kicks the goolies out of the top 6 if I power them all up.  I'm interested in solutions but some info on the heat I'm trying to remove would be useful.

I can relatively easily jack up the airflow, but I'm not convinced anything short of a wall of fans is going to do job.  I'm familiar with water cooling and could potentially run a loop through the neighboring room and from there vent to outside, but I'm not sure how much rad I'm looking at.  I could run through a water chiller, but I suspect they PC types although overkill for overclocking may be too small for this job.  Finally there's the aircon route, but I don't know the BTU's I'm looking for, and I suspect it'll kill the margins anyway.

My current intake is passive, I'd have a crack at firing air in from that and possibly something else firing upwards, just don't think it's gonna cut it.  So what do we do here guys?? Got 10 S5s sitting on the floor looking very sorry for themselves 

You have 2 options if your lucky you can look into evaporation cooling.   It is cheaper then regular ac.

If your not lucky you are like me and evaporation cooling does not work because of your humidity.    If this you use lots of CFM of fan's.   Were talking about a lot of cfm intake and exhust for 12 S5's.

Your cheapest option is find a hosting center for them.

[Tupsu]

.  If it's of any interest my current kw/h goes for about US$0.10. ...

You're going the wrong way. Miner does not need to be at home.

Read this topic

https://bitcointalk.org/index.php?topic=1137485.0

[chiguireitor]

As a pro-mining guy (in the sense that i think everyone should mine, not just datacenters) i applaud you.

Work your CFMs on the extraction point if you're on a sub 30ºC zone. Try to not mix your intake and exhaust as that will create a positive feedback of hot air and you'll be unable to cool the miners.

Each S5 needs 200 CFM of extraction power, when you break the 8 miners barrier free flow doesn't works.

[GigaBit]

I have a dozen miners too.... S1's, S3's, an S4 and 4 S5's... I am probably hotter than you due to my older process miners still running.
My secret weapon is a 3,000 CFM blower in a small room.  Takes me from 35C down to 30C in less than an hour with 35C outside.
If you have proper ventilation, you don't need AC... that's double true if you live in a Northern country.
All I have is a 3,000 CFM Blower which is sold by NorthernTool and a $50 24" walmart fan to feed the mine with airflow from another room.

Again, it's loud but if you're a miner, loud noises are part of the trade.

http://screencast.com/t/mjfFXEQIp - Blower

It is loud as hell but I only needed to run it a total of 28 hours since the summer started.
Without it, I would have had to shut everything down anything S3 and under...
I live in Canada though, where the hottest summer temps so far was 35C outside.
Even at the temp it was a battle keeping it cool in the mine, the blower was the hero.
For about $150, was a great investment for airflow.

[nevillebartos]

Thanks guys!  I do have use of relatively large amounts of space out of town where there's some good potential for power and cooling (none of it currently in place).  The general idea was to run these guys at home where I can keep a close eye on everything, and in the event of scale up move it all out there. 

It's winter now and ambients are 0 - 15c, noise becomes a non issue, for the minute I'm planning to move the 12 I have out there now. I'd be trading heat / noise issues for less reliable power (for the minute), and potential network issues, plus an hour drive if anything needed to be checked.  Come summer I'd be looking at ambients of 30 - 45c, large space and lack of insulation would bring new heat challenges to the table, but I've got a few ideas if I were to put em on water.  I expect much will have changed by then anyway, so probably best to cross that bridge when the time comes.

Hopefully will have everything online soon, will let you guys know what I come up with!

[notlist3d]

Thanks guys!  I do have use of relatively large amounts of space out of town where there's some good potential for power and cooling (none of it currently in place).  The general idea was to run these guys at home where I can keep a close eye on everything, and in the event of scale up move it all out there. 

It's winter now and ambients are 0 - 15c, noise becomes a non issue, for the minute I'm planning to move the 12 I have out there now. I'd be trading heat / noise issues for less reliable power (for the minute), and potential network issues, plus an hour drive if anything needed to be checked.  Come summer I'd be looking at ambients of 30 - 45c, large space and lack of insulation would bring new heat challenges to the table, but I've got a few ideas if I were to put em on water.  I expect much will have changed by then anyway, so probably best to cross that bridge when the time comes.

Hopefully will have everything online soon, will let you guys know what I come up with!

Power is not that hard to be honest.  But if you do not know what your doing get an electrician (in some places its required by code aswell).   

Cooling is a little harder.   You will have to refine it a little bit depending on the location and how you set up the miners.

[aarons6]

Hate to be a buzzkill but at 10 cents a killawatt you're better off on unloading them before the Antminer S7 comes out and it is a larger supply due to unloaders.

at 10c a kw/h he is good for around 120,000,000,000 difficulty.. so no worries there..

for OP..
first thing i would do with the s5s is disconnect the PWM (usually blue wire) from the fans. if you are careful you can pull them from the plug with a needle and not damage the wire so you can plug it back in later if you want..


this will set them to run full speed and lower the temps from what they would be if left alone.

[nevillebartos]

Hate to be a buzzkill but at 10 cents a killawatt you're better off on unloading them before the Antminer S7 comes out and it is a larger supply due to unloaders.

Well like I said let's not get caught up on the economics right now, few potential ways to toy with the kw/h price but difficult / not worth the effort without some results at the very least, and scale makes a big difference as well. That all really comes down to margins though and that isn't my focus at this time.  I'm after first hand operating experience and actual data of my own to work with, I've done web searches / calculators etc.. which have been a good guide, but it's not the same.  Of course, I don't really want to operate at a loss, so cost of kW is still a factor to some extent, as are the delays in getting up and running with the expected s7 release.  Unloading would definitely not be a terrible idea $ wise, but the s5s should get me much of the info I need, hopefully in time to make a decision come s7.

at 10c a kw/h he is good for around 120,000,000,000 difficulty.. so no worries there..

for OP..
first thing i would do with the s5s is disconnect the PWM (usually blue wire) from the fans. if you are careful you can pull them from the plug with a needle and not damage the wire so you can plug it back in later if you want..


this will set them to run full speed and lower the temps from what they would be if left alone.

Yeah chopped the PWM wire first thing, I'd seen some stuff about the s5 cutting the fan on network loss and boards getting fried, and had no real reason not to run them max anyway.   I've got a box load of 120mm collected up from PC builds over the years, they vary in quality but might chuck em on in pull for that little bit extra, not sure if the designs all that well geared towards push/pull, but can't hurt, I think.