Underpowered PSU...Will it Hurt Anything?

[ArchAngel777]

Originally posted by: saiga6360
I find it alarming that the rated wattage of PSUs have grown so much and yet there is still a lot of confusion and conflicting reports relying on individual experiences. All I can say to this is that with the money I spent for the Corsair HX520W, it better last me at least 3 years with these specs:

Q6600 @ 2.4 Ghz
2x1 GB DD2 800 mhz
1 250 GB SATA2 drive
1 PCI-E graphics card (will attempt a leftover X1900 XTX, but will buy an HD 2950 Pro or 8800 GTS later to upgrade)
1 500 GB HP Personal Media drive
1 SB Xi-Fi XtremeGamer


If I go by http://www.extreme.outervision.com/psucalculator.jsp , it comes to about 410W so I should be fine if it can be trusted. Then again, that's the problem isn't it. The PSU should be ok if I go by word of mouth, manufacturer reputation and of course, the price.

I used that 'free' version of the calcuator and my system recommend 778 WATTs at 90% load. :thumbsdown:

My actual wattage from the AC is 280ish at GFX and CPU Load

 

[Yellowbeard]

Originally posted by: SerpentRoyal

Originally posted by: Yellowbeard

Originally posted by: SerpentRoyal
As for the noobs harping about exessing ripples, let's post some scientific tests showing the relationship between amount of ripples vs. useful life of GPU, HDD, etc. Don't have them? Then you're still blowing smoke.

Let's put it this way. No individual that I know of is going to invest time and money in setting up a system and letting it run for 24 hours a day until something dies due to a junk PSU. Manufacturers may do that but, they are not publishing their reports for others to read. So, there are not likely any publicly accessible test results showing exactly how much ripple will do what type of damage over any given amount of time to a specific type of component. However, this does not mean that it is not a factor and it is not smoke. You also don't have scientific tests showing the opposite so your point is equally unsupported and must be anecdotal or empirical.

The opinions in this thread are VERY clear. I personally am not trying to change any of them as it seems to be degrading into an argument of semantics. Cheap PSUs still suck. Underpowered PSUs have their risks. It's that simple.

My point is the reduction in mean life expectancy is small due to higher AC ripple. If there is a significant gain to be had with lower AC ripple, then you can be sure that the PSU vendors would use AC ripple as a key selling point. As of today, I don't know of a single PSU vendor with a guaranteed max AC ripple on the 3.3V, 5V, and +12V rails.

It's easy to sell snake oil to noobs.

Right. And some in this thread seem to be perpetuating the idea that this thread is filled with the intent to sell low ripple PSUs. No one in this thread is selling anything. Read the first post again.

If the HP PSU ends up being underpowered for the system with the upgraded video card...will it actually hurt any of the hardware?

The answer potentially is yes, PERIOD! "IF" is the key word. We still don't know the full power draw of his system but, he does have a quad core CPU so that is approximately 100-120w right there.

I can't figure out why this turned into an argument of minute points about who knows the most about what might possibly happen if ripple or gremlins or space invaders give our PCs bad power. This has gotten to the point of ridiculous. You cannot support the argument that an underpowered PSU is a good idea and has no potentially bad side effects.

In high performance computing, a safety margin is a good idea. No one here is trying to sell a 5000w PSU. They have simply advised to buy a sensible amount of margin for error"IF IF IF"!!! the HP PSU is not enough. You don't need a Harvard study to back that up.

As a support rep for a manufacturer, I don't try to sell PSUs, I am not paid for that. Check my posts, no pimping here. If you want to argue technical merits, call some PSU engineer that cares to hear it.

 

[ArchAngel777]

An issue that no one really touched on that I seen while skimming over this thread was that a poor PSU may not damage anything physically, but it is possible that a poor or under powered PSU will cause reboots and or lockups on the system and cause data lost via hard drive crashing. So, you may have corrupted data as a result and if your work is important to you, you will want a good PSU that is not over extended. If you don't care about data, or could care less about random reboots, go ahead and give it a shot.

Even so, I am not advising someone purchase a new PSU as a 'just in case' method. I would say try out your card, test it thoroughly with the current PSU and if it meets your stability needs, then I would say go ahead and work with critical data. But until you determine if it will be sufficient, play it safe and backup data.

 

[saiga6360]

I will definitely give it a try. I bought the X1900 XTX last year and never got a decent run with it. Either my machine was underpowered (Dell XPS with a non-standard PSU) or the card itself is faulty. I ended up getting a lower powered X1950 Pro but could not return the X1900 card. If it does work, my next problem is heat as it barely fits and the case has one rear fan. I will keep a close eye on the temps.

As for choosing the right PSU, I understand that there is ambiguity but I agree that I would rather go over than under. I don't keep up with PSU trends so I rely also on word of mouth, manufacturer reputation, and looking over the specs. I really don't see myself buying a 1000W PSU but you never know what's down the road. Hopefully PC parts will get less power hunger as manufacturing process improves.

 

[Bosconian]

Originally posted by: ArchAngel777

My actual wattage from the AC is 280ish at GFX and CPU Load

I hear you, but there's been more than a couple examples out there showing how the APFC circuit in power supplies "plays tricks" on measuring devices ranging from simple Kill-A-Watts to expensive clamping ammeters and that the results provided by said device is actually a number LOWER than actual DC output (putting the actual AC number a good 45% higher than it should be. The only real way to know how much power your PC needs is to measure it on the DC end.

 

[dclive]

Originally posted by: AmberClad

Originally posted by: dclive
Sorry if I don't buy into the 'everyone needs to massively over-spec their PSU'. I ran the PSU within spec - the Kill-A-Watt gizmo read around 250W or so, typically, under load; read a few prior threads for exact numbers, as I posted all of those #s too.

If you want to ignore the fact that it's been proven by PSU reviewers as well as user experience that many of the lower quality PSUs heavily inflate their true wattage, and that PSUs operate at a lower efficiency level when they're on the edge of their capability, then that's your prerogative.

Translation: It may cost another dollar or two per year when you're at 72% efficiency rather than 80%. I have no problem with that in the world, particularly since I won't be amortizing $100-$300 PSUs over a 2 to 3 year lifetime.

But I wouldn't go around recommending that people go buy up CompUSA/Acer Aspire PSUs (or whatever system that PSU of yours came with) to power their high end parts. Many of us are DIYers and don't have the luxury of a prebuilt-system warranty in the event of a major failure.

Sorry to hear that. But it's hard to argue with the fact that it worked, it worked very well, and it had no issues at all....and KillAWatt says it needed only 250W or so, so the 300W PSU was obviously plenty.

 

[dclive]

Originally posted by: JEDIYoda
The issue really isn`t how much is too much.....
The issue really is where are all the lower wattage PSU with the massive 12volt rails?
I am not talking about the 500watts PSU.
I am talking the 250 and 300 and 350 watt PSU`s.

That article in and of itself is all well and good.
Yet-- Here is a comment`s from a different forum on this very subject -- mainly centered on dclves comments --

You should be fine with the LiteOn 300W PSU and the 79xx series. I had a LO300W PSU with 4x500G, 4x1G RAM, dual-tuner, and 8800GTS/320 - with Quad Core 6600 (Q6600). Don't believe the silly hype about 'needing' huge (over 300-350W) PSUs.

We have comments from people who actuallt test PSU`s on the above statement says--
Hmm...

The Lite-On PSU's are generally pretty good. One 79xx, I can see it. But I'd have to question the validity of the claim of running an 8800. Note the poster said "had." For how long? Had for one month before upgrading it? If it was "had for a year" I'd scream "shens."

I had what I wrote above in it....for perhaps 2 months or so, then decided I needed a media center, and I needed my VMs to have more headroom, so I put 6G into the quad box, ran a 64 bit os, and made a seperate MCE box. The all-in-one just didn't work out.

Please read AT's articles on actual wattage used with the 8800s. You'd learn a lot - namely that a 300W PSU is plenty.

 

[ArchAngel777]

Originally posted by: Bosconian

Originally posted by: ArchAngel777

My actual wattage from the AC is 280ish at GFX and CPU Load

I hear you, but there's been more than a couple examples out there showing how the APFC circuit in power supplies "plays tricks" on measuring devices ranging from simple Kill-A-Watts to expensive clamping ammeters and that the results provided by said device is actually a number LOWER than actual DC output (putting the actual AC number a good 45% higher than it should be. The only real way to know how much power your PC needs is to measure it on the DC end.

Oh the device I used is not a 'simple kilowatt meter'. It is an APC unit that actively reads the wattage change up to several times per second. it fluctuates immediately upon load and non-load and whether I turn something off or on...

Plus, it is rather ridiculous to think that a PSU pulls more power than what you are pulling the outlet... Do PSU's magically create energy? That isn't directed at you, BTW... Again, if I was using a simple meter designed to view wattage at a slower rate, I might buy (figuratively) it, but seeing as my APC unit is providing it VIA battery at a very high rate, I highly doubt they are going to configure the meter to UNDERRATE my system... IF anything, they will overrate it, trying to sell a bigger unit (talking UPS). Anyway, the unit is pretty accurate... I have seen these 'spikes' people speak of, but basically are only a 10-15% over standard wattage under load.

 

[JEDIYoda]

Originally posted by: ArchAngel777

Originally posted by: Bosconian

Originally posted by: ArchAngel777

My actual wattage from the AC is 280ish at GFX and CPU Load

I hear you, but there's been more than a couple examples out there showing how the APFC circuit in power supplies "plays tricks" on measuring devices ranging from simple Kill-A-Watts to expensive clamping ammeters and that the results provided by said device is actually a number LOWER than actual DC output (putting the actual AC number a good 45% higher than it should be. The only real way to know how much power your PC needs is to measure it on the DC end.

Oh the device I used is not a 'simple kilowatt meter'. It is an APC unit that actively reads the wattage change up to several times per second. it fluctuates immediately upon load and non-load and whether I turn something off or on...

Plus, it is rather ridiculous to think that a PSU pulls more power than what you are pulling the outlet... Do PSU's magically create energy? That isn't directed at you, BTW... Again, if I was using a simple meter designed to view wattage at a slower rate, I might buy (figuratively) it, but seeing as my APC unit is providing it VIA battery at a very high rate, I highly doubt they are going to configure the meter to UNDERRATE my system... IF anything, they will overrate it, trying to sell a bigger unit (talking UPS). Anyway, the unit is pretty accurate... I have seen these 'spikes' people speak of, but basically are only a 10-15% over standard wattage under load.

Acctally not true at all!

Think about what you just said.....lol.....APC unit that actively reads....rofl....think hard i am not going to tell you......hehehee

 

[ArchAngel777]

Originally posted by: JEDIYoda

Originally posted by: ArchAngel777

Originally posted by: Bosconian

Originally posted by: ArchAngel777

My actual wattage from the AC is 280ish at GFX and CPU Load

I hear you, but there's been more than a couple examples out there showing how the APFC circuit in power supplies "plays tricks" on measuring devices ranging from simple Kill-A-Watts to expensive clamping ammeters and that the results provided by said device is actually a number LOWER than actual DC output (putting the actual AC number a good 45% higher than it should be. The only real way to know how much power your PC needs is to measure it on the DC end.

Oh the device I used is not a 'simple kilowatt meter'. It is an APC unit that actively reads the wattage change up to several times per second. it fluctuates immediately upon load and non-load and whether I turn something off or on...

Plus, it is rather ridiculous to think that a PSU pulls more power than what you are pulling the outlet... Do PSU's magically create energy? That isn't directed at you, BTW... Again, if I was using a simple meter designed to view wattage at a slower rate, I might buy (figuratively) it, but seeing as my APC unit is providing it VIA battery at a very high rate, I highly doubt they are going to configure the meter to UNDERRATE my system... IF anything, they will overrate it, trying to sell a bigger unit (talking UPS). Anyway, the unit is pretty accurate... I have seen these 'spikes' people speak of, but basically are only a 10-15% over standard wattage under load.

Acctally not true at all!

Think about what you just said.....lol.....APC unit that actively reads....rofl....think hard i am not going to tell you......hehehee

Well, the way the APC works is that the power charges the battery while the battery feeds the power to the PC. It reports the watts drained from the battery at any given time on the LCD display that it has. So if the unit had to supply more than it's rating (780 Watts) then the unit would blow the built-in circuit breaker.

In any event, I am not sure why you would suggest that the wattage drain on my UPS would be incorrect.

So, if you have something to say, tell me... Because as I understand it, I am not sure how it could be wrong.


Edit ** And I already read through Johny Guru's thread that you posted. He didn't state all power supplys cause this 'issue' and implied it was a minority of PSUs. At least that is what I gathered from his multiple posts within that thread. He also said that there is a way to accurately measure AC consumption, just not something he has access to, or something he even knows about.

I suppose the way to test it wuld be to figure out how many Watt hours my battery backup is rated for and unplug it from the wall at idle and see how long the PC lasts then average it out.

 

[Bosconian]

Yeah... I don't think the APC measures any differently than a Kill-A-Watt, Power Angel, ammeter, etc. So who really knows?

 

[JEDIYoda]

Topic Title: Mythbusters--SLI Power Consumption Results - se link to Jonny`s site!
Topic Summary: Addendum to the Underpowered PSU...Will it Hurt Anything?
Created On: 10/16/2007 11:51 PM


http://www.jonnyguru.net/forums/showthread.php?t=3137

 

[dclive]

Anandtech reviews of power consumption, idle and under load:

http://anandtech.com/cpuchipse...howdoc.aspx?i=2903&p=7

Clearly shows under-load maxes at around 200-250W for a modern CPU and a normal 8800GTX setup.

Consider that Anandtech has no significant power supply advertisers to the extent JG has, has no vested interests in selling more power supply stuff, and these numbers generally agree with what most other internet links (and my own Kill-A-Watt results) say.

 

[ArchAngel777]

Originally posted by: Bosconian
Yeah... I don't think the APC measures any differently than a Kill-A-Watt, Power Angel, ammeter, etc. So who really knows?

That is a question I would like answered... But until someone can provide proof, I am more inclined to believe that the APC units read wattage differently, mainly because they are not just a pass through device, but are actually supplying the wattage via their battery.

 

[Yellowbeard]

Originally posted by: ArchAngel777

Originally posted by: Bosconian
Yeah... I don't think the APC measures any differently than a Kill-A-Watt, Power Angel, ammeter, etc. So who really knows?

That is a question I would like answered... But until someone can provide proof, I am more inclined to believe that the APC units read wattage differently, mainly because they are not just a pass through device, but are actually supplying the wattage via their battery.

I don't know the answer but, it seems as if it would depend on whether the measurement is being taken at the wall (AC) side of the UPS or at the output side (DC of course). Interesting question IMO. I guess you could unplug it from the wall and see what your power draw info indicates while the unit is powering the PC WITHOUT AC feed. Not sure what that would tell you but there may be some info to be gained.

Again, we have wandered VERY far from the original topic here. Cheap PSUs still suck and an underpowered PSU, regardless of its rating, does pose potential risks.

TYPO: Should read "without" AC feed.

 

[SerpentRoyal]

Antec Earthwatts 430/500 for $30 to $50. It doesn't get much cheaper than this unless it's a black friday special. Both should be able to run any single GPU rig. Watch for sale. Earthwatts was on sale last week at Frys for $50 out the door. 430 was on sale this week at NewEgg for $30 AR.

 

[Yellowbeard]

Good pricing on the Antec. One of those may be suitable IF the OP here finds his unit to be underpowering. Also, I enjoyed this thread:

http://forums.anandtech.com/me...id=84&threadid=2097209

 

[JEDIYoda]

Originally posted by: Yellowbeard
Good pricing on the Antec. One of those may be suitable IF the OP here finds his unit to be underpowering. Also, I enjoyed this thread:

http://forums.anandtech.com/me...id=84&threadid=2097209

Whats interestying about that thread is most people are using a 400--700watt PSU with the difference split down the middle...

Thus can we conclude that the sippose argument on this thread is shall we say a moot argument?

Plus it would seem those expousing using only a 200-300 watt PSU even for stuff now a days are a minority at best!!

 

[tomoyo]

Originally posted by: dclive
Anandtech reviews of power consumption, idle and under load:

http://anandtech.com/cpuchipse...howdoc.aspx?i=2903&p=7

Clearly shows under-load maxes at around 200-250W for a modern CPU and a normal 8800GTX setup.

Consider that Anandtech has no significant power supply advertisers to the extent JG has, has no vested interests in selling more power supply stuff, and these numbers generally agree with what most other internet links (and my own Kill-A-Watt results) say.

I'd have to agree with dclive on his basic point, a HIGH QUALITY 300-350w psu can handle a lot. Basically any non-sli system out there even. Now high quality to me means, Seasonic S12-II 330, Earthwatts 380, and the highly efficient FSP's.
Problem is that people are affected by all the junk psu's out there. I would say any junk psu should immediately be treated as half the wattage at best. So those fake 500W psu's should really be thought of as 250 or less. That would be a fair comparison imho, junk psu/2 = good psu wattage. The problem that continues, even for corsair and other makers...is nobody really puts out many high quality 300-400w psu's. Almost everything goes into the 500-750 range these days. The vx450 is still kind of overkill...but with how marketing is, It's understandable why corsair doesn't want to try to do anything lower.

 

[dclive]

Originally posted by: JEDIYoda
Plus it would seem those expousing using only a 200-300 watt PSU even for stuff now a days are a minority at best!!

A minority too is the number of people who understand how to read a memory dump, but that doesn't make it any less correct and useful. Popularity is a function of marketing, not need or anything else.

Read the Anandtech review URL I posted for actual specs on what's really required. Clearly, 300W is more than enough for the 8800GTX systems they tested.

 

[tomoyo]

Originally posted by: JEDIYoda

Originally posted by: Yellowbeard
Good pricing on the Antec. One of those may be suitable IF the OP here finds his unit to be underpowering. Also, I enjoyed this thread:

http://forums.anandtech.com/me...id=84&threadid=2097209

Whats interestying about that thread is most people are using a 400--700watt PSU with the difference split down the middle...

Thus can we conclude that the sippose argument on this thread is shall we say a moot argument?

Plus it would seem those expousing using only a 200-300 watt PSU even for stuff now a days are a minority at best!!

Ahem also I'd like to mention my secondary system DOES use a 300W psu.

 

[Yellowbeard]

The problem that continues, even for corsair and other makers...is nobody really puts out many high quality 300-400w psu's. Almost everything goes into the 500-750 range these days. The vx450 is still kind of overkill...but with how marketing is, It's understandable why corsair doesn't want to try to do anything lower.

Well, it's not just marketing. I don't sit in on the decision making for PSUs but, the VX series, the 450 in particular is a solid unit. It's quiet enough for use in an HTPC, and has plenty of power for any single system whether it is a Celeron based system with onboard video on the MOBO all the way up to a Quad 6850 and ATI 2900 XTX or whatever their top card is.

This makes sense from a manufacturing standpoint for several reasons. We don't have to build 30 different models to make every single niche user happy.

Also, there is a price point on the manufacturing end ( and no, I don't know what it is ) but, a PSU is going to cost a bare minimum of "X" amount of dollars to make. So, for example, a 300w PSU for example may cost as much to make as a 400w PSU due to this minimum cost. So, why not build a 400w unit as this covers a larger customer base?

This really isn't a problem per se. I can't understand why this is such a big deal to argue over. Why does Detroit kick out cars that can go 120mph when the speed limits around the nation are about 1/2 that? Why do I have a 3000' sqare house and there is only 3 of us? Why did you get your chocolate in my peanut butter? Who cares. Cheap PSUs still suck and underpowered PSUs do have their risks.

 

[SerpentRoyal]

A well-built 500 watts PSU shouldn't cost more than $30 to manufacture. We're talking about 100% Japanese electrolytic caps and a quality cooling fan. Knock-off another $3-$5 max for a 380 watter.

 

[Yellowbeard]

Originally posted by: SerpentRoyal
A well-built 500 watts PSU shouldn't cost more than $30 to manufacture. We're talking about 100% Japanese electrolytic caps and a quality cooling fan. Knock-off another $3-$5 max for a 380 watter.

Just curious, how did you come up with this number of $30?

 

[dclive]

I'd be surprised if it's anywhere close to $30. I would've thought more like $8-10, if that.

Note: Manufacture. There are many other costs besides that of manufacturing. Plant costs (or contract costs), transport costs, marketing costs, staff and support costs, etc. add up. I suspect non-OEM PSU manufacturing is highly profitable.