PSU Guide

[erixx]

AND.... Why did the processor suddenly go from 100% to 0% usage at the same time? was it a 'nearly fry'? Or just a 'hardware fart'?

 

[westom]

I have a hard time to understand, but appreciate the information! Stil don't know 'what has happened' and if I have to do something. (I don't have know-how of multimeter, sorry)

Then you have no business touching an Ipad or mobile phone. Those are far too complex and more dangerous. No I am not facetious. Your reasons are based only in fear. Multimeters are sold in stores that also sell hammers. Even sold in K-mart - because it is that complex. A good hammer is more expensive and often more dangerous. If you never swung a hammer, then a hammer is far harder to use. Not kidding. Get a meter without fear.

Processor did not go to zero at the time of that spike. It occurred later. Unknown and relevant is how much later.

 

[Swamp Monster]

Monitoring software tends to have glitches. Asus PC probe sometimes showed weird things on my PC too (for a few secs). But I will try Aida 64 to see what it shows. What version are you using?
Lag could have been because of CPU @ 100% and since there is no time on horizontal axis, I can't tell if voltage spike and CPU throttle happened at the same time at all.

*edit* Posted almost the same as westom, only I think that because of no time on axis, there is no way to tell what was first - Spike or CPU throttle, altough it looks like spike was first if we assume that software updates all tables simultaneously.

*edit2* I am testing with the same program now, and It's lagging when I write Lower table goes much faster than higher one (about 4.5 times faster).

 

[cdawall]

15 volts on 12 volt parts will not harm them. 7 volts also does not harm 5 volt parts. But safety functions required in a supply (and sometimes missing on supplies selling on price) must keep voltages lower.

BS it will blow parts i have seen it happen. No it wont kill something like a fan or a HDD even an ODD however when you get into more sensitive stuff like a video card or motherboard and you jump the voltage 25% you will see things fail. you cannot make a blanket saying this will not effect that. thing is UL says everything will be regulated within 15% there is a reason for that. The UL folks didn't go well this seem like a good number and use that.

+1 My PSU will cut output when 15,6V is on 12V; 7V is on 5V and 4.5V is on 3V rail. But under normal operation it regulates voltage +3/-3% on all rails. It should be similar on all other PSU's, varying only percentage and voltage a little. I know that on some PSU's they have 5% regulation.

+1 what that i am correct obviously someone thinks 15v is bad having yours cut just above it

 

[Swamp Monster]

1 what that i am correct obviously someone thinks 15v is bad having yours cut just above it

I think what misses in my post is "for short periods of time". I think 15V on 12V rail is bad too, just that components "should" last for short periods of time, like in OP's case. Then we are all right, because increased voltage shortens life of components, and it depends on components and how electronics is designed.

 

[cdawall]

I think what misses in my post is "for short periods of time". I think 15V on 12V rail is bad too, just that components "should" last for short periods of time, like in OP's case. Then we are all right, because increased voltage shortens life of components, and it depends on components and how electronics is designed.

even for short bursts it is bad that is why it is out of spec for UL if it was "OK" then it would not be out of spec.

 

[Mussels]

15 volts on 12 volt parts will not harm them. 7 volts also does not harm 5 volt parts. But safety functions required in a supply (and sometimes missing on supplies selling on price) must keep voltages lower.

15V will destroy half the parts in a PC. i've seen one go up in smoke at 13V. wiring started melting, and pretty much everything but the fans died that was connected at the time.

 

[erixx]

So if I am running this PC as if nothing happened I can call me a lucky boy? Or does the hypothesis that Aida64 is just fooling me still standing? ...

 

[Mussels]

So if I am running this PC as if nothing happened I can call me a lucky boy? Or does the hypothesis that Aida64 is just fooling me still standing? ...

quite likely.

some peoples arrogant comments aside, software readings are worthless. if you don't test it with a multimeter you'll never know for sure what the real readings are.

 

[erixx]

I understand Mussels, and yes. I have seen extremely erratic fan speed readings by software, so why not wrong voltages? I can only repeat this: I never saw that voltages before and I doubt I will see them again. They have always been okay and are okay.

 

[westom]

II can only repeat this: I never saw that voltages before and I doubt I will see them again. They have always been okay and are okay.

Routine is for a defective power supply to still boot a computer. If only assuming, well, many see a computer boot. Then 'assume' it is OK when problems exist. Problems morph into hard failures later.

No good answer exists without numbers from a meter. A tool used even by early teenage science student and K-mart shoppers. Explained is why hardware can report that less than 15 volts when no such voltage spike exists. Explained is why voltage spike can be more routine. And your software only reports it rarely. Just of few of many reasons why you cannot make that conclusion. Reasons that say you have no idea, yet, what your machine is doing.

It is your decision to either have no idea of that machine's integrity and not care. Or to know.

That motherboard hardware is a monitor. Once calibrated with the meter, then you can set it to alarm on problems. Problems that exist but are not yet severe enough to crash a computer. Problems that can be averted before it causes data or other losses. Monitor is for identifying the problem even months before it crashes a computer.

The monitor may also report that failure with greater frequency months later. Providing you time to learn and later avert a problem in advance. Too many possiblities exist to discuss here. But another reason to setup that monitor function.

 

[cdawall]

Routine is for a defective power supply to still boot a computer. If only assuming, well, many see a computer boot. Then 'assume' it is OK when problems exist. Problems morph into hard failures later.

No good answer exists without numbers from a meter. A tool used even by early teenage science student and K-mart shoppers. Explained is why hardware can report that less than 15 volts when no such voltage spike exists. Explained is why voltage spike can be more routine. And your software only reports it rarely. Just of few of many reasons why you cannot make that conclusion. Reasons that say you have no idea, yet, what your machine is doing.

It is your decision to either have no idea of that machine's integrity and not care. Or to know.

That motherboard hardware is a monitor. Once calibrated with the meter, then you can set it to alarm on problems. Problems that exist but are not yet severe enough to crash a computer. Problems that can be averted before it causes data or other losses. Monitor is for identifying the problem even months before it crashes a computer.

The monitor may also report that failure with greater frequency months later. Providing you time to learn and later avert a problem in advance. Too many possiblities exist to discuss here. But another reason to setup that monitor function.

you do understand what that monitor is correct? it is a PC chip on a motherboard one known to be incorrect by everyone on these forums. right this very second the one on my DFI UT nf3 says my 12v rail is running at 11.7 the -12 @ -6.2 and -5@ -11.14 PSU is fine checked it with a meter board is fine checked it with a meter turns out its the stupid monitoring chip like everyone else has that still does not work

 

[westom]

you do understand what that monitor is correct? it is a PC chip on a motherboard one known to be incorrect by everyone on these forums.

Therefore one starts with a multimeter to calibrate those numbers. Reposting what was posted,
> Once calibrated with the meter, then you can set it to alarm on problems.

 

[cdawall]

Therefore one starts with a multimeter to calibrate those numbers. Reposting what was posted,
> Once calibrated with the meter, then you can set it to alarm on problems.

How do you calibrate when they don't even read the correct rail or for that matter what about when they read nothing

 

[westom]

How do you calibrate when they don't even read the correct rail or for that matter what about when they read nothing

What is 'they'?

Read numbers from the multimeter. Post multimeter them and monitor numbers. Then have an answer. I don't understand what your question is, in part, because it is subjective. More specifically, what is 'they'? A better answer would have been possible if specific numbers were provided. I don't understand what you are asking.

 

[cdawall]

What is 'they'?

Read numbers from the multimeter. Post multimeter them and monitor numbers. Then have an answer. I don't understand what your question is, in part, because it is subjective. More specifically, what is 'they'? A better answer would have been possible if specific numbers were provided. I don't understand what you are asking.

They meaning programs how do you correctly calibrate something when it is never going to be correct just because you set a baseline doesn't mean the rest will be correct. Ie meter says 12v set program to say 12v doesn't mean 12.5v is 12.5v

 

[westom]

Ie meter says 12v set program to say 12v doesn't mean 12.5v is 12.5v

Multimeter says 12 volts. Monitor hardware measures 12.5 volts. (program only translates what monitor hardware reads into human readable numbers. Program does not measure anything.) Now you know 12.5 volts displayed by the program is really 12 volts in hardware. Measurement hardware on a motherboard is typically cheap - explains its poor calibration.

Set an alarm signal at 0.25 volts higher.

 

[cdawall]

Multimeter says 12 volts. Monitor hardware measures 12.5 volts. (program only translates what monitor hardware reads into human readable numbers. Program does not measure anything.) Now you know 12.5 volts displayed by the program is really 12 volts in hardware. Measurement hardware on a motherboard is typically cheap - explains its poor calibration.

Set an alarm signal at 0.25 volts higher.

no i am saying just because you get the initial setting IE multimeter says 12v program reads 12.5 off of the hardware lets say this is idle. now there is a load on the system so the multi now reads 11.75v but the program reads 12.4v off of the hardware how do you make up for that.

or better yet you set your dumb useless alarm to .25 higher PSU above bites the bullet and now has a 14v rail but the hardware still only reads 12.65v boom shits gone hardware didn't read high enough for an alarm

 

[Mussels]

Multimeter says 12 volts. Monitor hardware measures 12.5 volts. (program only translates what monitor hardware reads into human readable numbers. Program does not measure anything.) Now you know 12.5 volts displayed by the program is really 12 volts in hardware. Measurement hardware on a motherboard is typically cheap - explains its poor calibration.

Set an alarm signal at 0.25 volts higher.

it doesnt work that way. they arent an accurate reading thats miscalibrated, they're an inaccurate reading, period.

you could drop from 12V to 11V, and the readings might only change from 12.5v to 12.1v

 

[cdawall]

it doesnt work that way. they arent an accurate reading thats miscalibrated, they're an inaccurate reading, period.

you could drop from 12V to 11V, and the readings might only change from 12.5v to 12.1v

thats a much simpler version of what i was trying to say

 

[westom]

thats a much simpler version of what i was trying to say

Appreciate how these things measure. The A/D converter does not read 12 volts. A set of resistors convert 12 volts to something below maybe 2.4 volts. So how accurate are those resistors? Well 5% resistors are cheap. 0.1% resistors are extremely expensive. You will never know how inaccurate that voltage divider is until you measure it. If the 12 volts drops to 11, then the A/D converter may see 2.1 volts drop to 1.92. And will report the change accordingly. Your job is to recalibrate for electronic components that are fixed at wrong values. That requires a three digit multimeter.

If voltage drops from 12 to 11, then a monitor reading 12.5 volts (because the A/D converter sees 2.1) will drop to about 11.5 (because the A/D converter sees 1.92).

 

[cdawall]

Appreciate how these things measure. The A/D converter does not read 12 volts. A set of resistors convert 12 volts to something below maybe 2.4 volts. So how accurate are those resistors? Well 5% resistors are cheap. 0.1% resistors are extremely expensive. You will never know how inaccurate that voltage divider is until you measure it. If the 12 volts drops to 11, then the A/D converter may see 2.1 volts drop to 1.92. And will report the change accordingly. Your job is to recalibrate for electronic components that are fixed at wrong values. That requires a three digit multimeter.

If voltage drops from 12 to 11, then a monitor reading 12.5 volts (because the A/D converter sees 2.1) will drop to about 11.5 (because the A/D converter sees 1.92).

so how do you know joe-china manufacturer actually set the biased the chips so that they correctly read the drop from 12-11 volts. oh and you are still completely wrong about how a ADC works its going to take 12v and turn it into a series of 1's and 0's you know the things a computer reads maybe you should appreciate how these things work. this is my job i deal with these and gates all day long

you know what i don't know why i am arguing this you are just not getting it these chips are useless i have never ever seen a BIOS correctly monitor anything voltage rated all the way down to core voltage of the CPU.

 

[Mussels]

westom: you can use logic and theory all you want, we actually know from experience that these sensors are worthless. you cannot calibrate them to be accurate. its been this way since the sensors first started getting used in motherboards, and its likely going to be this way forever.

 

[erixx]

And then we have motherboards with direct measuring points for multimeters (3 or 4 high end boards per chipset).

Now back to the real world:
1) instead of working or gaming do we have to watch a multimeter, just in case? no..please!!!!!
2) so I take the hassle to do the multimeter part, let's say, once: what benefit do I get?
3) can we avoid funky voltages? not really
4) can we enjoy a 'smart' monitoring? apparently it's all lies, lies, lies, so why worry? just keep your backups up to date...

PD: I am imaging a The Simpsons scenary where you buy a hardware part, apply the multimeter at home, see a weird voltage, you return it, they give you another, then you use the multimeter again, you return it, and so on, again, and again.... untill the shop doesn't let you enter hahahahaa

 

[Mussels]

1. you do it the same reason you run benchmarks or do stress tests. to know how your system is really running. Do you rely on the guages in your car instead of taking it to a mechanic to be serviced?

2. you know what the voltages are. if your 12V rail droops a massive amount under load, you know to replace it BEFORE it dies and takes your hardware with it. its the kind thing you only test once, and after that only when you have problems.

3. yes. by replacing your PSU.

4. yeah most smart monitoring/automated warning systems are absolute shit. thats why you test it manually when you first setup the system for a baseline, and again when you have problems.

as for the PD: i've seen people do that. buying generic $20 PSU's that contiually fail after hours/days in high powered systems.