mobo Aopen AX4PER Intel845PE socket478 diagnosing

[AndreyS]

Good day there.
I'm trying to recover my old 478 socketed motherboard Aopen AX4PER. Back in that days I had some electricity spikes, and when another one spike happened, my PC turned off, and didn't turn on on power on. At that time I brought mb to the "repair master", on next day he gave back my mb and said the southbridge is dead. When I came home I noticed he soldered out two biggest capacitors and replaced one transistor (2) with other brand (NEC).
About 10 years later, I decided to fix that by myself. I fixed power supply by just replacing all capacitors. Now power supply is working 100% I tested it with workable PC. CPU is working too (I checked on workable PC).
But the motherboard, I soldered new capacitors, but one (1) of them I soldered in near slot (because the middle one was a bit damaged). Are those capacitors in parallel? Actually I checked with multimeter, it seems they are parallel.
North bridge gets very hot (I didn't let it to became too hot). Some time ago I used this mb as a donor, took some smd caps and resistors (3) (4). Recently I replaced resistors, but one resistor (3) I couldn't find the same nominal, so I used 221 instead of 201. Is that critical?
Also I didn't know the nominals of two capacitors (4), I bought any caps similar size (50v, 90 pF dont remeber how much picoFarads was, but it was picoFarads for sure). Is that critical? How to find out nominal of that caps? Maybe solder out some near caps and measure?
I have bought diagnosis card. When I power on mb the card shows following See photo attached.
Motherboard doesn't make any sound. I have read somewhere on the web, that it might be some transistors (which power CPU) failing.
Transistors (2) are 10N03L RN303 ane one NEC K3296 33N (or 33M, hard to read).
I have access to my high school lab (solder station, multimeters, oscillographs).
I'm studying programming, please don't be too rude.
Why do I want to hang out with this mobo:
1) forums say, that there is common problem with southbridge of this chipset -> it overheats. But in my case it doesn't hit at lot (maybe a little bit, hardly noticeable)
2) I have all the hardware workable, except mobo, it is useless w/o mobo
3) I love electronic stuff
Please help me to diagnose it. I will do everything you say in order to fix\diagnose my mb.

[AndreyS]

bump

[AndreyS]

Although I forgot to mention, I rang-out caps which are next to socket (where number '2' is located), and every cap has 0 voltage drop. I soldered out 2 caps and actually they work.
These caps look healthy. What if one cap is shorted, could it be a reason of all the problems above?

[AndreyS]

I checked almost every cap on short. Here is what I got.
Green - there is no short (some voltage drop measured)
Red - no voltage drop (very low ~0.001-0.002)
Light blue - outside this figure I didn't find any shorts of capacitors.
Purple - capacitors under the sink, on the i845PE

So based on my research, I could conclude that the i845PE chipset is dead (shorted). But I would like to see some thought of people who did have a deal with mobos.

[fubar.gr]

How exactly are you measuring the voltage drop across those capacitors?

It is quite uncommon for these capacitors to fail short. If you are measuring a short across the capacitor leads, it is possibly due to some other component in parallel with the capacitor, usually the MOSFETs

[AndreyS]

Thank you for the answer.

How exactly are you measuring the voltage drop across those capacitors?

It is quite uncommon for these capacitors to fail short. If you are measuring a short across the capacitor leads, it is possibly due to some other component in parallel with the capacitor, usually the MOSFETs

I measured across caps leads, and by putting multimeter's ground onto ground on the mobo and 'plus' of multimeter onto each of two contacts of capacitor.
In order to check if MOSFETs are working should I unsolder them?
It seems, two of them 2nd and 5th (if count from top to bottom, top of mobo is where cpu is located) are working:
if I put ground of multimeter onto pin '2' (drain) of MOSFET, 'plus' onto pin '3' (source) and vice versa, in this case multimeter shows voltage drop.
In case of another four MOSFETs there is no voltage drop (0.001-0.002 v). I feel like it is not ok.
Although every source is grounded (0.001-0.002 V drop). Is that as it should be?

It is quite uncommon for these capacitors to fail short. If you are measuring a short across the capacitor leads, it is possibly due to some other component in parallel with the capacitor, usually the MOSFETs

Is that possible that some circuits inside northbridge got shorted due to spikes comming from psu in the moment before it (psu) turned itself off when I had electricity spikes in the entire building, and that's way there are so many shorts when checking caps + so much heat from this chipset?

[AndreyS]

I made some research. Here is connection principle of MOSFET driver 5093MTC:


Here is what I measured/tracked the tracks on the mobo (it has black coating, very hard to track): image attached.
As I understood two High Side MOSFETs's sources mustn't be shorted. Am I right?
I guess I have to unsolder them and check them under load.

[AndreyS]

I continued some research in power cpu circuit, here is what I got: image attached.
So basically it is similar to reference, except on High Side FETs they (AOpen) put only one transistor to each channel, and drain & source are swapped.
Does it make difference connecting + to source, - to drain or - to source, + to drain?.
So the problem is that High Side MOSFETs's source and Low Side MOSFETs's drain do not have to be grounded

[AndreyS]

I'm gonna solder out those 6 electrolytic caps and check them, if they are ok, solder out Low Side MOSFETs and check them.


Would it be ok checking those MOSFETs with 4w 12v bulb?

[AndreyS]

Good day there,
I soldered out MOSFETs and those electrolytic capacitors. They are OK.
In the connection principle of MOSFET driver 5093MTC there are 3 pins which are connected to ground, but they must not be (SWB 10, SWA 15, VFB 24).
I believe the driver is OK and the problem is not in the connection principle of MOSFET driver 5093MTC.

I contacted with AOpen support, they said datasheet/schematics of this motherboard are available only for service centers.
Could anybody get that datasheet/schematics of AOpen AX4PER? I wnat to know whether the mobo can be repaired or not.

[AndreyS]

Can anybody help me to get the schematics/datasheet?

[Rasz]

how hot does the chipset get?
no point investigating vcore if its vio that fried chipset
find chipset power rail, im guessing its getting 5 or 12 v thru shorted regulator

[ovnr]

Why in the world are you even trying to repair that at all?

I usually thow away hardware that's newer.

[drussell]

I agree, but not so much because it is old tech but because Acer stuff is utter garbage.

[AndreyS]

how hot does the chipset get?
no point investigating vcore if its vio that fried chipset

Chipset gets pretty hot: in ~2-3min I can't touch it. I don't let it to become too hot though. Max temperature I let it go was ~70-80 C (measured with pyrometer). I turn on mobo only with heatsink on it.
If vio as far as I understand IO voltage, than I have Video Card and RAM fully working. Shouldn't they have also be damaged if it would be vio?

find chipset power rail, im guessing its getting 5 or 12 v thru shorted regulator

This is a problem, I could find nor this particular motherboard schematics, nor this RG82845PE (SL6H5) northbridge connection principle.
If this regulator(fdb6035al) located above AGP slot on the right side (you can see it on pics above, the regulator has high heatsink): it's drain is connected to 3.3V, source to ground

Why in the world are you even trying to repair that at all?

I usually thow away hardware that's newer.

I have 2 main reasons to repair it:
1) I have the rest of my PCs parts fully functioning (I've tested on another PC).
2) I want to make web server out of this PC.

I agree, but not so much because it is old tech but because Acer stuff is utter garbage.

well, actually it's a bit different brand (AOpen) garbage. While trying to repair it, I studied a lot of ICs connection principles and they (AOpen) cut corners on each circuit:
MOSFET driver (CPU power supply): 2 MOSFETs missing, about 30% of recommended capacitance of Cin/Cout missing, some SMD resistors missing, and I feel like a lot more which I didn't noticed.
RAM power supply: some SMD resistors missing.
And there are a lot of empty slots on the PCB, as you can see on pics I attached earlier.
All this is making me more picky when buying new motherboard.

I attached some pics of northbridge. What do you think, is it looking like burned? I googled something about burned chips, and they say such marks are not a problem.

PS: Thank you guys for answers.

PPS:
Northbridge: RG82845PE (SL6H5)
Southbridge: FW82801DB (SL6DM)

[Rasz]

Why in the world are you even trying to repair that at all?

I usually thow away hardware that's newer.

because you dont start learning how to fix highly integrated electronics on your $2k macbook motherboard, doh

how hot does the chipset get?
no point investigating vcore if its vio that fried chipset
find chipset power rail, im guessing its getting 5 or 12 v thru shorted regulator

Chipset gets pretty hot: in ~2-3min I can't touch it

yep, that pretty much means its dead

. I don't let it to become too hot though.

too late

If vio as far as I understand IO voltage, than I have Video Card and RAM fully working. Shouldn't they have also be damaged if it would be vio?

there is more than one vio
1.25
1.5
1.8
2.5
2.5_ddr
3
3s
vcore

find sources for all of those

I studied a lot of ICs connection principles and they (AOpen) cut corners on each circuit:
MOSFET driver (CPU power supply): 2 MOSFETs missing, about 30% of recommended capacitance of Cin/Cout missing, some SMD resistors missing, and I feel like a lot more which I didn't noticed.
RAM power supply: some SMD resistors missing.
And there are a lot of empty slots on the PCB, as you can see on pics I attached earlier.
All this is making me more picky when buying new motherboard.

meh, most of this stuff is nc for a reason. lower model = less redundancy

I attached some pics of northbridge. What do you think, is it looking like burned? I googled something about burned chips, and they say such marks are not a problem.

is that corner pink white? yep its dead
this chip is too big for homemade reballing with a hotair
maybe if you had a lot of experience and preheater, but for the first time its a lost cause

[ovnr]

I have 2 main reasons to repair it:
1) I have the rest of my PCs parts fully functioning (I've tested on another PC).
2) I want to make web server out of this PC.

Well, if you don't have any sentimental attachments to it and you're not doing it to learn, get a Raspberry Pi and use that as a web server instead. You'll probably save more money on the power required to run the PC to break even, not to mention the cost of replacing components.

Also, that chipset does seem a bit toasted.

[AndreyS]

is that corner pink white? yep its dead
this chip is too big for homemade reballing with a hotair
maybe if you had a lot of experience and preheater, but for the first time its a lost cause

I'd say it's white on one corner and the rest is purple with smooth border between them.

If vio as far as I understand IO voltage, than I have Video Card and RAM fully working. Shouldn't they have also be damaged if it would be vio?

there is more than one vio
1.25
1.5
1.8
2.5
2.5_ddr
3
3s
vcore

find sources for all of those

I hoped that chipset balls match with board cutouts, but nope, none of them does (1st_pic).

I ran through 845PE chipset datasheet and its Platform Design Guide, checked AGP (Vccq shorted) and PCI (everything OK) pins with multimeter, and there are VCC Core, VTT, VTTDECAP, VCCHI, VCCAGP connected to the ground. VCCSM (Memory Voltage) is OK.
Then I went through Voltage Distribution schematic, and I noticed Voltage supply for chipset (3rd_pic:A,B) is independent from CPU voltage supply (3rd_pic:C). In pinout table there are no VccCPU pins. I'm woundering how did VccCPU get shorted?

3rd_pic: there is 2 voltage regulators for the chipset, which take 5V input. I could find any voltage regulators which use 5V, but there is one (FDP6035AL) with quite big heatsink (4th_pic:A). The drain of it is connected to 3.3V and source is grounded. Is that really possible they (AOpen) used 3.3V instead of 5V to convert in to 1.5V?

Well, if you don't have any sentimental attachments to it and you're not doing it to learn, get a Raspberry Pi and use that as a web server instead. You'll probably save more money on the power required to run the PC to break even, not to mention the cost of replacing components.

Also, that chipset does seem a bit toasted.

Actually it's quite interesting. And yes, I thought about raspberry, it would be the best choice for my needs. But I wanted to know what is the real cause for damaged motherboard. As I said in very first post, a repairer said that ICH4 (Southbridge) burned. And most likely it's not.

Yes, northbridge is most likely burned. But I feel like there are some gaps in my understanding, which I'm trying to fill.

[Rasz]

I'd say it's white on one corner and the rest is purple with smooth border between them.

he's dead Jim

I hoped that chipset balls match with board cutouts, but nope, none of them does (1st_pic).

cutouts? you mean vias? via in pad are very tricky and most manufacturers flat out disallow using them, especially around ~2000-2005
usually you put vias between pads(balls)

checked AGP (Vccq shorted)

you mean 1.5V? yes, this is most likely the rail shorted inside the chipset

[AndreyS]

he's dead Jim

I'm not denying that.

cutouts? you mean vias? via in pad are very tricky and most manufacturers flat out disallow using them, especially around ~2000-2005
usually you put vias between pads(balls)

yes, vias. Thank you. I'm not electric engineer, not good at it's terminology. But all these stuff is quite interesting to me.

you mean 1.5V? yes, this is most likely the rail shorted inside the chipset

I have to unsolder that chipset. I have to know for sure, that it is faulty


Thanks to everybody for any help. I'm not going to throw it away, rather I'm gonna make my DIY infrared station 

[Rasz]

I have to unsolder that chipset. I have to know for sure, that it is faulty
Thanks to everybody for any help. I'm not going to throw it away, rather I'm gonna make my DIY infrared station 

you can unsolder with even the cheapest hotair, its soldering the new one that is tricky