Steven Whatley said:
Hello,
Is there a way to check the load on the power supply without having to
go out and buy a volt meter? I had a couple of sudden reboots
yesterday and would like to try rule out PS problems. Here is my
system config:
430W PS
P4 2.8GHz with HT (not overclocked)
1GB RAM (two 512MB DDR memory cards)
Albatron PX865 m/b
i865 chipset
3 case fans
4 hard drives (40GB WD IDE, 120GB WD IDE and two 250GB Maxtor SATA set
up as RAID 1 (hardware mirrored)
Sony DVD-ROM drive
HP DVD Writer 630i (does dual layer)
Matrox Parhelia video card
Audigy 2 ZS Platinum (with front side imputs in a 5.25" drive bay)
Hauppauge WinTV Theater tuner PCI card.
Promise FastTrax TX2200 SATA RAID card.
PalmOne PDA USB charging cradle (with external power connected)
USB conection to a HP psc 2210 MFC
As you can tell, it is a very full system. Just in the last couple of
weeks I added the second 250GB drive and upgraded my CD-RW to the DVD
writer. Do you think I need to upgrade the power supply?
Thanks,
Steven
Check to see if the motherboard maker has a hardware monitor utility for
download from their web site. You can then see if the voltages out out
of spec (i.e., too low). There is the MotherBoard Monitor (MBM)
utility, too. The author stopped supporting it a couple months back but
it still works for lots of motherboards, but I'd see if the mobo maker
already has one specifically to use with their mobo (and beware that
different mobos, even from the same maker, use different sensing methods
and have different translate tables for the thermocouple to equate to a
temperature, so you need to the right monitor utility for your
particular model). If you can't find a hardware monitor utility that
works with your mobo, get a multimeter (measures volts, amperes, and
ohms and hence the "multi" in multimeter).
Even if the voltages look okay when you happen to check them in the
hardware monitor, leave it running to alert you if voltages drop out of
range during use of your computer. You may, for example, find voltages
are okay for normal use but drop below minimums when you play a game.
There are utilities around for severe loading of the operating system in
an attempt to consume as much power as possible to let you know what
might be the worse case scenario for voltages. You might want to Google
for info on overclocking because I usually see such utilities mentioned
by overclockers wanting to figure out how much they have to tweak up the
voltages when they overclock the CPU and put it under load. I don't
overclock but I've been curious now and then and know such loading
utilities exist.
Volts measure pressure. Amperes measure flow rate. If you want to know
the load on your PSU, you need to measure amps. A decrease in the
voltage indicates an increase in the amperes but it is only an
indication. A 12V lead acid battery measure okay for voltage when
unconnected but may not provide any amps (i.e., it appears okay when
unloaded but fails when loaded). A good 12V battery with its terminals
shorted (with a really thick bus bar) will show zero volts but is
cranking out tons of amps.
If you want to measure load, you need to measure amps, which means you
have to insert the multimeter inline of the power. You'll need some way
of splicing into the power wire, like modifying a Y-adapter used for
power connectors, cutting the wire in the adapter, so you can attach the
multimeter between the cut wires. There might be specialty power
adapters that are made for this if you didn't want to make one from a
cheap Y-adapter power cable.
With AC current, you could measure with an amp probe that clamps around
the power wire. You can't do that with DC current so you have to
measure inline for DC volts/amps. Check that your multimeter can handle
the rated amperage output of the PSU. Many consumer-grade multimeters
only measure up to 2 amps (but might come with a shunt that lets you
multiply that max current).
Knowing the amps getting sucked out of the PSU doesn't mean much unless
you know how much that PSU is rated to push out each rail. If you were
sucking out 20 amps on a rail rated for 20 amps then you have no reserve
(unless the maker underrated their PSU). There's also the problem that
the rails are bridged and so the total wattage that can be drawn across
ac couple rails is less than their maximum amperage times voltage (and
how the cheapie makers will bogify their claims as to wattage). Some
PSUs have stickers on the side telling you what are the max amps per
rail but the cheapies usually try to hide or omit that info. Also, for
the cheapies, figure you really only have 75% of their rated wattage for
what you can actually draw from them. Some PSUs underrate themselves so
you can actually draw more power than they claim and do so safely. Some
PSUs rate themselves damn close to their max specs. And lots of
cheapies make claims based on bogus or misleading specs (much like LCD
monitor makers are doing today regarding response time).
There is quick and more non-intrusive way of determining how much your
PSU is spewing out for amps and thus its load *IF* you know the
efficiency factor of your PSU (some manufacturers will include that spec
on the PSU label or have it online). The efficiency is the ratio of
power into the power supply versus how much it pushes out (i.e., how
much is going in on the power cord and how much is going out its taps to
the connected devices). Use an amp probe to clamp around the AC power
line to see how many amperes are getting sucked into the PSU. However,
you need to clamp around only *one* of the power lines. If you clamp
around the hot and neutral (and ground) all inside the power cord,
you'll get a zero or inaccurate reading. Many amp probes come with a
receptable you plug into the wall and then plug the device into that
which allows the amp probe to clamp around only one of the power lines.
Or use an extension cord, cut it open to remove the outer protective
shell or split it so the power wires are separated, and then clamp
around each until you get one that measures amperage. Say you have 4
amps going into the PSU on the 120VAC hot side of the power cord.
That's 480 watts going into the PSU (P = I x E; RMS is not a factor
since line power is sinusoidal and probably in phase). At 70%
efficiency, the PSU is pushing out 336 watts. Remember that at 70%
efficiency for the PSU then 30% is getting dissipated within the PSU, so
the PSU is cooking off 30% of 306 watts, or 91 watts (think of a 100W
light bulb burning inside the case for the PSU). I have a Fortron PSU
and I consider it a good brand (because they underrate their products;
my 350W model can safely push out 430W) but it only is rated at 70%
efficiency. If it were a cheapie overrated PSU claiming 430W capacity,
I'd probably figure it could only safely provide 75% of that so I
wouldn't load it more than 322 watts. In this example, I'm already past
that so I would feel that I'm pushing that cheapie PSU close to its real
limits and probably can't add much of anything more to that system
without dropping voltages below their minimums and maybe adding too much
ripple.
You can find more or better hardware experts over in the
alt.comp.hardware newsgroup. Be more specific as to what brand and
model PSU you have. See if you can get Kony to respond. He's proven me
wrong lots of times. :-D I fab PCs for personal and work use and have
built many hosts but that's part of my hobby or setting up known
hardware for our alpha test lab. However, Kony seems to do this stuff
for a living.
