ECS K7VZA - strange behaviour after recap

[Dan81]

Another freebie from the strange package I've explained in the best cheap/free scores thread.


This one came to me with all the VRM G-Luxons bulging. Literally EVERYTHING around that place (aside from a small 1000uF 6.3v G-Luxon which wasn't bulged at all, but close to the caps that bulged ) was bulging. Other caps that were away from the VRMs were fine and not bulging or out of spec. Before recapping, the board would simply not POST or beep regardless if RAM or GPU was present, meaning it wouldn't detect the CPU at all.

I recapped the board and got rid of every G-Luxon. VRM low and high got a mix of OST RLX (2007 datecode, barely used) and Rubycon MBZ 3300uF 6.3v caps. (I used what I had available.) I used known good caps, to eliminate any possibility of using a dud cap. Most of the RAM caps are MBZ, with the AGP slot and PCI slots below having OST RLX.

Anyways the symptoms are these:

-will not POST with RAM installed - no beeps at all. Not even the missing GPU beep.
-will beep if there is no RAM installed - I get the continuous beeps signaling no RAM present in slots

I've used every known good (meaning one that POSTs with all my other boards I own) PSU I've had in the house. Same symptoms.

Anything I could possibly try? I'd guess it hangs somewhere on the C1 POST code ( I have no possibility to find a POST card here, so don't ask.) or after that.

I should mention I've also tried about 5-6 different SD-RAM sticks of various speeds (PC100 and 133.) with no result.

[Per Hansson]

Re: ECS K7VZA - strange behaviour after recap

What CPU are you using for testing?
Note depending on what revision the PCB is it supports different processors:
http://soggi.eu/motherboards/ecs-eli.../K7VZA-3.0.htm

[cpt.charlie]

Re: ECS K7VZA - strange behaviour after recap

Maybe the dc-dc converter doesn't like a too low ESR.

[Dan81]

Re: ECS K7VZA - strange behaviour after recap

What CPU are you using for testing?
Note depending on what revision the PCB is it supports different processors:
http://soggi.eu/motherboards/ecs-eli.../K7VZA-3.0.htm

Athlon 1133 T-Bird. It's a black Rev 3.0 board. (I only had Rev 3 boards through my hands, no 1.0s)

Maybe the dc-dc converter doesn't like a too low ESR.

All the caps that were bulged were G-Luxons 2200uF 6.3v.

I'll look up both the specs for the replacement caps and for the G-Luxons so I can find out what ESR they had.

But still... if it would have been the ESR, wouldn't the board actually not beep at all? (meaning no beep at all regardless if there's RAM or GPU - just as if the CPU wasn't installed.)

[momaka]

Re: ECS K7VZA - strange behaviour after recap

Inspect the motherboard carefully for any ripped or damaged traces, especially around the edges. Some people handle their boards very carelessly.

Also check the CPU output voltage. Should be 1.65V or 1.75V for that t-bird, if I am not mistaken.

That leads me to ask: are there any jumpers for setting the CPU voltage or frequency? If yes, double-check them. Also check any other jumpers that the board has - not only if they are in the right position but if they are functional as well. I had a board refuse to post because one of the jumpers was slightly damaged on the bottom and not making good contact.

Maybe the dc-dc converter doesn't like a too low ESR.

That shouldn't be an issue. Though I will note that most ECS motherboards from that era had a fairly low-frequency switching CPU VRM, typically controlled by a KA7500 PWM chip and having 6x 2200 uF caps on the output of the CPU.

As a test, Dan81 could also try just 3x 2200/3300 uF caps on the CPU output if their ESR specs are much more superior to the G-Luxons.

[Dan81]

Re: ECS K7VZA - strange behaviour after recap

Inspect the motherboard carefully for any ripped or damaged traces, especially around the edges. Some people handle their boards very carelessly.

I did that from the very beginning. No traces were broken. For a fact it came with foam under it in the box,and it was looking pristine (save for the bulged G-Luxon LZ caps.)

Also check the CPU output voltage. Should be 1.65V or 1.75V for that t-bird, if I am not mistaken.

Where can I measure the voltage? On the VRM low caps?

And by the way VCORE is present, or else I wouldn't get any beeps. (as I said, the board WILL beep if there is no RAM)

That leads me to ask: are there any jumpers for setting the CPU voltage or frequency? If yes, double-check them. Also check any other jumpers that the board has - not only if they are in the right position but if they are functional as well. I had a board refuse to post because one of the jumpers was slightly damaged on the bottom and not making good contact.

I used good jumpers that make contact. And no, there's no voltage jumper for the CPU - just for FSB - 100 or 133.

That shouldn't be an issue. Though I will note that most ECS motherboards from that era had a fairly low-frequency switching CPU VRM, typically controlled by a KA7500 PWM chip and having 6x 2200 uF caps on the output of the CPU.

As a test, Dan81 could also try just 3x 2200/3300 uF caps on the CPU output if their ESR specs are much more superior to the G-Luxons.

Here's the ESR specs for the caps I used, straight out their datasheets. The MBZ and RLX are the caps I used to recap the bulging LZ caps.

Rubycon MBZ: 0.012 ohms (4 used on mobo)
OST RLX: 0.014 ohms (8 used on mobo)
G-Luxon LZ: 0.025 ohms (the ones I pulled out and were bulging.

EDIT:

Voltages measured:

VRM high: 4.65V
VRM low: 1.78V (this must be VCORE as momaka said)
SD-RAM: 3.22V (makes sense since it runs at 3.3v)

[momaka]

Re: ECS K7VZA - strange behaviour after recap

And by the way VCORE is present, or else I wouldn't get any beeps. (as I said, the board WILL beep if there is no RAM)

That's true, but it is always worth checking aways. Sometimes, you just don't know if it too low but the board is still trying to boot with that. In which case, anything would be possible as an error.

I used good jumpers that make contact.

Good.

And no, there's no voltage jumper for the CPU - just for FSB - 100 or 133.

I assume you tried both.
In any case, use 100 MHz for now, as that is typically the failsafe FSB.

Here's the ESR specs for the caps I used, straight out their datasheets. The MBZ and RLX are the caps I used to recap the bulging LZ caps.

Rubycon MBZ: 0.012 ohms (4 used on mobo)
OST RLX: 0.014 ohms (8 used on mobo)
G-Luxon LZ: 0.025 ohms (the ones I pulled out and were bulging.

So looks like the caps have about half the ESR. In that case, try running with half the caps. I doubt the issue will be ESR-related, but if you run out of ideas and nothing else works, this would also be worth a try, I think. And in any case, if the board is dead, you probably would be pulling the caps from it anyways to use for something else.

Voltages measured:

VRM high: 4.65V
VRM low: 1.78V (this must be VCORE as momaka said)
SD-RAM: 3.22V (makes sense since it runs at 3.3v)

4.65 Volts on the VRM high side... really?!? That is quite a major drop. Double-check that. If that is correct, double-check the 5V rail coming out of the PSU. I know you said you tested several different power supplies, but if all drop down to less than 4.75V on the 5V rail under load, that's an issue right there! ATX spec says no more than +/- 0.25V on the 5V and 5VSB rails (5%).

Other than that, the VRM low and SD-RAM voltages seem okay.

The only other voltages you need to check is Northbridge Vcc and AGP. For the AGP, there should be a MOSFET or linear regulator near the slot somewhere outputting 1.5V. Looking at the pictures for this motherboard, I see there is one (FET/regulator) right under the lower-end of the slot.

The Northbridge may or may not have its own regulator. On some boards, the Northbridge is fed from the CPU V_Core (VRM low). And on others, depending on how power-hungry the NB is, it may have a dedicated buck VRM (with a inductor, like the CPU) or just a regular linear regulator / MOSFET. Again, looking at pictures of this board, I don't see any regulators or MOSFETs nearby, so either the NB is fed from CPU V_core *OR* one of the four phases near the CPU is actually a dedicated phase for the NB. Thus, check if the inductors near the CPU are all connected together by means or resistance test (but without the CPU). If connected, you should get less than an Ohm, or whatever is the lowest resistance your multimeter can show.

[Dan81]

Re: ECS K7VZA - strange behaviour after recap

4.65 Volts on the VRM high side... really?!? That is quite a major drop. Double-check that. If that is correct, double-check the 5V rail coming out of the PSU. I know you said you tested several different power supplies, but if all drop down to less than 4.75V on the 5V rail under load, that's an issue right there! ATX spec says no more than +/- 0.25V on the 5V and 5VSB rails (5%).

Yeah, with the MB as load, the voltage goes down between 4.75V (on main ATX cable) and 4.93V on other Molex plugs.

The only other voltages you need to check is Northbridge Vcc and AGP. For the AGP, there should be a MOSFET or linear regulator near the slot somewhere outputting 1.5V. Looking at the pictures for this motherboard, I see there is one (FET/regulator) right under the lower-end of the slot.

The Northbridge may or may not have its own regulator. On some boards, the Northbridge is fed from the CPU V_Core (VRM low). And on others, depending on how power-hungry the NB is, it may have a dedicated buck VRM (with a inductor, like the CPU) or just a regular linear regulator / MOSFET. Again, looking at pictures of this board, I don't see any regulators or MOSFETs nearby, so either the NB is fed from CPU V_core *OR* one of the four phases near the CPU is actually a dedicated phase for the NB. Thus, check if the inductors near the CPU are all connected together by means or resistance test (but without the CPU). If connected, you should get less than an Ohm, or whatever is the lowest resistance your multimeter can show.

I tested AGP and NB voltages:

AGP: 3.28v with no GPU
1.49v with FX5500 GPU
NB: 3.28v

I measured the pins on three 10uF 25v capacitors very close between the NB heatsink and the main 20 pin ATX connector. As for AGP, I did the same with the nearest 6.3v cap it had.

[momaka]

Re: ECS K7VZA - strange behaviour after recap

Yeah, with the MB as load, the voltage goes down between 4.75V (on main ATX cable) and 4.93V on other Molex plugs.

That is a pretty sad voltage drop. Definitely something suspicious going on with PSU or motherboard.

I tested AGP and NB voltages:

AGP: 3.28v with no GPU
1.49v with FX5500 GPU

Good.

NB: 3.28v

Hmmm. :\ I know this is an old motherboard, but I still wouldn't expect the NB to run on 3.3V. More like 2.5V max, if even that. Those 10uF 25V caps you were measuring may be connected to a secondary rail going to the NB (as this is an old AMD motherboard, the NB likely needs a secondary rail of 3.3V to communicate with the SD-RAM, because all SD-RAM transactions go trough NB as opposed directly through CPU with newer AMD chips with integrated memory controllers and DDR1/2/3/4 RAM.) Flip the board on the back and see if there are any large traces under the NB connecting it to the CPU V_core rail.

Other than that, I am out of ideas for now what the problem may be. With all of the heat waves going through your region right now, maybe give the motherboard a soapy bath and let it dry in the sun for a few days. If nothing else, at least you will have a shiny-looking motherboard to hang on the wall as an art piece.

[Dan81]

Re: ECS K7VZA - strange behaviour after recap

Hmmm. :\ I know this is an old motherboard, but I still wouldn't expect the NB to run on 3.3V. More like 2.5V max, if even that. Those 10uF 25V caps you were measuring may be connected to a secondary rail going to the NB (as this is an old AMD motherboard, the NB likely needs a secondary rail of 3.3V to communicate with the SD-RAM, because all SD-RAM transactions go through NB as opposed directly through CPU with newer AMD chips with integrated memory controllers and DDR1/2/3/4 RAM.) Flip the board on the back and see if there are any large traces under the NB connecting it to the CPU V_core rail.

Other than that, I am out of ideas for now what the problem may be. With all of the heat waves going through your region right now, maybe give the motherboard a soapy bath and let it dry in the sun for a few days. If nothing else, at least you will have a shiny-looking motherboard to hang on the wall as an art piece.

No traces that I can see go to the CPU V_core rail. I did find another cap inbetween the SDRAM and NB of the same value (10uF 25v) that measured about 1.12v. Might help you I guess?

[momaka]

Re: ECS K7VZA - strange behaviour after recap

1.12V is too low for anything on a motherboard that old. Might be worth investigating.

Research online shows that this motherboard has a VIA VT8363A for the NB. So I stand corrected: 3.3V for this NB is indeed in spec, as the VT8363A apparently can take up to 5.5V Vcc.

I suppose it's time to wash 'n' dry the motherboard and see what happens.

[John843883]

Re: ECS K7VZA - strange behaviour after recap

How's the power supply? Any bad caps in there? I ask as I've been through a similar experience when I recapped a similar motherboard (K7S5A), turned out the power supply was full of bad caps and also had that glue stuff which turned conductive over time.

How are the connections looking for the CPU power connector? Maybe its loosened up a bit causing the voltage drop?

[Dan81]

Re: ECS K7VZA - strange behaviour after recap

How's the power supply? Any bad caps in there? I ask as I've been through a similar experience when I recapped a similar motherboard (K7S5A), turned out the power supply was full of bad caps and also had that glue stuff which turned conductive over time.

How are the connections looking for the CPU power connector? Maybe its loosened up a bit causing the voltage drop?

I already chuckled the board in the trash. No, the PSU I used didn't have any bad cap. (recapped L&C LC-B350ATX used for testing)

[John843883]

Re: ECS K7VZA - strange behaviour after recap

I already chuckled the board in the trash. No, the PSU I used didn't have any bad cap. (recapped L&C LC-B350ATX used for testing)

I only just realized how old the last post was! Is that a re-branded deer supply and are you 100% sure there wasn't and of that brown conductive glue? The one from my old 2001 athlon XP machine was and it wasn't very stable until the brown glue was removed, I'm surprised it didn't go bang actually.

[Dan81]

Re: ECS K7VZA - strange behaviour after recap

I only just realized how old the last post was! Is that a re-branded deer supply and are you 100% sure there wasn't and of that brown conductive glue? The one from my old 2001 athlon XP machine was and it wasn't very stable until the brown glue was removed, I'm surprised it didn't go bang actually.

Nah,ironically it had some yellowish glue that was as hard as a rock to remove (kinda like old burnt hot glue, if you ever saw it).

Yes,that is a rebranded Deer PSU. It is one of the better ones. Still have it, and interestingly the Koshin caps seem to hold up no problem. When I got it only the 5vSB had two bad caps which I replaced with 2006 Nichicon HMs. It has a temperature controlled fan by the way.

[John843883]

Re: ECS K7VZA - strange behaviour after recap

Nah,ironically it had some yellowish glue that was as hard as a rock to remove (kinda like old burnt hot glue, if you ever saw it).

Yes,that is a rebranded Deer PSU. It is one of the better ones. Still have it, and interestingly the Koshin caps seem to hold up no problem. When I got it only the 5vSB had two bad caps which I replaced with 2006 Nichicon HMs. It has a temperature controlled fan by the way.

Oh I think that yellow stuff is conductive too at that point (in my experience), pretty sure it starts off white.

Is it still working to this day? Mines been in storage since about 2012 when I last poked around with it, will have to get it out at some point and see if it goes bang when I plug it in!

This one had a grab bag of weird and wonderful capacitor brands, all of them bulging on the secondary side. The mains input capacitors where actually a standard value lower than their spec! 470uF coming out as 330uF or 390uF IIRC, other than the scammy ratings the ESR and leakage tested fine. Can't remember the exact brand of those but Tosin rings a bell.

[momaka]

Re: ECS K7VZA - strange behaviour after recap

Nah,ironically it had some yellowish glue that was as hard as a rock to remove (kinda like old burnt hot glue, if you ever saw it).

Yellow hard glue does sound a lot like the conductive glue (though I can't recall now if Deer / L&C ever used it - I only seen them use hot glue in their older gutless units). In any case, if it has a bubbly texture, it very likely IS the conductive glue. And if it's darker near areas of heat (like heatsinks or hot resistors), then it is definitely the conductive glue.

I already chuckled the board in the trash.

Shame, lol.
Those older boards have a lot of TO-263 MOSFETs. Also bigger SMD resistors with the value often written on them (very useful when I need SMD resistor for a project/repair). Oh, and the ports too: audio, serial, VGA, and USB are always welcomed in my parts box.

Can't remember the exact brand of those but Tosin rings a bell.

Rings a bell to me as well. In particular, when I hear of Tosin caps, I can only think of tossin' them in the trash.

[Dan81]

Re: ECS K7VZA - strange behaviour after recap

Yellow hard glue does sound a lot like the conductive glue (though I can't recall now if Deer / L&C ever used it - I only seen them use hot glue in their older gutless units). In any case, if it has a bubbly texture, it very likely IS the conductive glue. And if it's darker near areas of heat (like heatsinks or hot resistors), then it is definitely the conductive glue.

No, it's just simple hotglue like the one from hotglue guns. I think you may be thinking about the kind of glue found in Enermax units, which isn't the case.

I did see it turn fully yellow on one unit where it was really hot, but other than that it wasn't the slightest conductive.

About the ports and mosfets - I don't have a hotgun to remove them nor do I find any use for the ports. (well,maybe USB)

Sad thing is a dead ASUS A7N8X-E Deluxe will follow the same way probably (it's the one mentioned in the General Computer Discussion) as I've did nearly everything to fix it. It will turn on for 3 seconds, then shut off,with the SB getting fiery hot. I measured the resistance on all rails (sans the negative ones since it wouldn't make too much sense to, would it?) and so far I got 0 ohms for 12V, 4 ohms for 3.3v and the one I suspect is the culprit is the 5v rail alone measuring 28 ohms. 5vsb gives a infinite reading.

[RJARRRPCGP]

Re: ECS K7VZA - strange behaviour after recap

T-bird, is 1.75 V for the minimum default Vcore in most cases, LOL.
And even for a late socket 462 Athlon, 1.75 V (or around there) is usually the minimum, when OC'ing, before you get a BSOD! A lot has changed over the years!

[momaka]

Re: ECS K7VZA - strange behaviour after recap

About the ports and mosfets - I don't have a hotgun to remove them nor do I find any use for the ports. (well,maybe USB)

You don't need a hot air gun to remove them. Just two more beefy irons (one of them needs to be rated for at least 50-60 Watts and have a big tip).

Also, considering the amount of hardware you go through and the number of broken stuff you've posted about here, I really don't see how you couldn't have a use for those parts. I bet most of the motherboards you trashed could be saved if you kept a few for parts and investigated the problem more thoroughly (though, I also understand that's a matter of having the time to do so and if you like to).

MOSFETs blow out fairly often. Granted TO-263 MOSFETs are almost never used anymore (even TO-252 is becoming rare), but you can still use them to make a repair (just not a very pretty one, perhaps).

As for the SMD resistors: they are great for video card / VRM voltage modding on older hardware that doesn't support BIOS modding (and even on hardware that does, I often find that a hardware voltage mod performs more reliably and is easier to customize to your liking over a software volt mod - case in point: under-volting older video cards to make them run cooler).

And SMD ceramic caps / multi-layer ceramic caps (MLCC): I can never have enough of those, even with my stash of broken parts. Practically almost every third or fourth used video card I buy comes with broken ceramic caps. Most of the time, they are not in spots that matter too much, so the video card will work without them too. But I've had a few cards where it did matter, and not having those caps would have caused VRM damage (due to the MLCCs having a different function for the VRM controller). And then there's the convenience of not having to buy really small electrolytic caps for motherboards: for example, a 100 uF cap can usually be replaced with one or two 4.7-10 uF MLCCs, depending on the circuit.