What happens with not enough Watts?

Associate
Joined
18 Oct 2002
Posts
211
Location
London
Hey all
What happens if you don't have enough watts from your power supply?
are we talking wires burning, locking ups, or just not booting?
Thinking of a new graphics card but don't want to fry everything..
k
 
Permabanned
Joined
15 Nov 2008
Posts
6,968
It wont fry everything dont worry, so long as its not a cheap psu which is gonna blow when overloaded (if it is cheap this may happen) assuming its a decent make then it depends on how close the limit is, if its near to limit it may well boot then crash under heavy loads, eg gaming, if its way past limit its likely not to boot.
Whats your system specs and psu?
 
Caporegime
Joined
14 Dec 2005
Posts
28,071
Location
armoy, n. ireland
I remember a few years ago i was having issues with a pc, i thought it might have been my enermax noisetaker 600w psu, to narrow down the issue i borrowed a cheap ez cool 500w psu from a mate, it wouldnt even power up the machine at all, my enermax was fine, turned out to be a dodgy memory controller on the mobo.
 
Associate
Joined
7 May 2009
Posts
343
What happens if you don't have enough watts from your power supply?
As others noted, no hardware damage results from the resulting low voltage. But the system becomes unstable. How to discover if it has enough power. Play complex graphics (ie a movie), while downloading from the internet, while powering a USB device, while playing sound loudly, while searching the hard drive, while reading a CD-Rom, etc. Only then are you ready to read voltages with the multimeter from any one purple, green, gray, red, orange, and yellow wires. Then post those three digit numbers here to learn more.

Undersized power supply will boot and run computers for months or years. A multimeter would have exposed the reason for very intermittent crashes back on day one. Above is the only test (without $thousands of test hardware) that says a supply is 100% powerful enough or 100% too undersized. Either you have numbers posted for a complete answer. Or don't know if the supply is sufficient.
 
Associate
Joined
20 Mar 2007
Posts
1,051
As others noted, no hardware damage results from the resulting low voltage. But the system becomes unstable. How to discover if it has enough power. Play complex graphics (ie a movie), while downloading from the internet, while powering a USB device, while playing sound loudly, while searching the hard drive, while reading a CD-Rom, etc. Only then are you ready to read voltages with the multimeter from any one purple, green, gray, red, orange, and yellow wires. Then post those three digit numbers here to learn more.

Undersized power supply will boot and run computers for months or years. A multimeter would have exposed the reason for very intermittent crashes back on day one. Above is the only test (without $thousands of test hardware) that says a supply is 100% powerful enough or 100% too undersized. Either you have numbers posted for a complete answer. Or don't know if the supply is sufficient.

That's got to be the worst test I've ever seen :D
A movie won't stress the gfx or cpu. Downloading doesn't use much. USB devices can only consume 2.5W each max and don't stress the cpu. You'd want to plug in every USB device you have
Turning up the sound doesn't add any stress as it's externally powered or you're on headphones which uses naff all.
The HD and DVD/CD don't uses a huge amount of extra power while reading.

Running lots of programs at once also reduces the maximum system stress you can acheive since the CPU spends more time switching between tasks.

I'd start with a game with gfx set to max to start with. If the cpu isn't maxed out, run orthos or prime95 at a lower priority while playing to soak up spare CPU time.
You could do a DVD copy to the HD in the background, it may or may not add any stress since it may slow other parts of the system, reducing the power consumption. I wouldn't bother with this tho.

Hardware damage can result in low voltage, it will need to draw more current to get the same power, stressing the PWM FETs for example.
 
Associate
Joined
24 Sep 2009
Posts
569
Location
Dublin
Just make sure it isn't this psu.

That little red switch that I just had to press to figure out what it did and what would happen if I changed it, caused the death of my old motherboard and probably reduced my hearing by about 50% and damaged my lungs with the smoke inhalation, We learn from these mistakes :D
 
Associate
Joined
7 May 2009
Posts
343
That's got to be the worst test I've ever seen
A movie won't stress the gfx or cpu. Downloading doesn't use much.
Then you are stressing the wrong thing. Please reread what was posted and what must be stressed to have valid numbers.

Meanwhile multitasking to all peripherals is maximum load on a CPU. Move - complex graphics is a heavier load on the GPU. Both are needed to have useful numbers. Prime95 stresses something that results in no useful numbers. Obviously.

If you had sufficient knowledge, then you also know low voltage only damaged electronics when an education comes from hearsay. Now, what component is damaged by low voltage. Good. Now you can post the manufacturer datasheet that says so. Not hearsay. Show me the numbers. You know this stuff. Then your next reply defines which electronic component is damaged by low voltage - and why.

BTW - you cannot. You only know low voltage is destructive because that is a very popular urban myth. Same reason why you did not understand how to put a system under maximum load. Popular myth is not a replacement for learning technology.

Provide is how one gets useful voltage numbers. If you understood the objective, you would not be wasting time discussing CPU stress. If you had basic electronics training, then you knew low voltage does not cause electronics damage. Either learn the technology or learn from hearsay. Your choice.
 
Soldato
Joined
11 Oct 2007
Posts
3,114
Location
London, UK
firstly, haveing an underpowered (cheap) PSU can and will kill components, I killed 3 gpus before i realised it was my PSU that was underpowered (first old build and i upgraded the GPU and had a cheapo 550w). It caused it to spike when under heavy load which damaged teh cards. Your right in low voltage in and of itself wont damage a part, but the fact the part will then draw more current to make up for it, means the psu can then fry your parts/itself.

Secondly, i could do all the things you have stated in your test westom, and my pc will be at about 25% cpu load, and searching a hdd, well i spose it puts it under load, but by no means will it max out my array.

the most load I have got my pc under is from prime95 on 3 cores + furmark at 2560. this causes my PC to draw 967 watts from the wall. Your tests, which wont be stressing any of my gpus, and only have minute load on one (meaning they are using idle power, not 3d power states) mean i draw 450w, which considering my idle is 427 (no speedstep etc) kinda proves you wrong.

How the hell does Prime stress "something" that gives you dodgy numbers, surely the cpu needs to be under to load to be drawing its maximum power, and without the peak loads, your numbers are worthless on identifying if a psu can handle it?

Multitasking on all peripherals is max lod on a cpu? Thats a lie, or your running a very old CPU that cant handle them.

Movie = complex graphics? Its just your card displaying 1 still image after another, generally at 24-30 fps. For this, it uses 2d clocks because it doesnt stress it any more than displaying the desktop with you waving a window around. 3D clocks only come into play when you load up a game that is 3D. Any game will stress your gpu to a similar amount if you turn vsync off and the game doesnt have a hard cap.
 
Associate
Joined
7 May 2009
Posts
343
firstly, haveing an underpowered (cheap) PSU can and will kill components, I killed 3 gpus before i realised it was my PSU that was underpowered (first old build and i upgraded the GPU and had a cheapo 550w). It caused it to spike when under heavy load which damaged teh cards. ...
Secondly, i could do all the things you have stated in your test westom, and my pc will be at about 25% cpu load, ...
Learn what Prime95 does. Maximizes stress on certain CPU intenal functions - especially using heat - as a worst case design test of that CPU. Once hot spots are identified in computer simulations, then heat generating parts (ie bipolar bus drivers) are moved to distribute heat so that timing changes cause no crashes. Prime95 mostly identifies thermal weaknesses that cause timing changes. Does nothing for what the meter must measure.

Well, I arrived when they kept blowing out graphics cards. Turns out the graphic cards were never damaged. They converted observation into automatic conclusions. One need only read specs to know why that damage did not occur. If low voltage damaged a GPU, then power off also does the same damage. When does a GPU learn power is going off? When voltage drops below a preset level. Power off or low voltage - the same circuit powers off the GPU without damage.

If voltage spikes exist, then you have numbers for those spikes. Spikes too exist because a reason must be invented. No numbers is a symptom of wild speculation and resulting myths. People who knew computers even twenty years ago know your 'voltage spikes' can not exist. Read specs that are standard even in the very first IBM PC. Internal circuits limit spikes well below what would damage a GPU. In fact, if that spike exists, then the entire power supply must lockout and shutdown. Learn about crowbars.

So when we trace voltage spikes, what do we find? Spikes cause no damage. But interrupt (interfere with) logic circuits. That would look like compete GPU failure to those using observation and speculation. For example, 0.7 volt spike on the ground pin means nothing can work ... until the missing and required sub- microfarad capacitor is installed. Just another example of how the naïve *know* something was damaged (convert speculation into knowledge) rather than first learn what is wrong.

Many computer assemblers have no electrical knowledge. Even A+ Certified computer techs need no electrical basics to be certified. Therefore many clone computers contain supplies that are missing essential functions.

Too many computer 'experts' only understand two numbers: watts and dollars. Therefore failure is directly traceable to that tech - not to assumed spikes or low voltage. Just another reason why so many computer techs ‘know’ low voltage causes damage. And, BTW, another reason why the Silicon Valley needs so many Chinese and Indian immigrants for designers.

Your post demonstrates too many ‘experts’ do not even have first semester training. Anyone with minimal training knows that anecdotal evidence only results in a speculation.

Too often, computer failures are traceable to "I speculate this must have failed; therefore this definitely caused the problem." In advanced science, your reasoning contains too much "I wouldn't have seen it if I hadn't believed it."

OP – appreciate that a majority will make recommendations without any electrical knowledge. Too many know only from anecdotal evidence – not even one semester of training. Will even attack the poster with a few generations of design experience only because they know better from shotgunning and hearsay. Even A+ Certified computer techs need no electrical knowledge to pass the test.

Appreciate how much knowledge is available to assist you – but only if you provide requested numbers from a 3.5 digit multimeter.
 
Associate
Joined
20 Mar 2007
Posts
1,051
Westom, you didn't understand what I was trying to accomplish, obviously.
The GFX and CPU are the most hungry parts of the system, so it's best to load those up first, hence running a decent 3D application. Now, if the CPU isn't maxed out with that, running a CPU intensive app in the background on a lower priority willl help there.

A movie does not stress the GFX or CPU significantly.
How can multitasking to all peripherals maximise the load on a CPU? tight software loops that fit in the cache and access memory at just the right time intervals to avoid wait states is the optimum, not task switching between multiple applications.

As for lower voltage damaging compnents, it can and does. I already gave an example of which components are put under stress by low voltage. Some motherboards don't have heatsinks on the PWM FETs for the CPU. lets say you have a CPU consuming 60A at 1.4V, not an unreasonable estimate if you're overclocking. That's 84W. Assuming 95% efficiency (it's probably less) of the switching, you've got 88W on the 12V line at 7.4A. Now if the PSU droops to say, 11V, You've just increased the current to 8A.
Obviously, it depends what cut off level the mobo manufacturer has put in and whether they actually got it right. That's assuming they didn't cut corners and left it out.
However you cut it, lower volts will certainly put more stress on the power FETs.
Anywho, it's not relevant, if the PSU is drooping, it's time to get another one.

I assume you have a watt meter? Use it, try your HD movies and drive thrashing. Then try some 3D apps.
 
Last edited:
Soldato
Joined
11 Oct 2007
Posts
3,114
Location
London, UK

In response to the GPU stuff, So your saying that my cheapo 550w that could only supply around 330w stably, when i put a new gpu in teh system, and it died, (i rpeeated this 3 more times as at first i thought it was faulty gpu's and had forgotten my psu was fail). The GPU's wouldnt die totally, but one whichever output the monitor was using, it would lose at least one color, so everything would be tinted another color. This would happen 100% reproducibly while playing crysis (highest gpu loading game/thing i had at the time). On the 4th card (bought second hand, proved to me to be working, in person) I linked up another monitor so i could monitor voltages while playing. 30 mins in, seemed fine, longer than all the others by a good margin. then at 34 mins in the screen goes a nice "teal" 9lack of red). Looking at the history on the monitoring the voltage for the gpu had spiked for around 2 seconds to 50% over what it should have been. I bought another PSU as this was undenibly the cause imo. Since that day I used the GPU for another year on the output that hadnt gone teal, and now its in another system with another psu and it has been able to play crysis etc with no further damage.

During all this time all temp monitors were well within margins, and my case was big and well cooled.

So your telling me that i imagined the graph showing a spike in voltage and that it did not cause any damage? The PSU wasnt rated to be at a sustained output that high, so it couldnt cope, and lost stability on the 12v rail (this is a guess, and something i AM assuming so that I have a reason for it, please correct me).

The other points you make go into a level of detail i dont understand. But I still dont understand why you would need to take readings when the parts arent drawing their maximum power draw, as surely you need to test the PSU is suitable at a peak load scenario, not just normal use?
 
Back
Top Bottom