Power supply replacement?

[ISI300]

Hey guys! It's been a while seen I've been active in this here TPU forums, so sorry about that.

I have this power supply that I've been using since at least 2010. The company is called Green, which you might never have heard of. The exact model is GP-685B and it's based on the Seventeam V-Force 650 modular design.
It's a decent enough power supply and the components look solid to my eyes. And upon closer inspection during dust cleanup, all the components at least look to be in good condition, the caps, rectifiers, coils, etc.
My question is, should I replace this because it's old? My current rig is this thing:
i5 4690K (stock), Asrock z97 Pro4, 2x4Gig Apacer DDR3, GTX 580 (ref.)
all the parts are running at their stock clocks except the 4690K which I've down-volted by 100 milivolts.
after a 10 min furmark run, HWinfo had this to say about the voltages:

The 580 is kinda old but it still does the job just fine. during gaming, the PSU fan ramps up a bit and it's audible (don't know the exact RPM though). it gets much quieter when I chuck in the older 5870 2G card that I have laying around which is obviously much slower.
Thanks in advance and any help is really appreciated!

 

[RejZoR]

550-650W should be more than enough based on this calculator: http://www.bequiet.com/en/psucalculator

Seems about right considering I had something similar in the past and it was running on 750W unit with tons of reserve.

 

[ISI300]

550-650W should be more than enough based on this calculator: http://www.bequiet.com/en/psucalculator

Seems about right considering I had something similar in the past and it was running on 750W unit with tons of reserve.

So the fan getting louder is normal? What about it's age? Should I worry about that? What if it fails? Is it gonna take everything else with it?

 

[LTUGamer]

Usually software voltage monitoring is not accurate so it means nothing. However these numbers are OK, nothing to worry about.

There is no point to upgrade your PSU. It is normal that PSU becomes louder when get stress because then fan spins faster. Only one point to upgrade is if you are silent computing enthusiast and that noise is too high.

BTW system with GTX 580 and HD 5870 don't have massive amount wattage difference. http://images.anandtech.com/graphs/graph4008/33853.png So maybe your GTX 580 card is bit noisier.

 

[ISI300]

Usually software voltage monitoring is not accurate so it means nothing. However these numbers are OK, nothing to worry about.

There is no point to upgrade your PSU. It is normal that PSU becomes louder when get stress because then fan spins faster. Only one point to upgrade is if you are silent computing enthusiast and that noise is too high.

BTW system with GTX 580 and HD 5870 don't have massive amount wattage difference. http://images.anandtech.com/graphs/graph4008/33853.png So maybe your GTX 580 card is bit noisier.

The 580 is definetly louder than the 580. But the PSU has this sort of distinctive hum to it. Thanks, that puts my mind at these. Especially since the sanctions and everything has skyrocketed the prices so bad.

 

[RejZoR]

Either buy a PSU with high quality fan (to my current knowledge, only Be Quiet use FDB fans) or you can replace it yourself if you know what you're doing. You'll most likely have to use a fan adapter or re-solder the wires to old ones. It's also smart to find fan that has low starting voltage and same amps so the fan controller in PSU can operate it the same as original one.

 

[ISI300]

Either buy a PSU with high quality fan (to my current knowledge, only Be Quiet use FDB fans) or you can replace it yourself if you know what you're doing. You'll most likely have to use a fan adapter or re-solder the wires to old ones. It's also smart to find fan that has low starting voltage and same amps so the fan controller in PSU can operate it the same as original one.

You're making a good point. I've thought about changing the fan but hesitated a bit because a quieter fan won't have the same airflow as the current one. However, I don't think the default ramp is justified, the air coming out of the back of the PSU is barely above room temperature.
Gotta find a decent, quiet fan first.

 

[Bill_Bright]

Usually software voltage monitoring is not accurate so it means nothing. However these numbers are OK, nothing to worry about.

It cannot "mean nothing" then be "OK" and "nothing to worry about".

The fact of matter is, the software monitoring software simply reports what the hardware sensors are reporting to the system. There are no proprietary algorithms or coding to make one better than the other. So if the reported voltage by the software is not accurate, it is because the hardware sensors are not accurately measuring the voltage. That is not the fault of the software. And there is nothing users can do about it unless they have an accurate volt meter, know how to properly load the circuit, and then know how to properly (and safely) measure the voltage.

That said, I have done exactly that with several motherboards (budget to high-end) and power supplies and have found the voltages shown by the BIOS and monitoring software (specifically, Speccy and HWiNFO64) to be quite accurate and consistent - typically within a point or two, if not spot on, indicating the sensors themselves are fairly accurate, and the reporting the system is fairly consistent too. Any differences between the BIOS readings and the software I feel safe to say is due to sampling times and rates.

The ATX Form Factor standard requires voltages to maintain ±5% tolerances. So,

12V ±5% = 11.40 to 12.60 V
5V ±5% = 4.75 to 5.25 V
3.3V ±5% = 3.14 to 3.47 V​

All of your voltages are well within that range except your +3.3Vsb standby voltage which is pushing the limit, but still okay, especially as a standby voltage.

According to your PSU specs, there is nothing wrong with your PSU's capabilities, except the hold up time. This is how long the PSU must continue to output power during the loss of input power. It says >12ms yet the ATX standards call for >16ms. I have to say it seems odd they would publish this shortcoming, but then again, maybe this show of honesty is a good thing.

I see no reason to replace your PSU unless you start experiencing some unexplained shutdowns, or your voltage readings indicate problems with regulation by falling out of the required tolerances.

 

[ISI300]

It cannot "mean nothing" then be "OK" and "nothing to worry about".

The fact of matter is, the software monitoring software simply reports what the hardware sensors are reporting to the system. There are no proprietary algorithms or coding to make one better than the other. So if the reported voltage by the software is not accurate, it is because the hardware sensors are not accurately measuring the voltage. That is not the fault of the software. And there is nothing users can do about it unless they have an accurate volt meter, know how to properly load the circuit, and then know how to properly (and safely) measure the voltage.

That said, I have done exactly that with several motherboards (budget to high-end) and power supplies and have found the voltages shown by the BIOS and monitoring software (specifically, Speccy and HWiNFO64) to be quite accurate and consistent - typically within a point or two, if not spot on, indicating the sensors themselves are fairly accurate, and the reporting the system is fairly consistent too. Any differences between the BIOS readings and the software I feel safe to say is due to sampling times and rates.

The ATX Form Factor standard requires voltages to maintain ±5% tolerances. So,

12V ±5% = 11.40 to 12.60 V
5V ±5% = 4.75 to 5.25 V
3.3V ±5% = 3.14 to 3.47 V​

All of your voltages are well within that range except your +3.3Vsb standby voltage which is pushing the limit, but still okay, especially as a standby voltage.

According to your PSU specs, there is nothing wrong with your PSU's capabilities, except the hold up time. This is how long the PSU must continue to output power during the loss of input power. It says >12ms yet the ATX standards call for >16ms. I have to say it seems odd they would publish this shortcoming, but then again, maybe this show of honesty is a good thing.

I see no reason to replace your PSU unless you start experiencing some unexplained shutdowns, or your voltage readings indicate problems with regulation by falling out of the required tolerances.

Thank you very much Bill_Bright! Very enlightening. Thanks for taking the time to respond.
It's good to know that software monitoring is so accurate, I've always had doubts about their accuracy.
I guess I'll keep this thing for a while until sanctions relief kick in.
Thanks everybody for your valuable input! What an awesome community.

 

[Bill_Bright]

You are welcome but gain, the software is just reporting what the sensor is measuring. And those sensors are not very high-tech so without proper testing with a real meter, there is no way to be 100% certain.

That said, high-speed digital devices are very sensitive to voltage anomalies so as long as your system is not experiencing sudden unexplained shutdowns, freezing, over heating, or other odd behaviors, it is pretty safe to assume your sensors are reporting fairly accurate voltages.

 

[ISI300]

You are welcome but gain, the software is just reporting what the sensor is measuring. And those sensors are not very high-tech so without proper testing with a real meter, there is no way to be 100% certain.

That said, high-speed digital devices are very sensitive to voltage anomalies so as long as your system is not experiencing sudden unexplained shutdowns, freezing, over heating, or other odd behaviors, it is pretty safe to assume your sensors are reporting fairly accurate voltages.

I realize that, but since I don't have the tools or the know-how to measure with leads this'll have to do.

 

[OneMoar]

psu should be fine

 

[Bill_Bright]

I realize that, but since I don't have the tools or the know-how to measure with leads this'll have to do.

Neither do most users and that is just fine.

 

[LTUGamer]

It cannot "mean nothing" then be "OK" and "nothing to worry about".

The fact of matter is, the software monitoring software simply reports what the hardware sensors are reporting to the system. There are no proprietary algorithms or coding to make one better than the other. So if the reported voltage by the software is not accurate, it is because the hardware sensors are not accurately measuring the voltage. That is not the fault of the software. And there is nothing users can do about it unless they have an accurate volt meter, know how to properly load the circuit, and then know how to properly (and safely) measure the voltage.

That said, I have done exactly that with several motherboards (budget to high-end) and power supplies and have found the voltages shown by the BIOS and monitoring software (specifically, Speccy and HWiNFO64) to be quite accurate and consistent - typically within a point or two, if not spot on, indicating the sensors themselves are fairly accurate, and the reporting the system is fairly consistent too. Any differences between the BIOS readings and the software I feel safe to say is due to sampling times and rates.

The ATX Form Factor standard requires voltages to maintain ±5% tolerances. So,

12V ±5% = 11.40 to 12.60 V
5V ±5% = 4.75 to 5.25 V
3.3V ±5% = 3.14 to 3.47 V​

All of your voltages are well within that range except your +3.3Vsb standby voltage which is pushing the limit, but still okay, especially as a standby voltage.

According to your PSU specs, there is nothing wrong with your PSU's capabilities, except the hold up time. This is how long the PSU must continue to output power during the loss of input power. It says >12ms yet the ATX standards call for >16ms. I have to say it seems odd they would publish this shortcoming, but then again, maybe this show of honesty is a good thing.

I see no reason to replace your PSU unless you start experiencing some unexplained shutdowns, or your voltage readings indicate problems with regulation by falling out of the required tolerances.

I want to disagree with you. My personal experience shows that according to software half PC has 7~9 volts on 12V rail so these measurements are not accurate at all.

"mean nothing" means that they are inaccurate. But it is OK because numbers are good, it would be big surprise that there are voltage failure but sensor error "converts" it into right numbers.

 

[OneMoar]

bill as per usual has it wrong
IC2 sensors simply send a value to the sm bus
its up to the software reading that to convert it into a human-readable temperature and or fan speed there is no high-level API
and different software can read the registers in different ways
different signal floors/resistance thresholds driving voltages ect ect as well as plain old querying the wrong pin/register

 

[ISI300]

So as long as I'm not getting symptoms of PSU failure it should work fine then? Like unexpected reboots, bsods or shut downs or buzzy sound coming from PSU?

 

[BarbaricSoul]

^yes

 

[Bill_Bright]

bill as per usual has it wrong

And OneMoar, once again, decides to degrade the technical discussion with denigrating personal insults instead of simply debating the facts. I guess it strokes his ego to put others down.

IC2 sensors simply send a value to the sm bus
its up to the software reading that to convert it into a human-readable temperature and or fan speed there is no high-level API

First, we are talking voltages, not temperatures.

But these values the sensors send is not some mystery number that each software (and BIOS) has to use some hocus pocus proprietary coding to come up with some arbitrary voltage, temperature or RPM! These values conform to published specifications. If 479AB means +5.02VDC, then it means +5.02VDC to every monitoring program.

You guys are suggesting that these sensors and monitoring programs (including the BIOS) are totally worthless and therefore the ONLY way to measure voltages, temperatures, or fan speeds is with a multimeter, temperature probe, and some optical fan speed counter respectively. That's ridiculous!

My personal experience shows that according to software half PC has 7~9 volts on 12V rail so these measurements are not accurate at all.

That IN NO WAY proves or even suggests the monitoring software is at fault. It could very well be the sensor. Did you measure the voltage at the sensor with a multimeter? What software? Did you use several software products that all showed different values?

Are these sensors 100% accurate? Of course not. They cost pennies, not big bucks. But this is the about the software that monitors those sensors. And the value sent by the sensor (faulty or not) is still a value that represents a real number. Not some arbitrary number the software does anything it wants with.

"mean nothing" means that they are inaccurate. But it is OK because numbers are good,

And the point of my comment on that is how can you say it means nothing then declare those same meaningless numbers show his voltages are OK? If meaningless, you don't know if okay.

(edit- fixed typo)