Author Topic: Garbage power supply (Read 23528 times)

Phaedrus2129 · « **on:** Fri, 25 June 2010, 23:19:14 »

As you may or may not know I'm also into power supplies. You know, for PCs. I know which ones are good, which ones are bad, what differentiates a good power supply from a bad in terms of build quality and performance, I know a good many of the OEMs and platforms and who builds what and even a bit about the more common topographies.

You know, normal kid stuff.

Anyway, I recently decided to pick up a cheap-o power supply to compare/contrast with a high end power supply. I have an Antec EA430 PSU sitting around, so I decided to get a garbage 430W power supply. Namely, the Linkworld LBJ2 430W:
http://www.newegg.com/Product/Product.aspx?Item=N82E16817165023&cm_re=linkworld_430w-_-17-165-023-_-Product

I'll be posting my review of this ****fest example of cost cutting and lying soon. But as a teaser:

The primary rectifiers for 115V input are two IN5406 rated at 3A. There are two other diodes for 230V input that I can't ID, but are presumably rated at 1.5A. For 115V (US/Canada) input, we get 3A*115V = 345W. Assuming 80% efficiency (which is VERY generous) this PSU can only deliver 276W before these diodes burn. With a more realistic assumption of 70% efficiency, this PSU could only deliver 241.5W before these parts go up in flame.

This doesn't bode well for this PSU's odds of being able to deliver its rated wattage of 430W.

More when I upload pics and analyze the primary and secondary transistors.

Nonmouse · « **Reply #1 on:** Fri, 25 June 2010, 23:34:08 »

Ummm.... No offense, but starting out your calculations with "presumably" pretty much trashes your hatchet-jo... err... review. Myself, when I read a paper that uses the the word "assume" or any synonym, I mentally multiply or divide the conclusion based thereupon by a factor of ten. It has been, in my experience, much more accurate than the reported numbers.

Not that I doubt that this is a crappy PSU- but support it, don't just pull **** out of your ass.

Phaedrus2129 · « **Reply #2 on:** Fri, 25 June 2010, 23:38:01 »

Never did. My calculations were based on the 115V diodes. I presume the 230V diodes are rated a 1.5A, because that's what these companies usually do. But I never did any math or pulled any conclusions from the 230V diodes, only the 115V, which I know, concretely, are 3A.

When I present something as fact, I know it to be fact as far as is possible to check it. When I'm fairly certain, I'll mention it but won't try to do math based on it. When I really don't know, but am guessing, I'll preface it in a way you know I'm not sure.

Nonmouse · « **Reply #3 on:** Fri, 25 June 2010, 23:43:36 »

Nice editing...

gr1m · « **Reply #4 on:** Fri, 25 June 2010, 23:44:25 »

Unless you're doing it for your personal enjoyment, generic power supplies have been proven to suck before.

Phaedrus2129 · « **Reply #5 on:** Fri, 25 June 2010, 23:45:33 »

Quote from: Nonmouse;196567

Nice editing...

Don't get nasty mate. I changed so it was more clear, to address your objection.

Quote from: gr1m;196569

Unless you're doing it for your personal enjoyment, generic power supplies have been proven to suck before.

My personal enjoyment, and also to show some idiots at OCN. Plus a few other reasons.

gr1m · « **Reply #6 on:** Fri, 25 June 2010, 23:52:42 »

The biggest difference between a quality and generic PSU is their weight. My Corsair VX550 is as heavy as a Fatman from Fallout 3. A generic PSU pulled from an HP computer that I have lying around is about as light as the cloth bag Corsair packages their PSUs with.

/OT

Phaedrus2129 · « **Reply #7 on:** Fri, 25 June 2010, 23:56:37 »

Quote from: gr1m;196573

The biggest difference between a quality and generic PSU is their weight. My Corsair VX550 is as heavy as a Fatman from Fallout 3. A generic PSU pulled from an HP computer that I have lying around is about as light as the cloth bag Corsair packages their PSUs with.

/OT

Often true, but not always. Makalu (another PSU geek, only with a $5000 testing lab the lucky bastard) opened up a generic and found a bunch of iron weights, to increase the weight.

But yeah, generic PSUs (made by Linkworld, Leadman, Sun Pro, etc.) often cut out nearly a third of the components, and use lightweight heatsinks and small coils and all that, so they end up a lot lighter.

However, once you hit the high end weight's no good, since some of the best power supplies can be relatively light.

Nonmouse · « **Reply #8 on:** Fri, 25 June 2010, 23:59:31 »

Quote from: Phaedrus2129;196570

Don't get nasty mate. I changed so it was more clear, to address your objection.

My personal enjoyment, and also to show some idiots at OCN. Plus a few other reasons.

Like I said, I don't doubt it's a **** PSU, but base it on numbers or empirical evidence. Find the actual draw on the unit or/and blow the mother-****er up, or find out the actual rating on the components (the former being, obviously, preferable to the latter), but don't post stuff with "presumably", unless you expect to get responses like mine- or much worse.

gr1m · « **Reply #9 on:** Sat, 26 June 2010, 00:02:03 »

Quote from: Nonmouse;196575

but don't post stuff with "presumably", unless you expect to get responses like mine- or much worse.

Na, you've got it covered in the "**** comment" compartment.

gr1m · « **Reply #10 on:** Sat, 26 June 2010, 00:09:59 »

Obama wants us to eat da poo-poo?

To lick da anoos!?

Phaedrus2129 · « **Reply #11 on:** Sat, 26 June 2010, 00:10:28 »

Quote from: Nonmouse;196575

Like I said, I don't doubt it's a **** PSU, but base it on numbers or empirical evidence. Find the actual draw on the unit or/and blow the mother-****er up, or find out the actual rating on the components (the former being, obviously, preferable to the latter), but don't post stuff with "presumably", unless you expect to get responses like mine- or much worse.

What do you think I'm doing? I ID'd the 115V diode, looked up its data sheet, found the amperage rating, did the math.

And testing power supplies isn't simple. A full testing suite costs a minimum of ~$4000, if I go cheap and improvise. JG's lab costs ~$8k, and [H]'s cost around ~$10k.

Want more hard info? Here's the secondary silicon

SBL2045CT = 20A @ +5V = 100W
F12C20C = 12A @ +12V = 144W
S10C40C = 10A @ +3.3V = 33W

Leading to a total secondary wattage of 277W. Which matches up with what I found for the primary as well (though I have yet to analyze the primary switchers; shall I do it now?).

EverythingIBM · « **Reply #12 on:** Sat, 26 June 2010, 00:17:09 »

I'm actually very interested in your research on these PSUs. This is a new area for me, so, I'm ecstatic about any info I can get my hands on: especially if it's conveniently in geekhack.

I got some old certified data computer, and the power supply was made by "avance technologies", it's a strange PSU which actually shoots out a hefty amount of air compared to other PSUs I have; know anything about them?

An interesting thing is the delta ones in my intellistations don't automatically turn the computer on when plugged in; whereas, in the thinkcentres, soon as you plug the stupid things in, it causes the computer to boot. I'm assuming it's a PSU issue, but it could be the computer itself.

Phaedrus2129 · « **Reply #13 on:** Sat, 26 June 2010, 00:20:22 »

Yeah yeah, but feeding trolls is like feeding guppies, I just feel guilty if I don't give them something.

The primary switchers are regrettably (or thankfully?) overrated, 16A pulse mode, though I'm not sure of the voltage they'll be running at. Nor do I know what temperature they were rated at. So I can't draw too much from it, except to say that they probably aren't quite as bad as the primary rectifiers and secondary silicon.

Anyway, IBM, I don't know about "Advance Technologies". However, it probably uses a high-end fan; Delta, Sanyo Denki, NMB, etc. That would explain the higher airflow. I don't know what PSUs they use in Thinkcenters either, but it may be done that way by design.

EverythingIBM · « **Reply #14 on:** Sat, 26 June 2010, 00:26:41 »

Quote from: Phaedrus2129;196585

Anyway, IBM, I don't know about "Advance Technologies". However, it probably uses a high-end fan; Delta, Sanyo Denki, NMB, etc. That would explain the higher airflow. I don't know what PSUs they use in Thinkcenters either, but it may be done that way by design.

Actually it's "avance" not "advanced" (don't ask me why, I think it should be advanced, avance isn't even a word).
The fan is "jamicon", I never heard of them before. Probably some obscure defunct company lol.

Well the thinkcentres and intellistations both use delta PSUs, but they both behave differently, so, I'm guessing IBM told delta to cheapen the thinkcentres.

Nonmouse · « **Reply #15 on:** Sat, 26 June 2010, 00:27:42 »

Quote from: Phaedrus2129;196579

What do you think I'm doing? I ID'd the 115V diode, looked up its data sheet, found the amperage rating, did the math.

And testing power supplies isn't simple. A full testing suite costs a minimum of ~$4000, if I go cheap and improvise. JG's lab costs ~$8k, and [H]'s cost around ~$10k.

Want more hard info? Here's the secondary silicon

SBL2045CT = 20A @ +5V = 100W
F12C20C = 12A @ +12V = 144W
S10C40C = 10A @ +3.3V = 33W

Leading to a total secondary wattage of 277W. Which matches up with what I found for the primary as well (though I have yet to analyze the primary switchers; shall I do it now?).

Really, I don't mean to be too agro- more pointing out that posting numbers before you can back them up is just going to lead to people getting all up up your back. Assuming is always a bad way to start- what does it do to your numbers if those diodes happen to have 2, 4 or 10 times higher ratings? Can you be sure you can tell the difference by eye?

EverythingIBM · « **Reply #16 on:** Sat, 26 June 2010, 00:31:05 »

Quote from: Nonmouse;196589

Really, I don't mean to be too agro- more pointing out that posting numbers before you can back them up is just going to lead to people getting all up up your back. Assuming is always a bad way to start- what does it do to your numbers if those diodes happen to have 2, 4 or 10 times higher ratings? Can you be sure you can tell the difference by eye?

I'm perplexed. Just wait until the tests on the PSUs can be done instead of making it trivial. Obviously the information will be corrected as the test progresses, phaedrus was just proposing an hypothesis based on the current information gathered.

Phaedrus2129 · « **Reply #17 on:** Sat, 26 June 2010, 00:35:33 »

Actually this unit was already tested. Or rather the "500W" version. I was hoping I'd have a virgin, untested platform on my hands, but Gabriel Torres already got his hands on this one. And it does indeed go up in flames (figuratively) around 265W, and the component ratings are all the same (though a few in the 500W version couldn't be identified). The only major difference was the 230V rectifiers were the same 3A diodes as the 115V. On mine the 230V diodes are half the size, which is why I believe they are 1.5A. They cut the current in half and manage the same effective rating at 230V, and save $0.025 in the process.

Nonmouse · « **Reply #18 on:** Sat, 26 June 2010, 00:40:11 »

Quote from: EverythingIBM;196590

I'm perplexed. Just wait until the tests on the PSUs can be done instead of making it trivial. Obviously the information will be corrected as the test progresses, phaedrus was just proposing an hypothesis based on the current information gathered.

No, he was posting numbers. One doesn't (or shouldn't, at least) present results based on assumptions. A hypothesis would be worded quite differently. Say- "If the diodes were 1.5A, the unit should produce about 277 watts. However, as we were unable to verify this diode amperage at this time, we cannot rate the actual output of the unit. When we are able to either rate the diodes or the actual failure output of the PSU, we will publish accurate wattage numbers."

Phaedrus2129 · « **Reply #19 on:** Sat, 26 June 2010, 00:52:04 »

Read the post, I did rate the diodes, at least for the 115V. And that's what I did my math on. Unless the manufacturer is lying (very unlikely; that crap isn't tolerated in manufacturing circles) then the 115V diodes are rated for 3A, which means they can do 345W, which means between 240W and 280W after losses. Above 3A the diode will behave erratically, and will usually maintain that erratic behavior even at "safe" levels again, and a bit higher it will burn.

Nonmouse · « **Reply #20 on:** Sat, 26 June 2010, 00:56:28 »

Quote from: ripster;196571

I think lavender text is kinda gay.

Nope- so sorry to disappoint you.

Oqsy · « **Reply #21 on:** Sat, 26 June 2010, 02:12:29 »

Did someone page me to this thread? I HEART COMIC SANS AND ITALICS AND LAVENDER!1!11! I sure could use a mansicle right now... Ta ta boys!

kishy · « **Reply #22 on:** Sat, 26 June 2010, 02:40:13 »

*facepalm*

Even if he did present assumptions as results or facts, it's the internet...

This whole 'exact science' idea of testing power supplies bugs me anyway, just get a ballpark load figure for when it blows up and don't run it near the limit. Good enough for me.

Anyone happen to know what's up with the newer Thermaltake TR2 430W units? All I can find is info on older versions of the same model that claim different specifications from each other.

I know it's a discount unit but I'd kind of like to know what to expect from it.

Rajagra · « **Reply #23 on:** Sat, 26 June 2010, 03:12:07 »

PSUs are a bit like amplifiers. Anybody can build one, but not all can make them sound great. The difference though is that it is not always apparent that a bad PSU is the cause of problems.

If anyone is pulling numbers out of their ass it is the manufacturers of cheap PSUs. Not only do they make claims that are literally impossible to reach with the rated components, they use dodgy math. (E.g. they can supply 200W on the 12V line or 200W on the 5V line, and claim total power of 400W even though they can't supply both at the same time.)

ch_123 · « **Reply #24 on:** Sat, 26 June 2010, 05:26:36 »

I saw a comparitive review of a bunch of different top quality PSUs in a computer magazine once. For the lulz, they threw in a white-box 400W from Maplin. They increased the load on it (the were using special equipment to test it) and as it hit 150W it started whine and smell odd. When it hit 170W or so, it actually exploded... Says it all really.

The other trick used by cheap PSU manufacturers is the temperature one. The ability of a PSU to deliver power is affected by how hot its internal components are. Cheap PSU manufacturers often claim wattage at a temperature lower than the PSU would ever operate at. Good ones will rate for the temperature it is likely to be on at near maximum load, and the better quality components will be able to stand the heat better anyway.

Quote

An interesting thing is the delta ones in my intellistations don't automatically turn the computer on when plugged in; whereas, in the thinkcentres, soon as you plug the stupid things in, it causes the computer to boot. I'm assuming it's a PSU issue, but it could be the computer itself.

A lot of Dells do this too.

Phaedrus2129 · « **Reply #25 on:** Sat, 26 June 2010, 08:01:01 »

Quote from: kishy;196614

Anyone happen to know what's up with the newer Thermaltake TR2 430W units? All I can find is info on older versions of the same model that claim different specifications from each other.

I know it's a discount unit but I'd kind of like to know what to expect from it.

It's an HEC 350W. They keep changing the casing and load tables, but as long as the model # is "W0069**" or "W0070**" the internals are the same HEC garbage.

Though it might as well be a SeaSonic compared to this Linkworld.

microsoft windows · « **Reply #26 on:** Sat, 26 June 2010, 08:17:37 »

Power supplies are a great reason to get old computers. By the time a computer's 15 years old, you've got to know it's got a good power supply. And old computers use much less electricity.

ch_123 · « **Reply #27 on:** Sat, 26 June 2010, 08:25:05 »

They use a lot of power relative to how powerful they are. My phone is probably faster than a Pentium 2 machine, and it lasts 2-3 days on a small battery.

microsoft windows · « **Reply #28 on:** Sat, 26 June 2010, 08:26:48 »

Yeah, but it costs an arm and a leg to actually use a phone for something.

Phaedrus2129 · « **Reply #29 on:** Sat, 26 June 2010, 09:27:56 »

Quote from: ripster;196649

I have a simple Power Supply test.

If it is PC Power & Cooling - buy it.

If it is NOT PC Power & Cooling - don't buy it.

Bought a high end OCZ LED blinkomatic once and it fried itself in 2 years. Ironically enough OCZ bought PC Power & Cooling so I made sure I stocked up with an extra before Quality goes downhill.

Already has. I think high end OCZ LED blinkomatic is a good description for the current PC P&C Silencer Mk. II lineup. I mean, they aren't terrible, but the build quality and electrical performance are nowhere near as good as the original Silencer units.

But there are other good brands. Currently my "always good" list includes Corsair, Enermax, Antec, and SeaSonic. And there are good units from other brands floating around as well.

Phaedrus2129 · « **Reply #30 on:** Sat, 26 June 2010, 09:36:23 »

Yup, SeaSonic S12. Same as the Corsair TX650 and HX620 (only those two). They're nothing great performance wise these days, just average, but they're still very reliable, usually the only part that wears out during normal use is the fan. Of course, I think most of them are still under warranty; we'll have to see how these units do a few years from now.

Rajagra · « **Reply #31 on:** Sat, 26 June 2010, 10:07:42 »

I just did something really stupid. I went to check Microsoft's hardware compatibility list to see what PSUs it approved. As if they would test anything that critical.

ch_123 · « **Reply #32 on:** Sat, 26 June 2010, 10:46:54 »

Corsair is my brand de choix, they're easily available, and most of the models are Seasonic units.

kishy · « **Reply #33 on:** Sat, 26 June 2010, 13:45:08 »

Quote from: Phaedrus2129;196635

It's an HEC 350W. They keep changing the casing and load tables, but as long as the model # is "W0069**" or "W0070**" the internals are the same HEC garbage.

Though it might as well be a SeaSonic compared to this Linkworld.

Well, at least it's better than something...

It reliably runs my desktop with my ridiculous 22 amp graphics card (despite that being significantly over the 12V max according to the sticker) so I'm good with it...for now...

(22 amps is the figure specified on the box, I can't seem to find that repeated online though. Google results are suggesting 17A which I believe is my 12V max if the PSU sticker is to be believed)

ricercar · « **Reply #34 on:** Sat, 26 June 2010, 14:35:20 »

Seasonic
PC Power and Cooling

gr1m · « **Reply #35 on:** Sat, 26 June 2010, 17:56:54 »

I hate when people say "Corsair don't make the PSUs man so don't support them!"

Corsair:
- Puts the internals that they do not make in an attractive package that they do make
- Has an outstanding warranty
- Prices their units very well

I have every right to like Corsair. Also, the XFX 750W has brilliant internals apparently. My Corsair VX550 has 52A combined. That's better than your ****ty Thermaltake Toughpower 700W+s, etc.

bionicroach · « **Reply #36 on:** Sat, 26 June 2010, 18:09:59 »

Quote from: gr1m;196754

Corsair:
...
- Has an outstanding warranty
...

Seconded. I got a bunk HX 620 on my last system build which burnt itself up in spectacular fashion after running fine for about a month, but one call to customer service later, a brand new unit was on its way. The second unit has been rock solid for a good couple years now, so I'd definitely buy Corsair again.

You can end up with a dud PSU from any brand (who knows what happens to it in shipping, etc) so the most important thing IMHO is how easy the exchange process is.

ch_123 · « **Reply #37 on:** Sat, 26 June 2010, 18:12:15 »

Quote from: gr1m;196754

I hate when people say "Corsair don't make the PSUs man so don't support them!"

Well, very few do. I mean, doesn't PC P&C subcontract their production out to Seasonic as well?

bionicroach · « **Reply #38 on:** Sat, 26 June 2010, 18:24:14 »

Quote from: ch_123;196764

Well, very few do. I mean, doesn't PC P&C subcontract their production out to Seasonic as well?

Not sure about Seasonic, but as Ripster mentioned, PC P&C sold out to OCZ a while back for sure. Not that that means they have a bad product now, just not the same as it used to be.

Just another sign of the times trending toward everything being owned and made by one giant corporation. (skip to 3m07s - can't get the youtube tag to do it...)

Not a valid youtube URL

audioave10 · « **Reply #39 on:** Sat, 26 June 2010, 22:37:21 »

my .02 worth....
Corsair HX620 - 4 years of serious abuse
Corsair HX520 - 3 years of serious abuse
Antec 550 - 4 years of serious abuse
all are going strong still

I had an Enermax 550 die after one year which led to me going to the Corsairs.

gr1m · « **Reply #40 on:** Sat, 26 June 2010, 22:40:50 »

How can you abuse power supplies? Do you toss them in the oven every weekend? Maybe connect some tri-SLI GTX 480s twice a week for a torture test?

audioave10 · « **Reply #41 on:** Sat, 26 June 2010, 22:44:48 »

No man...just heavy gaming like modded Stalker series, Crysis, GTA 4, modded Oblivion and Fallout3 and also with big hungry video cards while both CPU & GPU are overclocked. Usual stuff.

keyb_gr · « **Reply #42 on:** Wed, 30 June 2010, 06:06:26 »

PSUs are interesting stuff.

Has anyone ever taken a closer look at what's inside one of these old OEM Astecs that were used ca. '99 (110 or 145 W stuff)? My box here has been running with a 110 watter modified with a YS-Tech fan for 5+ years, only the dust had to be cleaned out every one or two years. I once met a guy on the interwebs who swore by them, running a rather grown-up Athlon system using a 110 watter with only an additional 60 mm fan installed for better cooling.

Phaedrus2129 · « **Reply #43 on:** Wed, 30 June 2010, 06:59:24 »

A 110W is fine if you have a dual core, no dedicated GPU, one HDD, and don't do anything really intensive on it. For web browsing it would work fine. Though I don't see the point when you can get a Corsair CX400 for like $30.

Edit: Also, those old units are all geared toward delivering +5V power, not the +12V power modern systems need. That means that the wattage is effectively lower. A system like I described would pull maybe 70W during internet browsing, with approximately 50W on the +12V. On an old school 110W that would be approaching the +12V maximum. Also, if it's group regulated then the voltage regulation will probably be a bit wonky, which could cause problems. It isn't a horrible idea to run a system like that on an old 110W, but I definitely do not, do not recommend it.

didjamatic · « **Reply #44 on:** Wed, 30 June 2010, 07:08:16 »

10 years ago you could get good power supplies quite inexpensively made by Sparkle. I just did a search and it appears they are now "SPI" and there are many of them on Newegg with good ratings. Anyone used them here? I put a ton of these into service over the years and never had one fail. Though back then power requirements were much lower than today.

Phaedrus2129 · « **Reply #45 on:** Wed, 30 June 2010, 07:11:14 »

They're owned by FSP now, and most of what they sell are low-end FSP rebrands. Far from the worst out there, but nowhere near the tier they were at 10 or 15 years ago.

didjamatic · « **Reply #46 on:** Wed, 30 June 2010, 07:28:13 »

When I built my current system I had a PSU fail. When I got a replacement I ended up with a half aluminum, half clear acrylic LED infused PSU that has a knob on the back to adjust the fan, I liked that feature but the flo-green switch and neon blue were things I just accepted. I'm not a gamer and typically try to stick to server-class hardware built for stability and longevity but in this case I went with the bling-bling neon disco-of-the-future affair.

Found the model, it's made by Apevia. I don't know how good this PSU is, but it's worked well for over 2 years and I run a lot of drives and devices off of my system and I like the wrapped cables.

ch_123 · « **Reply #47 on:** Wed, 30 June 2010, 07:40:14 »

Well, if it has worked so far I'm sure it's alright. I just wouldn't touch a PSU like that with a ten-foot poll.

gr1m · « **Reply #48 on:** Wed, 30 June 2010, 07:43:21 »

It's only worked because you haven't run anything intensive on it.

gr1m · « **Reply #49 on:** Wed, 30 June 2010, 07:48:11 »

Besides, cheap power supplies don't have to explode. They can do other subtler things like kill your hardware (a motherboard and a stick of RAM were the victims of a 250W Bestec HP PSU).

News:

Author Topic: Garbage power supply (Read 23528 times)