The Official PC building thread - 4th Edition

I've known for a while now, Just another post about a bottom of the barrel power supply Seem to be typically used in older HP's/compaq's. Although my 2001 also used a FSP PSU. They seem ok, provided you take care of them. I think my original one quit because the fan quit, causing it to heat up too much. I caught it too late...
That was when I was VERY green LOL! The air coming out of the current FSP seems somewhat warm, but not an eye opener. I'm gonna strongly recommend his upgrading it in the very near future.

 

Yes that would be accurate and is the best way to check total wattage but doesn't tell you what each rail is being stressed in your PS. For instance how many watts is the 5v pulling, 3.3v pulling, or any of the possibly 12v rails pulling? What are the max wattage for each rail and are they over stressed? And how efficient is the PS rated for?

Power supplies are very simple beast normally but can get to be involved with the newer switching PS's used today.

I've used 250w PS's in a dual core system without issue for a basic system, nothing fancy, but would prefer a 350w+ as a general rule. You can add up what the different devices draw and determine what you need pretty easily.

 

Could you recommend a device that could tell me just how much a PSU is being stressed?

I've used PSU calculators in the past

 

There is no device, single device, that will test all loads on your PS. To do so discretely without opening up you PS and adapting a test jig would require you to make several power adapters for each interface that would consist of a 1 ohm precision resistor that can handle high wattage in series with leads that a DVMM could read current and then parallel connected for voltage to get watts. This would be a real pain in the ars and I wouldn't suggest it. The PS would be much a easier approach but you would void your warranty and it still would be a bit of work. Again I wouldn't suggest doing this.

Here are the connectors, minimally, that you would have to make adapters for;






The best way to handle this is theoretically by manufacture specs for each device in your system i.e., normal hard drive draws around 10 watts. You would need to do this for all devices including mainboard, RAM, & CPU. Also if you have a bunch of USB stuff that would also have to be considered. It's not enough to know the device wattage you also have to know on what rails. Again a hard drive draws 10 watts typically on one of your 12v rails depending on your PS and your 5v, 10w total across the two rails.

EXAMPLES: (these spec's can be found on the drives themselves, normally)

Code:

Maxtor DiamondMax 10 Model: 6B250R0 IDE-133 (250GB)
 5V/ 740mA =  3.70w
12V/1520mA = 18.24w
TOTAL:       21.94w

Maxtor D540X-4K IDE (40GB)
 5V/ 800mA =  4.00w
12V/ 500mA =  6.00w
TOTAL:       10.00w

Seagate Barracuda 7200.10 Model:ST380815AS SATA-2 (80GB)
 5V/ 720mA =  3.60w
12V/ 520mA =  6.24w
TOTAL:        9.84w

Western Digital Caviar Blue Model: WD1600AAJB IDE-100 (160GB)
 5V/ 650mA =  3.25w
12V/ 500mA =  6.00w
TOTAL:        9.25w

Also consider how your PS handles all of it's voltage rails as some (most if not all) of them share max wattage rating(s). Your PS should tell you how it handles different power outputs and they may (should) derate these by 10% which is a design method. You should also try to stay within 10% of the rating if at all possible as it is a safe practice. Unless the manufacture has worst case ratings under high temperatures your case temperature can skew the device ratings as your devices will draw more as heat goes up.

I hope that helps,
Stevo

(Sept. 30th 2011) EDITED AFTERWARDS to correct 10w drive usage which was misleading.

 

It seems like I saw a video once upon a time, of somebody testing a PSU. It was very fancy stuff though. Something like you might see NASA using Obviously such an apparatus would cost more than would be worth it to me.

I agree about knowing exactly which rails are being taxed too. While I may not be taxing the 12V amperage/wattage, I may be borderline on the 3.3 or 5v line.

The bestec was really weird. I tested it with one of my voltage meters. Everything I tested showed 2.2v. Really weird...

I tested it by hotwiring the PSU. By bridging the green wire, and a black wire (on the 20/24 pin supply), it activates the power supply. Very good to know when one wants to test their power supply. It should be noted though, I've had PSU's test ALMOST perfect, but would not run a system. I did however note a flicker when running a 12v light bulb. Intermittent power.

 

Load testing banks are super expensive and complicated to use. Have had all of my PSUS tested at one time or another for voltage stability and max shut down wattage.

 

Not really a load tester, but a little cheapo unit like this has helped me diagnose bad PSU's in the past

http://www.newegg.com/Product/Product.aspx?Item=N82E16899705003

 

The cheapy testers are only positive when they actually test a PSU as bad. You can have a bad supply and those testers may show them as good.

Kevin, A switching supply will have a little flicker to as it is switching, not fully regulated like old power supply designs, and that is why you have regulators on your motherboards and tons of caps too.

 

Hmm, wonder why it was dead then :S I thought for sure the flickering had something to do with it. Oh well, it's in the past

 

The caps could be bad or the regulator in the supply amongst other stuff. Who knows without detailed diagnostics & testing....

 

Not necessarily as good, but working. I have seen where the tester reads that it is "working", but giving low readings. The PC wouldnt post, and swapping the PSU fixed it, So it "worked" but not as rated, or enough to power anything.

 

You can use a digital multimeter at certain points to test voltages but it can only tell you so much...

 

@Deadrum33, That proves my point.

@Estuansis, A DVMM is worst than the PSU tester as it only checks the voltage and doesn't put the PS under load like the tester does. So if you have a tester use it first, if you want to check if your voltages are fluctuating use a DVMM. The new testers even have digital read outs for voltage now so a DVMM wouldn't even be needed with them.

 

Toying with the notion of purchasing 1090t. I realize Bulldozer isn't far away, but i'd really like to see a performance boost with X264 encoding. I'd like the WD30EZRX, but I'm nervous about it operating with the supplied card that came with my WD30EZRS. Theoretically, it should work, given backwards compatibility. Surely WD30EZRX doesn't require a Sata III UEFI port...

I hope if I do get 1090t, it behaves similarly in regards to my Ram. Apparently my board supports DDR3-1600, but Phenoms do not. Which is why I'm underclocked to 1333. I gather the 1600 support was merely a gimmick, since no AMD processor at the time would support it.

I suppose the 1600 Support was limited to being overclocked though.

 

Ummm, processor memory speed support is just about null for AMDs. I'm using the same chipset as you Omega with the same CPU and my 1600MHz memory runs just fine.

 

Are you saying your memory is stable at 1600 speeds? I may have to try again :S

Sam, you ever have problems with your SD slot on your Dell? Mine dropped connection in mid transfer, now I have an error with that port. I'm hoping its either a registry or driver error. I'm gonna boot up my Velociraptor and see if the error is there as well. If the slot works, I'm not sure how I'm gonna uninstall it properly from my current OS/settings.

Booted up the Veloci, which was oblivious to my new Dell. The SD slot appears to be toast. Which is probably my own fault. It behaves similarly to a button. I was unaware of this. I slipped the card in til it stopped. So the connection was not tight like it should be. Because of this, I can never use the slot again I've never seen a slot quite like this one. I'm not returning the Dell, because of a simple SD slot! I'll live LOL!

I think it caused a problem on the controller card within the Dell. Because now I get a message in device manager that won't go away. No biggie.

 

Yep it runs perfectly in IBT for 10 runs on the high stress level. Plays all my games just fine too. Just running on the advertised settings as well. Mind you this is 1.52v memory meant for mid-year i5/i7 rigs, not 1.65v standard like yours. The only reason I can think that it would be unstable at 1600MHz is if your overclock is unstable. The memory speed itself should be entirely independent on a Black Edition. Try turning your CPU down to 3.8 when you try the memory at 1600. Of course I might try the opposite and downclock my memory and try hitting 4GHz.

 

Kevin, Dell drop off the monitor to you and take the defective one, iicr

 

I'm not sure what you mean :S

 

Shaff or myself?