MSI Z690-A Pro DDR4 (MS-7D25

[Weissnix4711]

Hi all,

My MSI Z690-A pro DDR4 motherboard is showing no signs of life. SIO is the NCT6687D, for which I have no datasheet. I'm just going off net names from the boardview I have for this mobo and the datasheet of the similar NCT6686D.

All relevant power rails seem good, incl. 5VSB and 3VSB. When PSIN# is pulled low, the PSOUT# signal is pulled low by the SIO. However, the chipset appears to do nothing. SLP_S3# and SLP_S5# always remain low.

The RTC crystal is working, RTCRST# and RSMRST# are high. Any ideas what else may be preventing the chipset from pulling SLP_S3 low? Other than a dead chipset, but I want to exhaust every other option before I come to that conclusion.

I have very little understanding of this power sequence, I'm just trying to follow the datasheet I have the best I can. Any info is appreciated.

Thanks,
Thomas

[DynaxSC]

Re: MSI Z690-A Pro DDR4 (MS-7D25

First dump the bios and check if it is not corrupted, eg due to power failure during update (load it eg. into UEFI tool to see if it shows normal data). Instead or in parallel You can also just program the bios with a HW programmer with a stock image to see if this changed something, but do not forget to backup the current bios, as it contains important data, eg. Mac address, uuid, board serial.

Second check carefully with 10x magnifier glass wether no components are ripped off the board, and if tracks are not cut.

Check wether any components are getting hot, eg. spraying isopropanol over the board, or just with the inner side of the hand layed on the back side of the board, and if you find a hot spot check the area on the front side - use finger tip but very short, hot components can burn your finger. Also you can measure temperatures with a distance thermometer, anything between 30 to ca 50 C in case od chipset is normal.

These are the most simple and easy actions to check in the beginning.

You can also check wether the chipset tries to read the SPI rom, attached a scope probe to the pin 2 of the spi rom, activate pwrbtn signal, and look if you see any activity on the scope. It should be visible just after the pushing of the pwrbtn.

Rule out the above, if this will not help, then its starting to get more complicated.

[Weissnix4711]

Re: MSI Z690-A Pro DDR4 (MS-7D25

Hey, thanks for the reply.

I don't currently have a 1v8 adaptor on hand, so I can't read out the spi flash. However, I have tried probing both DI and clock pins of the spi flash, there is no activity whatsoever. I do also believe this very early initialisation should all be done by the SIO, right?

The board is basically immaculate. No cut traces anywhere I can see, or any corrosion / water damage.

The alcohol trick is an interesting one, thanks for that. I'll try later. For now I can tell you I could not feel anything get worryingly hot, other than the chipset which I'd guess was sitting about 50C. Kinda hot for the fact it doesn't appear to be doing anything, but I have another Z690 board where the chipset also runs insanely hot during normal usage so I think that's just how it is.

[DynaxSC]

Re: MSI Z690-A Pro DDR4 (MS-7D25

Hi, usually the very simplified proces is:

1. User asserts PWRBTN# signal (pin 61 of SIO - input)
2. SIO forwards PWRBTN# active signal to PCH (SIO pin 60 - output)
3. PCH assets SLP_S3# - (pin 64 - input on SIO)
4. SIO assets PSON# (pin63) - this is routed to PSON# on ATX connector - check for connectivity - usually there is a low ohm serial resistor in between
5. SLP_S3# is usually used/derivated to generate all EN signals on the revelant VRM's

SIO monitors several voltages on the board (less in stand-by mode, more in powered mode, dependent on which VRM's/voltages are monitored, also ATX voltages can be monitored), if there is a missing voltage, SIO will not assert PSON# signal, or will deactivate it in fully powered mode - leading somtimes to power cycling) - check the relevant VIN signals on the SIO (the 8 VIN monitoring pins are somewhere in the range 100-120), they should be in the range 0-2,048V normalized with resistor dividers. Missing voltages will stop SIO to run the ATX power. I believe SIO is also able to force pull down the SLP_S3# signal, to prevent the board to start in case relevant monitored voltages are not present (on VIN monitoring inputs).

All the Nuvoton SIO's are the same related to these pinouts, so no matter which one you have on the board. You can w/o any problem use the NCT6686D datasheet for the NCT6687D - I did not discover any differences in pinout between these chips.

In your case it seems that PCH deos not generate SPL_S3# to some reason, either PCH is damaged, or bad spi rom, damaged spi contents, or PCH has a corrupted management engine/configuration or some other signal on the SIO is in an incorrect state. PCH usually also has a connection to the PWRBTN signal before the SIO pass-through, so if this network is not delivering the signal to PCH this can also stop the PCH to assert SLP_S3#. Check also if PCH has all the different supply voltages present - attached you'll find the boardview.

What you can do is also to check all the different control signals between SIO, PCH and CPU for shorts - to do this you need to disconnect ATX power and remove the battery, as the SIO can pull down some signals also powered only by the CMOS battery producing fake shorts. Check resistance vs. ground, and do a reverse diode test on the signals (plus probe to ground, negative probe to mearured signal) -> the voltages measured should be between 0,2V to ca 0,5V, if you find something near 0 ohms (below 10 ohms) or 0V this is a reason to investigate what causes the short. The most important signals to check you can discover from any intel power sequence diagram, they are generally very similar named on any boardview / power sequence diagram. Resistances at 10 - 50 ohms usually mean no short, but some device is pulling down the signal (output mosfet resistance in conducting state).

Check also all the resistances on the VRM power coils present on the board for any shorts (board unpowered, cpu/ram not installed).
Check for VRM's with asserted EN signal, but not producing any power. Check any PWRGOOD signals for right state.
Check also for any bent pins in the LGA socket, or any dirt/thermal paste on the LGA pins (dirt on LGA pin tops can make the board not working very easily). Check/clean with isopropanol also the pads of the CPU/RAM's.
Check the data busses on the board, eg. DMI bus between SIO and CPU (can test only on the LGA socket, so must be done very carefully to not bend the pins), SMB bus. Check all clocks are working.
Are you sure CPU/RAM's are OK ? Sometime the most trivial things can be the reason, however usually a broken CPU/RAM should not stop the board from starting. To go sure try to start the board with CPU/RAM's uninstalled. Some boards do not start w/o installed CPU, to overcome this you need to short the SKTOCC# signal (socket occupied) on the SIO (pin 102).


MOD EDIT: Boardview Moved to original Schematic/Boardview request thread -> https://www.badcaps.net/forum/showthread.php?t=114026
.

[Weissnix4711]

Re: MSI Z690-A Pro DDR4 (MS-7D25

Hi,

> 3. PCH assets SLP_S3# - (pin 64 - input on SIO)
Do you mean the PCH pulls SLP_S3# low after PWRBTN# signal is received? Ie. the signal should be high (inactive) when standby power is available but before power button is pressed?

I also just realised I made an error in my original post, SLP_S3# is always low and is never pulled high. This leads me to believe it shouldn't be an issue with the PWRBTN signal as the issue appears to be somewhere earlier in the power sequence.

CPU and ram are good, I've tested them in another of the exact same model board. Also, this board will usually light up an LED if CPU / RAM are not detected, though this does not happen either.

PCH_0P82_PWRGD is high.
CPU_PWRGD is at 0.4V. Relevant or is this simply a result of the fact we only have standby power?

I'm kinda tied up with other things right now, so I'll get back to you tomorrow with more findings.

Thanks for all the help, I really appreciate it,
Thomas

[DynaxSC]

Re: MSI Z690-A Pro DDR4 (MS-7D25

Usually ".....#" means active low, so this means if there is an active L on SLP_S3# the board is in S3 sleep status.

Because by chance I have such a board just repaired "on stock", I have checked how it behaves in reality, and it is little bit more complicated, i.e.:

1. Just after switching on PSU (board still in standby, not started) SLP_S3# is H (3,3V) - inactive,
2. after pushing power button SLP_S3# stays H (3,3V), but the board/all ATX powers are starting,
3. after keeping power button pushed (shorted) 6-8 seconds (forcing power off) board stops operating, and SLP_S3# is getting L (0V),
4. after pushing power button again SLP_S3# gets H (3,3V) - bord starts and goes to S0 state,
5. after keeping power button pushed again for 6-8 seconds (forced power off) it gest L (0V),
6. then 4. & 5. repeats until eternity, unless PSU mains is completely powered off (no +5VSB standby voltage).

I'm not an expert in the S states, but most probably this means that just after applying ATX power (stand-by) the board is not in S3 state, probably it's in S4 or S5 state, thats why SLP_S3# is inactive. After then switching the board on and off with power button, the board toggles between S0 (G0) and S3 (or maybe S4/S5 state) state - would need to check the other state S-signals to determine exactly.

You can also check then what the other S-signals are making, and have a look on the ACPI state definitions to understand what's going on, and if there is anything going on, or the board stays stubbornly in one state w/o any changes.

[Weissnix4711]

Re: MSI Z690-A Pro DDR4 (MS-7D25

I have found a short. Problem is, it appears to be on the chipset package.

I had initially measured PCH_0P82_VSB to be right around the correct voltage, but I did not until just now notice there's only about .6V between it and ground. Probing around other points in the area, I eventually discovered the lowest resistance to gnd of about .4V across one of the caps right on the chipset package. This is unplugged, with CMOS battery removed.

So either one of the smt components on the chipset package is creating this short, in which case it would have around 2A running through it and presumably not exist anymore. OR, the chipset is dead.

Nothing else in the area seems to be getting particularly hot, especially not 2A worth of hot, so unless you can think of anything else I guess that's it for this board.

[DynaxSC]

Re: MSI Z690-A Pro DDR4 (MS-7D25

According to my experience the short is most probably on the chipset, caps do go very, very seldom shorted, but it is possibble. And a shorted cap would result in 0V - 0.4V speaks for chipset. However chipset usually if shorted also produces resistances very close to 0 ohms, so 0,4V-0,6V is a little bit non typical. Check the resistance on the 0,82V line, normal chipset resistance should be around tens of ohms (below 50). So I would say there is still some chance, that the VRM is demaged, and produces a strange voltage. BTW you would be surprised how much current shorted caps can withstand. So if a cap has a total short, it can withstand 10-20 A without a problem, maybe not the smallest, but the 0802 case for sure. I have often looked for shorted caps on graphic cards with voltage injection method (or more precise would be current injection), and I tell you 20 Amps are sometimes not enough to "burn out" a shorted cap, so I'm still looking for some regulated PSU with current throughoutput of 20-40 Amps, to be able to do this. And if you have on such board 100 caps in parallel, it's quite difficult to find "the" one. However shorted caps can often be distinguished from correct ones, as they have often some very small visible changes (eg. a very thin scratch, or color change). It's often possible to find the shorted cap with a magnifier glass just with our sense of sight.

BTW the chipset is possible to replace, a new one can be ordered at Aliexpress, cost ca. 40 USD for a used one (50 for a new one), however its only possible if you have a lot experience with hotair gun, as it is very easy that the solder balls get joined together due to adhesive forces and board deformation due to heat and produce shorts, so sometimes one needs to reball 2-3 times before the chipset gets soldered correctly. I do it in my lab, but the distance makes the cost probably too high. Also sometimes there is some additional chipset programming to do, as the Management Engine of the chipsets needs to be activated, depending on the way how the bios handles an uninitilized chipset. I have just written a post about this on this forum (https://www.badcaps.net/forum/showthread.php?t=120068).

[Weissnix4711]

Re: MSI Z690-A Pro DDR4 (MS-7D25

I'm not sure why I was on about ".6V" between the 0V82 rail and gnd, when I meant a resistance of .6 ohms. I guess it was a bit late lol. Also, I can't for the life of me figure out how to edit my posts. Ah well.

I've checked every cap on that net, none look weird to me.

I have some experience with BGA rework, but it's mostly flash chips and whatnot, nothing with quite as many contacts as this chipset. Also yeah, I wouldn't make much of a profit if I had to spend 40-50 USD on a new chipset, let alone taking into account my time.

I wonder if the ME crap is necessary if I reflash the board's firmware with coreboot. Either way, prolly won't bother replacing the chipset, so whatever.

Thanks for the help, despite the non-result. I really appreciate it nonetheless.

[mosesayegun]

Re: MSI Z690-A Pro DDR4 (MS-7D25

any chance you could share your boardview.
looked everywhere and cant find one for msiz690 a pro

[Per Hansson]

Re: MSI Z690-A Pro DDR4 (MS-7D25

There is a boardview linked at the end of post #4

[SMDFlea]

Re: MSI Z690-A Pro DDR4 (MS-7D25

There is a boardview linked at the end of post #4

It wasn`t linked,it was an attachment,but its been moved now. @DynaxSC should have posted it in his schematic request thread and then linked to it,rather than posting it here.
.

[McWhitePink]

Re: MSI Z690-A Pro DDR4 (MS-7D25

I'm not sure why I was on about ".6V" between the 0V82 rail and gnd, when I meant a resistance of .6 ohms. I guess it was a bit late lol. Also, I can't for the life of me figure out how to edit my posts. Ah well.

I've checked every cap on that net, none look weird to me.

I have some experience with BGA rework, but it's mostly flash chips and whatnot, nothing with quite as many contacts as this chipset. Also yeah, I wouldn't make much of a profit if I had to spend 40-50 USD on a new chipset, let alone taking into account my time.

I wonder if the ME crap is necessary if I reflash the board's firmware with coreboot. Either way, prolly won't bother replacing the chipset, so whatever.

Thanks for the help, despite the non-result. I really appreciate it nonetheless.

Hey, I don't think it's a short on that pin, for 0.8V rail, reading low resistance is normal. You got be careful that you may misjudged by the resistance value, it could be something else.
Changing the chipset is the most expensive part on the whole board, could be potentially the same price as the board you can sell on market place. that means if you couldn't fix the board by changing the chipset, you will lose money for sure.

[McWhitePink]

Re: MSI Z690-A Pro DDR4 (MS-7D25

This is great note, do you know if the PCIE_RST pin should be 3.3V or close to 0V when it is in S0? I have some boards measured 3.3V, and some measured 0V when powered, I have got confused by it.

Usually ".....#" means active low, so this means if there is an active L on SLP_S3# the board is in S3 sleep status.

Because by chance I have such a board just repaired "on stock", I have checked how it behaves in reality, and it is little bit more complicated, i.e.:

1. Just after switching on PSU (board still in standby, not started) SLP_S3# is H (3,3V) - inactive,
2. after pushing power button SLP_S3# stays H (3,3V), but the board/all ATX powers are starting,
3. after keeping power button pushed (shorted) 6-8 seconds (forcing power off) board stops operating, and SLP_S3# is getting L (0V),
4. after pushing power button again SLP_S3# gets H (3,3V) - bord starts and goes to S0 state,
5. after keeping power button pushed again for 6-8 seconds (forced power off) it gest L (0V),
6. then 4. & 5. repeats until eternity, unless PSU mains is completely powered off (no +5VSB standby voltage).

I'm not an expert in the S states, but most probably this means that just after applying ATX power (stand-by) the board is not in S3 state, probably it's in S4 or S5 state, thats why SLP_S3# is inactive. After then switching the board on and off with power button, the board toggles between S0 (G0) and S3 (or maybe S4/S5 state) state - would need to check the other state S-signals to determine exactly.

You can also check then what the other S-signals are making, and have a look on the ACPI state definitions to understand what's going on, and if there is anything going on, or the board stays stubbornly in one state w/o any changes.