So I just noticed another thing.
I dunno why I thought I had the rigs running the cards at 1.15V. But I just logged into mgpumon and saw that both rigs are running at 1.25V.
I'm probably wrong but isn't the difference between 1.15V and 1.25V something like 50W per card?
Topic: Is my Kill a Watt wrong? says pulling almost 1300W from 1200 W PSU? - page 3. (Read 7816 times)
May 22, 2013, 01:32:59 PM
May 21, 2013, 11:06:02 PM
Was hesitant to mess with the rigs because they have been running strong for a couple days but today on one of the rigs I put 4 of the 7950's on the 1200W PSU and I left just one 7950 on the other PSU.
I have the kill a watt hooked up and it is reading 1250-1275 W. I dunno how long it will last or if it is gonna turn off but if it runs like this that means its drawing nearly the same amount of power from 3 7950's as with 4 7950's. It has been running for about an hours right now.
I guess I will see how long it lasts.
But about how the two PSUs are wired together. I ran the rigs at first with just a paperclip jumping the second PSUs on pin (green to black). Now I have them wired together but both ways it was/is drawing the same amount of power.
I have the kill a watt hooked up and it is reading 1250-1275 W. I dunno how long it will last or if it is gonna turn off but if it runs like this that means its drawing nearly the same amount of power from 3 7950's as with 4 7950's. It has been running for about an hours right now.
I guess I will see how long it lasts.
But about how the two PSUs are wired together. I ran the rigs at first with just a paperclip jumping the second PSUs on pin (green to black). Now I have them wired together but both ways it was/is drawing the same amount of power.
May 21, 2013, 06:21:26 PM
TL;DR: I ran through the calculations to find out how much extra power would be drawn if the power supplies have mismatched voltages. The result was about 3 watts.
Whoa. I don't understand any of that.
I do have the PSUs wired together correctly now. (I think)
I used this guide.
http://www.burningissues.net/how_to/power/adaptor.htm
I took a second look at that guide. For some reason, the have the Power Good signal (pin 8 ) tied to ground (pin 7). Power good is 5V, ground is 0V. Assuming a source resistance of 0.1 Ohms, that may explain 50 Watts of power draw. I would cut that loop. If some power draw is required, I would use a 10kΩ resistor.
May 21, 2013, 02:51:05 PM
I don't think the Kill-a-watt is wrong. Something is probably wrong with your dual-PSU setup.
I suspect the grounds should be tied together (especially if you are drawing from two branch circuits), but that may cause ground loops. The power supply box may have a hole threaded for a ground screw. You should not be continuously drawing more than 12 Amps from a 15Amp 120Volt circuit anyway (load is supposed to be de-rated to 80%).
If there is a voltage difference between the two supplies, you may be getting enough current flow to explain the results described. V=IR -> R =V/I -> (12V)/(1200W) -- (P=IV -> I=P/V) -> (12V)(12V)/(1200W) = 0.12 Ohms
Power supply voltages are supposed to be within 5%: 12V x 5%= 0.6V
Current flow if power supplies mis-match: V=IR -> I=V/R -> (0.6V)/(0.12 Ohms)= 5 Amps
Power = IV -> (5A)(0.6V)= 3Watts.
Hmm, never mind. That does not appear to explain it.
I suspect the grounds should be tied together (especially if you are drawing from two branch circuits), but that may cause ground loops. The power supply box may have a hole threaded for a ground screw. You should not be continuously drawing more than 12 Amps from a 15Amp 120Volt circuit anyway (load is supposed to be de-rated to 80%).
If there is a voltage difference between the two supplies, you may be getting enough current flow to explain the results described. V=IR -> R =V/I -> (12V)/(1200W) -- (P=IV -> I=P/V) -> (12V)(12V)/(1200W) = 0.12 Ohms
Power supply voltages are supposed to be within 5%: 12V x 5%= 0.6V
Current flow if power supplies mis-match: V=IR -> I=V/R -> (0.6V)/(0.12 Ohms)= 5 Amps
Power = IV -> (5A)(0.6V)= 3Watts.
Hmm, never mind. That does not appear to explain it.
Whoa. I don't understand any of that.
I do have the PSUs wired together correctly now. (I think)
I used this guide.
http://www.burningissues.net/how_to/power/adaptor.htm
May 21, 2013, 01:37:26 PM
I don't think the Kill-a-watt is wrong. Something is probably wrong with your dual-PSU setup.
I suspect the grounds should be tied together (especially if you are drawing from two branch circuits), but that may cause ground loops. The power supply box may have a hole threaded for a ground screw. You should not be continuously drawing more than 12 Amps from a 15Amp 120Volt circuit anyway (load is supposed to be de-rated to 80%).
If there is a voltage difference between the two supplies, you may be getting enough current flow to explain the results described. V=IR -> R =V/I -> (12V)/(1200W) -- (P=IV -> I=P/V) -> (12V)(12V)/(1200W) = 0.12 Ohms
Power supply voltages are supposed to be within 5%: 12V x 5%= 0.6V
Current flow if power supplies mis-match: V=IR -> I=V/R -> (0.6V)/(0.12 Ohms)= 5 Amps
Power = IV -> (5A)(0.6V)= 3Watts.
Hmm, never mind. That does not appear to explain it.
I suspect the grounds should be tied together (especially if you are drawing from two branch circuits), but that may cause ground loops. The power supply box may have a hole threaded for a ground screw. You should not be continuously drawing more than 12 Amps from a 15Amp 120Volt circuit anyway (load is supposed to be de-rated to 80%).
If there is a voltage difference between the two supplies, you may be getting enough current flow to explain the results described. V=IR -> R =V/I -> (12V)/(1200W) -- (P=IV -> I=P/V) -> (12V)(12V)/(1200W) = 0.12 Ohms
Power supply voltages are supposed to be within 5%: 12V x 5%= 0.6V
Current flow if power supplies mis-match: V=IR -> I=V/R -> (0.6V)/(0.12 Ohms)= 5 Amps
Power = IV -> (5A)(0.6V)= 3Watts.
Hmm, never mind. That does not appear to explain it.
May 19, 2013, 04:46:44 AM
I think I've weighed in on this before, but I will do so again. Dual PSUs are almost always bad juju, from what I've read of peoples experiences.
If you're worried about blowing your PSU, try the following, 4x7950 off a single PSU. 250Wx4 cards + 200W overhead for your Mobo + fans + CPU + whatever should be plenty. It's a silverstone which is typically a good quality PSU, so it should be able to run at peak for extended periods of time, just don't go crazy and OC like a madman for this phase.
I don't want to read back to what exact model you have, but I believe you said 80Plus Gold. So assuming you're hitting exactly 1200, which feels unlikely but possible, you'll be running 87% efficient, so you should see ~1380W on your kllawatts from the wall.
If so, it is as I suspected, your dual PSU set up is being a boner and causing you headaches. If you see more than 1380W from the wall, I'm a bad man potentially causing you to murder your system, shutdown ASAP (assuming nothing faulty you shouldn't be in this scenario). If you see less than 1380W from the wall, as I mentioned earlier you probably have even more headroom than you thought.
Somehow my Athlon 145 +misc uses ~180W, which is why I don't expect you'd use much more than that.
If you're worried about blowing your PSU, try the following, 4x7950 off a single PSU. 250Wx4 cards + 200W overhead for your Mobo + fans + CPU + whatever should be plenty. It's a silverstone which is typically a good quality PSU, so it should be able to run at peak for extended periods of time, just don't go crazy and OC like a madman for this phase.
I don't want to read back to what exact model you have, but I believe you said 80Plus Gold. So assuming you're hitting exactly 1200, which feels unlikely but possible, you'll be running 87% efficient, so you should see ~1380W on your kllawatts from the wall.
If so, it is as I suspected, your dual PSU set up is being a boner and causing you headaches. If you see more than 1380W from the wall, I'm a bad man potentially causing you to murder your system, shutdown ASAP (assuming nothing faulty you shouldn't be in this scenario). If you see less than 1380W from the wall, as I mentioned earlier you probably have even more headroom than you thought.
Somehow my Athlon 145 +misc uses ~180W, which is why I don't expect you'd use much more than that.
May 19, 2013, 02:24:53 AM
just a thought...try pulling one card out your 3card puter n put it in the 2 card system...see if the power jumps up...i dunno if you tryed that...good luck 
May 18, 2013, 01:24:42 PM
I dunno. I tried two different 1200W PSUs and I tried two different 5 x 7950 BAMT rigs.
I could start pulling cards to see if one is consuming too much power but I have two separate rigs each with they're own 5 cards. So both rigs would have to have a card or two with big power problems which seems very unlikely to me. Unless I'm wrong.
I could start pulling cards to see if one is consuming too much power but I have two separate rigs each with they're own 5 cards. So both rigs would have to have a card or two with big power problems which seems very unlikely to me. Unless I'm wrong.
May 18, 2013, 01:07:32 PM
Those 7950s should only pull 200-225W each, so your second PSU looks just about dead on. Your main PSU should pull <700W for the GPUs, so I can't figure out how you're pulling ~500W for your motherboard and CPU and nothing else.
May 18, 2013, 12:26:37 PM
So I did some more testing.
I have another rig identical to the one we are talking about here. Only difference is it has the ASRock 990 EXE4 instead of the 970 EXE4. Everything else is the same.
So I switched this rig over to BAMT yesterday and I hooked up one of the 1200W Silverstones as the main PSU. Running 3 cards and the rest of the system.
Same results 1150-1200 watts.
I tried hooking up a Cooler Master 1200W gold PSU as the main PSU and it actually reads more W. It reads 1200-1250W on my kill a watt.
If you look at this thread https://bitcointalksearch.org/topic/rig-turns-off-after-15minutes-208299 he is getting about the same results.
Only thing I dont get is that the second PSU which is only running 2 7950's and a case fan only pulls 450W.
So to summarize I have 2 seperate rigs with nearly identical hardware which both draw 1150-1250 W on the main PSU running 3 7950's and the rest of the system (no matter using Cooler Master or Silverstone) but which only draw 450 W on the second PSU running 2 7950's.
I have another rig identical to the one we are talking about here. Only difference is it has the ASRock 990 EXE4 instead of the 970 EXE4. Everything else is the same.
So I switched this rig over to BAMT yesterday and I hooked up one of the 1200W Silverstones as the main PSU. Running 3 cards and the rest of the system.
Same results 1150-1200 watts.
I tried hooking up a Cooler Master 1200W gold PSU as the main PSU and it actually reads more W. It reads 1200-1250W on my kill a watt.
If you look at this thread https://bitcointalksearch.org/topic/rig-turns-off-after-15minutes-208299 he is getting about the same results.
Only thing I dont get is that the second PSU which is only running 2 7950's and a case fan only pulls 450W.
So to summarize I have 2 seperate rigs with nearly identical hardware which both draw 1150-1250 W on the main PSU running 3 7950's and the rest of the system (no matter using Cooler Master or Silverstone) but which only draw 450 W on the second PSU running 2 7950's.
May 15, 2013, 02:20:52 PM
Only other thing I can think of then, is on your computer system, remove one GPU at a time. Unplug one graphics card so you are only running two, and see how many watts.
Then remove another GPU, so you are only running one, and see again.
Can also compare watts with using riser/molex, versus card plugged into motherboard directly and see.
Then remove another GPU, so you are only running one, and see again.
Can also compare watts with using riser/molex, versus card plugged into motherboard directly and see.
May 15, 2013, 12:54:54 PM
Thats another thing I was asking before. Should I hook the two PSUs together instead of the paperclip?
Would that make a difference? Or is that just so they turn on together?
Would that make a difference? Or is that just so they turn on together?
May 15, 2013, 12:53:22 PM
So you've got 2 PSUs:
PSU1 has the Motherboard & CPU, 3 GPUs, and it's pulling ~1300W?
PSU2 has 2 GPUs and it's pulling ~500W?
1800W is WAY too much for a 5 GPU system. Each card should pull ~250W, so PSU2 looks about right. However, PSU1 should only be running ~800-900W or so, not 1300W.
Try not connecting the two together, and just shoving a paperclip between the green and black wires on PSU2 to start it without the computer. Or if you've already cut off the green wire, run it right into the ground (on the same PSU, not the other one).
PSU1 has the Motherboard & CPU, 3 GPUs, and it's pulling ~1300W?
PSU2 has 2 GPUs and it's pulling ~500W?
1800W is WAY too much for a 5 GPU system. Each card should pull ~250W, so PSU2 looks about right. However, PSU1 should only be running ~800-900W or so, not 1300W.
Can it be the way I have the dual PSU setup?
All I did was short the green wire to ground on the ATX connector on the second PSU.
I don't see how this at all could make the cards draw this much power. But I dunno.
All I did was short the green wire to ground on the ATX connector on the second PSU.
I don't see how this at all could make the cards draw this much power. But I dunno.
Try not connecting the two together, and just shoving a paperclip between the green and black wires on PSU2 to start it without the computer. Or if you've already cut off the green wire, run it right into the ground (on the same PSU, not the other one).
That is how I have it hooked up right now. I do NOT have them hooked together, just a paperclip in the second one like you said.
I switched them around and its the same thing. The one with the system and 3 cards hooked up is drawing like 1200W.
But here is something else I found. When I was swapping them over I noticed something I forgot to list. I forgot I also have 2 molex's hooked up to the main PSU from the riser cables. So thats 75Wx2 = 150W right?
So 250 X 3 =750, + 150 from the riser cables = 900 + whatever power the system is drawing which is ? I dunno what like 150W. If so thats 1050W right? Or am I wrong?
May 15, 2013, 12:44:23 PM
So you've got 2 PSUs:
PSU1 has the Motherboard & CPU, 3 GPUs, and it's pulling ~1300W?
PSU2 has 2 GPUs and it's pulling ~500W?
1800W is WAY too much for a 5 GPU system. Each card should pull ~250W, so PSU2 looks about right. However, PSU1 should only be running ~800-900W or so, not 1300W.
Try not connecting the two together, and just shoving a paperclip between the green and black wires on PSU2 to start it without the computer. Or if you've already cut off the green wire, run it right into the ground (on the same PSU, not the other one).
PSU1 has the Motherboard & CPU, 3 GPUs, and it's pulling ~1300W?
PSU2 has 2 GPUs and it's pulling ~500W?
1800W is WAY too much for a 5 GPU system. Each card should pull ~250W, so PSU2 looks about right. However, PSU1 should only be running ~800-900W or so, not 1300W.
Can it be the way I have the dual PSU setup?
All I did was short the green wire to ground on the ATX connector on the second PSU.
I don't see how this at all could make the cards draw this much power. But I dunno.
All I did was short the green wire to ground on the ATX connector on the second PSU.
I don't see how this at all could make the cards draw this much power. But I dunno.
Try not connecting the two together, and just shoving a paperclip between the green and black wires on PSU2 to start it without the computer. Or if you've already cut off the green wire, run it right into the ground (on the same PSU, not the other one).
May 15, 2013, 12:33:30 PM
I guess I would switch the two PSUs to see what happens. Since they are both 1200W, you shuold have no problems.
The one that is just running 2 cards and a fan ---> hook it up to your computer setup
The one hooked up to your computer --> connect to just the 2 cards and fan
Curious to see, but it could very well be a bad PSU.
The one that is just running 2 cards and a fan ---> hook it up to your computer setup
The one hooked up to your computer --> connect to just the 2 cards and fan
Curious to see, but it could very well be a bad PSU.
May 15, 2013, 12:11:28 PM
So I tested the kill a watt. It works fine. I hooked a hair dryer to it that says it is 1575W and when I turned it on hi it was pulling 1500-1520W. I hooked a 100W light bulb lamp up to it and it read 95W.
So this is where it gets wierd. I think maybe I have a bad PSU. Instead of hooking the kill a watt back up to the same PSU that is running the system and 3 GPUs, I hooked it up to the PSU that is running the 2 GPUs.
This PSU is only running 2 GPUs and 1 case fan. thats it. The rig is mining right now and it is reading 490W.
So how can this be? Does this mean the PSU that was reading almost 1200 W on three cards is bad?
So this is where it gets wierd. I think maybe I have a bad PSU. Instead of hooking the kill a watt back up to the same PSU that is running the system and 3 GPUs, I hooked it up to the PSU that is running the 2 GPUs.
This PSU is only running 2 GPUs and 1 case fan. thats it. The rig is mining right now and it is reading 490W.
So how can this be? Does this mean the PSU that was reading almost 1200 W on three cards is bad?
May 15, 2013, 10:56:23 AM
Oh yea and I am not doing anything with the CPU. Just mining LTC with the GPUs.
May 15, 2013, 10:51:29 AM
Testing the KAW with a known load (bulbs) that someone above suggested is a good idea.
Other questions: give us full details of your system hardware (motherboard, CPU, how many SSD/HDD, etc). Does the wattage also reflect a monitor that you have hooked up? I presume not since you are plugging the PSU straight into the KAW right?
Also, are you using the CPU for anything (CPU mining, etc)?
As others have said, you should not be pulling that much wattage with just 3x 7950s.
Other questions: give us full details of your system hardware (motherboard, CPU, how many SSD/HDD, etc). Does the wattage also reflect a monitor that you have hooked up? I presume not since you are plugging the PSU straight into the KAW right?
Also, are you using the CPU for anything (CPU mining, etc)?
As others have said, you should not be pulling that much wattage with just 3x 7950s.
This is the system.
ASRock 970 EXE4
Athlon II 270
8GB DDR3 (4 x 2GB)
1200W PSU running 3 7950s Another 1200W at the moment is powering the other 2 7950s.
No HDD just a USB stick with BAMT on it.
The W is being measured with just the PSU plugged into it.
I am about to test the meter. I will report back what I find in a few mins.
May 15, 2013, 07:03:28 AM
Testing the KAW with a known load (bulbs) that someone above suggested is a good idea.
Other questions: give us full details of your system hardware (motherboard, CPU, how many SSD/HDD, etc). Does the wattage also reflect a monitor that you have hooked up? I presume not since you are plugging the PSU straight into the KAW right?
Also, are you using the CPU for anything (CPU mining, etc)?
As others have said, you should not be pulling that much wattage with just 3x 7950s.
Other questions: give us full details of your system hardware (motherboard, CPU, how many SSD/HDD, etc). Does the wattage also reflect a monitor that you have hooked up? I presume not since you are plugging the PSU straight into the KAW right?
Also, are you using the CPU for anything (CPU mining, etc)?
As others have said, you should not be pulling that much wattage with just 3x 7950s.
May 15, 2013, 04:54:24 AM
Off topic: I used to have a high school teacher called Mrs. Watt. What makes it even better is that she was an IT teacher 
Jump to: