Own Custom Build Computer With ECS Motherboard-No POST

[momaka]

Re: Own Custom Build Computer With ECS Motherboard-No POST

I didn't know that a PSU could destroy a mobo, and assumed that they would all have built in over-current protection as a basic feature. The PSU fan was a bit noisy before this and I disregarded it as being in danger of failing.

Yes, a PSU can indeed destroy a motherboard when it fails catastrophically. This usually happens with cheap power supplies, since they lack some protection circuits (or when the protection circuits are not implemented properly).

However, the worst are power supplies that seem to work "fine" (i.e. all voltages come up), but they have too much ripple on the output, causing the PC to crash or not boot reliably. And looking at the pictures you posted, this could very well be the problem here. Your PSU has undersized input caps, only a single output cap per rail and without PI coils... Basically, your PSU likely is outputting too much ripple, which could easily be causing the motherboard not to boot.

Those scorch marks on PSU metal case show some components catastrophically failed, but only the discolored on the PCB (PIC-D) will tell if something else might be overheated (PIC-C).

Could be dust too. I've seen similar "phenomena" in other electronics. Usually fine dust in combination with heat causes it. If this was due to components that failed in the PSU, it wouldn't have worked at all.

How can I check if my motherboard or the system wasn't sufficiently damaged before I replace a new PSU to test on it?

Aside from a few basic checks, you can't really. A good power supply is needed to test the computer. If you don't want to buy a new power supply, see if you can borrow one from a friend (and preferably a good one).

How can I check if the Processor and RAM can still be reused on the system as they can be easily toasted by PSU fault?

It is extremely rare for the RAM or the CPU to get toasted. The voltage regulation circuits on the motherboard provide fairly good isolation from the PSU.
That said, if the RAM is toasted and you try it on another computer, in almost all cases the PC just won't boot, but the motherboard won't be destroyed.
Trying a toasted CPU in a good motherboard is a little less safe (depending on how the manufacturers designed the motherboard), but again, in most cases the motherboard will survive the "bad" CPU. Probably will trip the PSU short-circuit protection if the CPU is badly shorted. However, if you are using a cheap power supply, as mentioned above, the PSU may not trip and blow up.

So all in all, it is good to have a reliable power supply on hand.

Do you recommend any advise to reapply the thermal compound on the Processor and SouthBridge chipset heatsink with better thermal paste?

Well, if you want to rule that out as a possibility of the PC not booting, then go ahead.

Normally, I don't reapply thermal compound when *testing* computers, unless the compound looks extremely dry or the CPU has an exposed-core (i.e. one without a heat spreader, such as AMD socket A/462, Pentium 3, or a mobile CPU).

You have socket 775 CPU, and it more than likely has a heat-spreader. So for testing, you don't necessarily need to reapply thermal compound - just leave the old one on there. But if you already have thermal compound on hand or can get it for cheap, then give it a try.

Normally what will be the watts of soldering iron that can be used to replace the capacitors on the PSU circuit board?

For most PSUs, 35-40W is about enough. But if you are new to soldering, around 50W might be a little better for you. However, if you are planning to recap motherboards, 60W may be better to have.

That said, I don't think a simple recap will help your PSU too much. If anything, it needs a few mods in there as well - and unless you can obtain the needed parts for cheap, it isn't really worth it. Of course, that also all depends on what kind of power supplies you have on your market in Malaysia. If you can get a new decent PSU for a good price, then that might be the best option. Otherwise, you may have to just get a bunch of junk power supplies and see if you can make a "decent" one by taking parts from the others - but that's something I would rather not have you do if you are inexperienced with electronics component level repair.

As it currently stands, your power supply needs, at the very minimum, an input filter choke (also known as EMI/RFI supression choke/coil), bigger input and output capacitors (for output capacitors, you may want to up the capacity to at least 3300 uF per rail, since your PSU does not have PI coils), and clean that likely-conductive brown glue.

*UPDATE: Here I have found some information about the brown crusty stuff on the PSU circuit board. This stuff is probably glue (like RTV), used to secure the large components like transistors heatsink and capacitor with high capacity from snapping their tiny little legs during soldering and assembly. As it ages, it can become conductive enough to short components and kill them.

Good job on your research . That brown stuff in your PSU is exactly like you found - cheap glue that turns conductive over time. Clean it off as much as you can - or at least from all metal parts and component leads. That could very well be why the board is starting to darken around the 5VSB section.

[Agent24]

Re: Own Custom Build Computer With ECS Motherboard-No POST

That is a cheap and horrible looking PSU and I would be suspicious of its capability. At this point I would try a known working\good PSU in the PC because:

The Asia'x capacitors are not known to be good, and in that PSU could well have dried out from age and heat causing high ripple levels which mean the board won't start and the +12v rail is low.

With an old, cheap and unknown condition PSU like that, the best thing to do is just swap with a known good one to start with.

[c_hegge]

Re: Own Custom Build Computer With ECS Motherboard-No POST

I think that there is a very good possibility that the PSU has just slow-cooked the motherboard. With only one cap per rail, I can almost guarantee that the ripple would have been way out of spec from day 1.

[LENOVO-A880]

Re: Own Custom Build Computer With ECS Motherboard-No POST

....As it currently stands, your power supply needs, at the very minimum, an input filter choke (also known as EMI/RFI supression choke/coil), bigger input and output capacitors

.... I have replaced my unit PSU with a decent one which is the FSP HEXA+ 550W recently but the ECS motherboard still unable to boot up. Well, since I have narrowed the problem down to a PSU or motherboard, likely 95% chances of power related problem which caused motherboard being unstable. How about a PSU that keeps its output well within the required voltages, but has a huge amount of ripple and kill those polymer capacitors on the motherboard? And likely these capacitors here took up the ripple current from a poor quality PSU, causing them to heat up badly. Then how about a crappy PSU with non-functioning OVP. If that puts out high voltages, or high voltage spikes, there is nothing to shut it off, so the voltages could fluctuate fairly high too. Those sudden jumps between voltages could give voltage regulators problems too, which could give high output spikes to the RAM, CPU, chipset, GPU, etc. I've heard and seen horror stories of PSUs failing, and killing components. So my question here will a polymod on my ECS G31T-M7 motherboard especially on the PSU sides fix this problem here?

Thanks

*Update*

POLYMER CAPACITOR PLAGUE
A capacitor can fail open so that it's disturbing the flow of electricity in a circuit and then the circuit is no longer working at all. Polymer type capacitor can also failed shorted unfortunately or fail open without any visible signs. Polymer capacitor, don't have much of the problems of electrolytic capacitors. They don't leak or swell, due to absence of electrolyte which would cause gases inside. However it's possible one of them is shorted or damaged without showing any signs due to extreme heat (ex: MOSFETs near them shorted or broken and heating every others component around) but this would happen with any series of polymer type capacitors. It's also unlikely. More likely you have some shorted MOSFETs / cracked ceramic capacitors due to high voltages

[Agent24]

Re: Own Custom Build Computer With ECS Motherboard-No POST

Forget polymodding or anything like that for now.
You've got a new PSU, and the board still doesn't work, so start troubleshooting the board.

Troubleshoot the motherboard and repair until it's in a working state, and then maybe do some polymod. Don't start modding things when they don't even work to begin with! If you screw up the mod you won't know if the reason it doesn't work is because of the original fault or the mod. It just confuses things.


So.... First of all, I would check all the system voltages on the board and power-on signals.

See page 29 of the schematic for the power distribution chart and page 30 for the power sequence. You need to check all these rails to be sure they exist and are the correct voltage. You need to check the power-on signals to SIO etc to see if they exist and are OK.

Do you need help finding where to measure them?

[LENOVO-A880]

Re: Own Custom Build Computer With ECS Motherboard-No POST

Well, I do agree with you Agent24 ....polymod can come as a last resort. I should start to check the standard voltages on the motherboard with a multimeter but I've lot of hard time finding those rails like CPU core; Ram VDD; Northbridge VCC and Southbridge VCC or others which I'm not familiar with them yet. Honestly, I'm lacked of electronic components level skill, so I need your help finding those rails and signals to identify the problem precisely. It will always be great if you can really provide me with some clear pictures on those rails like you have said before..

...You need to check all these rails to be sure they exist and are the correct voltage. You need to check the power-on signals to SIO etc to see if they exist and are OK..

And please help me on how to trace those power-on signals on the the motherboard by referring the sequential part of the schematic (page30). I'm completely noob about this here.

[LENOVO-A880]

Re: Own Custom Build Computer With ECS Motherboard-No POST

..lately I started going through all the rails and references voltages to see if there are any problems or shorts. This is what I have found for so far:

STANDARD SYSTEM VOLTAGES ON ECS G31T-M7

CPU_VTT VOLTAGES: 1.189V
RAM_VDD VOLTAGES: 1.89V
RAM_STB VOLATGES: 0.946V
NORTHBRIDGE_VCC VOLTAGES: 1.238V
SOUTHBRIDGE_VCC VOLTAGES: 0V (refer PIC)

It's clearly that here one of the VCC has been shut off to prevent any further catastrophic failure to other parts of the system. I decided to moved on to test the THERMTRIP voltage of the Southbridge to be sure that it's the cause of the shut down and without any doubts it reads a 1.189V. I think the TCC or the THERMAL CONTROL CIRCUIT is asserted so that the silicon of the CPU when core reaches 135°C can be well protected from extreme thermal damage. BTW for other issues like heatsink falling off where there is no BIOS code involved - the THERMTRIP signal is tied into the front power switch logic, and causes PS_ON# signal to be de-asserted. That remove the power of the VCC and VTT completely to avoid thermal runaway of the processor.

*UPDATE*:The advantage of the TCC scheme might even require the user to turn off power at the back of the computer, then turn it back on, before the system will allow the front power button the be used. If you then press the front power switch again, it would shut down just as quickly for second time.

....In my case here I'm likely having a short or a cracked resistor within the TCC scheme when I checked on the ceramic SMD chip(refer PIC) near to the SIO chip since the PS_ON# signal is also not present at all - my digital multimeter shows 0V when I probe it on the green wire of the new FSP PSU. I would like to know if this chip can be replaced or can it be purchased anywhere else? And can it be possible to de-soldered it from the motherboard and how? Does anyone here have any ideas ?

[Agent24]

Re: Own Custom Build Computer With ECS Motherboard-No POST

Sorry I think I missed the previous email from this thread, I didn't see you were still going with it. I'll try to look at the schematic tomorrow and give you some ideas to look at (bear in mind, I am not an expert either, though)

I can say though that I don't think there may be anything wrong with your TCC circuit etc...

The PS_ON# signal that goes to the PSU green wire is what is called an 'active low' signal. That means it is active (working) when it is at a logic low state (eg: zero volts). You can tell this is the case as there is a # at the end of the signal name.

Other methods of showing an active low signal line are ! or / often, eg: !PS_ON or /PS_ON.

Basically, the PS_ON# signal should be 0v in normal operation.


Also, I am just wondering where you measured the Southbridge VCC voltage. Your photo shows an arrow pointing to a capacitor. Did you measure the voltage on the positive lead of the capacitor, or on the capacitor body itself?

[LENOVO-A880]

Re: Own Custom Build Computer With ECS Motherboard-No POST

Also, I am just wondering where you measured the Southbridge VCC voltage. Your photo shows an arrow pointing to a capacitor. Did you measure the voltage on the positive lead of the capacitor, or on the capacitor body itself?

...wow do you mean that I'm getting into all this suspected TCC sensor broken will likely be wrong...? Do you have any ideas to actually locate the right Southbridge_VCC or other voltages on the motherboard? For the voltage of the Southbridge_VCC, I abruptly measure the positive legs of all those 3 Nichicon capacitor located between the Southbridge and the ram slots, then I just get one of the capacitors reading which is EC49, so am I measuring the correct VCC voltage here?

[Agent24]

Re: Own Custom Build Computer With ECS Motherboard-No POST

...wow do you mean that I'm getting into all this suspected TCC sensor broken will likely be wrong...? Do you have any ideas to actually locate the right Southbridge_VCC or other voltages on the motherboard? For the voltage of the Southbridge_VCC, I abruptly measure the positive legs of all those 3 Nichicon capacitor located between the Southbridge and the ram slots, so am I measuring the VCC voltage correctly?

I don't know for certain that the TCC circuit is OK, but the PS_ON# signal should be 0v in normal operation so that is not cause for worry.

Yes, measuring on the positive leg of the capacitor is good. I was just worried for a second when in your photo the arrow was pointing directly from the capacitor body itself. I just wanted to make sure you were checking in the right place.


OK. To the voltage reading...

The voltage at EC49 is called 'VDDQ' in the schematic (page 11). It's one of the power lines that supplies power to the ICH.
According to the power distribution chart it should be 1.5v. If it's 0v that's not good as you already know.

But note also that the VDDQ rail Is derived from the 1_8VSTR rail. See the schematic in page 11. They drop the 1.8v 1_8VSTR rail to 1.5v to make VDDQ, using Q21 and the opamp U4A in a linear regulator circuit.

The opamp U4A is enabled by the reference signal VREF1_05V

So, for VDDQ to come up to 1.5v, you need 1_8VSTR to be present at Q21's Drain and also the VREF1_05V rail to be present at the U4A opamp.

Check those and then let's see if they are OK too.



Also, there is something else I noticed, when re-reading your first post. You said the front panel Power LED does not turn on. I find this rather interesting.

The front panel\case Power LED is driven through lines GLED0 and GLED1 (there are two for when you use a dual-colour LED) which are in turn switched via Q22 and Q23, a couple of SMD 2N3904 transistors. These transistors get their control signals from the SIO chip.

So, if the front panel LED is not coming up, it makes me think the SIO chip may not be sending the LED0 and LED1 signals to those transistors.

Q22 and Q23 are near the front panel connector on the board. See the image I have attached.

Can you measure the voltage on all pins of both those transistors and report back with what you find?

[LENOVO-A880]

Re: Own Custom Build Computer With ECS Motherboard-No POST

.....But note also that the VDDQ rail Is derived from the 1_8VSTR rail. See the schematic in page 11. They drop the 1.8v 1_8VSTR rail to 1.5v to make VDDQ, using Q21 and the opamp U4A in a linear regulator circuit.

The opamp U4A is enabled by the reference signal VREF1_05V

So, for VDDQ to come up to 1.5v, you need 1_8VSTR to be present at Q21's Drain and also the VREF1_05V rail to be present at the U4A opamp.

Check those......

DAC-AS358 SOIC-8 (Pg.11 schematic)

PIN_03 (inverting input): 2.25V
PIN_02 (non-inverting) : 1.06V

......as U4A is part of the digital-analog converter, thus the ref signal 1_05V is pegged as HIGH by the +12V rail of the PSU. If the ref signal is available the 1_8VSTR will be present too and this voltage can be obtained frm the Q21's transistor Drain pin.

FET-APM2055N (Pg.11 schematic)

Q21's drain pin (output) : 2.94V
Q21's source pin (VDDQ): 1.53V

....about a thing that confused me was that the VDDQ that provides power to the ICH returned as normal which as you can checked frm the specs, 1.5V. I did noticed that the 1.5V can be unavailable without any reasons for a couple of times when I checked the EC49 for voltage. All voltages tested here are without the presence of the CPU. Could this be a CPU doesn't get turned off, but from time to time the clock signal stops. Do you have any ideas what is causing this...and as for the voltages that you mentioned frm the previous post which should be Q22 and Q23 I just don't have enough time to check them out yet, so the readings shall be available on the next post. Sorry for that now...

[Agent24]

Re: Own Custom Build Computer With ECS Motherboard-No POST

Hmm.. Well there are 3 things here that I find strange.

First is that Pin 3 of U4A is supposed to be 1.05v as per the schematic, yet you measured it as 2.25v.
The circuit with D9 (the 431 zener) on Page 11 generates the VREF1_05V and if those resistor values for ER10 and ER11 are correct then VREF1_05V should indeed be 1.05v. Can you check the cathode of D9 and see if it is 2.5v as it should be? Also, is the VCC3 input to ER9 correct?

Secondly, you say you measured 1.53v at Q21's source pin, but previously you said you measured 0v at EC49, which is connected to it. Can you check EC49 again, be sure you are measuring the correct lead?

Edit: Also if Q21 drain is 2.94v, that suggests the 1_8VSTR rail is too high. If you measured that correctly then perhaps U14 or some of its components are bad.

[LENOVO-A880]

Re: Own Custom Build Computer With ECS Motherboard-No POST

..........circuit with D9 (the 431 zener) on Page 11 generates the VREF1_05V and if those resistor values for ER10 and ER11 are correct then VREF1_05V should indeed be 1.05v. Can you check the cathode of D9....?

D9431-S SHUNT REGULATOR (HA17L431ALTP)

PIN_C (CATHODE): 2.47V

As for the resistors ER10 and ER11, I found out that both of the resistors are having the same 2.47V which is slightly lower here whenever I measured them towards the gnd or towards the Ref pin of the D9431-S SHUNT REGULATOR. BTW, I too tried to measure up ER10 towards the other end which also happened to be VREF1_05V output and both of the ER10 and ER11 showed 1.067V. Here it's slightly the same voltage that I got frm PIN_02 of the U4A opamp.

Does this shows that the mobo is unable to detect the presence of the VREF1_05V and it doesn't fully boot up? Would this be a faulty 5VSB rail that puts up a high surge on the EC controller circuit that drive the MOSFET and voltage controllers? I suspect that the voltages should be in HIGH state will likely be the +2.5V, +1.8V and +1.5V which also caused the +1.05V to be in LOW state or not present due to certain ic that control the signal breakdown at this point.

Also, is the VCC3 input to ER9 correct?

..it seems that the VCC3 input which should be 3V towards ER9 is now 2.86V. This voltage will likely unable to fully supply the other controllers. Can it be the VCC3 output voltage which tied to the 3VSB and 5VSB of the U12 controller is probably too low?

[Agent24]

Re: Own Custom Build Computer With ECS Motherboard-No POST

If the output from the middle point of ER10 and ER11 is 1.067v then that is OK. Maybe the schematic is just wrong as to which pin of the opamp are connected to that point. Try a continuity check between the VREF1_05V node at ER10 and ER11 and opamp pin 2 or 3. Maybe it's actually the other way around to what it shows in the schematic.

Are you still getting 0v on EC49's positive lead?

[LENOVO-A880]

Re: Own Custom Build Computer With ECS Motherboard-No POST

Are you still getting 0v on EC49's positive lead?

Hi Agent24,

It has been a while since the last post on testing the discrete signal through the capacitor EC49's positive lead. The EC49 capacitor is working perfectly but I found a badly overheated SMT package I.C. which happens to be the ST75185CT-S. From the schematic (page 25), it controls the COM port by getting a dual switching power deriving from the 5V and +12V PSU (flyback converters). It's a serial port controller, the one with label U1 from the picture I've attached...

Questions
I removed the ST75185CT-S SMT controller from the motherboard. I would like to ask if the picture here is the one which I can order for replacing the shorted one. I will be ordering component no./stock no. 686-4954 from http://my.rs-online.com/web/p/line-t...ivers/6864954/. Please do let me know if this is the real problem where the I.C. should be replaced with a new one?

Thank You...

[fzabkar]

Re: Own Custom Build Computer With ECS Motherboard-No POST

Your photo shows a UTC75232L:

https://cdn.badcaps-static.com/pdfs/...776b6b3f8f.pdf

The 75232 is a very common part that was manufactured by several companies:

http://search.datasheetcatalog.net/key/75232

[Agent24]

Re: Own Custom Build Computer With ECS Motherboard-No POST

Also you may want to look into reasons why the chip is overheating. The IC itself might have failed but also there could be an issue with its supply rails going out of spec or some part on an output pin is shorted to ground etc, like one of the capacitors on a data line maybe.

686-4954 should work as replacement.

[Merlin_AmiBay]

Re: Own Custom Build Computer With ECS Motherboard-No POST

I don't wish to devalue your efforts with this board, however, ECS (Elitegroup) bopards have always been cheap, nasty and troublesome. A long time ago, a friend of mine bought an ECS motherboard to upgrade (sic) his PC and the first message 5that we got on booting up with a Windows 95 CD in it was that the board needed a BIOS update before it would even install Win 95. We got the BIOS update and eventually got Win 95 installed, but the board crashed randomly and it was never fully stable whatever we did with it.

ECS Elitegroup motherboards are on my Top 10 list of crappy boards, right up there with the Abit ST-6 RAID board.

[fzabkar]

Re: Own Custom Build Computer With ECS Motherboard-No POST

I'm still using a 15-year-old PCChips motherboard on a daily basis. My previous PCChips motherboard lasted for 16 years or so. They both run/ran 24/7.

As for the RS232 IC, sometimes these fail when using long cable runs.

[ChaosLegionnaire]

Re: Own Custom Build Computer With ECS Motherboard-No POST

ecs seems to like using ost caps on their boards. my experience with these caps on p4 boards has been horrible/terrible to say the least. they often cause all sorts of strange issues on boards.

i once briefly had an ecs p45 board. it was a boardful of osts except for the cpu vrm out area which used polys. i got that board as part of a bundle of a sale of second hand parts. wanted the cpu, cpu cooler and ram but not the board, so i resold the board quickly. didnt trust those osts. would be a pain to recap all the 20+ caps on it too.

reviews of ecs boards have also shown they perform poorly in overclocking. this means that as the components age, there is very little headroom for allowance when things go wrong with the board components.

so if u want cheap, u got cheap. u get what u pay for.