JD said:
Thanks again guys for all your help.
I revisited that computer with the Silencer power supply (ps)and found
some odd things. First, my tester indicated that the main supply from
the ps to the motherboard was ok - green light. Then I noticed that the
fan was not running. I tried the ps in several electric outlets away
from the strip and there was no response. I also used my multimeter on
the 2x2 connection (4 contacts) that gets connected to the board near
the CPU. All contacts were dead. I opened up the ps to check if there
were blown fuses. It is very hard to see everything there but I saw no
trace of a fuse. This ps has a 5-year warranty and there is at least 3
years to go.
The Intel D845 board needs unusual plugs from the power supply - I have
several Enlights but none has the 2x2 connection.
Now, while this project is developing I have another - to revamp my
older AL440 box and get it into a circuit with a new (Samsung) NetBook,
a new D-Link router, and a new DSL modem that I bought recently to
replace my dead Siemens Speedstream 4100 - well, it's alive apart from
the DSL light. ;-) Will start this project with a new thread. My
toughest problem will be dealing with Windows 7 (stripped!). I have been
using Win2K for over 10 years and, from what I have seen so far, the 7
or even a stripped 7, will be quite a change.
Have a great weekend
When you say "tester", it isn't clear what kind of tester that is.
That could be a device that plugs into the main power connector,
shorts PS_ON# to COM, uses window comparators on the main supply
voltage rails, and tells you whether each rail is within +/- 5%
of the right value.
A multimeter does simple things, like measure one voltage at a time.
It is a general purpose tester. It isn't that specific for power
supplies, and may need some other things connected to the supply
to make a test setup.
To make a PSU work, away from the PC, you have to connect PS_ON# to
an adjacent COM pin with a paper clip. That will turn on all the
rails. A purpose-built tester dongle, has that connection already
in place.
Without the paper clip in place, you should be able to measure and
verify the +5VSB pin. If +5VSB is good, then you can move on to the
main rails.
With the paper clip connecting PS_ON# to COM on the main harness,
you can go about measuring +3.3, +5, +12, -12V, on the main
cable, as well as picking up the 2x2 ATX12V connector and
checking it for +12V. The fans in the computer, run from the 12V output
carried on the main connector. The CPU runs from the voltage on the
2x2 connector. The power supply fan should start to run, when the
paper clip is in place. Some supplies may choose to not run the fan
at all when they're cold, but then, as the owner of the supply, you're
probably familiar with its habits in daily usage. A change of habit
is enough to suggest something isn't working right in the supply.
Power supply specs, old ones to new ones.
http://web.archive.org/web/20030424...org/developer/specs/atx/ATX_ATX12V_PS_1_1.pdf
http://www.formfactors.org/developer/specs/atx/ATX12V_1_3dg.pdf
http://www.formfactors.org/developer/specs/ATX12V_PSDG_2_2_public_br2.pdf
At least one poster, has corrected my description above, by suggesting
the power supply should have a slight "test load" connected to the supply,
when the PS_ON# jumper is in place. I have a load box I built here, which
would meet that definition, and that is what I use to ensure the power
supply stays within regulation. I've also tested a couple supplies without
a load on them, and they behaved themselves.
I know that in the past, some power supplies have had a set of
"minimum loading" current values in amperes printed on the label,
and the supply may not be guaranteed to stay within +/-5 % if that
minimum specified load is missing. About the absolute worst power supply
design I've seen, had a "minimum load" of roughly 25% of the rail capacity,
which is absurd. Many computers might not manage to present that amount of
loading in day to day usage, let alone constructing load boxes to meet the
load.
While a modern supply may not need a load to behave well (some actually have a
switchable load inside, to handle this on their own), you'll have to use your
own judgment as to whether that is necessary or a good idea or not. While I could
insist to you to "build your own load box" or "connect a couple old disk
drives you're no longer using", you'll have to decide yourself, what you want
to use. And "light bulbs" are not the best loading solution, as the
cold filament may double or triple the load current for the first few
seconds. Light bulbs are a non-linear resistor. Attempting to build a
100W load from light bulbs, might cause a 300W load for the first
few seconds. My load box uses regular power resistors I got at my local
(good) electronics store. A place like Radio Shack, wouldn't have a wide
enough selection of components, to carry good power resistors.
The power supply has one other pin of interest, which is something
like "Power Good". That is a logic level signal. The supply could be
merrily humming away, but if Power Good is not asserted, the PC won't come
out of reset, and the screen will remain black. You can check the voltage
level on that signal, just like any other wire, by using your multimeter
set to volts. The power supply specs above, will tell you whether that
signal is active high or not, in which case you'd be looking for a level
closer to 5V than to 0V.
When using the multimeter for tasks like this, I try to place the black probe,
onto a pin on an idle cable somewhere. Basically, what I suggest to you,
is don't get the tips of the red and black leads close together, because
you can short them accidentally. As part of my test procedure, first I
make sure the black lead is well away from where I'm measuring (but still
connected to a black wire or to the chassis), before starting a series of
measurements. I haven't had a short yet, by doing so.
I had a friend working in the lab with me one night, and there was a bright
flash of white light coming from his side of the lab. I ran over to see what
happened, and he'd shorted a 5V @ 100A supply to ground. The short point
was the leg of an integrated circuit. With 100A available, the leg on the
integrated circuit was vaporized instantly, leading to the burst of white light.
Very impressive, and no harm to my buddy. Some modern PC supplies
can do stuff like that too, as PC supplies are now up into the 1200W
range. So try not to get any test leads or probe tips, close together.
Either that, or as they say "pack a change of underwear" for when the
fireworks start
Paul
