Il Sat, 21 Jan 2006 20:08:08 +0000, kony ha scritto:
Describe what is hooked up to this system. Does it have PSU2 or USB
keyboard connected "always"? If so, try unplugging them and all other
USB devices and retrying it. I am wondering if the 5VSB is not
sufficient for the system load. You might also see if the board has
[5VSB vs 5V] jumpers for USB or PS2, often near the effected ports or
pin-headers and should be outlined in the motherboard manual... and
change them to 5V, not 5VSB.
Thanks for your answer, but I have tried to start the mainboard with only
the system memory plugged in with no result... (if I use a 400w psu the
epia boots up in this condition)
I bought the power brick and the power supply board "in bundle", but I
have noticed that the psu need 10A at 10.8V
(
http://www.watt-power.co.uk/WPPS80.PDF) but the AC adapter is only 90W
(4.5A at 20V, 8.3A at 10.8V)
http://www.powerstream.com/z/EA10953.pdf
However I'm always convinced there is a timing problem between the psu and
the mainboard during the booting, but it's very hard to prove.
WPPS80:
3.3@5A 5V@5A
[email protected] [email protected] [email protected] [email protected]
80% efficient, which is why it needs 100 Watts in to make 80 Watts out.
Protection method is "latch off", which means at the first sign
of trouble, you'll need to cycle the input power to get it to recover.
Overload protection is at 110% to 160% of rated current on
any output. Input voltage range is 10.8V to 20V. It would
draw 10A at 10.8V (as stated), or 5.4A at its max 20V input.
That is because input power is a constant, and as volts rise,
amps drop. A good switcher will detect when the input voltage
is dropping below a certain level, otherwise the current
goes to infinity as voltage goes to zero (figuratively
speaking)
On your power brick, I hope you have EA10953(2)A, as its output
voltage range is 12 to 17V, and that fits inside the 10.8 to
20V input rating of the ATX converter. Your power brick shuts
off if its output voltage rises above 16 to 22V. This brick doesn't
seem to have a very tight reference, as it has a five volt output
variation (even though line and load regulation are 2% and 5%).
Now, the power brick has auto recover, which means as soon as
whatever problem it has detected, has disappeared, it returns
to normal operation.
So, the first question you ask, is what are the power requirements
for the ITX board.
http://www.via.com.tw/download/mainboards/3/4/OG_EPIA-PD_111804.pdf
"PD6000E
Running CC Winstone2001
3.3V @ 2.439A
5.0V @ 0.931A
5VSB @ zero
12V @ 0.097A
Total power 13.963W"
By using the 80% efficient WPPS80, that 13.963W translates to
17.45W from the EA10953(2)A. If the EA10953(2)A produces ~12V
output, then 17.45W means 1.45A flows from the EA10953(2)A.
Since the EA10953(2)A can make 6.6A, I doubt the limits of
the EA10953(2)A are being reached.
So, while the conditions of measurement are not stated (what
capacity DIMM was used?), I would say those numbers are far enough
from the 5A limit, to not be a problem. You have enough 5V and
12V to try to run a desktop hard drive (but possibly not a hard
drive and a DVD drive, unless the DVD is the laptop kind that
runs from +5V only).
Based on the protection method (latch off on the WPPS80,
auto-recover on the EA10953(2)A), I would say the WPPS80 is
most likely to be waving the surrender flag. Generally, how
this works, is most switchers have an interval, during which
the overcurrent protection on the output is disabled. I'll
call this 50mS as a rough guess. That is the time allowed
for the WPPS80 to charge its own output capacitors, and charge
the input filter caps on the PD6000. Now, lets imagine a
cascaded system response scenario. We plug in the EA10953(2)A).
The ~12V output comes up slowly, in say 35mS or so. The
WPPS80, being a heroic design, starts to do its switching
conversion, before the EA10953(2)A) output is fully developed.
The WPPS80 starts its 50mS timer. At the 50mS mark, the WPPS80
hasn't finished charging the caps. The overcurrent protection
is enabled. The WPPS80 instantly switches off. In summary,
a "soft" drawn out output on the EA10953(2)A), means the
assumptions put into the WPPS80 design are violated.
Powering the EA10953(2)A) from a "rigid" supply is another
alternative. One way to emulate this, is insert a switch in the
12V output path of the EA10953(2)A). Plug in the EA10953(2)A)
with the switch open. When the EA10953(2)A) is delivering its
12V output, then flip the switch to connect the "solid"
12V to the WPPS80. The WPPS80 starts its 50mS timer, with
full source voltage present. (If the WPPS80 has no inrush
limiter, and has input current sensing, it could still
switch off. But I don't see any mention of a feature like
that.)
A second rigid supply test you could attempt, is to use your
ATX power supply and its hefty +12V output. You could put
a switch in line with that as a source, and try the same
test to the WPPS80. That will require an ATX extender cable,
to give you some wire to hack up.
Based on the estimated power numbers, I don't see a reason
for this to be an output limit problem. It is likely
the protection on the WPPS80 tripping, because the input
power transient waveform from the EA10953(2)A) is too soft.
One way the WPPS80 could avoid a problem like this, is if
it refused to start switching, until the measured input
voltage was quite close to its rated minimum input voltage.
HTH,
Paul
