Re: Somewhat OT - can underpowered power supplies damage a system?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, 2010-08-18 at 00:27 -0500, Thomas Cameron wrote: 
> Say, hypothetically, I have 8 identical Intel DG45ID motherboard-based 
> system with 8GB memory and a single drive with a cheap-o generic 300w 
> power supply.  All running Linux, so sorta on-topic.
> 
> Say those systems have been running for over a year, but lately they've 
> been really flaky.  Random lockups, sometimes they boot with garbage 
> characters on the screen instead of the normal OS boot, sometimes they 
> hang at various parts of the boot sequence, spontaneous reboots when 
> kickstarting.
> 
> Say the Antec web site says I need more like a 350w power supply, and I 
> read somewhere that PSUs tend to age and get less efficient over time, 
> so I am guessing that I've been starving these things for juice for 
> weeks or maybe months.
> 
> Hypothetically, could I have actually damaged these systems if the PSU 
> was under-powered?  Or would a new, higher powered PSU fix the problems 
> I've seen?
> 
> I would guess that too much juice might hurt the systems, but would too 
> little actually damage anything?
> 
> I know it's not thermal, they're in a very cold room with dedicated HVAC 
> and the airflow is great.
> 
> Hoping someone who knows more about electronics can chime in.

Thomas,

Your suspicions about cheap-o PSUs may be well-founded. A power supply's
Achilles heel is often the electrolytic capacitors used to absorb and
filter out high frequency components on its main power output lines.
There are a lot of square-wave-to-sine-wave conversion operations that
take place in modern power supplies.

When electrolytic capacitors (the big ones that look like beer cans) age
the paste-like electrolyte inside begins to dry out. The heat inside a
PSU accelerates this aging process. The telltale sign is AC ripple on
the +5V and +12V outputs. You can see this on an oscilloscope and
measure it with an inexpensive voltmeter set to read AC volts.

When 120VAC at 60Hz leaks onto DC power rails, the instantaneous voltage
(the sum of AC and DC voltages) applied to delicate logic may be many
times higher than the nominal low DC voltage that should be there. This
can easily confuse circuitry trying to measure the electrical charge
stored in a memory cell that represents a 0 or 1 bit.

I've seen this cause spontaneous reboots, system freezes, and all sorts
of seemingly random behavior. It's quite common in home PCs whose PSUs
have been inhaling dust and pet fur for 4-5 years. It's less common in
data centers, but equipment close to the floor can still accumulate a
lot of dirt. Lower temps in a data center may slow heat-related
deterioration, but cheap PSUs will still age faster than well designed
units with quality parts.

BTW, this problem is not confined to PSUs. You've probably read about
motherboards with bad capacitors. A few years ago there was a major
scandal involving defective capacitors sold to many top tier motherboard
manufacturers. Tens of thousands of motherboards made with those bad
parts are causing headaches for sysadmins today.

If you have any doubts about a particular machine, go ahead and replace
its power suppply with a known good one. If you can afford it and can
find one that fits, you might try replacing a cheap-o PSU with one from
a vendor like PC Power & Cooling.

Good luck.

--Doc Savage
  Fairview Heights, IL

-- 
users mailing list
users@xxxxxxxxxxxxxxxxxxxxxxx
To unsubscribe or change subscription options:
https://admin.fedoraproject.org/mailman/listinfo/users
Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines


[Index of Archives]     [Current Fedora Users]     [Fedora Desktop]     [Fedora SELinux]     [Yosemite News]     [Yosemite Photos]     [KDE Users]     [Fedora Tools]     [Fedora Docs]

  Powered by Linux