Mike said:
I had been using the cord from the old PS. Now just to give a positive
report on /something/ - anything, including the multimeter, both the old
and the new PS cords show correct line voltage at the C14 coupler end,
also confirming the bench strip.
There is a 'fuzzy' observation point about the new PS which is different
from the old.
When plugged in and then turned on, it 'seems like' there is some
transient effect on a small incandescent lamp's brightness on the same
circuit, as if the line voltage had been very slightly and very very
transiently affected by something. That effect 'vanishes' - does not
occur - if the switch is turned off and back on.
Perhaps it is related to some internal capacitor in the new PS
being/getting charged up and remaining charged during the off phase of
re-evaluating.
The old PS does not demonstrate the same effect when initially turned
on, so maybe it is 'dead in a different way'.
That effect, is called "inrush current". For a couple of AC primary side
cycles, the rectifiers are pumping up the main capacitor. This draws a
*lot* of current from the AC line. Sometimes, you can get a spec for
that, and the value could be 40 amps for a couple cycles. 40 amps times
the resistance of the house wiring for that transient, translates into
a visible drop in voltage as viewed by the lamps lighting your house.
The rectifiers on the input, have a steady state current rating, and
a transient rating. And the transient rating will be a lot higher
(because it takes time to "blow up" the diodes
).
I have one new power supply, where the inrush effect is so pronounced, it
temporarily trips my UPS "overcurrent" indicator. Normally, the
period of time, times the current, isn't sufficient to do that,
but the Sparkle brand supply I bought, seems to draw more than
the other supplies.
Looking at the reference schematic for an ATX supply, the "NTCR1"
negative temperature coefficient device up near the power
inlet, is intended to produce an RC charging time constant.
When the supply is cold, and you flip the switch for the
first time, NTCR1 has a high resistance value R. Then, the charging
transient for the main capacitor, takes a longer time to complete.
That reduces but does not eliminate the "lamp dimming" effect.
NTCR1 gets hot while it's operating. In the hot state, the internal
value of R drops to a low value. That's the "negative temperature
coefficient" feature of the materials used in its design. With the
value of R dropping, there isn't a large effect on the "efficiency"
measure of the supply. So over the long term, the value of R stays
low, and efficiency doesn't suffer too much due to the inclusion
of inrush limiting.
If you toggle the power switch rapidly between ON and OFF, NTCR1
has no chance to cool off. Any capacitor charging transient then,
is not suppressed, and the lights will dim a lot more than if NTCR1
was cold. Which is why I recommend to people, wait a short time like
30 seconds, before flipping a power supply back ON again.
Now, another thing to think about, is if the line voltage is
dropping temporarily, why aren't the other computer(s) connected
to the power strip being reset ? They're not affected, because
their main capacitor functions as a "hold up" device, storing
enough energy to wait out the transient caused by starting the
other supply. It is also one of the reasons that cheap UPS devices,
with transfer times measured in the milliseconds, don't cause connected
computers to be reset. If an adapter doesn't have sufficient
charge storage inside it, then a short transient may well be
enough to reset something.
*******
The fact you see that transient, when you flip the switch, means
your primary side capacitor is successfully connected to the line.
You likely have the 300VDC or more, present at the inlet to the
secondary supply circuit. Now the question would be, why isn't the
+5VSB showing up ? Both the +5VSB terminal on the main power cord,
and the PS_ON# signal, should be at 5.0 volts measured value, if
the power supply is just sitting there on the bench.
There is no need to worry about loading the supply, until you're
ready to assert PS_ON# (connect it to ground, for the period of
time you wish to check the main supply section). For an initial
check, that the supervisory part of the supply is working, verification
of the presence of 5 volts on +5VSB is sufficient. And +5VSB should
be within regulation even at zero current draw.
It's better to have some load on the main connector, if you're
asserting PS_ON# and doing a test of the main rails. The label
on some supplies, states the minimum recommended current flow,
to allow the supply to maintain regulation. So if you intend
to assert PS_ON# (with the paper clip to COM), then you can connect
the main cable to something or connect a couple hard drives
to the Molex cables. The hard drives would load two of the three
"main, high current" outputs.
I purchased a set of power resistors, an ATX extender cable, and
build a small load for testing ATX supplies here. When I get a
new supply, I run it for a couple hours, while connected to that
dummy load, and check the rail voltages with a multimeter. That's
how I stay within the spirit of the "provide some loading" thing.
But that hasn't stopped me from temporarily testing a supply
with no load on it, to see if the internal fan still works, as
a quick test. (Note that, once a power supply emits "magic smoke",
I stop testing it. I try not to torture them and make them
blow up, if I can avoid it. But if a supply has given no signs
it's about to "go", I'll then still give it a test of some sort.
That operating philosophy, is to protect my eardrums
)
Paul
