Archived from groups: alt.comp.hardware.overclocking.amd,comp.sys.ibm.pc.hardware.chips,uk.comp.homebuilt,alt.comp.hardware,alt.comp.hardware.pc-homebuilt (
More info?)
On Fri, 17 Sep 2004 16:56:49 GMT, Wes Newell
<w.newell@TAKEOUTverizon.net> wrote:
>On Fri, 17 Sep 2004 12:24:27 +0000, kony wrote:
>
>> On Fri, 17 Sep 2004 10:35:01 GMT, Wes Newell
>> <w.newell@TAKEOUTverizon.net> wrote:
>>
>>>That's your choice. Personally I've used nothing but for the last 15 years
>>>with no boards lost to PSU failures. And I'm talking about in the
>>>hundreds.
>>
>> ... and others have had opposite experience.
>>
>Could you name someone that's used hundreds and had them take out MB's?
How about the local shop I mentioned in next sentence VVVVV.
>
>> I could go to computer shop right now and see a stack of
>> motherboards damaged by generics. It's at the point where
>> technicians there check what PSU is installed as one of the
>> initial steps in troubleshooting.
>>
>It's more than likely that the MB (or installed card) failed, overloaded
>the PSU, and it failed also. And the reason you'll see more failures with
>generics is simple. There's many more generics than name brands. Alomost
>every case you buy comes with a generic PSU to remain competitive.
Most OEM boxes come with name-brands though, so it's not
necessarily true that generics have such a large share of
the market as you'd imply.
It is not more likely that the MB or installed card failed,
excepting cases where the motherboard had known flaws like
defective capacitors. Quite often after PSU dies, different
PSU is installed and rest of box still works, but
unfortunately not often enough.
>
>> If you have a specific PSU to recommend, for very similar
>> platform, and have had it running long-term, over a year,
>
>How about hundreds that's been running over 5 years, and some as much as
>10 years. The only PSU failure I've had in the last 20years in my home
>systems was one I caused. And I've given the last 3 system to relatives
>and they are still runing after several years.
You completely miss the point.
SPECIFIC instances of specific PSU are all that matter.
Generic PSU, on average have not increased in capacity to
match the average increase in power used by todays systems.
Some can cut it (barely, for a little while), others can't,
and a few will be fine with lesser loaded systems.
Do the math. Add up the power requirements of that 5-10
year old system. If today's box uses twice as much power,
all other things being equal, you need find a generic with
twice the true output for the same margin.
If you've only had one failure in 20 years, you either have
had very few systems over that time, good luck exuding out
of every pore of your body, or are full of it.
Again, it is not relevant how long yesteryear's systems ran
if the base design of the generic does not change but the
system power requirement does.
>
>> Given the brief description of the parts planned, many generics would be
>> of suitable true capacity, but so would a Fortron 300W, very quiet with
>> 12cm fan and only $27,
>
>And so would a $12 550W.
>Got one of these that worked fine with my power hungry A64 system.
A64 is not particularly power hungry. A Sparkle 250W can
run one unless it has more than a couple HDDs and a vidcard
also needing a few 12V amps. However, your system hasn't
been running for 2 years even, it's not old enough to be
telling of much... just about any generic that's sold will
run a box for a "while", else it wouldn't be marketable AT
ALL.
In fact, P4 uses as much power and there are SFF systems
with P4 in 'em, running off PSU not rated for half of that
"550W" fraudulent rating the generic wears.
As I mentioned previously, if you have a specific PSU you
can, with confidence, recommend based on it running a very
similar system (to the extent that power distribution among
the different rails is also similar), for over a year, that
might be relevent... at least it would suggest same
make/model might suffice, for a year. Success with same or
different generics running lower-powered old systems is not
relevant.