wrote:
> Hi,
> I'm trying to solve an issue with my PC, it doesn't boot and I don't
> get POST messages. So, it's hard to solve the issue if you are lost
> without any hints...
>
> My PC didn't boot once in a while, after a few minutes waiting and
> sometimes pressing reset, it worked fine. No hardware changes or
> anything, system configuration is stable since >1 year. On
> installation, I disabled the POST speech, as it was annoying
> Yesterday, it didn't boot at all anymore. After hints that it could be
> a power supply issue, I replaced the psu. So, it *seems* to boot, but
> I don't get anything on the screen.
>
> I tried the following:
> - unplug everything, re-plug step by step - no changes
> - Re-enable the POST speech blindly, as I had no display signal - no
> changes, don't know if I've been at the right setting.
> - Clear the CMOS using various methods: remove battery+jumper, battery
> for half an hour, unplug ATX-cord+remove battery+jumper, unplug ATX-
> cord+remove battery+short-circuit the battery-contacts
> - Replace the speakers with headphones, some very muted *noise*, but
> nothing like POST-speech
> - All ATX-cables and graphics-power are plugged in correctly
>
> What happens: Green mobo LED is always on, press Power Button, all
> Fans start, HDD does something (a few seconds, doesn't sound like the
> usual OS-boot), Keyboard lights are on, GPU fan starts.
> What doesn't work: No POST-messages, My display says: No Signal.
>
> => Can anyone help to get the POST-messages re-enabled, so that I can
> find out which component doesn't work? And/Or can someone tell whether
> this is a mainboard issue?
>
> Thanks,
> rawaho
>
The Vocal POST chip is documented here. Not every aspect of how it works
is defined here. The device uses a serial EEPROM, and that EEPROM contains
a sequencer program, plus a series of voice samples.
http://www.winbond-usa.com/products/...83791SDSGa.pdf
The chip has two sides to it. The SDA/SCL ties to the SMBUS, and the processor
can write to the thing, very slowly. The processor clears timers in the chip,
to prevent certain error messages from being emitted. The other side of the
chip, interfaces to a serial EEPROM. You might see an eight pin DIP next to
the speech chip, and a sticker on top of the DIP will indicate what version
of code is stored in the serial EEPROM.
The chip does not seem to be tied to the reset logic on the motherboard.
It has an analog monitoring pin, but the function of that pin is not
explained. Usually, a chip triggers off something. The chip has a 5 volt
powering requirement, and the power for it could come from +5VSB or from
+5V main rail.
I would have expected, that clearing the BIOS, would re-enable the Vocal POST.
I think the chip is basically disabled, by the processor writing to
the W83791, via the SMBUS. So it should take active intervention from
the processor, to completely stop the Winbond chip.
So the question is, what triggers the W83791 into thinking the motherboard
has started ? Is it the +3.3VIN terminal ? Would it be based on an internal
reset generated from the rising edge of the 5V supply ? The datasheet doesn't
make that clear (there are several references to "MR" or master reset, but
no explanation of where it comes from).
For an example of the sequencer code that runs the W83791, that can be found
in the large download for the Winbond Voice Editor software. There are some
psuedo-code source files in the install directory, that seem to represent source
code for the sequencer. The sequencer does things, like wait a timed interval
before declaring the CPU test has failed. If the BIOS code that clears the
sequencer condition is reached, before the timer expires, then the user won't
hear the CPU failed code. The CPU present is a bit more tactile, in that a
simple logic signal (SLOTOCC) is used to detect whether something is plugged
into the CPU socket. Many of the error messages in the list, are commanded by
the motherboard BIOS writing the desired event number, into the Winbond
chip. Only a couple failure conditions are detected by timers.
The audio output of the Winbond chip, is capacitively coupled to the Lineout
on the back of the motherboard. Even if you remove the jumpers on the FPAUDIO
header, the signal will still be present (someone else figured that out for me).
You should use amplified speakers, and plug them into the rear of the computer,
into the green Lineout jack. The Vocal POST should come out on the left channel.
It is possible there isn't enough drive on that signal, for you to hear it with
headphones. I get good volume, when I have my stereo's Line-In connected to
the Lineout.
I think Vocal POST is a great feature, and when they stopped using it, they should
have included something to replace it. Some motherboards, for example, have a
POST code display, which isn't as useful as Vocal POST, but does show that
the designer did care about the user's experience.
HTH,
Paul
Linear Mode
