Have you replaced the CMOS battery?

Yes and nothing changed... :\

You replaced it with a NEW battery, I presume. Do a visual inspection of the board. Any bulging caps? Also, after clearing the CMOS and successfully turning it on, try leaving it running for a day or two. Does it stay running or crash?

Well can be some different things.
If clearing CMOS works then my first hint would be that the ROM is damaged somehow and can't save or keep it's data.
Reflashing the BIOS might do the trick then. If it is a loose module then reseating the whole bios module often resolves this issue.
If this doesn't help.. well my next guess would be a bad Bios chip 😵

Somehow it often happens that some residue power gives the same behaviour, this often happens if the board builds up a static charge. Seen this alot the last years.
If you build the motherboard + vga + psu + cpu + ram out of the case and put it outside the case this often resolves this issue. The case needs to be grounded better or the screws interfere somewhere with the board circuitry. 😠 It's a real bugger since it happens alot with cheaper cases..

Hope any of this helps.

Thank you both!
Clueless1: no bulging caps and no crashes.
Nvm1: My tests were done outside of computer case. I am going to test flashing the Bios...

I like Nvm1's idea of re-flashing the BIOS. I hope that works for you!

Caps dont have to be bulged to be bad... They also can dry-out.. And then wont reach its normal values anymore..
Changing caps could be a good investment. Specially does power the processors and the Incoming current lines.

Normally when the caps are bad the computer freezes during the usage, it's not the case.
I haven't made the BIOS re-flashing yet but at this moment I can turn on the PC inconsistently, sometimes works sometimes don't.

mp10 wrote:

Normally when the caps are bad the computer freezes during the usage, it's not the case.
I haven't made the BIOS re-flashing yet but at this moment I can turn on the PC inconsistently, sometimes works sometimes don't.

If I am correct the Bios chip is removable on this board. Take it gently out en insert it again first before trying to reflash the bios. Little bit of corrosion or dirt can cause these issues. 😢

mp10 wrote:

Normally when the caps are bad the computer freezes during the usage, it's not the case.

That is not true.
I've been recapping 100's of motherboards a year for a decade or so. That's what I do.
Bad caps present a wide variety of different symptoms on motherboards and some of them are really off the wall.
The symptom depends on which circuits the bad cap(s) is(are) in.

I am not convinced bad caps are the cause of your current problem but it is a possibility.
.

mp10 wrote:

I can turn on the PC inconsistently, sometimes works sometimes don't.

That is a VERY common symptom of bad caps. Is not proof though.

I agree with the others that a good thing to try next is to reflash the BIOS.

Also, if you haven't yet, try reseating the RAM.
Same deal with a bit of corrosion or dirt causing issues.

Does the problem show itself on cold-boot, hot-boot or both?

Are you using the same front panel switches out of the case as you did in the case?
I have twice seen this caused by bad front switches.
.

I added this variable in my analysis and I found something: when the PC doesn't boot I leave it ON for a while (only fans running) and then I turn off. After that, the PC boot without any problem. Any hints about this behavior?

Yes, I'm using the front panel switches of my PC case.

mp10 wrote:

I added this variable in my analysis and I found something: when the PC doesn't boot I leave it ON for a while (only fans running) and then I turn off. After that, the PC boot without any problem. Any hints about this behavior?

Yes, I'm using the front panel switches of my PC case.

To me that kind of sounds like cracked solder? When you leave it on for awhile, it heats up and expands and the bad solder connection becomes good, until it cools down again.

clueless1 wrote:

mp10 wrote:

I added this variable in my analysis and I found something: when the PC doesn't boot I leave it ON for a while (only fans running) and then I turn off. After that, the PC boot without any problem. Any hints about this behavior?

To me that kind of sounds like cracked solder? When you leave it on for awhile, it heats up and expands and the bad solder connection becomes good, until it cools down again.

Indeed that is a possibility. - The OP should inspect for it.

If the motherboard would boot after a CMOS reset, it sounds to me like there might be a BIOS setting set incorrectly.

I had a similar problem recently on an Opteron-based motherboard, but instead of merely clearing the CMOS, I had to pull out all the removable components and put them back in, as well as clear the CMOS, and it would timetimes boot up. I recapped the whole motherboard, which did not resolve the problem. The board eventually stopped turning on entirely and was moved to my dead motherboard box. The last time it booted, I ran MemTest, which ran for 4 hours, then hung up, never to be heard from again. After that experience, I am somewhat hesitant to blindly recommend recapping the board without further evidence that it is needed. That particular board contained really stubborn solder. In my particular case, there were about 10 bulging 1000 uF capacitors, which I took as "evidence", but it still did not help. Is it possible that the board was damaged by the previous owner from over use with bad caps?

mp10 wrote:

Yes, I'm using the front panel switches of my PC case.

Disconnect both switches and try starting it by shorting across the pins for the power switch with a paper-clip, screw driver, unused jumper or whatever.
The switch is a momentary contact type so don't leave the pins shorted for more than 1 second.
- Any change?

mp10 wrote:

I added this variable in my analysis and I found something: when the PC doesn't boot I leave it ON for a while (only fans running) and then I turn off. After that, the PC boot without any problem. Any hints about this behavior?

There are a number of things that can do that including (as given) bad solder joints and bad caps in either/or the PSU and motherboard.
-
When a system starts the chipset continuously sends a reset to the CPU.
This is to allow the BIOS time to complete the POST checking before the CPU starts.
Once the BIOS completes the POST it tells the chipset to stop resetting the CPU and the CPU starts.
That's if everything works right.
-
Another concern is the 'power good' signal. (Sometimes called other things like pwrgd.)
There are one or several points on a motherboard where the voltage is checked via IC chips. (May be one chip with several sensors.)
Those chips "report to" the main chipset. When the all the voltages are good the chipset "tells" the PSU and the PSU stays running.
There is timing involved. These numbers are approximations because it varies with different PSU specifications.
When the PSU first starts the voltages have to be within 80% of nominal within x-many milliseconds or the power good signal won't go and the PSU turns back off.
A typical time is like 50-200 milliseconds so all this can happen so fast you won't even notice the PSU tried to come on.
-
Bad solder or bad caps (even just one) can muck up either one of those routines.
For example if a cap is partially shorted it will take too long to charge and the PSU won't stay on due to no power good signal.
Bad solder can simply break the signal path.
.
As the system is warm-booting it's possible some marginal cap is able to build up enough charge to hold it over for a short time until the next start,,, but it can't hold the charge long enough to hold it over until the system cools off.
.

feipoa wrote:

If the motherboard would boot after a CMOS reset, it sounds to me like there might be a BIOS setting set incorrectly.

Or a falsely locked-in reset (or switch) from the BIOS that changes state when the BIOS is reset.

PCBONEZ wrote:

Disconnect both switches and try starting it by shorting across the pins for the power switch with a paper-clip, screw driver, u […]
Show full quote

mp10 wrote:

Yes, I'm using the front panel switches of my PC case.

Disconnect both switches and try starting it by shorting across the pins for the power switch with a paper-clip, screw driver, unused jumper or whatever.
The switch is a momentary contact type so don't leave the pins shorted for more than 1 second.
- Any change?

Thank you very much for you technical explanation. I tried to short the pins with a screw driver and nothing changed.

feipoa wrote:

If the motherboard would boot after a CMOS reset, it sounds to me like there might be a BIOS setting set incorrectly.

But when we do a CMOS reset the BIOS settings go to the default, right? Which settings are you talking about?

I begin to doubt about the relation of CMOS reset with this problem. If I turn on the computer (cold) and wait for 5/10 seconds and then turn it off, I can ear the hard drive spinning in the next try and then I need to wait for 1 minute to turn it on the computer again in order to boot completely.

edit: After a successful booting, he goes directly to the BIOS in "advanced" tab where we can choose the CPU internal frequency, CPU Frequency Multiple, etc.

PCBONEZ wrote:

Or a falsely locked-in reset (or switch) from the BIOS that changes state when the BIOS is reset.

What do you mean with "falsely locked-in reset from the BIOS"?