Cheap/generic PSUs - any good? (was: Low power socket A)

[JAD]

<current-gen Intel chips use more power>


like I said one or the other in any given quarter........... until
right NOW...you guys love to spank the moment..then all goes silent on
the 'power' front and we start on the 'performance' kick
again....round n round it goes. Hey lets throw in 'die' size
now.............


OTSO PSU's, NEVER have I bought into the expensive PSU's and never
have I had a 'cheapo' one fail in fours years+(IMC systems rarely stay
in the same configuration for more than that, except for my 'museum'
pieces). I have been LUCKY. And in this market, with all the
precautions you take, there are no guarantees. Codegen has been the
absolute best for me, and whenever I can, I use them. However, I have
not used them for AMD machines. Sometimes its because after all the
'savings' the customers insist on expensive cases. Along with that
PSU output issue. Now I have not put a P4 "current-gen" together yet,
as it seems most people are waiting for the dust to clear, clearly
common sense.

 

[kony]

<current-gen Intel chips use more power>


like I said one or the other in any given quarter........... until
right NOW...you guys love to spank the moment..then all goes silent on
the 'power' front and we start on the 'performance' kick
again....round n round it goes. Hey lets throw in 'die' size
now.............

NO, it is not this quarter only.
It was true WITH NORTHWOODS too!
Due to motherboard design/development decisions, many Athlon
boards didn't make use of bus-disconnect HALT idling by
default so Athlons might 'Idle' hotter, but they did not use
more power. They have not used more power since the
Palomino days, and earlier Athlons... it's not one sided
though, previously the K6 used less than Intel's higher end
at the time. Specific chips must be compared instead of
making assumptions about power over a manufacturer's entire
line of chips, over multiple generations and revisions.

OTSO PSU's, NEVER have I bought into the expensive PSU's and never
have I had a 'cheapo' one fail in fours years+(IMC systems rarely stay
in the same configuration for more than that, except for my 'museum'
pieces). I have been LUCKY. And in this market, with all the
precautions you take, there are no guarantees. Codegen has been the
absolute best for me, and whenever I can, I use them. However, I have
not used them for AMD machines. Sometimes its because after all the
'savings' the customers insist on expensive cases. Along with that
PSU output issue. Now I have not put a P4 "current-gen" together yet,
as it seems most people are waiting for the dust to clear, clearly
common sense.

Early issues were that a cheap generic PSU mostly had it's
12V rail unloaded, with only a few hundred mA of fans and a
couple drives there was a lot of reserve capacity for a P4's
12V VRM power input. THEN it became more popular to have
more drives with them becoming more inexpensive, video cards
using 12V power arrrived in the market, and Athlons also
switched over to using 12V power for for CPU VRM. Today's
power needs are close enough that if a given PSU can't power
EITHER platform (to ignore which CPU is used), it's not
really suitable for the other, either.


Generic PSU had a golden era, when systems used less power
up through P6 and early K7, then marginal with newer CPUs
and video cards not needing external power. Today they show
more weaknesses than ever before simply due to increased
power usage of the *average* system... ignoring other
issues like fan longevity and failsafe measures.

 

[JAD]

sigh motherboard design sun spots Martian memory blah...never a
clear cut and never the same scenario...stop it already you know as
well as I that either of us could come up with a configuration that
would dispute the other.........except for the NOW....then it will be
later....

 

[David Maynard]

Every packaged PSU module we have used has worked that way - Our
product goes into industrial sites, and it is an unfortunate fact of
life that accidental human-induced faults do happen occasionally, and
in those cases we do want the SMPS to keep trying to bring up the
line, as long as there is no damage. If it needed power removed before
power would be restored, THAT would be a "logic error" and a major
PITA for our customers.

I see. Well, that would, or IMO 'should', be a second 'feature' rather than
inherent in the OVP/OCP circuit itself. I.E. some kind of timer as, IMO,
having the thing simply oscillate at the natural frequency of 'whatever'
doesn't sound like a 'design'.

What kind of application is this where 'humans' are routinely inducing PSU
faults?

 

[larrymoencurly]

Codegen has been the absolute best for me, and whenever I can, I use them.

Codegen is among the worst and contains some of the smallest
heatsinks, transformers, capacitors and output chokes you'll find in
any ATX PSU of a given power rating, and one person who dissected his
found that one of the voltage rails had a much higher amp rating than
the diode pack for that rail

I can understand using a Codegen by necessity, but why do that by
choice when so many really good PSUs are cheap? I haven't paid more
than $15-20 for any PSU in the past 2-3 years but have never bought
junk, unless you count a Soyo/Key Mouse/MaxPower that was free after
rebate, although I bought it for the case it came in (PSU was actually
OK, only it lacked an EMI line filter). The other brands were
Fortron/Sparkle, Delta, and, most recently, a 350W Antec.

 

[JAD]

I suppose I should have clarified 'the cases' but I assume they use
their brand PSUs in their cases. Either way, I never had one fail on a
Intel system. The oldest one I have now is a P3V4X and it runs very
well, that one has been running everyday for almost 5 years. I don't
buy into the expensive 'quiet' dual fan' 500w with lights and LED's,
and rarely have to buy a PSU separate of the case. again I know that
I have been 'lucky' as everyone ends up being...or not. Without doubt
YMMV it always does, for me it has been a good decision. They don't
look half bad...well most of the simpler designs look ok.

 

[kony]

I suppose I should have clarified 'the cases' but I assume they use
their brand PSUs in their cases. Either way, I never had one fail on a
Intel system. The oldest one I have now is a P3V4X and it runs very
well, that one has been running everyday for almost 5 years. I don't
buy into the expensive 'quiet' dual fan' 500w with lights and LED's,
and rarely have to buy a PSU separate of the case. again I know that
I have been 'lucky' as everyone ends up being...or not. Without doubt
YMMV it always does, for me it has been a good decision. They don't
look half bad...well most of the simpler designs look ok.

Lights, LEDs, cable sheaths, etc, have little to nothing to
do with PSU quality, as a "power supply" rather than a case
ornament... which is a 2nd function some people want, but
IMHO, never should a cent be spent on incorporating these
frills into a unit before it's actually capable of
sustaining the wattage on the label with acceptible
lifespan.

There are VERY few really fancy looking PSU that are even
any good, the average one is indeed the same POS that has a
fancy casing and then costs 50% more. Case in point-
Enermax. They've never had any models competitive with
Antec, Sparkle/Fortron, Delta, PC P&C, when comparing the
rated wattage, or amperage, yet they've been one of the
first to set the trend for PSU beautification, and where
priced according to that. They're not as bad as many
(really cheap) generics, but worse than other,
aforementioned name-brands at similar price points.

When dealing with unknown PSU, seek an online review.
Unknown generics made by same company, may wear quite a few
labels. Upon seeing the inside one can often note the
origin or at least assess certain components even from very
poor pictures, like size of transformer, inductors,
capacitors, and perhaps brand of fan. Ironically enough,
often the brand of fan is telling by itself, crap PSU almost
never use major fan manufacturer's products. One exception
there is that Sparkle used to use crap Yate Loon sleeve
bearing fans on their mid-grade OEM, lower wattage power
supplies. They were a failure liability unless relubed, but
so it is with almost every generic sleeve-bearing fan.

 

[kony]

sigh motherboard design sun spots Martian memory blah...never a
clear cut and never the same scenario...stop it already you know as
well as I that either of us could come up with a configuration that
would dispute the other.........except for the NOW....then it will be
later....

.... but it gets even worse for Intel now that AMD has 90nm
cores... 80 Watt difference in favor of AMD is nothing to
sneeze at;

http://techreport.com/onearticle.x/7417

 

[Tony]

Tony wrote:

I see. Well, that would, or IMO 'should', be a second 'feature' rather than
inherent in the OVP/OCP circuit itself. I.E. some kind of timer as, IMO,
having the thing simply oscillate at the natural frequency of 'whatever'
doesn't sound like a 'design'.

What kind of application is this where 'humans' are routinely inducing PSU
faults?

"accidental human-induced faults do happen occasionally" is not really
the same as "humans routinely inducing PSU faults". We have a large
installed base and we can't always control the quality of the
installation staff, and wrong connections still happen occasionally,
even during on-line maintenance. In all our cases, the result was
simply a relaxation oscillation at the protection circuit's natural
frequency - in the earlier units, about a 1 second pause between
re-tries, but in the later units (different power supply modules) it
was more like 100ms. But as long as the system came up without
intervention and nothing got too stressed, that was still fine with
us. I don't imagine that the designer set out to produce a specific
frequency - he/she is more likely to have simply made it safe; but
either way, it's really of no consequence to the application, and it
doesn't render the design any less valid.

Tony (remove the "_" to reply by email)

 

[larrymoencurly]

Lights, LEDs, cable sheaths, etc, have little to nothing to
do with PSU quality, as a "power supply" rather than a case
ornament...

When dealing with unknown PSU, seek an online review.
Unknown generics made by same company, may wear quite a few
labels. Upon seeing the inside one can often note the
origin or at least assess certain components even from very
poor pictures, like size of transformer, inductors,
capacitors, and perhaps brand of fan. Ironically enough,
often the brand of fan is telling by itself, crap PSU almost
never use major fan manufacturer's products. One exception
there is that Sparkle used to use crap Yate Loon sleeve
bearing fans on their mid-grade OEM,

Do you know of any good PSU review sites, other than Tom's and
SilentPCReview? Almost everybody else's "full load" tests consists of
just connecting the PSU to a computer and running a game, but that
rarely applies more than 250-300W.

 

[kony]

Do you know of any good PSU review sites, other than Tom's and
SilentPCReview? Almost everybody else's "full load" tests consists of
just connecting the PSU to a computer and running a game, but that
rarely applies more than 250-300W.

Unfortunately, no, the reviews I've seen do same, barely
loading a power supply then giving it 2 thumbs up! Most of
the reviews I've seen were not too good, mainly I prefer
those with hi-res pictures so we can see if there's been any
iimprovement or if they're still just slapping higher
wattage labels on same units.

While I don't claim the reviews are any good, there is a
large list here,
http://www.amdboard.com/psu.html