Why everyone is so PSU fanatic?

[Darkseid]

Why whatever problem any one has you guys ask him about his PSU and tell him to get more than 400w?? :eh?:
i have a Tualeron 1.0a@1.43 and never a problem. also 2 disk, cdrom sound card nic, etc... and my psu is only 250w and a friend has an athlon xp 1800+@1610 2 disk, cdrom, cdr-scsi with his own controller, soundblaster live, nic, and a Radeon8500 and a pinnacle tv decoder and never a problem either, i´m probably wrong but i think you are too extremist about PSU power, can someone give me some "real" explanation???? please?
Thanxs, this is something i always wonder

[OC Detective]

Well maybe the reason your friend hasnt had a problem is because he has a high quality PSU! Your 250W PSU is suitable for the system you have - newer chips need more juice

[rogerdugans]

Well, it may not fully explain it to you, but this is the way I see it:

AMD specs a MINIMUM of 300w for XP cpus. And as long as you have a decent one, that is enough to run at stock speeds- however:)

AMD chips since the Thunderbirds have been very susceptible to problems if the 5v rail of the power supply runs low: anything below 4.78 volts will usually cause the computer to be unstable; it simply needs more electricity than that, even at default speed.

Many of us here do NOT run our cpus at the speed AMD rates them for- we want MORE. If the cpu is sensitive to low 5v readings at low speed (default) it gets even more sensitive when overclocked: it uses more power and makes more demands on the power supply; it is common for the voltages to drop when overclocking, and if you cause the 5v to go below 4.78v...again stability goes away.

A power supply that is rated higher will usually perform better: capable of higher output, so less strain is put on it to deliver the needed power. Generally speaking you will be less likely to have a power supply problem with a 500w psu than a 300w one. Not a guarantee, but less likely.

That said, there are exceptions: I ran an XP1800@1700mhz with a 250w psu just fine for 6 months:D Power supply did end up dying, but since it was 4 years old or so and running almost continuosly the entire time, I think it did its duty just fine;)
I have also seen and heard of 450w power supllies that could handle an XP at default- bad equipment does occur- thats what RMAs are for!

Its a matter of playing the odds or wanting headroom for overclocking: a 300w unit MAY be enough for mild overclocking too, but it may not. A 400w or better is far more likely to be fine then a 300w, and give plenty of room for extra mhz:D

I do use 300w units in my Farm pcs: but, as an example I have two XP that have been run by my one "good" psu and now have 300w units:
XP1700@1710 did 1751 with a 430w psu but is MAXED out now
XP1700@1751 did 1800 with a 430w psu but is MAXED out now

Those computers have no cd roms, floppies, etc. Barebones running computers with 300w.

When they were in the primary pc they had floppy, cdrom, cdburner and two hard drives.

When you really push the hardware, it makes a big difference.

And it also sounds like you guys got lucky with a couple of good "little" units:D

[Darkseid]

yes that´s what i thoguht too, agree 100% but there is still something wrong here... 300 may fail, 400 probably good, that´s 100w more!! i think is just too much, i think that the real problem is in the design of the PSU´s and thery are to lazy or is a good selling point and don´t want to fix the 5v line, what do you guys think about that? think what other thinks you can do with 100w and here is just for some stability, for me that´s wierd

[OC Detective]

The ratings of 300W and 400W might not necessarily relate to the combined power output across the 3.3V and 5V this is why sometimes a good 300W might be better than a poor 400W - and like you said perhaps the rating of the PSU is a selling point and people sometimes dont read the information contained on the sticker on the PSU

[litghost]

typlically an increase of 100 w only increase the line about 33 w each which is little. but nowadays more psu a not fixed, and can redistrubet wattage accross different line. most people dont drain all lines evenaly

[ntpphong]

will the Antec True Power 430W do fine? =) it came with my AMG1080 server

[Darkseid]

so you guys agree that what should be done is a better design of the PSU and not just an increase in the rating? like really use the power where it`s needed, correct me if i`m wrong but i think that newer systems use way more the 5v and 3.3 line right? so maybe a 300w PS with better design could be enough for almost any system, it`s like the P4 Mhz vs Athlon PR, people likes big numbers...

[FunkDaMonkMan]

I just don't understand why people say you need at a minimum 400w when running a 1 600xp for example.

I run two 1600xp's, two video cards, one sound card, one network card, 11 fans, 2 cd roms, one hard drvie, and one CC on a 400w antec.

[papounet]

when in doubt, read the doc !!
here is the link
please check the total budget equation in the AMD doc
http://www.amd.com/us-en/assets/con..._docs/26003.pdf
It includes typical values for most things you plu into the PSU

you'll see you rarely need more than 12 Amps on the 12V line (use by CPU)
the reason the 5V rail is often a weakness IMHO, is because it used by the PCI/AGP acrd, most overclockers boost their FSB and overclock on top of that video card which are power hungry (ATI)

see the full series of post in http://forum.oc-forums.com/vb/showthread.php?s=&postid=1118922#post1118922

to see how one can attempt to narrow wether or not it is the PS which is the limiting factor.

[Darkseid]

ok i'll read that

[james.miller]

Originally posted by papounet
when in doubt, read the doc !!
here is the link
please check the total budget equation in the AMD doc
http://www.amd.com/us-en/assets/con..._docs/26003.pdf
It includes typical values for most things you plu into the PSU

you'll see you rarely need more than 12 Amps on the 12V line (use by CPU)
the reason the 5V rail is often a weakness IMHO, is because it used by the PCI/AGP acrd, most overclockers boost their FSB and overclock on top of that video card which are power hungry (ATI)

see the full series of post in http://forum.oc-forums.com/vb/showthread.php?s=&postid=1118922#post1118922

to see how one can attempt to narrow wether or not it is the PS which is the limiting factor.

cd-roms and hdd's also pull power from the 5v line

[daS_mAgg0t]

i was under the impression that the CPU was powered by the 5V+ rail ?

Someone told me that AMD state fluctuation allowanes on the Rail by 0.25v, i.e. maximum of 5.25v, minimum of 4.75v, anything above or below those numbers causes the CPU to 'cut out' and reboot. Which is why alot of people get random reboots when there CPU is under load.

I dont know too much about it, but i know that when i increase the load, my rail drops from 4.89 to 4.78...then shortly after that it reboots (its limiting my overclock :(), but when i use my bro's psu, the issues are gone, as the rail is steady at 4.96v under load.

[Darkseid]

yes i think something like that too

[daS_mAgg0t]

Well, if the 5v+ rail didnt power the CPU, then why would it lower when I load the CPU with prime95, thats what im thinking. I don't doubt that you guyz know more than me but...i'm kinda confused lol ?

[papounet]

Well from AMD, it clears the CPU is from the 12V.

but 2 plausible hypothesis:
a/ 5V is used to generate 3.3V used by RAM and Northbridge.
soo what are you are seing is usage by the CPU peripherals
b/ all lines in PSU/MB power circuitry are linked, if you pulled a lot on the 12V lines, It could be that voltage is lowered across all lines as the PSU/MB stress to feed the draw on the 12V.

We need a PSU specialist here.

[Darkseid]

"We need a PSU specialist here."

Yes Please!

[papounet]

To go back to your initial question:
"Why whatever problem any one has you guys ask him about his PSU and tell him to get more than 400w?? "

my own answer would be that :
for some (but not all), it is "when in doubt, increase the PSU" reflex.

It is due IMHO to a combination of good and less good reasons:
a/Power requirement of PC has increased over the last years because of CPU consumption and because of added peripherals (cd-writer, AGP-pro....), thus many of the budget overclockers have been obliged at one point or another to shell out more dough for a larger PSU, (I remember my 486-S25 PC had PS of 80W)
b/ PC components are very sensitive to quality of power (this is why the ATX specs was created and we have now a ATX-12 specification which provide for a separate plug to deliver all the amps for 12V to the MB (at leats one person wrote some time ago about the benefits of soldering the PSU to the MB to avoid the flaky contacts of the standard ATC connector)
c/PC components (CPU and RAM) operating beyond specs (overclocked/overvolted) need even more clean and stable power supply. By design, cheap PSU are less clean, thus you have to go for brand-name PSU, which do not come in the mass-market volume small PSU market
d/ when a PSU is pushed close to its limits or just above, it doesn't fail neatly, it just deliver less clean power, but how many people have the equipment, the time, the knowledge or the willigness to investigate such an issue
e/ bragging right
f/ easy answer "solve all, cure all"

in conclusion, I believe you are right to question why this pseudo-answer is so freely given whenever a problem occur.
I have recently posted several times on PS issues, including the link to the official CPU docs as I tend to prefer facts over "common knowledge".

I beleive some people have actually measure the real consumption of their "loaded" PC and have come with figure < 250 Watts mean consumption.

Shall we create a faction for "Quality PSU yes, bloated PSU NO" ?

[daS_mAgg0t]

You speak the truth, but your considering only watts, and not voltage control on the rails.

Consider comparing the following

Budget 350watt vs Antec 350watt

Both have the same combined total output, yet the fluctuation on the budget 350watt would be MUCH more severe on the rails than that of the antec more than likely.

When people like myself suggest it could the PSU, I refer to the built-in voltage regulation for the rails, as stability is 99% due to the rails when concerned with PSU's

For instance, i'm not even close to using 350watts, yet i need a new PSU cos of bad voltage regulation ;(

[nikhsub1]

Originally posted by daS_mAgg0t
You speak the truth, but your considering only watts, and not voltage control on the rails.

Consider comparing the following

Budget 350watt vs Antec 350watt

Both have the same combined total output, yet the fluctuation on the budget 350watt would be MUCH more severe on the rails than that of the antec more than likely.

When people like myself suggest it could the PSU, I refer to the built-in voltage regulation for the rails, as stability is 99% due to the rails when concerned with PSU's

For instance, i'm not even close to using 350watts, yet i need a new PSU cos of bad voltage regulation ;(
You are dead on. voltage regulation by the majority of PSU makers is 5%, hence seeing voltages at load at around 4.75, and still in spec. Antec True Power have a 3% tolerance, meaning you should never see below 4.85 and above 5.15. PC Power and Cooling is king with just 1% regulation (note the high price) so you should never see anything below 4.95 nor above 5.05. My PCP&C 5V NEVER MOVES, steady 5.01, Idle, load whatever. I have 4 hard drives, 2 CD Drives, Floppy, and an XP 1600 at 2.05V, 1.9G. My unit is a 450W and can power plenty more. Never underestimate a good PSU, after all it is the heart of the system.

[Darkseid]

Thanks to all for your replies, i started this thread because i`ve read too often people telling each other to get a bigger PSU for any problem of stability, we got a lot of technical details here but i`m still looking for something else, like a real way to check if the PSU is the problem, looking the 5v seems to be a good one, but anything else?? i mean, don`t just make pepole spend money in a new PSU if we can check before what is the problem, if is the PSU, then yes, but not "just in case"

[DemonicAardvark]

Would it be better to get a 430w antec psu or a 350w pc power and cooling psu?

[lee]

hehe, i had read here about PSU probs b/4 and determined i had a bad PSU............my 300W PSU was supposed to be AMD-compliant but when i had all my peripherals loaded up, turned out to be crap - i kept getting random blue screens under Win98SE......so i went and got an enermax 350W PSU and viola, never had a prob again........sometimes, after years of playing w/computers, things happen for no explained reason that can sorta clue you in on the power problem...........

[nikhsub1]

Originally posted by Darkseid
Thanks to all for your replies, i started this thread because i`ve read too often people telling each other to get a bigger PSU for any problem of stability, we got a lot of technical details here but i`m still looking for something else, like a real way to check if the PSU is the problem, looking the 5v seems to be a good one, but anything else?? i mean, don`t just make pepole spend money in a new PSU if we can check before what is the problem, if is the PSU, then yes, but not "just in case"
Easy answer here, use a multimeter to check the lines on the PSU. MOBO's can rob power, but, the PSU may be putting out the correct amount of juice. MBM may show one value but a multimeter is going to show a true value right off of the PSU. Bad voltage via multimeter = bad PSU. Bad voltages via MBM or bios = check PSU with multimeter, it is the only way.

[papounet]

in response to daS_mAgg0t and nikhsub1,

It may have seen like I was only discussing watts, but I was attempting to stress what I believe is a popular misconception :
"in case of problem, get a PSU with more watts".

Watts don't make a good PSU, good voltage regulation does
(thanks nikhsub1 for providing % target)

Put this PSU quality is not easily perceived:
Generic PSU with simple design and common-specs components will output slightly fluctuating voltage, which won't bother most systems.
a/ Some system pushed beyond design specs will be more sensitive to this so-so quality,
b/ the more the system will tax the PSU, the higher the fluctuation.

By getting a higher wattage generic PSU, people only push back the level at which fluctuation will arise. (a level which they'll never reach).
By going to a high wattage high-quality PSU, people get both the cleaner voltage and the capability to run extra large system.

My take is, based on typical max consumption for the devices we know, most systems don't need a large PSU but you can't a small wattage high quality PSU.

On a related subject, one should not forget that the M/B is also responsible for Voltage regulation (VDIM (3.3V) and Vcore).
A good M/B voltage regulation will barely filter some of the noise from low-quality PSU.
a so-so M/B won't do well even from a good PSU.

for some data, have a look at the Hammer board picture from
thom's hardware (yehh, I know)

http://www17.tomshardware.com/mainboard/02q4/021017/images/via_referenz2.gif

[papounet]

Here is an article from THG

http://www.tomshardware.com/howto/02q4/021021/index.html

we may discuss their evaluation of consumption of typical devices, but all in all, the high-end specs PC gobbles up power as a real one

Could a electronics specialist say if their way of loading the Powersupply is 100% similar to the actual load experience with active components ?

It is a pity the article does not address in details the fluctuations in voltage which, IMHO, occur when PSU are loaded.

One issue tHG points out correctly is combined load :
I have seen PSU able to deliver a lot on either 5V or 12V unable to do both at the same time. And this is perhaps the reason why people see more variation int heir 5V rail than in the 12V.
But my preferred reason for 5V irregularities is AGP cards :-).

[Darkseid]

i like the article but i don`t trust much in thg, they make too many reviews with too many things, but the article is good

[nikhsub1]

Originally posted by papounet
in response to daS_mAgg0t and nikhsub1,

...By getting a higher wattage generic PSU, people only push back the level at which fluctuation will arise. (a level which they'll never reach)...
This is absolutely not true. I had a 550W Enermax in my current set up. I got TERRIBLE fluctuation between idle and load. I also tested a 400W Antec (not TP) and had terrible fluctuation as well. With the 450W PCP&C, I get NONE and it is has LESS wattage output than the 550W Enermax.