wrote:
>> So tell us how the motherboard LED is behaving.
>>
>> If the motherboard LED is blinking as well, I would try another power
>> supply. Maybe you are using the same model and brand of supply on
>> both computers ? Try something different.
>>
>> Paul
> After the shutdown everything seems normal. About 2 minutes later the
> speakers start clicking in sequence with the LED hard drive caddy.
> The LED on the M/B is solid green, no blinking. What I just noticed
> is the CPU fan is jerking with the speakers!! It does not spin, it
> just moves about 5 degrees and then returns back. If I leave the
> system off the speaker and hard drive will finally stop in about 2-3
> minutes. The two LED's on the NIC are solid orange during this whole
> issue.
>
> The system I have open right now has 1 IDE hard drive, ATI 9800 Pro,
> Antek SL400 power supply and no other cards. 1gig ram and no USB
> devices.
>
OK, how this works, is the power supply has a logic input called
PS_ON#. It isn't really a logic input, as you aren't likely to find
a TTL logic chip actually in the power supply. But the input is
supposed to roughly follow TTL logic levels.
+5VSB
|
/ Pullup
\
PS_ON# / |\
+------->>----+---| \--- PSU
/ |/ Enable
Super_I/O --- Southbridge ----|< | Receiver
\ <-Mobo|PSU->
GND |
The PSU has a pullup resistor on the PS_ON# signal. What that means
is, if the main connector was disconnected from the motherboard, and was
just sitting in the air, the pullup keeps a logic 1 signal on
the PS_ON# signal. The PS_ON# signal is active low (i.e. at ground
level) and won't be on if the main cable is disconnected. The resistor
helps keep it deactivated.
The job for the motherboard is pretty simple. The Southbridge and
the Super I/O contribute reasons why the power should be on or off,
depending on whether something is waking the computer, or shutting
it down. There will be a driver chip or even a simple transistor circuit
of some sort, suited to driving that pulled up logic signal. Usually the
device will be open collector. What this means, in fact, is that you
could ground the PS_ON# signal without hurting anything. That is because
the driver cannot make a logic one on its own, and relies on the PSU to
provide a pullup to a logic 1 level.
So, how could we explain an oscillating power supply ? The signal on
the PS_ON# wire is supposed to be in one of two states. Either it should
be less than 0.8V or so, or greater than 2.0V. These might be values
a logic gate would use. If the motherboard cannot either pull the signal
all the way to ground, or the motherboard, when it releases the signal,
doesn't manage to let it drift up to 2.0V or higher, the PSU doesn't
know what to do. The PSU logic input functions almost like it is
in linear mode. Maybe the PSU will run a bit and maybe not. Maybe
it will slowly oscillate. (A smart designer would stick a Schmitt
trigger inside the PSU, but nobody would waste $0.25 doing that.
Every penny counts in a PSU design.)
So, we're left with several theories. The motherboard isn't completely
releasing the signal. The PSU isn't pulling the signal up hard enough.
The level of the +5VSB supply is not actually +5VSB, but some lower
value. If the +5VSB is lower than normal, that might cause some
logic to malfunction.
If you take a multimeter, and measure the voltage on the PS_ON#
signal, with respect to ground, that will tell you something about
what is going on. If the signal is a solid 0 volts, or a solid 5 volts,
then the motherboard is off the hook in terms of meeting the requirements.
If the voltage is kinda floating around, where it should not be, then
I'd have to swap power supplies and try another. If another PSU does it,
then it could be the motherboard driver.
It is possible to test the motherboard, under laboratory conditions, by
connecting a bench supply in place of the normal PSU. You would energize
the +5VSB on the main connector. (No other voltages would be needed for
this test.) Then press the computer case power switch. Then, with a
resistor to +5VSB, you would investigate the drive characteristics of
the motherboard driver output. You could, for example, connect a 470 ohm
resistor to +5VSB, and see if PS_ON# could pull it down. That would be
essentially a "strength test" for the motherboard. AFAIK, the driver
should be good for 8 to 10 milliamps or so, and should pull the signal
to 0.4V or lower.
You could do a similar test for the PSU. You take a multimeter, put it on
the DC current scale, connect the meter between PS_ON# on the main cable
and ground. What you're doing in this case, is measuring the value of
the pullup resistor being used. The pullup is probably designed to
source a couple milliamps, so 2.4K ohms on the PSU end might be a good
value. When the PSU puts +5VSB on its end of the resistor, and the multimeter
on the amps scale pulls down the PS_ON# signal, then a current of 2mA
or so would flow. If the number is grossly different than that, then
you'd suspect the PSU is at fault.
But not many people are equipped to do stuff like that, so swapping the
PSU as a first test, then getting a warranty repair on the motherboard
if there is still a problem, is the best you can do.
There is a reference schematic from Intel, but unfortunately, they
show a custom ASIC driving the PS_ON# signal, so there is no additional
educational value in the schematic. Still, this doc will give you some
idea what is inside an 865PE/875P generation motherboard. Page 79
at the top, shows the PS_ON# signal. Intel has a 22K ohm pullup on the
motherboard, in addition to the pullup on the power supply. Page 68
shows the Port Angeles chip driving the PS_ON# signal, but there are
no docs for Port Angeles. This is one of the reasons I hate this
schematic - they should have used components in the design, for
which datasheets could be downloaded. Note that Asus doesn't follow
this schematic, and the power distribution is quite different on
an Asus board. The logic connections for data busses and the like,
would be the same as the schematic.
http://www.intel.com/design/chipsets...ics/252812.htm
Paul
Similar Threads
Linear Mode
