Can ASICS kills power supplies?

[paramind22]

I was just reading this thread and I was wonder if ASICS can be hard on power supplies.

Server is Killing Power Supplies
http://www.dslreports.com/forum/r29071814-Server-is-Killing-Power-Supplies

According to a guy who sells HP 1200 revamped for six pin outlets for Antiminers, it could be something
else besides the power supplies that is creating the problem of dying power supplies.

[1715311615]

1715311615

[1715311615]

1715311615

[gt_addict]

I don't really understand what you're going on about. Asics use electricity, servers use electricity, computers in general use electricity. The main point is that if you have a 1000w computer you don't buy a 1000w psu, you buy a 1200-1400w psu so you are not putting the psu under too much strain. You overcompensate for your needs to avoid premature failures and unsafe scenarios.

[paramind22]

I don't really understand what you're going on about. Asics use electricity, servers use electricity, computers in general use electricity. The main point is that if you have a 1000w computer you don't buy a 1000w psu, you buy a 1200-1400w psu so you are not putting the psu under too much strain. You overcompensate for your needs to avoid premature failures and unsafe scenarios.

I was using a HP 1200 for an old Antimer S1.   The only possible abuse that it might have got is that it was on a wood surface and probably should have been on a grill or had more bottom surface exposed to air circulation.

[NotFuzzyWarm]

And what are you powering with the supply now? Still the s1 or now an s7 or s9? Please do remember that the HP 1200 CS server supply is derated to 800w when fed 110vac.

The early s7's pull around 1,100-1250w at stock speeds which is at the limit of the HP's fed 220v and powering an s9 with one even if fed 220v is out of the question.

[paramind22]

And what are you powering with the supply now? Still the s1 or now an s7 or s9? Please do remember that the HP 1200 CS server supply is derated to 800w when fed 110vac.

The early s7's pull around 1,100-1250w at stock speeds which is at the limit of the HP's fed 220v and powering an s9 with one even if fed 220v is out of the question.

Thanks.  It was only on the S1.  I really think I may have killed it with overheating.  It's good to know it has a limit at 800w at 110vac as there is that temptation to use the two other plugs it has.

[takagari]

miner will not kill it directly
But anything used at a heavy load for long lengths will be at risk of dieing.

server grade psu's are better equipped for it.
but none are really meant to be run even at 80% for months on end...

[Amph]

chinese psu surely will burn first, but a good psu should automatically shutdown when you exceed the maximum power that cna be handled by your units

i tried to go above the maximum power rated on my psu and the system reboot, this was tried with gpu rig

it should work with asic too, but maybe with asic it does not work because you have not an entire pc connected to it, not sure about this

but i remember some pcie cable fried(here ont he forum) because someone exceeded the power a single rail

[NotFuzzyWarm]

chinese psu surely will burn first, but a good psu should automatically shutdown when you exceed the maximum power that cna be handled by your units

The jab @ being made in China is pure wrong considering all of the IBM, HP, Dell, etc server supplies are sourced from China. A crappy PSU can be made anywhere. Just depends on what the customer (IBM, HP et al) specifies it's performance, lifetime, design max operating conditions, etc to be and how thorough the design testing called for is.

If the customer tells the likes of Delta (the major supplier of IBM & HP branded server PSU's) to make a marginally rated PSU for them they will be more than happy to do that. Tell them to make it bulletproof and they will do that and do a damn fine job of it.

Frankly, for mining the only PSU's I steer clear of are all of the Consumer grade ones -- even the best of them such as Corsair, Thermaltake, Lepa etc. They just are not made for the operating envelope that miners require.

[Finksy]

chinese psu surely will burn first, but a good psu should automatically shutdown when you exceed the maximum power that cna be handled by your units

The jab @ being made in China is pure wrong considering all of the IBM, HP, Dell, etc server supplies are sourced from China. A crappy PSU can be made anywhere. Just depends on what the customer (IBM, HP et al) specifies it's performance, lifetime, design max operating conditions, etc to be and how thorough the design testing called for is.

If the customer tells the likes of Delta (the major supplier of IBM & HP branded server PSU's) to make a marginally rated PSU they will be more than happy to do that. Tell them to make it bulletproof and they will do that and do a damn fine job of it.

Frankly, for mining the only PSU's I steer clear of are all of the Consumer grade ones -- even the best of them such as Corsair, Thermaltake, Lepa etc. They just are not made for the operating envelope that miners require.

+1 on everything, especially the last line.  There's not a make of off-the-shelf ATX PSU's, even EVGA, that will stand the abuse of 90+% load in hot ambient temperatures to the extent that nearly any server-grade PSU will. Some better than others, perhaps, but even ignoring the obvious $ for $ comparisons, server PSU's will stand up to harsher more demanding environments, and the voltage output is considerably more stable with load.

[VRobb]

Hey Fuzzy, Fin,
I really respect your opinions.  What would you say to using a 'high end' consumer psu - one listed above - running at most 60% of max load instead of 80+%?  Like a 1600W psu running a 900W miner?

[NotFuzzyWarm]

Hey Fuzzy, Fin,
I really respect your opinions.  What would you say to using a 'high end' consumer psu - one listed above - running at most 60% of max load instead of 80+%?  Like a 1600W psu running a 900W miner?

Not ideal but if you already have the PSU then it should do well and be in its sweet-spot for efficiency. Again, even the high-end ATX supplies are not made for supplying only 12vdc at rated wattage. Part of that rating includes a couple hundred watts of 5v and 3,3v power expected to be used by CPU's, memory, etc. Throw in the usually too-thin PCIe cables used, the price of ATX supplies, the fact that many high wattage ones have 2 or more 12v rails vs a single one, and things are not good.

[VirosaGITS]

Hey Fuzzy, Fin,
I really respect your opinions.  What would you say to using a 'high end' consumer psu - one listed above - running at most 60% of max load instead of 80+%?  Like a 1600W psu running a 900W miner?

Not ideal but if you already have the PSU then it should do well and be in its sweet-spot for efficiency. Again, even the high-end ATX supplies are not made for supplying only 12vdc at rated wattage. Part of that rating includes a couple hundred watts of 5v and 3,3v power expected to be used by CPU's, memory, etc. Throw in the usually too-thin PCIe cables used, the price of AXT supplies, the fact that many high wattage ones have 2 or more 12v rails, and things are not good.

My 1600w PSU does 100% 12v just fine. Its rated for 1599.6w on the single 12v rail. Don't buy a shitty ATX psu and you will be plenty fine, minus the 300% markup/watt versus a server PSU, that is.

[VRobb]

Indeed, but as fuzzy surmised I already have the psu and so it goes...
Thanks all for the input. Mine on!

[Finksy]

Virosa, I don't doubt your experience one bit.  We're making broad generalizations here, and generally speaking ATX PSU's don't last as long or as consistently as server PSU's at any level of load in my and most others' experiences (hello S2 PSU's!). You are definitely right that quality goes a long way, but you pay dearly for it.  Also, ATX PSU's are typically built for peak draw demands and aren't expected to operate at over 90% load for extended periods of time. Server PSU's are designed to operate in steady states of load, and I imagine even in server settings that load level is expected to be considerably higher than that of a typical PC. It's about intended usage.  Just out of curiousity, what is the voltage at the plug that your ATX PSU puts out under 100% load? In my experience the voltage starts to drop off above even 50-60% load. I wouldn't be surprised that @ 100% you are at or even slightly under 12.0V.

One thing ATX PSU's have as an advantage over server PSU's is their fans/cooling and resulting ATW efficiency.  80+ efficiency ratings do not take fan draw into account, and since ATX PSU's aren't as restricted in form factors as their server counterparts, they can use lower power fans to manage similar wattage ratings.

[VirosaGITS]

Virosa, I don't doubt your experience one bit.  We're making broad generalizations here, and generally speaking ATX PSU's don't last as long or as consistently as server PSU's at any level of load in my and most others' experiences (hello S2 PSU's!). You are definitely right that quality goes a long way, but you pay dearly for it.  Also, ATX PSU's are typically built for peak draw demands and aren't expected to operate at over 90% load for extended periods of time. Server PSU's are designed to operate in steady states of load, and I imagine even in server settings that load level is expected to be considerably higher than that of a typical PC. It's about intended usage.  Just out of curiousity, what is the voltage at the plug that your ATX PSU puts out under 100% load? In my experience the voltage starts to drop off above even 50-60% load. I wouldn't be surprised that @ 100% you are at or even slightly under 12.0V.

One thing ATX PSU's have as an advantage over server PSU's is their fans/cooling and resulting ATW efficiency.  80+ efficiency ratings do not take fan draw into account, and since ATX PSU's aren't as restricted in form factors as their server counterparts, they can use lower power fans to manage similar wattage ratings.

Was more to fill the gap than throw out the previous recommendations.

12.04~ ish, they dont budge much at load, +- 0.02v, And that may just be the multimeter i use. These PSU have 10years warranty, i've been using the f out of them for years(often at TRUE DC 100% load), its a love story. These are EVGA G2/P2/T2. I pity anyone that get cheap ATX PSUs, it seem a lot of people cheap out on their ATX PSU, pay the price, and are better told to just not do it. Thats not ATX PSUs as a whole , dont give me dearie a bad name! But i do have hard reasons to use ATX, and its not to save money lol.

 

[adaseb]

I got some cheap Corsair 750W/Thermaltake 850W/Revolion 87+ PSUs that are running for 3 years straight without any issues.

However I usually run them at 80% load or less.

Only issue I had was with those modular PSUs which split into 2 PCIe connectors and if you drew close to 150W each on each connector, it drew 300W at the modular side and had some melting issues there.

With GPUs you shouldn't have any issues with those but they aren't good if you are pulling the entire 150Watt from each connector.


https://i.imgur.com/dOS17tM.jpg

[gt_addict]

In reference to the ATX vs Seever psu, I'm currently making the switch to a server psu. Primarily because of the cost of it compared to an ATX and secondly because of the reasons listed above.

My current EVGA 1000w P2 is drawing 650w at the wall so roughly at peak efficiency but the voltage has dropped to 11.93v. With the server psu I'm able to adjust the voltage underload so provided a true 12v to the S9 I have.

[gt_addict]

In reference to the ATX vs Seever psu, I'm currently making the switch to a server psu. Primarily because of the cost of it compared to an ATX and secondly because of the reasons listed above.

My current EVGA 1000w P2 is drawing 650w at the wall so roughly at peak efficiency but the voltage has dropped to 11.93v. With the server psu I'm able to adjust the voltage underload so provided a true 12v to the S9 I have.

Server PSU's are great but only if you aren't having them in your home, the noise is just not home friendly.

Hence my thread tying to devise a way to make them quieter

[takagari]

In reference to the ATX vs Seever psu, I'm currently making the switch to a server psu. Primarily because of the cost of it compared to an ATX and secondly because of the reasons listed above.

My current EVGA 1000w P2 is drawing 650w at the wall so roughly at peak efficiency but the voltage has dropped to 11.93v. With the server psu I'm able to adjust the voltage underload so provided a true 12v to the S9 I have.

Server PSU's are great but only if you aren't having them in your home, the noise is just not home friendly.

Hence my thread tying to devise a way to make them quieter

Everyone keeps saying this, but forgetting the kits finsky and others sell
Dual dps-2000 can give you 4000WATTS and is essentially silent.
you can put 2 or 4 120mm fans of your choice on them EASILY and push what ever amount of air for which ever noise level you select, without issue.

And they are cheap to replace, so having spares sitting around for when one drops offline is a no brainer.

[gt_addict]

I fully appreciate what you're saying about kits but I don't need that much power and can't use that much either. I'm trying to keep within the sweet spot for efficiency of the psu. Hence why I've only bought a dps1200 platinum. Well two of them one to experiment on and another to use. Already found the voltage adjustment on it.