What happens with not enough Watts?

Ch3m, just re-read your post #16, the power draw at the wall says it all, westoms test is complete cr@p. Obviously someone with a little knowledge, but not enough to be useful :D
I still can't get over the bit about turning the sound up....
 
Westom,

westom said:
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.
Click to expand...

Power going off is a completely different scenario to just low volts. Continuous low input volts causes a rise in input current on GFX/CPU PWM supplies.
As the power goes off, the same thing happens but the PSU can only deliver the energy stored in the caps so the power drops to zero quickly, it doesn't have a chance to overheat.
 
yea sound being turned up was pretty funny, since im sure turning up my speakers (which will increase their power draw marginally) will affect the voltages in my pc lol.

Chers dave :)
 
This reminds me of when I was working a techie at a shop, a guy brings his computer saying something is wrong, he couldnt boot up. Well I take it on the bench plug it in and *BOOOOM* a black cloud all over my face :D

A tip, if it happens don't be close by! :D
 
I assume you meant low voltage?
A lower voltage will cause more heat to be generated on any switching supplies, like the FETs on the mobo or GFX card. Damage as a result is down to how good the design is to prevent it. There's usually some headroom to allow for lower volts, but there is a limit and it's even worse when you overclock.
It might not boot, or it might be unstable.
So, to summarise, the answer is possibly to all 3 :)
 
weird_dave said:
I assume you meant low voltage?
A lower voltage will cause more heat to be generated on any switching supplies, like the FETs on the mobo or GFX card.
Click to expand...
Amount of heat necessary to cause hardware damage is so much above those levels as to not be relevant. Heat does not cause lower voltage. In fact, heat may even create higher voltage. weird_dave has no basic electrical knowledge – is only preaching a very popular myth. Guarantee - he never designed or built computer hardware. Only the least informed would claim a hot CPU causes sufficient power supply loading to see what the OP is asking.

Heat is a problem because it changes signal timings and voltage thresholds. Heat causes computer crashes - not hardware damage. Maximum heat in a CPU does not even heat the power supply let alone create a sufficient current load. Supply load must be significantly larger to identify it as undersized.

Loading is the problem. Accurate measurements of the power supply 'system' are only relevant when maximum load is applied. Prime95 obviously does not create anywhere near enough load to result in relevant numbers. But then weird_dave is so confused as to create more CPU heat – not put a large load on the supply ‘system’. Large numbers of self-proclaims computer experts only understand ‘more fan’. Have so little electrical knowledge as to not understand the difference between CPU loading and supply loading.

How to identify insufficient wattage in a power supply. Create a load that is much larger – as big as possible. OS multitasking to the hard drive, USB device, internet, sound card, CD-Rom, video subsystem, DMA subsystem, timing subsystem, processing numerous interrupts from that sub-system, etc. A multimeter will immediately identify a power supply 'system' that is undersized.

Computer may operate just fine even with an undersized supply. Only meter numbers can identify that supply as undersized. Normal is for an undersized supply to power a computer - and then have failures months or years later. Only way to see that problem today is a meter and sufficient loading. Obviously Prime95 does nothing to create what is needed to answer the OPs question.

What is low voltage? Incandescant lamps can dim to less than 40% intensity and the properly designed computer works just fine. Even Intel standards required all computers to work with voltages that low. What happens when voltage drops even lower. The same circuits that triggger on power off (when voltage drops so low) also trigger to during low voltage to power off the computer. To that power control circuit, low voltage or power off is the same thing. Only popular myths easily promoted to the electrically naive say low voltage is hardware destructive.
 
westom said:
Loading is the problem. Accurate measurements of the power supply 'system' are only relevant when maximum load is applied. Prime95 obviously does not create anywhere near enough load to result in relevant numbers. But then weird_dave is so confused as to create more CPU heat – not put a large load on the supply ‘system’. Large numbers of self-proclaims computer experts only understand ‘more fan’. Have so little electrical knowledge as to not understand the difference between CPU loading and supply loading.
Click to expand...

Higher cpu load = higher psu load. its quite simple and seems like you're trying to over-complicate it. multi tasking wont do it. Chipset, ram, hard-drives, dvd-roms....they are all insignificant in all honesty and everybody here knows that. If you want to get a cpu working, you use a program designed to do just that. Trying to read from a dvd rom whilst playing a bluray from another wont do a thing, but 1 instance of prime or toast per cpu core sure as hell will, just like furmark's ability to mince ATi gpu's into a current-limited submission when you never EVER hit that limit while gaming.

Its not quite simple actually, it's very simple. furmark and 1x instance of toast per core will see the maximum power draw for any system,. beyond that, if you are really gunning for numbers then its overclocking and playing with voltages. reading from a dvd-rom wont test a thing expect the dvd-rom drive.

Amount of heat necessary to cause hardware damage is so much above those levels as to not be relevant. Heat does not cause lower voltage. In fact, heat may even create higher voltage. weird_dave has no basic electrical knowledge – is only preaching a very popular myth. Guarantee - he never designed or built computer hardware. Only the least informed would claim a hot CPU causes sufficient power supply loading to see what the OP is asking.
Click to expand...

Careful, they are switchmode power supplies remember. The cpu will need x voltage, the supply on the board will draw y amps to meet that demand. unless the heat output of those fet's is exactly linear to the power produced (ie, irrespective of the voltage/current relationship) then a drop in supply voltage and increased current draw as a result will lead to hotter running fets. Weird dave didnt mention hot running cpu's, he said hot running supply.
 
Last edited:
james.miller said:
Careful, they are switchmode power supplies remember. The cpu will need x voltage, the supply on the board will draw y amps to meet that demand. unless the heat output of those fet's is exactly linear to the power produced (ie, irrespective of the voltage/current relationship) then a drop in supply voltage and increased current draw as a result will lead to hotter running fets.
Click to expand...
Scary are so many posts without first learning basic technology.

For example, how to increase current. Short all power supply outputs together. That will destroy any power supply - according to popular myth and what was posted here. Reality - Intel even defines how large the wire must be to short all outputs together. No power supply must be damaged even when all outputs are shorted together. Any excessive current that would cause overheating cannot happen - if you had designed power supplies. Only popular myths worry about FET heating.

Now - let's say we use Prime95 to maximize CPU load. Therefore we have done nothing to sufficiently load the 3.3 and 5 volts. You call that maximizing load? CPU is at 100% internally when multitasking to everything. And creating additional maximum loads on the North and South bridges. And every else is also drawing power from the 12 volts, 5 volts, and 3.3 volts - maximum load. Prime95 does not do these important tasks - does not maximize the load.

CPU gets massive power from the 12 volts. But a power supply loading test must draw power from 12 volts, 5 volts, 3.3 volts, and 5 VSB. That means multitasking. Obvious with basic design knowledge and with actual hardware experience. Completely misunderstood by so many told to use Prime95 - only because they were told it is better.

But then the same myths also worry about power supply FET overheating - because the naive only understand heat - not electronic design or what a power supply actually does.

Prime95 is a CPU stress test mostly to maximize heating inside the CPU - mostly to confirm CPU timings and voltage threshold changes do not cause a CPU crash. It does not and cannot properly load the power supply. But again I must use the word obvious because so many are posting the popular urban myth - posting only what they were told. Posting without even confirming this with experience. How does Prime95 load the 3.3 volts for useful meter numbers? It *obviously* does not. Obvious when one has basic computer electrical knowledge, designed electronics, and from a few decades of that experience.

Scary also is that myth of overheating power supply FETs. Total nonsense if any power supply design knowledge existed. A perfect example of subjective reasoning resulting in a classic junk science myth.
 
If it wasn't such a pain, I'd shove a current clamp on the 5V and 3.3V lines and see just how little current is being drawn, but I'd have to fiddle with the braiding too much. Any takers?
A bit of board level IO isn't going to use that much power, neither do drives, ooh, half an amp on the 5V per drive.
By far the biggest chunk of power is on 12V which is why most sane people try to load that up first.

westom said:
How to identify insufficient wattage in a power supply. Create a load that is much larger – as big as possible. OS multitasking to the hard drive, USB device, internet, sound card, CD-Rom, video subsystem, DMA subsystem, timing subsystem, processing numerous interrupts from that sub-system, etc.
Click to expand...

I do have to agree that it would be nice to load up the 5 and 3.3, but it's tricky to do that without compromising the 12v load.
Back in post 13 I did actually say "You could do a DVD copy to the HD in the background".
Which is better than just playing back the oh so complex graphics in a movie as the drives will be working harder.
The highest throughput of IO is going to be when running benchmarks or stress tests of any sort, not browsing the web while watching a film and webcamming the budgie.
Sure, it doesn't load up on all parts of the system, but think about it, if I move CPU time to other parts of the system, I'm losing it where it's being used most effectively. Multitaking wastes CPU time and therefore less IO happens.

To say I have no electrical knowledge is rather amusing.
"he never designed or built computer hardware"
Well, I have, and designed switch mode supplies.
Also designed the engine management for this:
http://www.youtube.com/watch?v=tkV2xNGVP8o
 
westom said:
Scary are so many posts without first learning basic technology.

For example, how to increase current. Short all power supply outputs together. That will destroy any power supply - according to popular myth and what was posted here. Reality - Intel even defines how large the wire must be to short all outputs together. No power supply must be damaged even when all outputs are shorted together. Any excessive current that would cause overheating cannot happen - if you had designed power supplies. Only popular myths worry about FET heating.

Now - let's say we use Prime95 to maximize CPU load. Therefore we have done nothing to sufficiently load the 3.3 and 5 volts. You call that maximizing load? CPU is at 100% internally when multitasking to everything. And creating additional maximum loads on the North and South bridges. And every else is also drawing power from the 12 volts, 5 volts, and 3.3 volts - maximum load. Prime95 does not do these important tasks - does not maximize the load.

CPU gets massive power from the 12 volts. But a power supply loading test must draw power from 12 volts, 5 volts, 3.3 volts, and 5 VSB. That means multitasking. Obvious with basic design knowledge and with actual hardware experience. Completely misunderstood by so many told to use Prime95 - only because they were told it is better.

But then the same myths also worry about power supply FET overheating - because the naive only understand heat - not electronic design or what a power supply actually does.

Prime95 is a CPU stress test mostly to maximize heating inside the CPU - mostly to confirm CPU timings and voltage threshold changes do not cause a CPU crash. It does not and cannot properly load the power supply. But again I must use the word obvious because so many are posting the popular urban myth - posting only what they were told. Posting without even confirming this with experience. How does Prime95 load the 3.3 volts for useful meter numbers? It *obviously* does not. Obvious when one has basic computer electrical knowledge, designed electronics, and from a few decades of that experience.

Scary also is that myth of overheating power supply FETs. Total nonsense if any power supply design knowledge existed. A perfect example of subjective reasoning resulting in a classic junk science myth.
Click to expand...

Westom, your posting style is infuriating.. You do realise somthing like a TX650 corsair has somewhere around 28-30a available on the 5v rail. You tell me how much a hard drive draws from the 5v rail then tell me how many you'd need to stress a psu. It's insignificant and a waste of time trying to load those rails at the expense of maximum draw from the main components on a modern psu and I'll happily stand by this statement.

you take a very dim view on other people here and that annoys me when actually you've shown nothing of any real use.

To say I have no electrical knowledge is rather amusing.
"he never designed or built computer hardware"
Well, I have, and designed switch mode supplies.
Also designed the engine management for this:
http://www.youtube.com/watch?v=tkV2xNGVP8o
Click to expand...

Face it dave, you're a noob lol
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom