Two Dead Power Supplies - Unsure of Cause

[superman859]

I haven't had much luck with PSUs lately...About 1.5 months ago my 500 Watt Enermax power supply (few years old) went out when I was out of town. Not sure what happened. Computer simply wouldn't boot or power up or spin fans or anything.

Yesterday, it's replacement (only 1.5 months old) went out - Corsair TX750w if I remember the model correctly while I wasn't in my room. Again, wouldn't boot or anything.

Today I got some tools and have been trying to figure out what is going on. I think the Corsair overheated somehow since it kind of smells funky, although the case is pretty ventilated (Antec P180) and it's not running much (older stuff, 1 basic dual-monitor graphics card).

First, tried the simple paperclip test - shorted green wire and ground on 24-pin connector, and they both spun up this way.

Next I tried an Antec power supply tester. Plugged in the Corsair, and it didn't even turn on the LCD screen or respond at all. Completely dead. Now it won't even spin up with paperclip method.

The Enermax passed tests by the Antec PSU tester. Voltages were fine, molex was fine, etc. However, it still doesn't power on the machine or even begin to spin fans when actually plugged in.

Also got a multimeter. Didn't bother on the Corsair yet, but the Enermax seems to pass multimeter manual tests as well when not in motherboard. using paperclip for 20-pin (or Antec tester, which also connects the pins), I tested the remainder pins and they all seem to have the correct voltage. I tested the molex, and it too has the correct voltage.

However, if I plug the Enermax into the motherboard and then test the molex, it is way off. 12v comes off as like 140v - 150v, and 5v is kind of jumpy around 0.

Plugging a much older, 250w backup PSU into the motherboard does start it up and CPU fans and everything, although it's much too old and weak to really use as the real PSU.

So I understand that the Corsair is probably dead. It doesn't respond for any tests and smells funky. Overheated somehow.

But I dont understand the Enermax. It passes the Antec tester. It passes paper clip only. It passes manual inspection via multimeter in conjunction with paperclip/Antec tester (unplugged from motherboard), but when plugged into the motherboard (and only the motherboard), it is way off. Is this common for a bad PSU, or is this an indication that my motherboard is evil?

I also don't understand how two decently rated power supplies died, both within a couple months of each other. Is it possible my motherboard is the bringer of death and killing all PSUs that hook in? Or can the other hardware not really harm the PSU (besides asking for too many watts - I need about 375w I believe for everything)?

I bought a new PSU, again, but worry if my motherboard or any other hardware is pure evil that it should be replaced as well.

Any thoughts?

 

[God Mode]

Did you check your wiring and outlets? Put a heavy load on the outlet and see if it fluctuates a lot. Also, I've never had a recent PSU that didn't smell bad or have some industrial cleaner type odor to it. Sometimes its strong enough to stink up the room for days.

 

[superman859]

Did you check your wiring and outlets? Put a heavy load on the outlet and see if it fluctuates a lot. Also, I've never had a recent PSU that didn't smell bad or have some industrial cleaner type odor to it. Sometimes its strong enough to stink up the room for days.

Coming up with a heavy load without my machine might be hard...just have lamps and my monitors and smaller things....but with this basic stuff on (everything thats normally on except computer) it seemed fine at 119.5-119.6 AC and wasn't jumping.

As for the Corsair smell, it didn't used to smell (I don't think)...it actually smells like burnt plastic or something, especially the 24-pin connector. It made the end of my Antec tester smell as well after plugging it into that (without getting any readings)...A few days ago I actually smelled the same smell (and so did my dad) but couldn't find the cause...However, I didn't open the computer up either then. The PSU just now stopped working.

I did a few more checks related to the +140v. I tested the old working PSU and it also gave ~140v when plugged into motherboard but BEFORE hitting power switch. After hitting power switch, fans spun up and voltage dropped to correct numbers. On the Enermax, I see the 140v, but pushing power button doesn't do anything at all. Nothing starts, and readings stay at 140v.

 

[C1]

The 140V is coming from the peak charge accumulated on the PS smoothing capacitors and would be normal for a plugged in PS just sitting there not energizing anything.

It could be chance that two PS went bad, but to be safe, I would pull the MB from the case & and remount it ensuring that it is mounted correctly (ie, no possibility for intermittent shorts to ground). Also carefully reinstalling and custom re-arranging the wires/connections to the MB provides an opportunity to better inspect for any wiring/connector issues which might lurk but not be easily recognized otherwise.

 

[superman859]

I rewired everything, but ended up deciding to replace anyways. This stuff is about 4 years old and well, I think it's time to upgrade. Have a new MB, 3.4ghz quad-core processor (up from 2.4 dual core), RAM, and a pair of matching monitors coming my way this week...

I think this time around though I will be doing a lot more monitoring if possible, since I haven't been doing much in the past. I installed SpeedFan and it seems to detect GPU and CPU stuff just fine (GPU seems a little warm at 65C but perhaps that's expected - CPU fine), but how can we monitor the PSU? Afterall, that's what has been causing issues lately. SpeedFan displays voltage, but from what I've read there isn't really any way to monitor the temperature / watts used, etc. How can I know if things start looking bad for the PSU?

In my case, the PSU is in a lower chamber with the hard drives, whereas everything else is in a separate upper chamber. I need to figure out how to monitor the lower chamber well, since it doesn't seem to be designed as well for the air flow.

Thought maybe I could buy a separate temp sensor to put down there, but I couldn't find what I was looking for (one that plugs into MB so I can monitor it). Perhaps I don't know the right keywords to search for if they exist.

 

[Zap]

A couple thoughts...

Does the system have to be running if you aren't around? If so, are you allowing it to downclock itself?

Have you considered purchasing a line conditioner? Friend of mine lived in the country and he would lose a PSU every few months. After losing around 5-6 of them, he got a line conditioner and I don't think he's lost one in years.

 

[superman859]

A couple thoughts...

Does the system have to be running if you aren't around? If so, are you allowing it to downclock itself?

Have you considered purchasing a line conditioner? Friend of mine lived in the country and he would lose a PSU every few months. After losing around 5-6 of them, he got a line conditioner and I don't think he's lost one in years.

I have needed the system running many times when I wasn't around since I remote desktop in for research purposes. However, it wasn't downclocking itself (I know very little about changing clock speeds on machines and doing it properly).

If this one goes too, I may look into some additional hardware. I wouldn't expect it to be needed though. I've got it running on a good surge protector, and there are a couple other machines in the house and they never have issues. This is the only one.

The only other thing I can think of to test now is related to the breaker. The outlet / surge protector seems fine and stays pretty steady between 119v and 121v depending on load, but I'm wondering if something happens when the breaker gets switched for the circuit in that room. For some reason that circuit breaker has a test button (shouldn't need it) and the door on the box bumps it every few months if I / someone else is working down there and is not careful closing the door, and I have to reset it. But I would expect this to just have the same effect as plugging in the PSU or unplugging it, or losing electricity temporarily, etc. I would expect a quality PSU to be able to handle something like that without being damaged after one or two instances.

This happened a week or two ago and the PSU was still working fine after that until this weekend, so I'm not sure if it's related or not.

 

[Zap]

You don't have to manually downclock the CPU. Just enable power management (for Intel, enable EIST in BIOS). This has been a feature on Intel desktop CPUs since Core 2 Duo (except Atom).

You may want to look into protection beyond a surge strip. I purchased a Tripp Lite AVR550U for myself on jonnyGURU's recommendation. It is a low priced (I may have paid $80?) unit that provides surge protection, AVR (automatic voltage regulator, which few units provide) and UPS (battery backup). There are others on the market as well.

 

[Puffnstuff]

A battery backup with avr goes a long way to preserving your sensitive electronic components.

 

[westom]

The Enermax passed tests by the Antec PSU tester. Voltages were fine, molex was fine, etc. However, it still doesn't power on the machine or even begin to spin fans when actually plugged in.

The Antec power supply tester and the paper clip test (when supply is not connected to the computer) are wasted efforts. Say nothing useful. For example, a complete defective supply can still spin fans when started with a paper clip. The Antec tester is best when binned.

The supply must be connected directly to a motherboard. Powered off but connected to AC mains. Nothing - not even one wire - is disconnected.

Use a multimeter on 20 VDC range (you should have never measured 120 or 140 volts). Measure a purple wire from supply to motherboard. It should measure about 5 volts. The second and third digits contain valuable information.

Next, measure the green wire before and when the power switch is pressed. Again, record numbers. And finally measure the gray wire before and when the power switch is pressed. Note gray wire behavior when the switch is pressed.

And finally measure any orange, red, and yellow wire as the power switch is pressed. Observe their behavior and record a final number.

I cannot even begin to define how much can be learned from those numbers. The numbers report on things you did not even know where inside your machine.

A computer is at ideal temperature even when the room is at 100 degrees F. Only myths blame hardware damage on such temperatures. Heat is how timing and threshold failures are easily identified. A diagnostics tool that can report a defect before the defect morphs into a failure. Heat did not cause your failures.

Anything that might be on a power supply's power cord is already accomplished inside every minimally acceptable supply. A supply does more than any UPS or power strip might do. The UPS only added battery backup. A power strip protector can sometimes make electronics damage easier. Ignore those solutions. If electrical transients were causing damage (overwhelming protection that is always inside all supply), then an effective solution is located where wires enter the building.

What is this 140v? 140 volts AC or DC? Where (between what two points) did you measure that? You should not have.

First learn what has actually failed. Use the multimeter with each supply connected to a motherboard. Those numbers will report what is without doubts and without any 'it could be' speculations.

 

[westom]

A battery backup with avr goes a long way to preserving your sensitive electronic components.

Battery backup is often the dirtiest power to electronics. Power so dirty as to be potentially harmful to power strip protectors and small electric motors. Since all electronics are required to be more robust, then a battery backup UPS with excessive harmonics and spikes is ideal perfect power to electronic appliances. AVR is what retailers promote. See through that scam by demanding numbers.

Normal power for electronics is even when incandescent bulbs dim to 50% intensity. When urban myths exist, well, read the spec numbers for every appliance. How often do your light bulbs dim to 50%? If bulbs dim, then the refrigerator - not electronics - is at greater risk.

 

[superman859]

The Antec power supply tester and the paper clip test (when supply is not connected to the computer) are wasted efforts. Say nothing useful. For example, a complete defective supply can still spin fans when started with a paper clip. The Antec tester is best when binned.

Well, good thing I still have the receipt then! I was a little disappointed when I found it told me my old dead PSU was good when I knew it wouldn't power up the machine...What a waste of $35.

The supply must be connected directly to a motherboard. Powered off but connected to AC mains. Nothing - not even one wire - is disconnected.

Use a multimeter on 20 VDC range (you should have never measured 120 or 140 volts). Measure a purple wire from supply to motherboard. It should measure about 5 volts. The second and third digits contain valuable information.

Next, measure the green wire before and when the power switch is pressed. Again, record numbers. And finally measure the gray wire before and when the power switch is pressed. Note gray wire behavior when the switch is pressed.

And finally measure any orange, red, and yellow wire as the power switch is pressed. Observe their behavior and record a final number.

I cannot even begin to define how much can be learned from those numbers. The numbers report on things you did not even know where inside your machine.

A computer is at ideal temperature even when the room is at 100 degrees F. Only myths blame hardware damage on such temperatures. Heat is how timing and threshold failures are easily identified. A diagnostics tool that can report a defect before the defect morphs into a failure. Heat did not cause your failures.

Anything that might be on a power supply's power cord is already accomplished inside every minimally acceptable supply. A supply does more than any UPS or power strip might do. The UPS only added battery backup. A power strip protector can sometimes make electronics damage easier. Ignore those solutions. If electrical transients were causing damage (overwhelming protection that is always inside all supply), then an effective solution is located where wires enter the building.

What is this 140v? 140 volts AC or DC? Where (between what two points) did you measure that? You should not have.

First learn what has actually failed. Use the multimeter with each supply connected to a motherboard. Those numbers will report what is without doubts and without any 'it could be' speculations.

As for the 140v, I'm not sure if it was AC or DC. It came off of the power supply molex / 4-pin cable when another 4-pin and 24-pin were plugged into motherboard. This was with the PSU hard switch on, but before hitting the power button (with both a good and bad PSU - same results). After hitting the power button, the numbers changed to correct numbers for the good PSU (computer now actually running with fans and stuff), but remained at the 140v readings for the bad PSU (no fans or anything running). In a previous post, C1 said this was normal for the PSU not energizing anything.

I should mention my multimeter is auto ranging and handles up to 600v for both AC and DC if I remember correctly. While it was my first time using a multimeter, 600v sounded high enough to touch stuff, although I was certainly a little scared some of the time!


Further investigation shows me that I don't think the circuit offers enough amps. Tonight I unplugged my surge protector from wall (computer off, although hard PSU switch on) due to stormy weather and when I unplugged it my whole bedroom power went off (circuit breaker). I investigated my outlets and circuit a little more, and I have one 15amp circuit that serves my bedroom lights, TV / cable box, and computer / monitors. My new 850watt PSU says it uses 120-240v 12amps input. The previous one that died after a couple months was 750watts and took 10amps. One monitor says it takes 1.5amps and I couldn't find the reading on the second monitor. However, my guess is just from my computer and two monitors, I'm either at 15amps or just below it (with previous PSU). Adding in bedroom lights and a couple other things such as TV probably doesn't help....

Does my 850watt PSU use a full 12 amps by simply being plugged in? Or does it only use what is needed like watts? Again, if I sound like I don't know what I'm talking about - it's because I don't! Only started reading about this stuff. But I don't quite understand those numbers either. If watts = amps x volts, then 850w = 12a x volts, that puts volts = 71, and that number seems random to me. But then I might be mixing up input and output values, since 850w is output (DC?) and 12a from box was input (AC).

Anyways, as of now I'm blaming my bedroom circuit for causing all my frustration. Tomorrow morning I intend to add a new 20amp circuit in my bedroom solely for my computer (with dads help, who has done it before). I want it fixed and ready before all my new equipment arrives.

Does my weaker 15amp circuit shed some light on some of these issues? Am I just guessing that could have caused my two PSUs to go? (first to go was a 500watt not sure about amp input, second was the 750watt 10amp input)? If that does sound like it could cause some problems, I dont expect it would be long before my even bigger PSU at 850w 12a dies. I guess I should be more careful about buying big PSUs (I dont really need 850w, but it was on clearance for $2 more than the 750watt one - both have great reviews)

 

[westom]

What a waste of $35. ...
As for the 140v, I'm not sure if it was AC or DC. ...
Further investigation shows me that I don't think the circuit offers enough amps.

Anything that tests a power supply without a major load (motherboard and all computer peripherals connected) cannot accurately report on a supply. Notice how so many reviews do not discuss critical features such as full load, testing the power supply by short circuiting all outputs together, or test other essential features such as over voltage protection. Meanwhile a multimeter provides numbers that can also identify what has failed, or why. Can even report on your AC line suspicions.

There should be no 140 volts inside a computer. Ever. Red probe was on the molex connector. What was the black probe connected to? All voltages are a number between two points. Between what two points was 140 volts?

Computers only consume about 200 watts or about 2 amps. If your machine was consuming 700 watts, well, put seven 100 watt incandescent bulbs inside. Is your computer's inside that so hot as to burn skin when you touch it? Of course not. 700 watt supplies are often recommended when one has no idea why a failure exists.

Power supplies must be selected based upon current for each voltage. A 500 watt supply is more than sufficient wattage. An example. If the 5 and 12 volts are rated for more than excessive current and if the 3.3 volts is too low, then most computer assemblers will blame failure on wattage. Maybe selected a supply with too much wattage but too little 3.3 volt current. Instead they will recommend a 700 or 800 watt supply so that the 3.3 volt is barely enough current. Power supplies are best selected based upon current for each voltage.

700 or 900 watt supply is still only outputting 200 watts and never more than 350 watts. Your 15 amp circuit is massive power. You are not consuming anywhere near to what that circuit can provide.

If using a multimeter to measures those DC voltages, then even 'too little AC power' is one of maybe 30 things identified by those voltage numbers. Those numbers are chock full of useful facts.

Now, how to check that 15 amp circuit. Put an incandescent bulb in one receptacle. Does the bulb dim when the computer powers on? If yes, then the 15 amp circuit has a wiring defect. Fix the defect. Do not cure symptoms with a new (and unnecessary) circuit.

Or connect the multimeter to that receptacle and read voltage when the computer is powered on. Voltage should never drop by more than 2 volts. If drop is larger, again, then fix the defective 15 amp circuit. Use the meter to trace where the defect existing inside walls.

BTW, if a new 20 amp circuit is in a bedroom, then you are required by safety codes to install an AFCI breaker - not a conventional 20 amp breaker.

Most troubling number was that 140 volts. Potentially destructive to digital logic circuits either immediately or months later (overstress).

 

[superman859]

For the 140v, here is a step by step of what I did:

1. Plug old working 250w PSU 20-pin P1 cable into 24-pin slot. Plug 4-pin CPU cable into motherboard as well.

2. Plug PSU into wall. There was no hard switch on this PSU, so the green light on MB turned on at this point, although nothing started up since case power switch had not been pressed.

3. Turn multimeter to DCV (auto ranging, so I do not select a value). Grab molex cable. Put black probe on one of the two black cables on molex (two center pins) and red probe on either side. On one side (5v if I remember correctly) multimeter was around 0, although it wouldn't stay steady. On the other side (12v?) the multimeter jumped up to about 130-140v. If the red pin was NOT touching either side (but had previously touched the 140v side since turning on multimeter), multimeter read around 70v.

4. Push case power switch. fans spun. CPU started booting (no hard drives were connected so it never fully booted into anything). Re-tested molex using steps above. This time, the readings were what they were supposed to be (12v and 5 or 3.3 or whatever the other was supposed to be)

5. Repeat 1-4 with older 500w bad power supply that broke 2 months ago. But, also perform step 3.5 seen below. Results were the same as with good PSU, except step 4 in this case did not result in correct multimeter readings or machine booting.

3.5 (between 3 and 4): PSU has 8-pin connector, only 4-pin half was being used for MB. Take other 4-pin half and put black probe on black wire and red probe on colored wires. Same readings as on the molex.

I'm not sure about the high reading either. I just worry about hooking up my new PSU and soon-to-arrive MB / Processor / RAM / Monitors if something isn't right and damaging these items.

As for voltages, thanks for the tips on selecting a PSU. Online tools tell me I need about 400 watts PSU for my hardware, but I just read online that the Corsair models that were at a store nearby on clearance have great reviews and selected based on that. So far, the voltages are within recommended range in one of the stickies here. I set up SpeedFan to alert me if they ever get near the borders with plenty of room to spare, and no alerts have shown up. They seem quite steady at this point.

But the 15amp circuit breaker does get tripped sometimes, like when I unplugged the surge protector tonight. Other times I've seen it trip are when trying to use one of those electric heaters you plug into the wall. The electric heater also said it took 12amps, and the breaker tripped nearly everytime I tried to use it until I gave up.

With the computer on, here are some results from the wall outlet with multimeter: Between hot and neutral, it was at 119.5v for the majority of the time (for about 2 minutes). Sometimes it waivered by .1v. At one point, it dropped to about 118.9v for about 1-2 seconds then returned to around 119.5v.

Between hot and ground, it was between 120.9v and 121.3v pretty much the whole time, usually at 121.1v.

Between ground and neutral, it stayed at about 1.44v.

However, you mention it should never drop by more than 2 volts ever with CPU on. But I don't know how I would measure this - I can only sit by the outlet staring at the multimeter for so long watching the number! Can I assume that if its steady for 1 minute or so then it should be fine? 2 minutes?

For the incandescent bulb test, I just grabbed my lamp with 60w bulb and plugged into same outlet as surge protector that has computer connections. Turned light on, hit power switch on case. Lamp flickered quite fast (blink of an eye quite literally) right when I hit power button, but otherwise no change. More of a flicker than a dim... It did not have that "dimming" effect that lasts for like a second or so when power issues are happening (maybe due to poor weather)

As for installing a new circuit - I will have to see and consider it I suppose. I'm still tempted since any time I use the electric heater the breaker trips and now this stuff...the wire run is actually pretty easy and shouldn't take more than an hour.

And perhaps I should mention that this circuit I've been testing and plug into currently is on a AFCI breaker - all bedrooms in this house are AFCI (but also all 15amps).

 

[westom]

For the 140v, here is a step by step of what I did:

140V: Well there is nothing (or should be nothing) that can create 140 volts on the yellow wire. (Please do not discuss what you think the wire is. State what you know: its color.) Are you sure it was 140 volts; not 14.0 volts or 140 millivolts?

That 70 volt number is also significant. All of your observations do not agree. But that 70 volts implies the 15 amp circuit is missing its safety ground. For example, someone replaced a two prong receptacle with a three prong - a code violation.


DC voltages: Defining a voltage as OK is insufficient. Numbers you might think are OK may be reporting a failure. Every number to three significant digits. And voltages from the purple, green, gray, and from any one red, orange, and yellow wires. The word good means nothing to me. Only numbers have value (pun not intended).

Also necessary are same numbers from the defective supply. For example, what were purple, green, and gray wire voltages on the defective supply both before and when switch was pressed? Did any voltage on the orange, red, or yellow wire attempt to rise? Also useful.


Tripping breaker: Disconnecting something does not trip a conventional circuit breaker. The breaker that trips in the circuit breaker box - is it an AFCI type? Does it have a test button?

What could be happening? Yanking a power cord might be shorting wires intermittently inside the receptacle box. To trip a 15 amp breaker means maybe 20 amps for 5 minutes, or 60 amps for 3 seconds, or 100 amps for half a second. Where is all that massive current going? It would make lots of noise. If a breaker in the circuit breaker box is a special (non-conventional) type with a test button, then maybe that tripping makes sense. And the breaker is also detecting a defect in the electric heater. Otherwise currents that large must be sparking somewhere - with noise.

Disconnecting something does not trip a conventional 15 amp breaker. Only shorting - consuming massive amperes - would trip that breaker. This discussion will continue assuming it is a conventional breaker.

(Notice, a 15 amp breaker does not trip at 16 amps. And twenty amps takes a long time.)


Monitoring for defective wiring: You measured 121.5 with no current. That means the street transformer is at 121.5 volts during our test. (There is nothing else on this circuit drawing power? That was important but not stated.) The computer draws only two or three amps at most. So a full two volt drop says the 15 amp circuit is a long way back to the breaker box (at least 70 feet). Or there are a few receptacles between your receptacle and the breaker box where wires could be better attached. Either way, that is still a good connection.

Demonstrated is how much more when numbers replace 'good or bad'. Actual numbers mean a better reply. How many tens of feet (inside walls - a ball park number) from the receptacle to the circuit breaker box?

How to inspect other receptacles between computer and breaker box by removing only the cover plate. Wires should be attached to the receptacle side screws, wrapped around the screw, and tight. If wires attach to the back, well you want to fix that. But that would not cause a computer crash or trip a conventional circuit breaker. Just something to fix when convenient.


Summary: First, reason to believe the safety ground is missing. With the computer plugged into the lower receptacle, measure AC voltages between each of three holes in upper duplex receptacle. For example, narrow to wide rectangular hole should be something around 120 VAC. Wide rectangular to semi-round ground may be a volt. Report all three numbers when the computer is drawing power from the other duplex to better understand where that 70 volts is coming from and why the other strange 140 volts (which may be 140 millivolts).

Second, voltage to that receptacle is more than sufficient even if some intermediate connections could be better.

Third, no reason for conventional circuit box breaker to be tripping due to disconnecting. Unless an intermittent short exists somewhere or if the breaker is a non-conventional type. Above amp numbers say a short (and loud noise) must be obvious to trip a conventional breaker.

Fourth, find a GFCI outlet (kitchen or bathroom). Does the heater trip a GFCI? If not, we might be discussing replacement of a defective circuit breaker.

Fifth, VDC numbers from both supplies are needed as defined previously.

 

[superman859]

Thanks for all the help and detailed responses.

I understand that to me 'good' and 'bad' don't mean much, as I know little about this. I was going by the numbers listed in one of the stickies:

For 3.3V, 3.125 to 3.465
For 5V, 4.75 to 5.25
For 12V, 11.4 to 12.6

Those were the ranges I was checking, so anything mentioned as good was in those ranges (most were actually much closer to the correct value. Example, 12V would be something like 12.1v but not any further away).

I will check for more specifics tomorrow morning, especially regarding the 140v. I'm pretty sure it's 140 and not 14.0 since I triple checked when I first saw such a large value, but perhaps there was a small symbol somewhere saying the units were different.

You mentioned checking the outlet values. I previously did this and had them in previous post but they may have been overlooked.

"With the computer on, here are some results from the wall outlet with multimeter: Between hot and neutral, it was at 119.5v for the majority of the time (for about 2 minutes). Sometimes it waivered by .1v. At one point, it dropped to about 118.9v for about 1-2 seconds then returned to around 119.5v.

Between hot and ground, it was between 120.9v and 121.3v pretty much the whole time, usually at 121.1v.

Between ground and neutral, it stayed at about 1.44v."

These seem to fit what you said. I did NOT see a full 2v drop. Largest was less than 1v drop between hot / neutral.

Easy way to check if the safety ground is missing? As far as I know, all outlets in this house are three-prong. The house was built new about 6-7 years ago and they were all installed as three-prong I believe (I haven't seen any two-prong anywhere)

As for the layout of everything: the circuit breaker is in the basement. It uses an AFCI breaker WITH a test button on the switch that I flip to turn everything back on. The cable runs to the second floor of the house, maybe 50-75ft, perhaps less (my dad estimated 70 ft for the NEW 20amp run, but that would have to run into the attic and back down for easy wiring - further than the wire that is already in place).

The electric heater runs fine if it's on a 15amp circuit with few other items. I used it in the basement on one circuit and it was fine, but on a different one (with a different computer on) then the breaker tripped. The basement has two different circuits so I just put the computer on one and the heater on a separate one without issues.

 

[superman859]

Continuation from previous post:

I just checked my current good power supply that's all hooked up and I'm pretty certain you are right about the units. While this one did not show 140, it did show 70mV when I wasn't touching anything, and I'm pretty sure I just missed the unit change. It's small in the top right corner and I thought the multimeter only did volts instead of mV.

For the new good PSU, here are some numbers:

Before turning on:
Molex yellow: 0.33V, red 36mV
24-pin:
Green 3.55V
Yellow 0.335V
Orange 92mV
Blue 0.32V

After turning on:
Molex yellow: 12.12V, red 5.11V
24-pin:
orange 3.36V
blue -12.18V
green 35mV
Red 5.10V
yellow 12.13V
gray 4.92V

So it seems like I just wasn't reading the numbers right on the bad PSU, although I will probably unplug this one and plug a bad one in later for more testing when I have the time, but it's extremely possible that I just missed the unit change (it's automatic and doesn't stand out on the LCD screen).

So now I'm back to thinking about the breaker / issues related to that. Again, it's AFCI (says so on the breaker) and has a test button on it. But from what I've heard, it sounds like it is supposed to be AFCI since it is a bedroom, so switching to a conventional breaker wouldn't work.

 

[Modelworks]

One thing that can give you the odd voltage reading is if the ground connection to the volt meter is not good. Often if you stick a probe in a connector it has wiggle room and provides a poor connection. Depending on what other devices are connected to the pc it can give you all kinds of crazy readings .

For testing PSU outputs without connecting it to something it could damage you can connect any of the outputs to a regular 60watt light bulb. The light bulb will work as a load and you can measure the supplies output . An old lamp with the cord cut off or one of those cheap utility lamp sockets work great.

AFCI is required by the NEC in all new construction so no way around that.

 

[westom]

I was going by the numbers listed in one of the stickies:
For 3.3V, 3.125 to 3.465 ....
Between hot and ground, it was between 120.9v and 121.3v pretty ...
Between ground and neutral, it stayed at about 1.44v." ...
These seem to fit what you said. I did NOT see a full 2v drop. Largest was less than 1v drop between hot / neutral. ...
It uses an AFCI breaker WITH a test button on the switch that I flip to turn everything back on. ...

The label is where numbers start. The label means 3.3 volts measuring 3.2 volts is defective - for reasons that include how meters measure voltage.

If that was 0.07 volts (not 70 volts), then a missing safety ground is no longer a concern. But appreciate how much information is within every number (and therefore why the numbers must be accurately read). Even the 70 foot length number was predicted because numbers were provided.

Your voltages appear good. If the ground to neutral is 1.44v, then either other appliances are also drawing power on that circuit, or the one loose (but still sufficient) wire connection is probably a white wire somewhere between the receptacle and breaker box. A sufficient connection and not a concern. But a loose connection may exist and might be located - assuming computer is the only power consumer.

AFCI means anything disconnected by making a spark will trip the breaker. 15 amps is massive power. Doubtful that anything was drawing too much. However, standard is to put all receptacles on 20 amp circuits - not 15 amps. Rare for a new house to power receptacles with smaller wire. You do not have insufficient power. The breaker was doing what it was supposed to do when you disconnected and created sparking.

If the house is seven years old, then no two wire receptacles exist. Another important number (6-7 years) provided so much information that much of the previous posts were unnecessary.

Heater should be tested on a GFCI outlet in the bathroom or kitchen. Does not matter if it works on another circuit. It must also work without tripping any GFCI circuit. If it trips a GFCI, then the heater is 100% defective - must be fixed or disposed.


IOW a 15 amp breaker is more that sufficient power (although 20 amps is the standard in most brand new homes like yours). Circuit breaker tripping would be due to sparks created during disconnecting and maybe because the surge protector is leaking excessive current. Circuit breaker must be an AFCI type because the receptacle is in a bedroom. It must not be replaced by a conventional breaker. Breaker is doing was it was suppose to do. Tripping when you create a spark when disconnecting. And tripping maybe because the power strip protector is defective - leaking current. Should also be tested (along with the heater) on a GFCI circuit. BTW a surge protector next to a computer provides no effective protection and sometimes can contribute to computer surge damage.

A concern about a missing safety ground was created by numbers read in error. Numbers contain far more information that most realize. Concern no longer exists based upon latest numbers measured from the receptacle.

 

[westom]

For the new good PSU, here are some numbers:
Before turning on:
Molex yellow: 0.33V, red 36mV
24-pin:
Green 3.55V
Yellow 0.335V
Orange 92mV
Blue 0.32V

After turning on:
Molex yellow: 12.12V, red 5.11V
24-pin:
orange 3.36V
blue -12.18V
green 35mV
Red 5.10V
yellow 12.13V
gray 4.92V

Those numbers suggest the supply is massively oversized or was not connected to a serious load. Provided numbers imply no problem exists. But take numbers as described earlier. For example, nothing should have been read from the blue wire. Nothing need be read from the orange, red, and yellow wires when not powered. And the so important purple wire number is not provided. Also, not necessary to read voltages from the 'disk drive' molex. Any red or yellow wire to the motherboard reports same.

Again, without a maximum load, a defective supply will measure good (which is why power supply testers are ineffective).

Let's summarize what you reported. Power switch requests the power controller to start the computer. Power controller used a green wire to order the supply on. When that green wire voltage drops to well below 0.7 volts, then the power supply should begin a power up sequence.

Once a power up process completes (about a second later), a gray wire should increase from near zero to well above 2.4 volts. At 4.92v, it is telling the power controller that power is stable. One requirement met before a power controller will let the CPU do anything. Also essential to this process are numbers for the purple wire both before and 'when' power switch is pressed.

Again, those numbers imply no load. To test a supply means everything possible must be drawing maximum power. If the orange wires cannot provide sufficient current, only then will 3.3 volts drop to 3.2. The computer will work. But the power supply is completely defective. Either has a hardware failure. Or is undersized; that voltage is too little current for the load. Your numbers imply the supply is not being properly tested (from information hidden in the numbers).

To test a supply, first the system must fully multitask to all power consumers. For example, the video controller must process complex graphics (ie a movie), while CD-Rom is being read, while hard drive is being searched, while a file is being downloaded from the internet, while the sound card plays loudly, while devices (ie a camera) draws power from a USB port, while .... now a computer is drawing power from the supply. Now voltages on the orange, red, yellow, and purple wires accurately report on that supply. Now a defective supply that is still booting a computer can be identified. For example, a voltage less than 3.23 on the 3.3 volts would report that voltage as defective.

I understand your computer will not do that stress test. So you load up the supply as much as possible without crashing the computer. Then read the orange, red, yellow, and purple wires. If those voltages get reported here as sufficient, then the power supply is not causing computer crashes. (I forgot what your original failure was.)

One final point. Newer video cards want power provided directly from the supply. I did not read about a wire from the supply to a power connector on the video card. Does a video controller card have a receptacle for connecting directly to the supply?

 

[superman859]

Thanks once again for the detailed response. I am beginning to realize just how important each number can be.

To review my status - away from the numbers for a second - may clarify what my question is about.

I had a PSU die 2 months ago and replaced it with a new one. The replacement died last week. I replaced that (now on the 3rd PSU) a couple days ago.

The current PSU probably is massively oversized. It's 850 watts and my machine probably doesn't need nearly that much. It has a single PCIe video card (with 6-pin power connector), 2.4ghz dual core processor, 2 hard drives, 4gb ram, and 1 dvd drive. Later this week I will be upgrading the processor, motherboard, and ram, but the video card will remain the same.

The current PSU hasn't failed yet either. It probably can handle loads, although I haven't stress tested the new one too much.

The first PSU that died can spin up without any load, but anything else fails. The second PSU that died does not supply any power no matter what (even the paperclip test fails).

So, I have a PSU that seems good and hasn't failed yet, but I was wondering if there could have been a relation between the two PSUs failing - I don't want my new one to be damaged or anything like that if something could be damaging my PSUs. However, it could have just been coincidence that two good PSUs went out within a couple months of each other.

Note: I also have an old, 250w PSU that I originally did some testing on before buying this new 850w one. It booted and worked, but I don't think it would pass stress tests since it is pretty old and small.

 

[ArtricMCE]

Sorry to hear about your misfortune. Lots of good info in this thread. Shouldn't you be able to replace those power supplies under warranty?

If they're going to sit in a closet or get thrown out, drop me a line. I'd be interested in their cases, as I use them to make grid chargers for hybrid cars. Good luck either way!

 

[superman859]

Sorry to hear about your misfortune. Lots of good info in this thread. Shouldn't you be able to replace those power supplies under warranty?

If they're going to sit in a closet or get thrown out, drop me a line. I'd be interested in their cases, as I use them to make grid chargers for hybrid cars. Good luck either way!

Hey - the first one I'm not sure about warranty (Enermax). It may or may not be out of warranty as I cannot remember exactly when I bought it or find the receipt. The second one I'm definitely going to try and get fixed / replaced under warranty, though, but am still waiting for an RMA number. However, I'm really pushed these next two weeks to get some work done and couldn't do without a PSU for the time, so I had to buy a new one to use in the meantime...I'll have to figure out what to do with the replacement when it arrives.

 

[Martimus]

Coming up with a heavy load without my machine might be hard...just have lamps and my monitors and smaller things....but with this basic stuff on (everything thats normally on except computer) it seemed fine at 119.5-119.6 AC and wasn't jumping.

As for the Corsair smell, it didn't used to smell (I don't think)...it actually smells like burnt plastic or something, especially the 24-pin connector. It made the end of my Antec tester smell as well after plugging it into that (without getting any readings)...A few days ago I actually smelled the same smell (and so did my dad) but couldn't find the cause...However, I didn't open the computer up either then. The PSU just now stopped working.

I did a few more checks related to the +140v. I tested the old working PSU and it also gave ~140v when plugged into motherboard but BEFORE hitting power switch. After hitting power switch, fans spun up and voltage dropped to correct numbers. On the Enermax, I see the 140v, but pushing power button doesn't do anything at all. Nothing starts, and readings stay at 140v.

The bolded part makes it sound like there was a short on the board. The burnt plastic smell usually comes from high current burning plastic (or silicon), which is usually caused by a short. Considering the location of the smell, it sounds like the short is actually on the MB, perhaps at the 24 pin connection.

 

[westom]

To review my status - away from the numbers for a second - may clarify what my question is about.

I had a PSU die 2 months ago and replaced it with a new one. The replacement died last week. I replaced that (now on the 3rd PSU) a couple days ago.

That is what I am working for. But I needed numbers from the purple, green, gray, orange, red, and yellow wires as described in the original post. And from each of those supplies.

Before anyone can report what is bad, those numbers are required. As noted previously, the power supply tester can report nothing useful. Useful numbers are those wires when each supply is connected to the computer. And not just a final number after the power switch is pressed. The expresssion was 'when the switch is pressed'. How each wire on each supply responds when the switch is pressed is important for knowing why a defective supply failed. I still do not have facts. So I cannot say anything about any of those supplies.

Numbers are from a third supply (but not yet the purple wire). Numbers that hopefully were taken when the load was maximumized. Those numbers from the third supply implied the computer was not drawing any significant power - the supply was vastly oversized or computer was not multitasking to all peripherals. To learn why other two supplies have failed, the same numbers are required when each other supply is connected to a major load - the computer.

I assume you confirmed those other numbers (ie 140v) were acutally 140mv. Power strip and heater have been tested in a GFCI circuit (to eliminate a possible human safety threat). Computer is not overloading a 15 amp circuit. And its safety ground does exist. Speculations originally created because an AFCI that was operating correctly. So peripheral issues have been eliminted. Now back to the primary question. But that means voltages each listed wire (which includes a purple wire and not a blue wire) from each supply. Those numbers should identify a pattern if failures are common.

Peripheral issues have been resolved. But not yet provided are numbers for the original question.