Pages:
Author

Topic: Too dangerous to draw more than 1200w on Corsair 1200 PSU? (Read 2871 times)

sr. member
Activity: 348
Merit: 251
WTF is going on with the 180-190V brownout though?

OK, my shelf rig uses a different approach to mackminer's, and he's pulling 8.7 GH/s against my 7.4 GH/s (according to arsbitcoin, oddly enough we're both using the same pool unless there are two mackminers...) but I have ONE cable running into my triple-board triple-PSU rig.

My PSUs are cheap, for all the usual reasons. Quality manufacturers, but anything over 1000W gets really expensive really quickly. So Shelf 1 runs two Cooler Master Silent Pros (1000W, 800W) and a Corsair 850W unit. None of the three PSUs individually pull more from the wall than their rated output, and since the inefficiency is on the conversion side, 10% inefficiency means the safe rated 1000W pull from the biggest PSU would pull 1,111W from the wall. That'd be OK. But none of the rigs do.

All three go into one power cable, which has the UK equivalent of a Kill-A-Watt stuck on the end. When ArsBitcoin isn't being DDoS'd, and the shelf rig can kick arse, that 'kill-a-watt-alike' shows between 2200 and 2300W.

Given the room has a single dual-socket outlet, each rated at max 13A at nominal 240V, I'm not 100% comfortable about the current draw on those cables. There's a 13A extension cord (20M) between the socket and the rig too!!! Yes, the cable gets mildly warm, but not as hot as when I tried to run the same sort of rig on a 10A home-made extension cable - the wire was too hot to touch and the female socket nearly caught fire. I'm using solid-conductor ring-main stuff now...

But even with the idiotic 10A extension, the smoking sockets and the floppy, melting wires... the voltage NEVER dropped lower than 220V. You'd expect a pretty hefty voltage drop from the resistance of the under-spec extension cable and the heating of the wire causing further resistance. But it wasn't much.


Mack... if your supply is showing 180-190V on a 240V main then either your 'kill-a-watt' is fucked or you need to get the circuit the miners are on looked at. Especially when you're investing in quality kit and 8 kW air conditioning units. Yes, the best quality PSUs are probably the only things that can cope with brownouts like that, but what happens when the input voltage drops? Well, the input current draw is increased to provide the same power. The PSU may be built to ramp up the amperage from your '240V' supply which has suddenly dropped to 180V... but that's the voltage dropped by a quarter. Hence to maintain the same power, the PSU will draw 4/3 more amps of current (sorry if maths is duff but I'm a bit wasted).

Going back to my situation, with around 10A steady on a 13A mains line, I'm in the safe zone. If my 240V supply dropped to 180V, then the PSUs would pull 10 * 4 / 3 = 13.333A. Hopefully the fuse in the extension cord would pop and 'boom' my miners would stop. Hopefully the PSUs would cope with a sudden loss of power and not fuck up all my GPUs.

But 13.3A on a domestic circuit is above spec. If you're really seeing 180V input and you're doing the same sort of thing as me, then you're either popping fuses every time it happens, or your kill-a-watt is fecked, or you're reading it wrong. That sort of voltage brownout would make your PSUs work like bastards to deliver the power your rigs are pulling. Fuck knows if the air-con could even cope with a voltage drop like that.

Where are you reading those figures from? I've got cheap (£10) Maplin own-brand 'kill-a-watt' devices and I've not had a duff one. I'm seriously taking the piss with the electrical setup in my 200 yr old UK house (the consumer unit still has Bakelite fuses with hand-coiled wire in them, for example) and I've built my own ring main for the office... I'm eating a LOT of power and *some* cables are getting warm. But I've *never* seen 180V at the outlet. The only time I have problems is when the knackered battery in my UPS decides to try to kill my Mac Pro and all the big screens - even a random spike like *that* doesn't affect the miner rigs - every one on a surge protector and kill-a-watt, either a power strip version or an individual socket.


OK, I've had enough booze, TL;DR - when your volts go down, a regulated PSU will pull more amps. If you're currently (hehehe) pulling 10A on a domestic main, and your voltage drops from 240 to 180, then you WILL either blow a fuse, trip a breaker, blow a PSU, blow an entire rig, or set fire to something. You've got me worried now - I'm checking the voltage on all my kill-a-watts.  Shocked Shocked Shocked

I really find these power supplies confusing. I understand the basics of electricity and stuff, I don't really find psu's that interesting to go to the trouble of understanding them fully. I'm waiting for the electricity company to get back with their findings after measuring it for a week or so. I will read your post more carefully tomorrow when I have slept though because I am concerned about some of the things you have mentioned.
legendary
Activity: 1260
Merit: 1000
I don't think you guys understand how the efficiency rating on a PSU works.

If you are drawing ~1220W from the wall on a 1200W PSU, you are more than ok.  At 85% efficiency, your PSU is delivering:

1220 * .85 = 1037w to all your components.  Your PSU is rated to 1200W, so you can draw up to ~1400w from the wall on a 1200w rated PSU.  As your efficiency increases, you can draw less from the wall, since you'll be delivering more of the power drawn from the wall to the components.  The power rating on a PSU is how much power it OUTPUTs not how much power is going in.

*EDIT* OOps... didn't see the second page.  Someone already beat me to it!
donator
Activity: 1218
Merit: 1079
Gerald Davis
Yes, but the 1200W DC rating is the addition of loads on the 12v, 5v and 3.3v rails. You will basically only be drawing power from the 12v rails.  To be sure, check the maximum specced load on the 12v rails (in watt or ampere, its always written on the side).

Then again, for an el cheapo 350W PSU, this might make a big difference as up to half of that  "350W" might only be delivered on the 5v and 3.3v, but a decent 1200W is probably all about 12v anyway, Id be shocked if it was specced for less than 90 amp (1080W DC)

Good points.

One should always look at the power distribution chart (often put on a sticker right on the PSU).

Here is an example of a near perfect PSU (Enermax 1200W platinum.  80Plus-platinum.  Up to 94% efficient and 100% of load available on 12V rails)
http://www.enermax.com/home.php?fn=eng/product_a1_1_2&lv0=1&lv1=52&no=174

Hopefully someday we will see ATX standard updated to drop the near worthless -12V rail, and 5V rails.  The requirement for 12V rails could be increased (i.e. at least 75% of rated load).  No reason that future storage drives can't be driven on just 3.3V  rails.  It would simply PSU designs.  We have changed PSU design requirements multiple times in the past.
hero member
Activity: 518
Merit: 500

That's really good to know. Just to clarify the max is 1200 dc.

Yes, but the 1200W DC rating is the addition of loads on the 12v, 5v and 3.3v rails. You will basically only be drawing power from the 12v rails.  To be sure, check the maximum specced load on the 12v rails (in watt or ampere, its always written on the side).

Then again, for an el cheapo 350W PSU, this might make a big difference as up to half of that  "350W" might only be delivered on the 5v and 3.3v, but a decent 1200W is probably all about 12v anyway, Id be shocked if it was specced for less than 90 amp (1080W DC)

Quote
Would I be correct to assume that a good PSU would have a cutoff aswell as opposed to letting itself blow up.

As mentioned above, yes. It even works fine on my lowly antec 500W
donator
Activity: 1218
Merit: 1079
Gerald Davis
Are you drawing 1220w at the wall? Keep in mind the ratings are for power delivered to the PC at 12v (and some at 5 and 3.3v). Power supplies are not 100% efficient, assuming 85% efficiency if you deliver the rated 1200W to the PC, it will by pulling  1411W at the wall without exceeding its specs.


This.  1220W AC at the wall @ 85% efficiency is 1037W DC.  The powersupply is rate to provide 100% of rated load (1200W) continuously

You are fine.

That's really good to know. Just to clarify the max is 1200 dc.

It makes sense because I assume the PSU can cope with dissipating extra ac anyway?

Would I be correct to assume that a good PSU would have a cutoff aswell as opposed to letting itself blow up.

Powesupply designer knows 1200W DC will require >1200W AC and should design powersupply accordingly. 

Yes a good power supply should trip both on load and on heat. 

On my Thermaltake 1000W powersupply (which is a piece of junk) I witnessed both.
A powerload trip will result in hard power off and if bios is set to power on automatically it will a moment later.
A thermal trip will result in power off but it will not be possible to power on the PS until it has cooled down.

The thermal trip was kinda scary.  I have my PS upside down in my workstation case.  It draws cool air from underneath the case.  Foolishly while changing things I moved it to carpet and that partially blocked the intake.  The computer hard shutdown but trying to power up computer resulted in no response... dead.  I notice the power supply case was hot to the touch so I laid the case on its side, pointed a fan at it and when I tried to power the computer up a minute later it did just fine. 
sr. member
Activity: 348
Merit: 251
Are you drawing 1220w at the wall? Keep in mind the ratings are for power delivered to the PC at 12v (and some at 5 and 3.3v). Power supplies are not 100% efficient, assuming 85% efficiency if you deliver the rated 1200W to the PC, it will by pulling  1411W at the wall without exceeding its specs.


This.  1220W AC at the wall @ 85% efficiency is 1037W DC.  The powersupply is rate to provide 100% of rated load (1200W) continuously

You are fine.

That's really good to know. Just to clarify the max is 1200 dc.

It makes sense because I assume the PSU can cope with dissipating extra ac anyway?

Would I be correct to assume that a good PSU would have a cutoff aswell as opposed to letting itself blow up.
donator
Activity: 1218
Merit: 1079
Gerald Davis
Are you drawing 1220w at the wall? Keep in mind the ratings are for power delivered to the PC at 12v (and some at 5 and 3.3v). Power supplies are not 100% efficient, assuming 85% efficiency if you deliver the rated 1200W to the PC, it will by pulling  1411W at the wall without exceeding its specs.


This.  1220W AC at the wall @ 85% efficiency is 1037W DC.  The powersupply is rate to provide 100% of rated load (1200W) continuously

You are fine.
donator
Activity: 362
Merit: 250
I'm running 2x 5970 and 1x 5870 with mild overclocks on an AX1200 and it's barely even warm.  Everything else on the system is undervolted/underclocked and it turns out 2.0GH/s.

I haven't measured at the wall but I don't think it would be wise to add any more load.   It is running on 120V but I've been considering switching to a 240V line for better efficiency.
hero member
Activity: 518
Merit: 500
Are you drawing 1220w at the wall? Keep in mind the ratings are for power delivered to the PC at 12v (and some at 5 and 3.3v). Power supplies are not 100% efficient, assuming 85% efficiency if you deliver the rated 1200W to the PC, it will by pulling  1411W at the wall without exceeding its specs.

Those voltage swings are something else though. If any appliance can handle input voltage swings, its probably a PSU, but its still not good I think.
sr. member
Activity: 280
Merit: 250
Firstbits: 12pqwk
I'm willing to Bet that you're probably drawing less than 1000W combined at the wall.

I thought I was drawing close to 1000W when I had my rig but in reality it wasn't even close.
I had a Corsair 1000HX with:
1x 5970 ~ 300W
2x 5870 ~ 200W x 2
Core2Quad 6600 150W+
HD, Fans, Mobo, etc.

Turns out I was only drawing around 700W at stock speed measured at wall, and only 60W more when OCed.


hero member
Activity: 756
Merit: 500
In fact I am using a UPS with backup generator (provided by datacenter) but the costs still outweigh the bitcoins generated atm Sad
sr. member
Activity: 348
Merit: 251
Not really useless, but at the very least it gives you a peace of the mind Smiley .  Dealing with power companies can be tedious and pain in the neck.

If you want real peace of mind then a UPS with pure sine wave output is what you want, good stable power to the box at all times using one of them.


I don't think this is required either. While it will condition the incoming current, its deliberatly made dirty within the psu and and won't affect the dc. Replacement of a psu is gonna be cheaper than a ups given u lose power efficienty aswell.
sr. member
Activity: 348
Merit: 251
Not really useless, but at the very least it gives you a peace of the mind Smiley .  Dealing with power companies can be tedious and pain in the neck.

I'm not sure they do tho, I think there a scam from.what I've read.
hero member
Activity: 756
Merit: 500
Not really useless, but at the very least it gives you a peace of the mind Smiley .  Dealing with power companies can be tedious and pain in the neck.
sr. member
Activity: 348
Merit: 251
Maybe you want to get a surge protector or something while waiting for your power company reply, such variance is not common.

The electricity mains in my country has a variance of +/-7%.  It is impossible to get 180-190V

I measured (with good equipment) anything between 150 and 270 V in a timeframe of about 3 weeks. I was curious why my equipment froze at random, UPS-es made strange noise (and one died) and so on. After that I decided not to measure anymore...

I really hope that my PSU's don't get damaged with this variance. Can't cope with more downtime and putting more money in.

From what I've read those protectors are useless. The current needs to travel from live to neutral to have flow. I.e. through the psu where its converted to dc - I assume the converter is a complete circuit in and of itself so the psu should absorb the surge.
hero member
Activity: 756
Merit: 500
Maybe you want to get a surge protector or something while waiting for your power company reply, such variance is not common.

The electricity mains in my country has a variance of +/-7%.  It is impossible to get 180-190V

I measured (with good equipment) anything between 150 and 270 V in a timeframe of about 3 weeks. I was curious why my equipment froze at random, UPS-es made strange noise (and one died) and so on. After that I decided not to measure anymore...

I really hope that my PSU's don't get damaged with this variance. Can't cope with more downtime and putting more money in.
sr. member
Activity: 348
Merit: 251
The electricity mains in my country has a variance of +/-7%.  It is impossible to get 180-190V

I measured (with good equipment) anything between 150 and 270 V in a timeframe of about 3 weeks. I was curious why my equipment froze at random, UPS-es made strange noise (and one died) and so on. After that I decided not to measure anymore...

I really hope that my PSU's don't get damaged with this variance. Can't cope with more downtime and putting more money in.
sr. member
Activity: 252
Merit: 250
The electricity mains in my country has a variance of +/-7%.  It is impossible to get 180-190V

I measured (with good equipment) anything between 150 and 270 V in a timeframe of about 3 weeks. I was curious why my equipment froze at random, UPS-es made strange noise (and one died) and so on. After that I decided not to measure anymore...
sr. member
Activity: 348
Merit: 251
I'm pulling about 5 or 6kW/h, our voltage should normally be at 230 but while I have everything running it's down at about 180 to 190.

lol that's made up on the spot b.s.  Cheesy
but whatever  Roll Eyes

How do you think I'm mining at 8GH/s. 4 rigs at over 1200 watts each with an air conditioner using up to 2kW.

I haven't seen anyone getting 8GH from one PSU drawing more than over the 1200 watt mark. Not sure why you decided to try and take the thread off topic though with this ill thought out and pointless post.


Just to re-iterate though, my reader is showing a variance of between 180 and 190v and it should be 230. The power company thought it was very odd when I told them and they put a measurement device into my house for the last week.

We did have voltage problems in the past aswell which were "apparently" sorted out. We only found there was an issue when a new electric shower was destroyed. The transformer is close to us if that should make any difference.
hero member
Activity: 756
Merit: 500
The electricity mains in my country has a variance of +/-7%.  It is impossible to get 180-190V
Pages:
Jump to: