T. Ment wrote:
On Thu, 1 Aug 2019 22:43:56 -0700, Banders wrote:

I followed this link
https://www.ifixit.com/Answers/View/448789/Blinking+Orange+Light+Only
because it specifically mentioned 755.

I'm not going to watch the video because I don't have a Dell, don't want
to ever have one,

I don't have one either but I was curious. He performed witchcraft with
the power supply wires. Scary.

Dell likes to make non standard changes to standard things like the ATX
power connector. Maybe they wanted to be proprietary like IBM mainframes
back in the 70s.

That strategy helped IBM corner the mainframe market, but drove them out
of the PC market. Too bad Dell didn't learn the lesson.



Dell only used non-standard ATX wiring for a couple
of years. They switched back to standard after that.

The video referenced, is a horrible idea.

The dude is playing with PS_ON# , that's an input
to the PSU, and is a "level". You ground PS_ON# for
as long as you want the PSU to run and the fans to turn.

The computer has a button on the front. It is a normally
open, momentary contact, push button. When you depress the
switch for a fraction of a second, one of the "edges" of
the pulse signal causes the logic on the motherboard to
"latch" the pulse and present a "constant level"
on the PS_ON# signal. That's the green wire.

This stuff is all detailed in the ATX spec, of which
there are three versions, and this picture is from
the newest version. The newest version, with 24 pins,
was created to suit the PCI Express era and the need
for a bit extra +12V current. And that's carried on
a yellow wire in the four pin section added to the
connector.

https://i.postimg.cc/4xY5d4XK/standa...-pin-power.gif

*******

This does not explain why the light is flashing on
the front panel.

Front panel LEDs can use one or two-color LEDs.

The front panel LED flashes when the computer is
asleep. On older computers, the steady level on the
LED might be a red LED, and when the computer is
sleeping, that red LED blinks once per second.

You can, however, use a two-color LED, two wires,
and two GPIO signals to control it. For example,
a green-yellow LED.

GPIO_0 ----+----+ GPIO_0 GPIO_1
| |
--- --- 0V 0V (no light)
green \ / /\ yellow 5V 0V green
--- --- 0V 5V yellow
| | 5V 5V (no light)
GPIO_1 ----+----+

This makes it possible for a two wire pair,
running from two GPIO signals, to run one
bi-color LED. And make two colors of light.

If GPIO_0 is held at 0 volts, you can "flash"
a 5V signal on GPIO_1 and make a "flashing yellow"
light.

*******

Well, this is great and all, but the mystery part is:

"What feature of a Dell power supply is different
enough, to flash the front panel at startup?"

I don't know the answer to that. There was at least one
Dell model, where the power supply itself had a LED
on the back, and it was a "self-test" LED. It was
never clear what the self-test was doing. If that supply
was bad, it would light the LED on the back.

The supply has the "PWR_OK" signal, which is
supposed to assert when all levels are "close
to full value". When the 12V signal is 11V or greater,
that would be a good time to show "PWR_OK". That
typically happens about 35 milliseconds after
PS_ON# is asserted to the supply.

Now, is there an additional signal from the power supply ?

Is the additional signal carried on a separate wire pair ?

Dunno.

These are things a clever repair person would be looking
for. Differences. Differences between "bog standard"
and "what Dell has done".

But the grossly wrong wiring pattern (wires moved around
on the 20 pin nylon shell), that was a few years before
the Optiplex 755. And if you miss that, plugging in
the wrong supply, can ruin stuff.

If this is a self-test feature, then there needs to be
a means to get the intelligence from the supply to the
motherboard (or directly to the front panel LED). All the pins
on the main connector have a function, so Dell can't really
hijack them for fun.

Paul