• Please review our updated Terms and Rules here

Why you don't use $20 power supplies

aaron7

Experienced Member
Joined
Aug 24, 2003
Messages
456
Customer came in today with a Compaq that someone put an "OKIA" power supply in.

Check the pics... yeah, those are capacitors floating around. And the ceramic choke is actually crumbling.

2010-12-23101720.jpg
 
Just out of curiosity, did the owner say how long it was in service when the PS blew?

Looking at user reviews from places like NewEgg, even good name brands have a small percentage that fail shortly after being put into service. This one looks like it had very poor circuit protection. And you never know if the owner did something they shouldn't have like forgetting to check the voltage selector (in the wrong position), or installing a video card with high power requirements, or a RAID setup with numerous hard drives.
 
Nah, brought in for no power.

System is very basic. Sempron, one stick of RAM, one optical, and one hard drive.

PSU was used for a month I guess.
 
Lol oh boy.

I've had some fun experiences with Bestec...but the components all stayed in their respective locations (not necessarily intact, but at least still soldered in place).
 
It used to be when capacitors could flop around, they'd be secured somehow (e.g. Ty-rap, hot glue, RTV) so they didn't flop around and cause trouble.

And that's the gripe I have with the "made in the Orient" designs. Electrically, they may be okay, but it's apparent that few manufacturers perform the needed environmental testing (e.g. shake table, oven, humidity, hipot etc. work). That takes money, I guess and delays production.

It isn't just the white-box off-brand stuff either. I've seen plenty of HP-branded LCD displays where the inverter and PSU are in a tightly shielded closed box and eventually cook the electrolytics--or one of the power semiconductors fails. Pop one open and look how many electrolytic caps nearly bump up against heatsinks.

Where have the design and QA engineers gone?
 
Hi
It may have used 'hot glue'. I've seen that quite often.
If you look carefully you'll notice that the PC board got hot
enough to melt solder. I'm not all the surprised that
the cap fell out and that the choke was a little flaky.
What I'd like to know is how it could have had this much power
going through it without popping the fuse ( assuming
it had a fuse ).
Dwight
 
Extended testing is something done on the higher end (expensive) power supplies. Generally when I buy a PS I just make sure it is rated for the application (for specific amp draw for each voltage) and that is has been approved by the regulatory agencies. Some makers don't even bother to have the supply UL/CSA aproved, so I asume they cut corners with the design itself and stay away. Properly designed supplies should fail without taking everything else in the computer with them, thats pretty much all I would ask for. There are plenty of cheap supplies in the $20-30 range (plus shipping) that would work in simple systems for years and allow for hardware upgrades along the way, but there are also supplies used by OEMs that will blow up if you add too many cards or drives to their system (designed on the edge of use).

I kind of wonder these days how many companies actually make power supplies and how many of the brands are just sombody elses product in different packaging.
 
I kind of wonder these days how many companies actually make power supplies and how many of the brands are just sombody elses product in different packaging.

There are some good sites about that subject, for instance:
http://www.jonnyguru.com/modules.php?name=NDArticles&op=Story&ndar_id=24

I noticed the very bad brands often change their names after a year or 2, presumably because their reputation starts being tainted because of failed units.

I had a convo with a friend of mine and his father a while ago where I tried to explain why I thought the problems he was having with his computer were caused by an insufficient PSU (It was a good make PSU, just very underpowered).

I told him some companies design and build a PSU the cheapest way they can, test it to find out it can draw at most 450W for a couple seconds before it self destructs, then they stick a label on it saying "600W POWER SUPPLY MADE BY HAPPY-SMILE"...they got the point and let me put a 600W FSP in his rig ;)
 
http://www.geeks.com/details.asp?InvtId=PS575XBK&cpc=RESX

575W

110V, 220V switchable power supply
50/60 Hz, 10A/5A
+3.3V, 38A
+5V, 40A
+12V, 25A
-12V, 0.8A
-5V, 0.5A
+5Vsb, 2A

http://www.geeks.com/details.asp?invtid=ESTAR-580-24&cat=PWR

580W
115VAC 7.0A 50/60Hz
230VAC 3.5A 50/60Hz
+3.3V, 35A
+5V, 45A
+12V, 19A
-12V, 0.8A
+5Vsb, 3A

2 supplies from different makers almost the same wattages (575 to 580) but the specs of each Voltage are a bit different. Notice how different the +12V line ratings are.
 
A couple of years ago I followed an Usenet discussion on power supplies. One guy who used to work as a parts purchaser at a major computer manufacturer retold his stories about choosing suppliers of power supplies. Just like it is mentioned in this thread, the number of watts specified on the cover doesn't have to mean a thing as to how it behaves under load. The guy who purchased parts soon came to learn to weigh the power supplies. His rule of thumb was the heavier supply, the more likely it contained useful electronics and was better built.
 
Actually, I think that that's not a bad rule of thumb.

I've got a couple of industrial PC supplies here and they're made with substantially heavier steel in the cases, so they do weigh quite a bit more than the average cheap PSU.
 
I have a $5 power supply I picked up at the local recyclers. It's oversized and heavy. But I can't find any information on the pinouts! It's definitely not ATX, even though it has what looks like a 24-position ATX connector on it.

It's a Compaq WTX460-0505. I think it was originally part of a dual-Xeon workstation.

Anyone familiar with this thing?
 
Yes, Chuck. Ten years ago I came across two of those Compaq PSU's. Thankfully I knew they were not ATX compatible so I didn't ruin any hardware by plugging it in on chance. I even had a pinout, but I'm not sure if I still can obtain it. The verdict was that some signals you expect to find on a regular ATX supply were missing and others were added, so it wasn't easy to rewire it.
 
I found this file, dated May 10, 2002:

Code:
 _________.-._________
| t s r q p o n m l k |         seen from above
| j i h g f e d c b a |
`---------------------'
 
Pin Dell Colour/Usage    ATX Colour/Usage       Dell shift+3
---------------------------------------------------------------
a   red     +5V          orange  +3.3V          n/a
b   black   GND          orange  +3.3V          n/a
c   red     +5V          black   GND            n/a
d   black   GND          red     +5V            +5V
e   orange  LO to HI *   black   GND            GND
f   purple  +5V          red     +5V            +5V
g   yellow  +12V         black   GND            GND
h   blue    -12V         gray    PWRGOOD        LO to HI
i   black   GND          purple  +5V stand-by   +5V
j   black   GND          yellow  +12V           +12V
k   gray    n/c **       orange  +3.3V/sense    n/a
l   black   GND          blue    -12V           n/a
m   black   GND          black   GND            n/a
n   black   GND          green   PSON           PSON
o   white   -5V          black   GND            GND
p   red     +5V          black   GND            GND
q   red     +5V          black   GND            GND
r   red     +5V          white   -5V            -5V
s   brown   n/c          red     +5V            +5V
t   red     +5V          red     +5V            +5V
 
*) may be PWRGOOD or may require a time delay circuit to power on
   ([URL]http://www3.sympatico.ca/mike.thang/mp3.htm[/URL])
 
**) may be PSON# ([URL]http://docs.us.dell.com/docs/systems/dmum/techovu.htm[/URL])
 
ATX Stand-by is 5V, 720 mA.
Dell purple is 5V, ~400 mA, while Dell gray (PSON) is 5V, 1.5 A

I haven't checked if the URL's still work, and take no responsibility for any mistakes in the above pinouts. "Dell shift+3" means what would happen if you on purpose misalign the connector into a regular ATX input, leaving 2x3 pins outside of the connector and another 2x3 pins not connected.
 
No, note even close--also note that mine's a Compaq, not a Dell. In addition, mine has a 6-pin +12 connector for CPU supply. In addition, there are 5 brown (3.3V) leads in the 24-conductor plug harness. The Compaq part number is 189643-004.

The big problem is that there are several Compaq "looks like ATX, but isn't" pinouts, so it's hard to know what's what.
 
http://www.geeks.com/details.asp?InvtId=PS575XBK&cpc=RESX

575W

110V, 220V switchable power supply
50/60 Hz, 10A/5A
+3.3V, 38A
+5V, 40A
+12V, 25A
-12V, 0.8A
-5V, 0.5A
+5Vsb, 2A

http://www.geeks.com/details.asp?invtid=ESTAR-580-24&cat=PWR

580W
115VAC 7.0A 50/60Hz
230VAC 3.5A 50/60Hz
+3.3V, 35A
+5V, 45A
+12V, 19A
-12V, 0.8A
+5Vsb, 3A

2 supplies from different makers almost the same wattages (575 to 580) but the specs of each Voltage are a bit different. Notice how different the +12V line ratings are.

Nothing too wrong there. I'd use those for anything up to pentium3/athlon. The important thing between both of those is the 12v. Buy the 25A one obviously.

But for a modern system, just buy either a corsair (rebranded seasonic bronze/silver), or a seasonic gold rated and be done with it. No need to fart around, just buy the best and steer clear of marketing that the other "gamer" power supplies have.
 
Back
Top