what are generally considered safe power.....how much sag is ok and what is not?

primetime

primetime

Supreme [H]ardness
Joined
Aug 17, 2005
Messages
7,338
just trying to give good advice to a long time member here and noticed some comments he made in a thread regarding his new card and recommended power supply versus what he is actually using. I made a comment to keep a close eye on those voltage rails while the system is under full load and told him to check with hwinfo64 since he did not have a meter at least yet to double check. here is the data he gave me in a pm and i advised the 5 volt rail was to low for safe keeping in my opinion.

Tzzird [H]ardness Supreme, 8.5 Years

Status: Tzzird is offline
Re: HWiNFO
So the stress test brought my 12 volt rail from a max of 12.049 to 11.795
5 V rail from a max of 4.758 to 4.677
3.3V rail from a max of 3.328 to 3.296

Not perfect, but not as bad as the examples you listed. What do you think?
Click to expand...

let me repeat.....i said the 5 volt was too low unfortunately to keep it that way and the 12volt seemed borderline but ok.....but i wanted other peoples thoughts on this.....
edit: for the time being can we keep the topic in regards to the reported measurements for now....most all of us know its best to double check with a working meter to verify the levels are what we actually believe them to be. thank you:)
btw this is the thread that caught my attention..http://hardforum.com/showthread.php?t=1837157&page=3....and i by no means wish to embarrass anyone.....just like to help people when possible in case there running a risk they may not otherwise be aware off
 
Last edited:
You cant rely on the figures from onboard sensors unless they have some guarantee of performance.
Get a good multi meter.
 
Nenu said:
You cant rely on the figures from onboard sensors unless they have some guarantee of performance.
Get a good multi meter.
Click to expand...
well yes and no......my board correctly reports exactly what my fluke does.....but yea i agree it best to have a second look with a meter.....but with that being aside...can you comment on the rails as if they are in fact accurate for the time being?;)
 
Its not possible to determine anything without a baseline.
ATX spec allows +/- 5% on most rails, but that doesnt take into account fast variation.
 
Nenu said:
Its not possible to determine anything without a baseline.
ATX spec allows +/- 5% on most rails, but that doesnt take into account fast variation.
Click to expand...

thank you for the reply...but you lost me there....whats this about needing a baseline to determine if voltage is dropping to much (meaning the ps is undersized to much or is malfunctioning) please give a more detailed explanation for those of us who need it, when you get time of course;) when you said fast variation i guess your implying how long the voltage dipped for? as in was it the hole time while stressed or for only a second? i guess i see where you going with this/ would an average value give you a baseline? Also going by your 5% then 4.75v would still be inside atx spec?
Capture_zps249ab0fe.png
[/URL][/IMG]
 
Last edited:
and ok lets get off topic a bit...i personally own a fluke meter but what would you recommend for the average joe who would only be using it once in a while but still needing to be accurate enough to get good measurements for a pc? any suggestions since you brought it up?
 
primetime said:
and ok lets get off topic a bit...i personally own a fluke meter but what would you recommend for the average joe who would only be using it once in a while but still needing to be accurate enough to get good measurements for a pc? any suggestions since you brought it up?
Click to expand...
A multi-meter is enough:
http://www.amazon.com/dp/B000EVYGZA/?tag=extension-kb-20

Also, Wikipedia shows what's the min and max range for the voltage (Scroll down to "Generally, supply voltages must be within ±5% of their nominal values at all time")0:
http://en.wikipedia.org/wiki/ATX#Power_supply

I think Nenu might be talking about ripple.
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
Primetime, I really appreciate your dedication to my problem. Never had anyone make a thread for me... I'm flattered. :p

After all these years on the forum, this may be my first time ever to the psu sub forum. I guess owning a psu has always been an afterthought for me. When I built my system eight years ago, the corsair hx520 was all the rage for non sli systems less than a thousand bucks.

I don't game as extensively as I used to, so I haven't been too worried about upgrading. Last time I was going to build a system was this summer. I went to the general hardware forum and was told to wait a month until the new chipsets came out. After a month, I guess I didn't want it that bad. Fast forward half a year and there's a game I've been itching to play in dragon age inquisition. Benchmarks with the 5850 were fairly pathetic. In fact, I couldn't even get my system out of a 1 fps crawl on high for my native 1440p.

With my new upgrade to the r9 280, I've been quite excited to so easily breathe life into such an old system. I don't think I've ever had a motherboard last me eight years before and still be able to play current games. In my research, I got some mixed opinions as to whether it's OK to use this card with my current psu, being that it's only 520 watts and is eight years old. I ended up just going for it as I figured I would finally have the motivation to build a new system of I ended up frying my old one.

It's good to hear that manufacturers typically allow a 5% deviation threshold for the rails. Looks like my my 50 just barely makes the grade.

Should I buy that multimeter just to be extra sure? I guess twenty bucks isn't the end of the world. And the ability to test batteries I'll be able to use forever.
 
The wattage is sufficient for a 280, but its age may be affecting it.

A multimeter will allow you to tell if there's droops, especially if it's getting close to borderline or out of spec.

The fast variations (if I'm interpreting it right) is referring to ripple. Ripple cannot be measured by a multimeter, but with an expensive oscilloscope.
 
primetime said:
just trying to give good advice to a long time member here and noticed some comments he made in a thread regarding his new card and recommended power supply versus what he is actually using. I made a comment to keep a close eye on those voltage rails while the system is under full load and told him to check with hwinfo64 since he did not have a meter at least yet to double check. here is the data he gave me in a pm and i advised the 5 volt rail was to low for safe keeping in my opinion.



let me repeat.....i said the 5 volt was too low unfortunately to keep it that way and the 12volt seemed borderline but ok.....but i wanted other peoples thoughts on this.....
edit: for the time being can we keep the topic in regards to the reported measurements for now....most all of us know its best to double check with a working meter to verify the levels are what we actually believe them to be. thank you:)
btw this is the thread that caught my attention..http://hardforum.com/showthread.php?t=1837157&page=3....and i by no means wish to embarrass anyone.....just like to help people when possible in case there running a risk they may not otherwise be aware off
Click to expand...

The 5 volt rail is probably for logic switching, where the threshold might be as low as 2V anyways. So I'd say 4.5 or whatever is fine.

primetime said:
and ok lets get off topic a bit...i personally own a fluke meter but what would you recommend for the average joe who would only be using it once in a while but still needing to be accurate enough to get good measurements for a pc? any suggestions since you brought it up?
Click to expand...


I use a cheap craftsman multimeter and it seems to be accurate enough (at least within a few %). Even though I'm an EE, I don't use a DMM enough to justify a fluke. DC voltage measurements especially it seems even the cheapest multimeters will have really good accuracy.
 
ok i have one final question for Tzzird....do you have a r9 280 or a r9 290.....somehow i thought you were using a 290 which i believe is a huge difference in power usage when at 100% usage compared to a 280....still and all these kind of threads are educational for everyone and when you have EE's commenting it don't hurt one bit.

One final recommendation is look into maybe getting a outlet power meter like a kill a watt or what ever brand floats your boat. its dam nice to be able to check total system draw especially if your kinda cutting it close on the supply or you just want to know what the dam thing cost to run:)
 
Even though I'm an EE, I don't use a DMM enough to justify a fluke.
Click to expand...
you probably already suspected i am an electrician at least by trade...lol has to be a fluke cause last thing i need is a malfunctioning meter...in these case's it can be life or death if the meter doesn't work correctly.;)
The 5 volt rail is probably for logic switching, where the threshold might be as low as 2V anyways. So I'd say 4.5 or whatever is fine.
Click to expand...
im sure your math is better than mine but is not 4.677 outside the 5% threshold.....not to even mention 4.5? (not sure if i did it right but 4.75 seems to be the low end of the 5% tolerance on 5 volt rail correct) or am i still missing something?
 
Last edited:
primetime said:
ok i have one final question for Tzzird....do you have a r9 280 or a r9 290.....somehow i thought you were using a 290 which i believe is a huge difference in power usage when at 100% usage compared to a 280....still and all these kind of threads are educational for everyone and when you have EE's commenting it don't hurt one bit.

One final recommendation is look into maybe getting a outlet power meter like a kill a watt or what ever brand floats your boat. its dam nice to be able to check total system draw especially if your kinda cutting it close on the supply or you just want to know what the dam thing cost to run:)
Click to expand...

280 not 290!

Whats the price for a outlet wattage meter?
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
As an Amazon Associate, HardForum may earn from qualifying purchases.
Capture_zps09905119.png
[/URL][/IMG]
Ran a power supply stress test using bionic for a few hours earlier to warm my room up then closed that and ran kombuster to max out the gpu and Intel burn test at max to max out the cpu then copied a blueray from one partition to another, just to make sure the drives are pulling power as well and here are my results. Im really NOT a fan of thermaltake power supplys but my enthusiast grade power bird went up in smoke while traveling for work and it was the strongest power supply best buy sold.....cause i hate down time lol
Anyways as you can see even with a less then perfect choice for a supply at least the rails seem very stable, course god knows how much ripple it has. Oh and one last thing , this was my max recorded power usage during the test.
Untitled_zpsa36c984b.png
[/URL][/IMG]
 
Last edited:
Dangman said:
A multi-meter is enough:
http://www.amazon.com/dp/B000EVYGZA/?tag=extension-kb-20

Also, Wikipedia shows what's the min and max range for the voltage (Scroll down to "Generally, supply voltages must be within ±5% of their nominal values at all time")0:
http://en.wikipedia.org/wiki/ATX#Power_supply

I think Nenu might be talking about ripple.
Click to expand...

only thing that concerns me about the meter you picked out is its not true rms. it can only measure perfect sine waves so that could be an issue......other that that i kinda like it...i think im going to buy this to replace my ageing fluke http://www.amazon.com/Extech-EX505-...16717350&sr=1-1&keywords=Extech+True+RMS...no battery tester but my god look at the other features and it looks like it wont break first time its dropped
Capture_zps2c360c88.png
[/URL][/IMG]
 
Last edited:
As an Amazon Associate, HardForum may earn from qualifying purchases.
Why have you recommended a mains meter?
A multi meter was recommended for measuring the output of the PSU, not its input.
I didnt see anything that suggests the mains voltage is at fault.

Dangman said:
I think Nenu might be talking about ripple.
Click to expand...
Yep,anything that can cause a change in DC that a meter wont show including transients and noise.
Not useful in its basic context, but so the op can know there is more can go wrong with a PSU than its avg DC level.
 
Nenu said:
Why have you recommended a mains meter?
A multi meter was recommended for measuring the output of the PSU, not its input.
I didnt see anything that suggests the mains voltage is at fault.


Yep,anything that can cause a change in DC that a meter wont show including transients and noise.
Not useful in its basic context, but so the op can know there is more can go wrong with a PSU than its avg DC level.
Click to expand...

i post here to learn....if a true rms meter is unnecessary for pc checks then i stand corrected. course for me i would use it in the line of work as well, and it would serve other purposes. My statement was made more to get a response from you experts in this catagory as to weather it post any benefits for pcs. If i was wronge, thanks for correcting me;)(i totally forgot all the voltages inside the pc are dc and as such true rms would be useless) but i guess checking for accurate power from the power source could come in handy for some
Thanks Nenu for the info:) but answer me this, if a person ever bought a ups then wouldn't a true rms meter be required, at least to check the output of the ups?
 
Last edited:
A good mains meter can be handy, not that helpful for this case.
I use one for many reasons, the most recent was to find how low idle power I can get.
I'm a bit miffed that running 2 screens makes my 290x use another 40W at idle because my old GTX580 could run 2 screens without extra power use.

The results of a "multi meter" are also RMS unless it has a peak setting that is in use.


UPS are rated in VA, using RMS figures when purchasing might not get you what you need.
Its up to you if you want to measure its output, but it will only supply (you will measure) what is needed, up to its max output.
 
Nenu said:
A good mains meter can be handy, not that helpful for this case.
I use one for many reasons, the most recent was to find how low idle power I can get.
I'm a bit miffed that running 2 screens makes my 290x use another 40W at idle because my old GTX580 could run 2 screens without extra power use.

The results of a "multi meter" are also RMS unless it has a peak setting that is in use.


UPS are rated in VA, using RMS figures when purchasing might not get you what you need.
Its up to you if you want to measure its output, but it will only supply (you will measure) what is needed, up to its max output.
Click to expand...
thanks for the info...i guess i assumed the power coming from a ups might not be perfect sine waves and as such a non rms meter might give an inaccurate reading of voltage or power. Thanks for your reply's, the place is full of great info;)
 
just a bit curiouse if this came in to doublecheck all the rails? how close was the software reported numbers compared to the multimeter? Also curiouse how much load it pulls in boinc like milkyway

let us know how it tuns out since mine peaks over 550 running boinic and could be a problem for a 500 watt supply
 
You must log in or register to reply here.
Back
Top