GA-P35-DS4 lost Ethernet

[pfrcom]

I can't work out what's happened to on-board Ethernet with this GA-P35-DS4 board

Doesn't even show CMOS setup option "Onboard H/W LAN" Enabled/Disabled

So I surmise its Realtek 8111B SMT chip isn't working, even though it appears undamaged, as pictured

Any suggestions, apart from PCI card ?

[goontron]

Re: GA-P35-DS4 lost Ethernet

i believe this is a problematic chip that can cause your mobo not to boot and the way to fix that is to pull the chip and use an Ethernet card. im probably wrong, but if im not.................

[shovenose]

Re: GA-P35-DS4 lost Ethernet

Was it working before, then just disappeared? Or did you get it like this?

[pfrcom]

Re: GA-P35-DS4 lost Ethernet

I got it like this

Board does boot OK, and other on-board stuff, like Firewire, seems to work

Although on-board sound has a bit of crackle

[tmiha71]

Re: GA-P35-DS4 lost Ethernet

If board has WOL, chip has to have some connection to 5V-standby, or 3.3V standby...

Possible reason could be lousy power distribution of +5V (presumed, that after power on - standby loads are switched to +5V rail )...

As a first step I would check if ALL caps near USB (rear connectors) and USB headers are in parallel, if answer is YES - than I would serach for MOSFET (or two) near USB headers - one mosfet leads to +5V rail, other to +5V standby...

If it's so, than I would start replacing most distant capacitors from that mosfets - one by one, even if they're polymers, with something like Rubycon XYG 1000uF 8x11 (entery lewel - low ESR, with at least 2000h endurance /the more is better/ )...

The drill goes,
1) with "Kill 'a Watt" measure consumption, leave it for minute or two (best on BIOS health screen in BIOS) write lowest and highest peak form "Kill 'a Watt"
2) repeat step one in Windows idling, if mobo boots (wait few minutes to stabilize)
3) replace capacitor (most distant from MOSFET)
4) repeat step 1
5) repeat step 2
6) check stability...

Kill 'a Watt will show if you're on right track - lower or steadier draw from wall socket should be seen, if you see increased consumption, or more fluctuating - abort and go back one step (put back original capacitor)

Asked objective not fulfilled, do it again (network chip awaken )...

Actually this procedure is not harmless, you can kill PSU, or mobo -> in general messing with +5V rail can introduce a lot of side effects, so without "kill 'a Watt" to verify improvements -> NO NO...

[momaka]

Re: GA-P35-DS4 lost Ethernet

If it's so, than I would start replacing most distant capacitors from that mosfets - one by one, even if they're polymers, with something like Rubycon XYG 1000uF 8x11 (entery lewel - low ESR, with at least 2000h endurance /the more is better/ )...

Sorry for the blunt question, but... why do you keep suggesting this?? I really don't see the logic / reason behind this. If there is a valid one, please share it here.

And how is Kill-A-Watt meter going to help? That thing is not precise enough to draw any kind of conclusions. Again, please share where you are you are reading up on these suggestions as the theory behind them does not appear very sound to me at all.

Now as far as the OP's topic... was the BIOS ever flashed on this board?

[pfrcom]

Re: GA-P35-DS4 lost Ethernet

Now as far as the OP's topic... was the BIOS ever flashed on this board?

Ohhh yes ...

... when I got it, was using F11 - seller said "Refused to boot on F14 (BIOS). Downgraded to F11 and works but network card is not detected"

Seeing fools rush in where angels fear to tread, I re-flashed it to latest F14 and got a boot loop

Downgraded to F13 - boot loop

Downgraded to F12 - after some trying, got stability - so I treated it to a new CMOS battery, and left it at that

BTW, I quoted seller's exact words - however a generic Realtek 8169 PCI card works fine, so I presume he meant on-board network

[pfrcom]

Re: GA-P35-DS4 lost Ethernet

I may have found the problem

LC3 shown in original pic, near the edge of the board, appears to be a component broken off, not just empty solder pads

Now to find a replacement

[pfrcom]

Re: GA-P35-DS4 lost Ethernet

I'll never know if missing LC3 was the problem - tore off a pad while trying to solder a replacement into place

At least the board still works

[tmiha71]

Re: GA-P35-DS4 lost Ethernet

Sorry for the blunt question, but... why do you keep suggesting this?? I really don't see the logic / reason behind this. If there is a valid one, please share it here.

On few boards, I've noticed that +5V rail does make some problems which wont go away with usual troubleshooting (changing PSU, etc...) and caps were of reliable brands :-)), changing one or few caps on +5V rail (near rear USB connector, or PS/2 connector, is best place to start, max 0.5A - really there's no need to have ultra low esr caps there) to higher ESR ones (wild guess /ESR/, can not find info about self-resonant frequency - which is probably real reason for improvements /in this case lower self resonant frequency should help/ or both ).

Why it helps?
Because of parasitic inductance or (more likely) higher resistance (max distance), those FJ's seems to have /beside very low esr/ higher than usual self-resonant frequency (which is always good for vrm-out, on other places it should be confirmed with measurements), so they produce some unwanted effects (LC effect or RC effect - caps not having same charge/voltage - but theory says that they want to have same charge/voltage /parallel connection - same capacitance/ and will to anything to achieve that ). If such effects exists Kill a watt can show them (after replacing single cap consumption will drop). I'm not interested in absolute value, but if some drop comes out (after replacing cap) - there were problems (or problems were introduced if consumption rises)...

Example:
A drop of single Watt on wall socket: wall socket 1W drop, output of PSU = 1W*efficiency, 0.8W, if that is on 3.3V rail I = 0.8/3.3 = 0.25A, we have lowered current on 3.3 rail for 0.25A, and if computer works OK (stability test) -> that was unnecessary current -> pure waste, it's only task was to kill caps :-)...
To note, here's not about saving money on e. bill, it's about mobo which draws to much current (without need for it), and that's bad...
So yes Kill'a Watt can help in some situations, but only as confirmation (after change is done), with it problems can not (or can?) be detected (problems should be presumed and investigated )...

Why FJ's?
Seems that majority of troublesome boards do have FJ's...

It's the same story for poly's on that place ( problematic +5V rail)...

OP asked for ideas, I wrote what I think could be problem - bad +5V rail...

So, I'm asking forum members to try to recall if boards with ASDS/ Axxx Sudden Death Syndrome / and from other manufacturers too, did had FJ's, and how many there were on +5V rail...

[momaka]

Re: GA-P35-DS4 lost Ethernet

Why it helps?
Because of parasitic inductance or (more likely) higher resistance (max distance), those FJ's seems to have /beside very low esr/ higher than usual self-resonant frequency (which is always good for vrm-out, on other places it should be confirmed with measurements), so they produce some unwanted effects (LC effect or RC effect - caps not having same charge/voltage - but theory says that they want to have same charge/voltage /parallel connection - same capacitance/ and will to anything to achieve that ).

But see, that's where this theory falls short. Capacitor ESR increases with time. So if there was even such a resonant frequency that was major enough to severely impact the operation of other circuits on the board, then the motherboard would have never worked right from the factory! But that's not the case here, so what gives? If anything, the increased ESR of the aging FLs/FJs would lower the amplitude of the resonant frequency. So it just doesn't make much sense to me.

Also, I doubt you will see a drop of 1W in power usage just from changing the caps. That's a substantial amount for just a few caps on the 5V rail on the motherboard. Now, in PSU, that could be a different story.

[tmiha71]

Re: GA-P35-DS4 lost Ethernet

1)

the increased ESR of the aging FLs/FJs

2)

would lower the amplitude of the resonant frequency

Agree on first statement, second is somehow troublesome...

To first statement would also add, lowered capacitance...

As I wrote, if some change is observed after replacing cap, this change should be interpreted, as
- increased consumption -> bad move was done,
- reduced consumption -> right direction (some unwanted effects are reduced)

And for observing single W drop from wall socket, buy yourself Kill'a Watt and try it (unfaithful Thomas :-)... I really mean it, it can be seen, improvements /or degradation/...

[momaka]

Re: GA-P35-DS4 lost Ethernet

Agree on first statement, second is somehow troublesome...

It's a series RCL network. Raise the series resistance and the amplitude of the resonance decreases. Do I need to review you linaer circuits theory more?

And for observing single W drop from wall socket, buy yourself Kill'a Watt and try it (unfaithful Thomas :-)... I really mean it, it can be seen, improvements /or degradation/...

No, no. I'm gonna pull an old "PCBONEZ" here and say this:
This is YOUR theory, so YOU have prove it. I'm not gonna go spend all day playing "Mythbusters".

First off 1W is quite a bit of power to be dissiapted just into RF. At the same time, 1W of power is practically NOTHING for power electronics. You can have that "extra" 1W of power just simply if the PSU was cooler when you performed the experiment the first time (schottky diode forward voltage drops with temperature increase, for example). And let's not even mention drops in the line voltage, frequency, and etc. - all of these can easily make the PSU loose or gain efficinecy in excess of 1W. If you've done this with different motherboards and PSUs, you have a TON of variables that come into play.

If I were you, I'd do the following: get at least 5 identical motherboards and perform your experiment on them with the same PSU, at the same room temperature, with an isolated line. Can you do all of that? I doubt it, but I'll still say maybe... and when you do, make sure to write down everything and create a thread showing your procedures and how exactly you got to your results. But even then, I will NOT blindly belive in a forum post. "unfaithful Thomas" - maybe (but not without reason ).

[tmiha71]

Re: GA-P35-DS4 lost Ethernet

It's a series RCL network. Raise the series resistance and the amplitude of the resonance decreases. Do I need to review you linaer circuits theory more?

How about pumping serial RLC (or RC) circuit with current (current source instead voltage source) ?

And let's not even mention drops in the line voltage, frequency, and etc.

How about using PSU which minimize those effects...

Actually forum is a place to say things, and people can take it (or not) for granted , or research it and/or TRY it, but some knowledge is needed...

And you're right about experiment, it should be repeatable, no arguing about that...

This thing with Kill'a Watt is off topic (sorry to bring it once more), I only suggested way of checking if things are going in right way, but should add that, time is also factor...

As you noticed 1W is nothing, PSU in free-running (no load) consumes aprox 20W (some more, some less)...

And computer at idle (doing nothing) draws from socket - do not know exactly, seen from 50W and up, usually with integrated gfx, lowest consumption with gfx card approx 60W to 100W and even up...

And to repeat, I do not hunt Watt's, but wast majority of mobo's (graphics card especially ) do have problems with consumption...

If we presume, that for same family of CPU-s (same process of production ) all mobo's are designed in similar way, have same tasks in idle (same OS, same circuits - chips may be different, same power saving setup) - they should have pretty similar consumption, ( thinking of word identical - but we're not living in ideal world )... But they do not have...

And to add - this is also valid for graphics cards - type of graphics card should not play any role (consumption wise) in idle (as noted above same tasks, or no task at all), but it plays - bad design of mobo, or graphics card, or both...

And to repeat once more, if consumption went up after replacement of capacitors -> something bad was done, recap will faill, problem could be wrong caps picked, bad soldering, or something else...

And to repeat once more, I do not expect that if I see 1W difference it is exactly 1W (with accuracy 1% ), i do expect that sign is correct, consumption went up, or down...