VOGONS


ECS 761GX-M754 no POST

Topic actions

First post, by Ydee

User metadata
Rank Member
Rank
Member

Hi, I bought mobo ECS 761GX-M754 for my lazy XP build (5 €). Unlucky, when I want to test it in the configuration of:

1 stick DDR400 RAM 512 MB (Kingston or PMI),
onboard VGA (SiS Mirage),
CPU Sempron 2800+,
PS/2 keyboard,
Seasonic 380W,
(all fully working in another rig)
New CMOS backup battery

only fan on CPU cooler spins - no beep, no POST. CPU is cold, northbridge is slightly warm. LED on keyboard do not shine. On transistors near of the CPU, I will not measure any voltage on any leg, the same on the back on any of the capacitors or coils near socket. I have ordinary cheap multimeter, measure + on component, - on the cape PS/2. Electrolytes don't look bloated, whether they're dry I don't know. Why should I focus on finding the cause? Thank you for the advice.
Photo is illustrated only, not mine - no rust or dust on mine. I am sorry for bad quality 😒

Attachments

  • ecs.jpg
    Filename
    ecs.jpg
    File size
    1.45 MiB
    Views
    453 views
    File license
    Fair use/fair dealing exception

Reply 1 of 24, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

I'm sure you have, but just to double check, have you plugged in that 4 pin CPU 12V power connector? Also, 11V across that inductor coil doesn't sound right at all, the current would be massive. I'd check what the resistance is with the board unplugged, it should be pretty close to 0. If it's gone open-circuit that might explain why there's no voltage to the CPU voltage regulators.

Reply 2 of 24, by Ydee

User metadata
Rank Member
Rank
Member

Thank You very much for Your help! Yes, CPU 12V was plugged in and functionally, I am sure. I took the photo of my mobo and marked measured parts.

As You can see, without CPU in socket, power on, I measure:

1) NIKO-SEMI P75N02LDG - gate+source 0V, drain 0,49V
2) same as nr.1
3) K3918 - gate+source 0,49V drain 12,19V
4) Nippon Chemicon? KZJ 1800uF 6,3V - 0,49V
5) Coils - 0,03V
6) Inductor coil - 12,19V
7) TK 1800uF 16V - 12,19V
😎 Coil (near chipset) - 1,79V
9) NIKO-SEMI P45N02LDG - gate 6,39V source 1,79V gain 3,33V
10) same NIKO - gate 11,19V source 1,16V gain 1,16V

With CPU in socket, power on:

1) gate+source 0V, drain 0,01V
2) same as nr.1
3) gate+source 0,01V drain 12,19V
4) 0,28V
5) 0,01V
6) 12,19V
7) 12,19V
8, 9,10) same values as without CPU

Unlucky, my skill is strictly limited (like my english too), therefore I dont know, where is problem. It could be dry electrolytics capacitors? Or something other? Would be nice fix it and have functional mobo with SiS chipset, which I like.😀

Attachments

  • 761gx.jpg
    Filename
    761gx.jpg
    File size
    1.57 MiB
    Views
    415 views
    File license
    Fair use/fair dealing exception

Reply 3 of 24, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

That inductor coil reading 6 is odd, where are you putting the meter probes? Unplug the PSU cables, then put your multimeter in resistance mode and measure the resistance across it. Also measure the resistance from one of the +12V pins (where the yellow CPU power wires go to) to both sides of the inductor. Inductors should have very low DC resistance so all those readings should be near to 0 (if your meter isn't auto-ranging, set it to 2k or less).

The useful formula is Voltage = Current * Resistance (V=IR), so if the resistance is very low, then to get 12V across the inductor the current must be very high. Given the size of the wire in that inductor, to get 12V across the inductor then things (inductor, cable, PCB traces) would get extremely hot. Carefully look at the solder joints on the back of the PCB in case one has cracked.

(When power is on then the inductor might have some voltage across it like your measurement 8, as it's there to filter out the high frequency current switching noise from the regulators, but not 12V)

Reply 4 of 24, by Ydee

User metadata
Rank Member
Rank
Member

Thank You for help anyway. I measure this inductor on both ends of wire (voltage). Resistance I measure according to your advice, between both ends of wire and then between CPU +12V and wire ends, resistance is nearly zero, as You can see on photos.

Attachments

  • crossresistance.jpg
    Filename
    crossresistance.jpg
    File size
    132.5 KiB
    Views
    388 views
    File license
    Fair use/fair dealing exception
  • inductorresistance.jpg
    Filename
    inductorresistance.jpg
    File size
    119.62 KiB
    Views
    388 views
    File license
    Fair use/fair dealing exception
  • inductora.jpg
    Filename
    inductora.jpg
    File size
    150.1 KiB
    Views
    388 views
    File license
    Fair use/fair dealing exception
  • inductor.jpg
    Filename
    inductor.jpg
    File size
    162.56 KiB
    Views
    388 views
    File license
    Fair use/fair dealing exception
  • back.jpg
    Filename
    back.jpg
    File size
    198.9 KiB
    Views
    388 views
    File license
    Fair use/fair dealing exception

Reply 5 of 24, by Ydee

User metadata
Rank Member
Rank
Member

So, I looked close at my board and if I understand correctly, then this coil inductor is powered from ATX 12V connector and is connected with high side mosfets (K3918) and capacitors (TK 1800uF 16V), then NIKO P75N02LDG are low side mosfets (2 paralel for each of three phases) and ferrite chokes ("coils") with capacitors Chemicon 1800uF 6,3V - this all is VRM? Then I think, that value 12,19V on coil inductor is correct - it is voltage from PSU 12V rail, or am I wrong?
I was watching carefully on PCB, but I could not see any damage or missing components like SMD, caps or so.
For better look I take out socket frame - there are two ICs, what is this? I googled, but no result...
And finally - it looks, like Chemicon caps are little bulging at the bottom (I can measure ESR, if needed and replace them all), but I dont know, if this could be cause of problem with no posting mobo? Can someone say?

Attachments

Reply 6 of 24, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
Ydee wrote on 2021-05-11, 16:01:
So, I looked close at my board and if I understand correctly, then this coil inductor is powered from ATX 12V connector and is c […]
Show full quote

So, I looked close at my board and if I understand correctly, then this coil inductor is powered from ATX 12V connector and is connected with high side mosfets (K3918) and capacitors (TK 1800uF 16V), then NIKO P75N02LDG are low side mosfets (2 paralel for each of three phases) and ferrite chokes ("coils") with capacitors Chemicon 1800uF 6,3V - this all is VRM? Then I think, that value 12,19V on coil inductor is correct - it is voltage from PSU 12V rail, or am I wrong?
I was watching carefully on PCB, but I could not see any damage or missing components like SMD, caps or so.
For better look I take out socket frame - there are two ICs, what is this? I googled, but no result...
And finally - it looks, like Chemicon caps are little bulging at the bottom (I can measure ESR, if needed and replace them all), but I dont know, if this could be cause of problem with no posting mobo? Can someone say?

Ah, sorry, I misunderstood. I thought the 12V was a measurement across the inductor, not a measurement from Ground to the inductor. In which case, yes, it should be about 12V on both pins for the inductor. So the problem isn't there. I'll have a look back and see what I've missed.

I think the AA-8B is an RT8800 PWM controller for the CPU power, based on a photo I found. The datasheet isn't very helpful about the chip markings. Don't know about the other one yet.

[could be an RT9605 or similar]

Reply 7 of 24, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

Ok, so perhaps the FETs aren't switching since it doesn't look like you're getting a signal on the Gate. If I'm right about those two chips then the smaller one generates the timing based on the voltage feedback and the larger one actually switches the FETs on and off. You should find the 3 phases going between them (I've marked where I think the pins should be), and then the 3 switching lines going to the Gates on the FETs, one for each group of 3 FETs. Check the resistances with the board unplugged to see if I've marked the connections correctly, then check voltages with the board on to see if anything is switching.

Attachments

  • PWM.jpg
    Filename
    PWM.jpg
    File size
    176.85 KiB
    Views
    330 views
    File license
    Fair use/fair dealing exception

Reply 8 of 24, by Ydee

User metadata
Rank Member
Rank
Member
snufkin wrote on 2021-05-11, 16:53:

Ah, sorry, I misunderstood. I thought the 12V was a measurement across the inductor, not a measurement from Ground to the inductor.

My bad, as I mentioned above, my english is everything, just not good - I am sorry and I know, it's hard to advise by far. I really appreciate Your help, thank!
Now, I have measured values (using a needle) and as you predicted, is connection between B3-7L and AA-8B. Without CPU voltage on P1,P2 and P3 is 0,49V (same, as I had measured on elyts and G+S on HI-SIDE and drain LO-SIDE FETs), but with CPU this voltage disappears = 0V. I dont know why and I am affraid, my knowledge is too little to fix it. I can replace capacitors or solder appropriately sized components, but not much more. If I had know, that replacing capacitors will help, I'll replace them. But I had a lot mobos with bulging, leaked and inflated caps, that passed the POST - that baffles me.

Attachments

  • PWM.jpg
    Filename
    PWM.jpg
    File size
    164.23 KiB
    Views
    308 views
    File license
    Fair use/fair dealing exception

Reply 9 of 24, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

Good work on the measurements, I think that confirms that the chips are RT8800 (AA-8b) and RT9605 (B3-7L). I think that the 2.2V you measured on the PWM signals going from the 8800 to the 9605 shows that there is a switching signal there. Those signals should be switching between 0V and 5V, but too fast for the meter to measure correctly, so it ends up showing something between. And you're right that the P1-3 outputs from the 9605 come from the mid-point of the upper and lower fets.

...(I wrote a bunch of stuff focussing on the 9605, then read the following in the 9605 datasheet, and changed my mind)...

If the PWM signal floating, the pin will be kept at 2.1V by the internal divider and provide the PWM controller with a recognizable level.

Is the meter reading stable at 2.2V or does it bounce around? If it's stable, then maybe there's a problem with the 8800 and it's not output anything on the PWM lines, which means they're floating and the 9605 pulls them to around 2.1V, which turns off both the upper and lower FETs.

With the CPU out, try to get a measure of Vcore, or one of the FET midpoints (drain of lower, source of upper), then at the same time press a finger around the 8800 and see if anything changes. Don't press hard enough to flex the board, but don't be too gentle either. Maybe there's a dry joint around there. If nothing changes then we'll have to probe all the pins on that chip.

At the moment I don't think it's the big electrolytic caps. They look ok, and normally when they start to fail it just means the Vcore voltage becomes noisy, rather than disappearing completely. I suppose it's possible that they failed in such a way that the feedback voltage to the 8800 spiked and damaged the 8800, but that seems very unlikely.

Reply 10 of 24, by Ydee

User metadata
Rank Member
Rank
Member

Sorry for the delay, I´ve been bussy. Again thank you so much for Your help and for Your patience.
I have voltage measured on Richtek 8800 and 9605, with power on and without CPU. If I measure correctly, it looks like this (see picture). Will it help anything?

Attachments

  • 9605volt.jpg
    Filename
    9605volt.jpg
    File size
    33.33 KiB
    Views
    271 views
    File license
    Fair use/fair dealing exception
  • 8800volt.jpg
    Filename
    8800volt.jpg
    File size
    24.7 KiB
    Views
    271 views
    File license
    Fair use/fair dealing exception

Reply 11 of 24, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

Looks useful. So I'm still assuming that the 8800 is holding the 9605 in shutdown mode, so all the UGATE and LGATE pins are never turning on. Looking at the 8800 then one thing that stands out is DVD should be +ve; from the reference design in the datasheet it should be around 1.2V. That it's 0V means that the 8800 thinks the 12V supply isn't ready to start switching, so never starts up.

Assuming that the motherboard designer roughly followed the reference design in the data sheet, then we can test a few things about DVD.

There are a couple of resistors near pin4, one marked 333 (33 kohms) and one 302 (3 kohms), just above the silkscreen marking for L59. You should find that one side of those resistors (probably nearest the 8800) are shorted together, and connected to pin 4. The other side of the 3 kohm resistor should go to ground, and the other side of the 33kohm resistor should go to 12V. Possibilities for pin 4 reading as 0V are that it's not actually connected to its pad (dry joint) or the 33k resistor isn't connected properly.

With the board off, I'd measure the resistance from pin 4 to Ground (should be something around 3k, maybe a bit less) and the resistance across the 33k resistor (should be around 33k, maybe less), then the resistances between the sides of the resistors nearest the 8800, and those resistors to pin 4. Then with power on, measure the voltage from ground of both sides of the 33k resistor. I'll try and sketch it up in a bit.

[sketch added. I've had to guess a bit about which resistors should be connected, so if the two I've marked aren't connected to pin 4, then check around the area to try and find some that are.]

Attachments

  • RT8800_DVD.jpg
    Filename
    RT8800_DVD.jpg
    File size
    232.44 KiB
    Views
    259 views
    File license
    Fair use/fair dealing exception

Reply 12 of 24, by Ydee

User metadata
Rank Member
Rank
Member

Thank You for Your other helpful advice and efforts to help. I measured the required values and it looks okay:

Pin 4 RT8800 (DVD - divider?) and ground - 2,99 kOhm
Pin 4 and near end R333 and R300 - 0 Ohm
Pin 4 and R333 - 33,0 kOhm
Pin 4 and R300 - 2,99 kOhm
Resistor 333 - 32,8 kOhm
Resistor 300 - 2,99 kOhm
R333 - 12,17 V
R300 - 0 V

But unfortunately, the problem will be more complicated, than I thought and I'm afraid we won't solve this remotely. When I measured DDR voltage, it is not 2,5V as i expected, but about 1,05V only (on most SMD capacitors next to slots).
The seller stated that it was an item removed from the PC Acer, nothing more and I thought, if it is not stated as non working or damaged, then it will work. As I see it, it's not always true.
I measured the parts, that I think form a phase for DDR voltage, but I am not sure. I wanted a mobo with PCI-E and s.754 with SiS chipset (are not so common here), but if the repair was complicated, I don't want to take up your time.

Attachments

  • ddr1.jpg
    Filename
    ddr1.jpg
    File size
    1.81 MiB
    Views
    237 views
    File license
    Fair use/fair dealing exception
  • ddr.jpg
    Filename
    ddr.jpg
    File size
    1.95 MiB
    Views
    237 views
    File license
    Fair use/fair dealing exception

Reply 13 of 24, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
Ydee wrote on 2021-05-16, 08:30:
Thank You for Your other helpful advice and efforts to help. I measured the required values and it looks okay: […]
Show full quote

Thank You for Your other helpful advice and efforts to help. I measured the required values and it looks okay:

Pin 4 RT8800 (DVD - divider?) and ground - 2,99 kOhm
Pin 4 and near end R333 and R300 - 0 Ohm
Pin 4 and R333 - 33,0 kOhm
Pin 4 and R300 - 2,99 kOhm
Resistor 333 - 32,8 kOhm
Resistor 300 - 2,99 kOhm
R333 - 12,17 V
R300 - 0 V

So the thing that's odd there is that previously you've measured the pin4 voltage as 0V, but with those resistors and voltages you've just measured should make the pin 4 voltage around 1V. So something else must be holding it at zero, and given what you've just found about the RAM voltage then possibly the motherboard keeps the CPU voltage off until other supplies are ok. Looking at one of your original photos, then it looks like there may be a trace going from pin 4 to somewhere else on the board. I've tried following it as far as I can (marked in red), can you tell where it goes?

DVD_trace.jpg
Filename
DVD_trace.jpg
File size
1.76 MiB
Views
223 views
File license
Fair use/fair dealing exception

But unfortunately, the problem will be more complicated, than I thought and I'm afraid we won't solve this remotely. When I measured DDR voltage, it is not 2,5V as i expected, but about 1,05V only (on most SMD capacitors next to slots).
The seller stated that it was an item removed from the PC Acer, nothing more and I thought, if it is not stated as non working or damaged, then it will work. As I see it, it's not always true.
I measured the parts, that I think form a phase for DDR voltage, but I am not sure. I wanted a mobo with PCI-E and s.754 with SiS chipset (are not so common here), but if the repair was complicated, I don't want to take up your time.

If this was an ebay thing then returning it might be an option (stuff listed as 'used' should still work). Problem with any trouble shooting is that until you find the problem you don't know if it's going to be something simple or complicated. At the moment your board power supplies aren't starting up, and until that's fixed we can't know if there are any other problems. Looking at a photo of someone else's board, there are a couple of big FETs in the corner between the PCIe slot and the DDR slots. Could you measure the voltages on those pins and photograph them?

Reply 14 of 24, by Ydee

User metadata
Rank Member
Rank
Member
snufkin wrote on 2021-05-16, 11:38:

Looking at a photo of someone else's board, there are a couple of big FETs in the corner between the PCIe slot and the DDR slots. Could you measure the voltages on those pins and photograph them?

Yes, if You click on a picture in my previous post to enlarge it, You can see the values written there (sorry, red numbers were not the happiest choice ). There are 3 mosfets (2x NIKO SEMI P45N02LDG, 1x D1703L), voltage regulator AT9173CG and DC/DC converter controller GS34063S.

I've tried following it as far as I can (marked in red), can you tell where it goes?

It looks like the trace leads to minus pole of capacitor 1800uF/16V or somewhere below it. I can measure it, if You need.
Mobo was not from Ebay, only from private advert, but it's not about money, only that we don't really know if it was functional and whether it's worth fixing...

Attachments

  • trace.jpg
    Filename
    trace.jpg
    File size
    1.57 MiB
    Views
    217 views
    File license
    Fair use/fair dealing exception

Reply 15 of 24, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
Ydee wrote on 2021-05-16, 12:40:
snufkin wrote on 2021-05-16, 11:38:

Looking at a photo of someone else's board, there are a couple of big FETs in the corner between the PCIe slot and the DDR slots. Could you measure the voltages on those pins and photograph them?

Yes, if You click on a picture in my previous post to enlarge it, You can see the values written there (sorry, red numbers were not the happiest choice ). There are 3 mosfets (2x NIKO SEMI P45N02LDG, 1x D1703L), voltage regulator AT9173CG and DC/DC converter controller GS34063S.

Oops, I missed those. It looks like the AT9173 is a low current termination voltage (not sure exactly what it does) for the DDR slots and should have 2.5V on pin 1 (VIN) and 1.25V on pin 4 (VOUT). That just sounds like another case of Vddr being wrong. I need to have a bit more of a think about the other components there, but I notice that the D1703L looks like it's been resoldered/replaced before. Soldering the large plate on the back can be a problem, so there's a chance that it's not making good contact there, could be worth giving it a prod.

I've tried following it as far as I can (marked in red), can you tell where it goes?

It looks like the trace leads to minus pole of capacitor 1800uF/16V or somewhere below it. I can measure it, if You need.
Mobo was not from Ebay, only from private advert, but it's not about money, only that we don't really know if it was functional and whether it's worth fixing...

I think perhaps the trace goes under those capacitors and carries on a bit further. Could you check the resistance between pin 4 and some of the components near the 4pin 12V CPU power connector? I think the 14pin chip is a comparator and it's possible that it's what switches the 8800 on. It might signal via the small 3 pin device just under the large inductor coil.

Attachments

Reply 16 of 24, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

I still haven't looked at those FET voltages properly, but I've found another possibly interesting chip to look at. U13, IT8282M, between the AGP and PCI slot. I can't find a good datasheet for it, but the one I can describes it as a power sequencer controller, and it has outputs that can apparently be used to enable/disable various supplies, including Vcore and Vdimm. Could you check what the voltages are on those pins, particularly pin 7 which should be the Power Good signal from the PSU, and should be 5V.

There's what looks like a better datasheet for an equivalent part here: https://www.digchip.com/datasheets/download_d … -number=ATPA02G

Reply 17 of 24, by Ydee

User metadata
Rank Member
Rank
Member

You have incredible patience, sir!

...but I notice that the D1703L looks like it's been resoldered/replaced before. Soldering the large plate on the back can be a problem,...

I don´t think, that this FET was resoldered - if You're thinking about that area (solder pads), probably was planned largest (TO-220 pack) FET to use, I think. Now is DPACK size.

I think the 14pin chip is a comparator and it's possible that it's what switches the 8800 on. It might signal via the small 3 pin device just under the large inductor coil.

I measured whole area, You can see voltages on pictures in attachments.

U13, IT8282M... Could you check what the voltages are on those pins, particularly pin 7 which should be the Power Good signal from the PSU, and should be 5V.

Of course I can, everything, what You need. Look at the photo (click to zoom in).

Attachments

  • VRM.jpg
    Filename
    VRM.jpg
    File size
    1.91 MiB
    Views
    165 views
    File license
    Fair use/fair dealing exception
  • D1703L.jpg
    Filename
    D1703L.jpg
    File size
    1.79 MiB
    Views
    165 views
    File license
    Fair use/fair dealing exception
  • IT8282.jpg
    Filename
    IT8282.jpg
    File size
    1.66 MiB
    Views
    165 views
    File license
    Fair use/fair dealing exception

Reply 18 of 24, by Ydee

User metadata
Rank Member
Rank
Member

Update: with CPU is increased voltage on ST D1703L (gate) 1,60V - was 1,38V. The same on pin 1 AS324M-E1 and there was voltage on reference pin (R) on GS431N - 2,49V as on cathode pin (C). Anode (A) stay at 0V.

Reply 19 of 24, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
Ydee wrote on 2021-05-17, 13:19:

...but I notice that the D1703L looks like it's been resoldered/replaced before. Soldering the large plate on the back can be a problem,...

I don´t think, that this FET was resoldered - if You're thinking about that area (solder pads), probably was planned largest (TO-220 pack) FET to use, I think. Now is DPACK size.

Ok, I thought the soldering didn't look as neat as the other pads, but it's probably just that the pads are oversized.

U13, IT8282M... Could you check what the voltages are on those pins, particularly pin 7 which should be the Power Good signal from the PSU, and should be 5V.

Of course I can, everything, what You need. Look at the photo (click to zoom in).

Great, thanks. So reading that datasheet I posted last, it looks like the power sequence is:
SB_PSON# (input to pin 5) triggers ATX_PSON (pin 9 output to PSU), which turns on the PSU. Once its voltages are all correct, it signals ATX_PWRGD back to this chip (pin 7). This chip then turns on other regulators, one at a time, and stops if any of the voltages are wrong. Only when all the voltages are correct does it release the CPU_PWRGD line (pin 11) and the computer can finally start to boot.

Also from the datasheet, it looks like Vdimm is the first board level regulator that the power sequencer tries to turn on.

Pin 13-15 are the output enable lines for Vdd (cpu core voltage), Vdda and Vldt (other cpu voltages, I think clock and bus termination), and you've measured those all enables as around 0V, so they're all turned off, which you can see because the input sense lines (Pins 1-3) are all around 0. Pin 12 is the Vdimm output enable, so that should be turned on, but the input to Pin 4 (the Vdimm sense pin) is only 1.13V.

So we've got confirmation that there's a problem with the Vdimm voltage, and an explanation why the CPU voltage is missing: the power sequencer keeps it turned off until everything else (including Vdimm) is right.

That means we can now start tracing forward from there, and it's likely once we fix the Vdimm problem, the other voltages will then come up correctly. Two things that would be really useful to find at this point are where pins 4 and 12 connect to, which will need your metre in ohms mode (I'd suggest the 2000 ohm range) and the board unplugged. You're looking for anything that measures close to 0 ohms between pin 4 and something else, and then pin 12 and something else. Give the meter a second on each point to see if it gets a reading or not.

If you can find one of the FETs connected to pin 4, then that's the FET that should be supplying Vdimm, probably best to start testing on those FETs near the RAM slots. Put either probe on pin 4, then start touching pins and tabs of other components with the other probe.

It might be more tricky to find where pin 12 goes to, but I'd start probing around that AS324 by the CPU12V connector. This measurement one will tell us what should be controlling the Vdimm regulation. If we can keep tracing these forward and backward then eventually we find what's failed.