I picked up a used Z390 AORUS PRO WIFI that fails to boot after short
power off situations, requiring resetting to defaults in the BIOS to
restart. It seems to fairly quickly deplete the battery done to 2.8
volts or less over a several weeks time.

I just tried patching a voltmeter across the battery to measure the
voltage in various states.

The current battery shows 3 V when not connected. Connecting to the PC
with the power switch off, the battery drops to 2.475 V at first,
counting down the last digit every few second.

If I then turn on the power supply, the voltage of the battery
immediately drops to below 2 V, counting down rapidly at first down to
about 1.9 V, then slowly back up and stabilizing around 2.16 V. With the
battery removed the power supply turned on and the computer shut down,
the battery socket shows 0 V.

When powered up, the PC starts and stops a few times before it gets to
the BIOS. Setting to default at that point boots it up normally. It will
continue to boot normally as long as I do not shut off the power supply.

I tried a different power supply, with the same results.

The power supplies I have only have 8 pins of the 12 pin secondary power
connector to the motherboard.

I have not yet pulled the board back out of the case for further inspection.

If anyone has seen anything like this and has ideas about the cause and
solutions, I would greatly appreciate suggestions.

Z390 AORUS PRO WIFI
I7-8700K Processor
8Gb memory
500 MB m.2 drive
2TB Hard drive

On 4/6/2022 5:39 PM, Bob F wrote:
> I picked up a used Z390 AORUS PRO WIFI that fails to boot after short power off situations, requiring resetting to defaults in the BIOS to restart. It seems to fairly quickly deplete the battery done to 2.8 volts or less over a several weeks time.
>
> I just tried patching a voltmeter across the battery to measure the voltage in various states.
>
> The current battery shows 3 V when not connected. Connecting to the PC with the power switch off, the battery drops to 2.475 V at first, counting down the last digit every few second.
>
> If I then turn on the power supply, the voltage of the battery immediately drops to below 2 V, counting down rapidly at first down to about 1.9 V, then slowly back up and stabilizing around 2.16 V. With the battery removed the power supply turned on and the computer shut down, the battery socket shows 0 V.
>
> When powered up, the PC starts and stops a few times before it gets to the BIOS. Setting to default at that point boots it up normally. It will continue to boot normally as long as I do not shut off the power supply.
>
> I tried a different power supply, with the same results.
>
> The power supplies I have only have 8 pins of the 12 pin secondary power connector to the motherboard.
>
> I have not yet pulled the board back out of the case for further inspection.
>
> If anyone has seen anything like this and has ideas about the cause and solutions, I would greatly appreciate suggestions.
>
>
>
> Z390 AORUS PRO WIFI
> I7-8700K Processor
> 8Gb memory
> 500 MB m.2 drive
> 2TB Hard drive

It's probably not a standoff post shorting to the battery area. The
nearest standoff hole is to the right.

https://c1.neweggimages.com/ProductImageCompressAll1280/13-145-091-V08.jpg

The coin cell has a series resistor, intended to prevent any electrical
activity on the right, from pushing the battery around too much.

+---- 1K ohm resistor -----------> to one leg of ORing diode
|
----- +3.0V
---
|
GND

To start with, even if, at the load end, you short the battery
to ground, no more than 3mA can flow.

This means it is not possible to drag a (healthy) battery all
the way to zero volts. You have to beat the piss out of the
battery, weaken it, then you can make larger voltage swings,
when it's just about exhausted.

*******

Power comes from two paths. The paths meet at one of these (BAS40W-05).
I know which part it is, because I have three motherboard models
with a "K45" next to the CMOS cell (all Asus motherboards).
It's a Schottky with a 0.2V or so Vfb. In any further descriptions,
I do not clutter any maths by subtracting 0.2V from everything :-)
The text below assumes "an ideal behavior" for simplicity.

https://www.diodes.com/assets/Datasheets/BAS40W_-04_-05_-06.pdf

Product marking = "K45"

You would be looking for a CR2032, a 1K ohm 0603 SMT resistor,
then, the "three legged transistor thingy" with the legend on
top "K45".
|\ |
+3.3V (derived from +5VSB, NOT FROM PSU) -------| >|----+
|/ | |
+---+-------> VBAT mon on SuperIO
|\ | | | -----> 3.0V moat on PCH
VBAT 3.0 ------ 1Kohm -------------------------| >|----+ |
|/ | --- Cbypass
--- 0.1u
Now we have to ask ourselves, how can we possibly "meet" all |
the characteristics you describe ??? GND

When the upper side of the ORing diode has a hard 3.3V source,
the output node is at 3.3V .

When the 3.3V source disappears (user pulls ATX PSU cord from wall),
the node drops to 3.0V, as only VBAT is available.

The SuperIO draws zero microamps. Documentation is simply not
available, on what measurement circuits are on the monitoring
pin. Some SuperIO list a 9 input mux for monitoring,
but they don't describe what "weight" of load is on the
3V pin. It could be a ten megohm input impedance cct.
But we don't really know. It's not specced in any way.
We know that some SuperIO, prevent the mobo from starting,
if the *monitored pin* (not the actual moat voltage) reads
precisely zero volts.

The PCH draws 10uA. The 10uA might include about 2uA to
run a 32768 Hertz quartz crystal oscillator. "Spiky" consumption
from some flip flips, comes from a ripple divider which divides
by 32768 to make a 1PPS pulse. When the watch circuit, counts
the time, there are larger current spikes. The spikes would
flatten the rail, unless holdup cap "Cbypass" was not present.

The moat can draw more power, when the computer is computing.
However at that time, "hard source 3.3V" through the upper side
of the K45, easily supplies the power for the "additional spikes"
of consumption. Reading or writing the CMOS memory, causes
those spikes. When you read or write the CMOS memory, zero current
is coming from the CR2032, because the upper +3.3V biases off
the lower diode. This implements the "OR" function of voltages.
The higher side of the ORing diode, cuts off the other side.

If the part with the "K45" label is defective, part of it is
shorted, then bad things could happen. The 1K ohm resistor is
a safety resistor, intended to prevent raw 3.3V from causing
the CMOS coin cell from exploding or such.

If the regulator that takes +5VSB and makes +3.3V fails,
you *could* have a hard source of +5VSB present. This would
ruin the PCH and you would never boot again.

When a good engineer (which is most of them) designs a board,
the VBAT components are near the battery. I don't know what
the Chinese place-and-route practices are, to comment further.

I would say there is a short to ground, where the 1K resistor
meets the "K45". The battery drops to 2.475V because 3mA
is drawn from the battery. (This could even be a short on
an internal copper plane, so does not have to be visible
externally.) A standoff short would nicely work for this...
but the standoff is a little far from the cct block.

But to me, that seems a little low, because some childrens toys
use those batteries and draw more current from them. If you can
find a spec sheet for a CR2032, the V-I curve may tell you what
the internal resistance of the cell is like.

Paul

On 2022-04-06 4:39 p.m., Bob F wrote:
> I picked up a used Z390 AORUS PRO WIFI that fails to boot after short
> power off situations, requiring resetting to defaults in the BIOS to
> restart. It seems to fairly quickly deplete the battery done to 2.8
> volts or less over a several weeks time.
>
> I just tried patching a voltmeter across the battery to measure the
> voltage in various states.
>
> The current battery shows 3 V when not connected. Connecting to the PC
> with the power switch off, the battery drops to 2.475 V at first,
> counting down the last digit every few second.
>
> If I then turn on the power supply, the voltage of the battery
> immediately drops to below 2 V, counting down rapidly at first down to
> about 1.9 V, then slowly back up and stabilizing around 2.16 V. With the
> battery removed the power supply turned on and the computer shut down,
> the battery socket shows 0 V.
>
> When powered up, the PC starts and stops a few times before it gets to
> the BIOS. Setting to default at that point boots it up normally. It will
> continue to boot normally as long as I do not shut off the power supply.
>
> I tried a different power supply, with the same results.
>
> The power supplies I have only have 8 pins of the 12 pin secondary power
> connector to the motherboard.
>
> I have not yet pulled the board back out of the case for further
> inspection.
>
> If anyone has seen anything like this and has ideas about the cause and
> solutions, I would greatly appreciate suggestions.
>
>
>
> Z390 AORUS PRO WIFI
> I7-8700K Processor
> 8Gb memory
> 500 MB m.2 drive
> 2TB Hard drive

My Sons PC is a Z390 Prime A M/B with nearly the very same setup as yours.

Looking through the manual I see his has only 8 pins on the secondary 12
volt connector, which shouldn't have any bearing on the problem.

Right near the front panel connector, lower right corner it shows a CLEA
RTC 2 pin connector, to clear Cmos you short these 2 pins together.

Make sure there is NO jumper installed on these 2 pins.

I am sure you have probably tried another new battery, if not give a
new one a shot.

Rene

On 4/6/2022 8:06 PM, Rene Lamontagne wrote:
> On 2022-04-06 4:39 p.m., Bob F wrote:
>> I picked up a used Z390 AORUS PRO WIFI that fails to boot after short power off situations, requiring resetting to defaults in the BIOS to restart. It seems to fairly quickly deplete the battery done to 2.8 volts or less over a several weeks time.
>>
>> I just tried patching a voltmeter across the battery to measure the voltage in various states.
>>
>> The current battery shows 3 V when not connected. Connecting to the PC with the power switch off, the battery drops to 2.475 V at first, counting down the last digit every few second.
>>
>> If I then turn on the power supply, the voltage of the battery immediately drops to below 2 V, counting down rapidly at first down to about 1.9 V, then slowly back up and stabilizing around 2.16 V. With the battery removed the power supply turned on and the computer shut down, the battery socket shows 0 V.
>>
>> When powered up, the PC starts and stops a few times before it gets to the BIOS. Setting to default at that point boots it up normally. It will continue to boot normally as long as I do not shut off the power supply.
>>
>> I tried a different power supply, with the same results.
>>
>> The power supplies I have only have 8 pins of the 12 pin secondary power connector to the motherboard.
>>
>> I have not yet pulled the board back out of the case for further inspection.
>>
>> If anyone has seen anything like this and has ideas about the cause and solutions, I would greatly appreciate suggestions.
>>
>>
>>
>> Z390 AORUS PRO WIFI
>> I7-8700K Processor
>> 8Gb memory
>> 500 MB m.2 drive
>> 2TB Hard drive
>
> My Sons PC is a Z390 Prime A  M/B with nearly the very same setup as yours.
>
> Looking through the manual I see his has only 8 pins on the secondary 12 volt connector, which shouldn't have any bearing on the problem.
>
> Right near the front panel connector, lower right corner it shows a CLEA RTC 2 pin connector, to clear Cmos you short these 2 pins together.
>
> Make sure there is NO jumper installed on these 2 pins.
>
> I  am sure you have probably tried another new battery, if not give a new one a shot.
>
> Rene

From the manual.

CLR_CMOS (Clear CMOS Jumper)

Use this jumper to clear the BIOS configuration and reset the CMOS values to factory defaults.

To clear the CMOS values, use a metal object like a screwdriver to touch the two pins for a few seconds.

•• Always turn off your computer and unplug the power cord from the power outlet before clearing
the CMOS values.
•• After system restart, go to BIOS Setup to load factory defaults (select Load Optimized Defaults) or
manually configure the BIOS settings (refer to Chapter 2, "BIOS Setup," for BIOS configurations).

Open: Normal
Short: Clear CMOS Values

*******

So it's the normal "suicide type". If you ignore the instructions and
leave the PSU running, when you short the contacts, the "K45" is burned.

Some boards leave two pads with no pins installed.

Some may have pins there (mine does, but it ships with no jumper).

But it is an absolute no-no to ship a retail motherboard
with a jumper across the two pins, for months on end.
That flattens the battery before the customer has a chance
to use it, as well as offering the opportunity if not
caught, to damage the K45 on the first power up.

Intel made a separate RESET terminal, for jumper usage,
but no designer seems to want to use it, so they use the
suicide method instead (jumper is across Cbypass node).

Paul

On 4/6/2022 6:32 PM, Paul wrote:
> On 4/6/2022 8:06 PM, Rene Lamontagne wrote:
>> On 2022-04-06 4:39 p.m., Bob F wrote:
>>> I picked up a used Z390 AORUS PRO WIFI that fails to boot after short
>>> power off situations, requiring resetting to defaults in the BIOS to
>>> restart. It seems to fairly quickly deplete the battery done to 2.8
>>> volts or less over a several weeks time.
>>>
>>> I just tried patching a voltmeter across the battery to measure the
>>> voltage in various states.
>>>
>>> The current battery shows 3 V when not connected. Connecting to the
>>> PC with the power switch off, the battery drops to 2.475 V at first,
>>> counting down the last digit every few second.
>>>
>>> If I then turn on the power supply, the voltage of the battery
>>> immediately drops to below 2 V, counting down rapidly at first down
>>> to about 1.9 V, then slowly back up and stabilizing around 2.16 V.
>>> With the battery removed the power supply turned on and the computer
>>> shut down, the battery socket shows 0 V.
>>>
>>> When powered up, the PC starts and stops a few times before it gets
>>> to the BIOS. Setting to default at that point boots it up normally.
>>> It will continue to boot normally as long as I do not shut off the
>>> power supply.
>>>
>>> I tried a different power supply, with the same results.
>>>
>>> The power supplies I have only have 8 pins of the 12 pin secondary
>>> power connector to the motherboard.
>>>
>>> I have not yet pulled the board back out of the case for further
>>> inspection.
>>>
>>> If anyone has seen anything like this and has ideas about the cause
>>> and solutions, I would greatly appreciate suggestions.
>>>
>>>
>>>
>>> Z390 AORUS PRO WIFI
>>> I7-8700K Processor
>>> 8Gb memory
>>> 500 MB m.2 drive
>>> 2TB Hard drive
>>
>> My Sons PC is a Z390 Prime A  M/B with nearly the very same setup as
>> yours.
>>
>> Looking through the manual I see his has only 8 pins on the secondary
>> 12 volt connector, which shouldn't have any bearing on the problem.
>>
>> Right near the front panel connector, lower right corner it shows a
>> CLEA RTC 2 pin connector, to clear Cmos you short these 2 pins together.
>>
>> Make sure there is NO jumper installed on these 2 pins.
>>
>> I  am sure you have probably tried another new battery, if not give a
>> new one a shot.
>>
>> Rene
>
> From the manual.
>
> CLR_CMOS (Clear CMOS Jumper)
>
> Use this jumper to clear the BIOS configuration and reset the CMOS
> values to factory defaults.
>
> To clear the CMOS values, use a metal object like a screwdriver to touch
> the two pins for a few seconds.
>
> •• Always turn off your computer and unplug the power cord from the
> power outlet before clearing
>    the CMOS values.
> •• After system restart, go to BIOS Setup to load factory defaults
> (select Load Optimized Defaults) or
>    manually configure the BIOS settings (refer to Chapter 2, "BIOS
> Setup," for BIOS configurations).
>
> Open: Normal
> Short: Clear CMOS Values
>
> *******
>
> So it's the normal "suicide type". If you ignore the instructions and
> leave the PSU running, when you short the contacts, the "K45" is burned.
>
> Some boards leave two pads with no pins installed.
>
> Some may have pins there (mine does, but it ships with no jumper).
>
> But it is an absolute no-no to ship a retail motherboard
> with a jumper across the two pins, for months on end.
> That flattens the battery before the customer has a chance
> to use it, as well as offering the opportunity if not
> caught, to damage the K45 on the first power up.
>
> Intel made a separate RESET terminal, for jumper usage,
> but no designer seems to want to use it, so they use the
> suicide method instead (jumper is across Cbypass node).
>
>    Paul
>

You can just put a jumper on one pin for storage. It will stay there fine.

I must have messed up my measurements earlier. I just tried connecting
directly to the battery + connector and chassis rather than removing the
battery and running jumpers from each battery socket connector to the
battery and meter outside the PC. This shows me that the battery stays
close to 3.028V, but a meter on the CMOS RESET pin starts at close to
that before and after I shut down the PC, but drops to 2V about 7
seconds after I turn the power supply off. The voltage at the battery
drops to 3.0 V at the same time. On power up it then reboots 2 or more
often 3 times before it starts and needs to have the default setting
loaded each time.

This board has "DualBIOS". I am wondering if it is showing a problem
with the primary BIOS. Or is the 2 V to the CMOS too low and it is
losing the settings. Is that drop to 2 V normal when the power supply is
off?

On 4/7/2022 1:45 AM, Bob F wrote:

> You can just put a jumper on one pin for storage. It will stay there fine.
>
> I must have messed up my measurements earlier. I just tried connecting directly to the battery + connector and chassis rather than removing the battery and running jumpers from each battery socket connector to the battery and meter outside the PC. This shows me that the battery stays close to 3.028V, but a meter on the CMOS RESET pin starts at close to that before and after I shut down the PC, but drops to 2V about 7 seconds after I turn the power supply off. The voltage at the battery drops to 3.0 V at the same time. On power up it then reboots 2 or more often 3 times before it starts and needs to have the default setting loaded each time.
>
> This board has "DualBIOS". I am wondering if it is showing a problem with the primary BIOS. Or is the 2 V to the CMOS too low and it is losing the settings. Is that drop to 2 V normal when the power supply is off?
>

The dual BIOS is just two flash chips for redundancy. It has
nothing to do with our little analog voltages problem :-)

|\ |
+3.3V (derived from +5VSB, NOT FROM PSU) -------| >|----+
|/ | |
+---+-------> VBAT mon on SuperIO
|\ | | | -----> 3.0V moat on PCH
VBAT 3.0 ------ 1Kohm -------------------------| >|----+ |
|/ | --- Cbypass
--- 0.1u
|
GND

So what you're saying, is when the upper section is not powered
(by pulling out the power cord), the Cbypass node is 2V.

3V minus 2V, divided by 1K ohm, is 1mA of current flowing into
that node and going somewhere. When the current flow number
is supposed to be 0.010 mA. Off by a factor of 100.

The lowest allowed voltage at Cbypass is 2.0 volts!

So you're getting close to the limits of CMOS clock
and CMOS RAM integrity, while the cord is unplugged.

Remember, that there are very few component loads
operating from Cbypass node. There is the clear CMOS
jumper. There is a high impedance pin on the SuperIO.
And then there is the Southbridge/PCH CMOS clock and CMOS RAM.

OK, so what leaks 1mA at 2V. A 2.0K resistor to ground would do that.
If Cbypass was accidentally stuffed with an 0603 2.0K ohm
resistor, maybe that would explain it.

You choices are:

1) Examination of board with 10x stereoscopic microscope,
looking for a component of the wrong type stuffed in
that area of circuit. You need that much magnification to
read the K45 lasermark off the three-pinner.

2) File a warranty claim with manufacturer. The board you get
will not be a new one. It could be an NFF board
(No Fault Found written on the toe tag). If you are under
30 days, sometimes a retailer will accept a return.
And sometimes the rules vary with the times. My
computer store changed the rules on a few product types,
so only the manufacturer warranty was a remedy. A HDD the
other day (returned the next day), was handled by the store.
That was my first infant mortality from the store.

That means you have to visually scan all the way from the CMOS
battery in the picture, over to the two CLR CMOS pads on the
far right. You can just barely see the two CLR CMOS pads. If
a resistor was place on the pads... etc.

https://c7.alamy.com/comp/2FX8J4A/z390-aorus-pro-wifi-high-end-motherboard-gigabyte-aorus-logo-and-bios-cmos-battery-detail-macro-kts-cr-2032-closeup-top-view-from-above-pc-comp-2FX8J4A.jpg

You can also see there are via holes right next to the
standoff screw mounting point. The standoff screws, on
purpose, are at GND potential.

Paul

On 4/6/2022 11:35 PM, Paul wrote:
> On 4/7/2022 1:45 AM, Bob F wrote:
>
>> You can just put a jumper on one pin for storage. It will stay there
>> fine.
>>
>> I must have messed up my measurements earlier. I just tried connecting
>> directly to the battery + connector and chassis rather than removing
>> the battery and running jumpers from each battery socket connector to
>> the battery and meter outside the PC. This shows me that the battery
>> stays close to 3.028V, but a meter on the CMOS RESET pin starts at
>> close to that before and after I shut down the PC, but drops to 2V
>> about 7 seconds after I turn the power supply off. The voltage at the
>> battery drops to 3.0 V at the same time. On power up it then reboots 2
>> or more often 3 times before it starts and needs to have the default
>> setting loaded each time.
>>
>> This board has "DualBIOS". I am wondering if it is showing a problem
>> with the primary BIOS. Or is the 2 V to the CMOS too low and it is
>> losing the settings. Is that drop to 2 V normal when the power supply
>> is off?
>>
>
> The dual BIOS is just two flash chips for redundancy. It has
> nothing to do with our little analog voltages problem :-)
>
>                                                          |\ |
>          +3.3V (derived from +5VSB, NOT FROM PSU) -------| >|----+
>                                                          |/ |    |
>
> +---+-------> VBAT mon on SuperIO
>                                                          |\ |    |   |
> -----> 3.0V moat on PCH
>          VBAT 3.0  ------ 1Kohm -------------------------| >|----+   |
>                                                          |/ |       ---
> Cbypass
>                                                                     ---
> 0.1u
>                                                                      |
>                                                                     GND
>
> So what you're saying, is when the upper section is not powered
> (by pulling out the power cord), the Cbypass node is 2V.
>
> 3V minus 2V, divided by 1K ohm, is 1mA of current flowing into
> that node and going somewhere. When the current flow number
> is supposed to be 0.010 mA. Off by a factor of 100.
>
> The lowest allowed voltage at Cbypass is 2.0 volts!

I miss-remembered the reading when I wrote that. It actually gets down
as low as 1.72V.

That is at the 2 pins next to the "Front panel" connector that show in
the manual as CLR_CMOS. But there are 2 solder pads about 3/4"
diagonally toward the center of the board from those pins clearly
labeled CLR_CMOS with "short - clear CMOS" and "Open - Normal". Those
pads are at 4.79V and 5.05V with the supply on, and computer off. The
front of those pads drops to near 0V in less than a minute after turning
off the supply. The other pad, nearer the battery drops way slower
towards 0V. When the start button is pushed, the front pad goes to 5V
and the other to 12V. This does NOT appear to be a CLR_CMOS location,
unless something if really wrong. There should not be 12V on CLR_CMOS
pads, right?

<- Battery
______________plastic cover________| |
|
O O <- Pads |
CLR_CMOS |
Short | CLR_CMOS | Front of board
OPEN | NORMAL |
Batt Voltage |
| |
O O Pins |
------------------------ |
| Front Panel | |
|
________________________________________|

>
> So you're getting close to the limits of CMOS clock
> and CMOS RAM integrity, while the cord is unplugged.

At 1.76V at the pins, it seems I'm there.

>
> Remember, that there are very few component loads
> operating from Cbypass node. There is the clear CMOS
> jumper. There is a high impedance pin on the SuperIO.
> And then there is the Southbridge/PCH CMOS clock and CMOS RAM.
>
> OK, so what leaks 1mA at 2V. A 2.0K resistor to ground would do that.
> If Cbypass was accidentally stuffed with an 0603 2.0K ohm
> resistor, maybe that would explain it.
>
> You choices are:
>
> 1) Examination of board with 10x stereoscopic microscope,
>    looking for a component of the wrong type stuffed in
>    that area of circuit. You need that much magnification to
>    read the K45 lasermark off the three-pinner.
>
> 2) File a warranty claim with manufacturer. The board you get
>    will not be a new one. It could be an NFF board
>    (No Fault Found written on the toe tag). If you are under
>    30 days, sometimes a retailer will accept a return.
>    And sometimes the rules vary with the times. My
>    computer store changed the rules on a few product types,
>    so only the manufacturer warranty was a remedy. A HDD the
>    other day (returned the next day), was handled by the store.
>    That was my first infant mortality from the store.

I got the board used, so warranty is not available.

>
> That means you have to visually scan all the way from the CMOS
> battery in the picture, over to the two CLR CMOS pads on the
> far right. You can just barely see the two CLR CMOS pads. If
> a resistor was place on the pads... etc.

I have not yet removed the board from the case. But I did finish my
taxes today, so I'm getting closer.

>
> https://c7.alamy.com/comp/2FX8J4A/z390-aorus-pro-wifi-high-end-motherboard-gigabyte-aorus-logo-and-bios-cmos-battery-detail-macro-kts-cr-2032-closeup-top-view-from-above-pc-comp-2FX8J4A.jpg

That is not the same as my board, which is at
https://download.gigabyte.com/FileList/Manual/mb_manual_z390-aorus-pro_wifi_1001_190708_e.pdf?v=99546c46cc78df8f6a13dfb6f642193a
>
>
> You can also see there are via holes right next to the
> standoff screw mounting point. The standoff screws, on
> purpose, are at GND potential.

I do not see obvious through holes near that screw, and the standoff
under neath the board is way smaller than the screw head, which does not
nearly cover the ground pads under the screw head.
>

Could a fused diode on the (K45?) line from the power supply be draining
current back towards that supply from the battery, resulting in that
drop to 1.72 volts (On the pins, not the pads).

On 2022-04-07 9:23 p.m., Bob F wrote:
> On 4/6/2022 11:35 PM, Paul wrote:
>> On 4/7/2022 1:45 AM, Bob F wrote:
>>
>>> You can just put a jumper on one pin for storage. It will stay there
>>> fine.
>>>
>>> I must have messed up my measurements earlier. I just tried
>>> connecting directly to the battery + connector and chassis rather
>>> than removing the battery and running jumpers from each battery
>>> socket connector to the battery and meter outside the PC. This shows
>>> me that the battery stays close to 3.028V, but a meter on the CMOS
>>> RESET pin starts at close to that before and after I shut down the
>>> PC, but drops to 2V about 7 seconds after I turn the power supply
>>> off. The voltage at the battery drops to 3.0 V at the same time. On
>>> power up it then reboots 2 or more often 3 times before it starts and
>>> needs to have the default setting loaded each time.
>>>
>>> This board has "DualBIOS". I am wondering if it is showing a problem
>>> with the primary BIOS. Or is the 2 V to the CMOS too low and it is
>>> losing the settings. Is that drop to 2 V normal when the power supply
>>> is off?
>>>
>>
>> The dual BIOS is just two flash chips for redundancy. It has
>> nothing to do with our little analog voltages problem :-)
>>
>>                                                           |\ |
>>           +3.3V (derived from +5VSB, NOT FROM PSU) -------| >|----+
>>                                                           |/ |    |
>> +---+-------> VBAT mon on SuperIO
>>                                                           |\ |    |
>> | -----> 3.0V moat on PCH
>>           VBAT 3.0  ------ 1Kohm -------------------------| >|----+   |
>>                                                           |/ |
>> --- Cbypass
>>
>> --- 0.1u
>>                                                                       |
>>                                                                      GND
>>
>> So what you're saying, is when the upper section is not powered
>> (by pulling out the power cord), the Cbypass node is 2V.
>>
>> 3V minus 2V, divided by 1K ohm, is 1mA of current flowing into
>> that node and going somewhere. When the current flow number
>> is supposed to be 0.010 mA. Off by a factor of 100.
>>
>> The lowest allowed voltage at Cbypass is 2.0 volts!
>
> I miss-remembered the reading when I wrote that. It actually gets down
> as low as 1.72V.
>
> That is at the 2 pins next to the "Front panel" connector that show in
> the manual as CLR_CMOS. But there are 2 solder pads about 3/4"
> diagonally toward the center of the board from those pins clearly
> labeled CLR_CMOS with "short - clear CMOS" and "Open - Normal".  Those
> pads are at 4.79V and 5.05V with the supply on, and computer off. The
> front of those pads drops to near 0V in less than a minute after turning
> off the supply. The other pad, nearer the battery drops way slower
> towards 0V. When the start button is pushed, the front pad goes to 5V
> and the other to 12V. This does NOT appear to be a CLR_CMOS location,
> unless something if really wrong. There should not be 12V on CLR_CMOS
> pads, right?
>
> <- Battery
> ______________plastic cover________|    |
>                                    |
> O  O  <- Pads                                     |
> CLR_CMOS                                |
> Short | CLR_CMOS                        | Front of board
> OPEN  | NORMAL                          |
>                 Batt Voltage            |
>                     |                   |
>                     O  O  Pins          |
>            ------------------------     |
>            |   Front Panel        |     |
>                                         |
> ________________________________________|
>
>>
>> So you're getting close to the limits of CMOS clock
>> and CMOS RAM integrity, while the cord is unplugged.
>
> At 1.76V at the pins, it seems I'm there.
>
>>
>> Remember, that there are very few component loads
>> operating from Cbypass node. There is the clear CMOS
>> jumper. There is a high impedance pin on the SuperIO.
>> And then there is the Southbridge/PCH CMOS clock and CMOS RAM.
>>
>> OK, so what leaks 1mA at 2V. A 2.0K resistor to ground would do that.
>> If Cbypass was accidentally stuffed with an 0603 2.0K ohm
>> resistor, maybe that would explain it.
>>
>> You choices are:
>>
>> 1) Examination of board with 10x stereoscopic microscope,
>>     looking for a component of the wrong type stuffed in
>>     that area of circuit. You need that much magnification to
>>     read the K45 lasermark off the three-pinner.
>>
>> 2) File a warranty claim with manufacturer. The board you get
>>     will not be a new one. It could be an NFF board
>>     (No Fault Found written on the toe tag). If you are under
>>     30 days, sometimes a retailer will accept a return.
>>     And sometimes the rules vary with the times. My
>>     computer store changed the rules on a few product types,
>>     so only the manufacturer warranty was a remedy. A HDD the
>>     other day (returned the next day), was handled by the store.
>>     That was my first infant mortality from the store.
>
> I got the board used, so warranty is not available.
>
>>
>> That means you have to visually scan all the way from the CMOS
>> battery in the picture, over to the two CLR CMOS pads on the
>> far right. You can just barely see the two CLR CMOS pads. If
>> a resistor was place on the pads... etc.
>
> I have not yet removed the board from the case. But I did finish my
> taxes today, so I'm getting closer.
>
>>
>> https://c7.alamy.com/comp/2FX8J4A/z390-aorus-pro-wifi-high-end-motherboard-gigabyte-aorus-logo-and-bios-cmos-battery-detail-macro-kts-cr-2032-closeup-top-view-from-above-pc-comp-2FX8J4A.jpg
>
>
> That is not the same as my board, which is at
> https://download.gigabyte.com/FileList/Manual/mb_manual_z390-aorus-pro_wifi_1001_190708_e.pdf?v=99546c46cc78df8f6a13dfb6f642193a
>
>>
>>
>> You can also see there are via holes right next to the
>> standoff screw mounting point. The standoff screws, on
>> purpose, are at GND potential.
>
> I do not see obvious through holes near that screw, and the standoff
> under neath the board is way smaller than the screw head, which does not
> nearly cover the ground pads under the screw head.
>>
>
> Could a fused diode on the (K45?) line from the power supply be draining
> current back towards that supply from the battery, resulting in that
> drop to 1.72 volts (On the pins, not the pads).

Sorry for any misinformation I gave you.

I made the stupid assumption that you had an Asus board, Now I see
clearly that you have a Gigabyte board. so some things are different,
although the principal is the same.

Rene

On 4/7/2022 10:23 PM, Bob F wrote:

> Could a fused diode on the (K45?) line from the power supply be draining current back towards that supply from the battery, resulting in that drop to 1.72 volts (On the pins, not the pads).

It's possible the original owner cleared the CMOS with ATX
power present, and damaged the item marked K45. Sometimes
there is visible damage (enough burnt to obscure the K45 label).

One poster fixed a ruined K45 with a couple 1N914 switching
diodes. And those have more Vfb (forward bias) and that
causes the battery to apparently look like it is worn out,
a bit early. The K45 thing is a dual Schottky, with low Vfb
for best performance in the battery life domain. The reverse
leakage is supposed to be less than 1 ua or so.

Diodes can fail short, so yes, the ORing function could be
ruined. And the various nodes in the circuit are then
not isolated in the desired way, and when one side of the K45
drops, it drags down the other side. The other side having
the 1K ohm resistor in the path, and having "weak" impedance
characteristic and being easy to drag down.

I think you can see the disadvantages of this "suicide design"
on the two CLR_CMOS pads, when the users don't heed the real
reason the paragraph in the manual is worded the way it is.

The instructions make it seem almost "optional" to remove
PC power while fooling around. No, it's not optional.

And another bizarre part of this 20+ year old problem, is the
number of times that specific page in the manual was WRONG!
This resulted in PDF files, for a one page insert with
the corrected text telling people to remove the line cord
before touching CLR_CMOS. Various incorrect manuals had
people leaving the jumper on the pins, not removing the
power and so on. To their credit, a lot of copy and paste
today of manual content, means a correct description can
be copied without too much trouble.

And maybe sometimes this is apparent, when a manual talks
of "pins" and you only find "pads". The manuals can at times
be a bit too "generic", and the copy paste should have been
the "pads" version and not the "pins" version.

I hope you can find that broken three pinner. You can try
to do front-to-back ratios with your meter, to spot the
defective side. Since there are two diodes, you can compare
their conduction behavior as a means to spot that there
is damage present.

And nothing in the circuit should involve high voltages.
A 12V on the board, could be for a fan header.

Paul

On 4/7/2022 9:05 PM, Paul wrote:
> On 4/7/2022 10:23 PM, Bob F wrote:
>
>> Could a fused diode on the (K45?) line from the power supply be
>> draining current back towards that supply from the battery, resulting
>> in that drop to 1.72 volts (On the pins, not the pads).
>
> It's possible the original owner cleared the CMOS with ATX
> power present, and damaged the item marked K45. Sometimes
> there is visible damage (enough burnt to obscure the K45 label).
>
> One poster fixed a ruined K45 with a couple 1N914 switching
> diodes. And those have more Vfb (forward bias) and that
> causes the battery to apparently look like it is worn out,
> a bit early. The K45 thing is a dual Schottky, with low Vfb
> for best performance in the battery life domain. The reverse
> leakage is supposed to be less than 1 ua or so.
>
> Diodes can fail short, so yes, the ORing function could be
> ruined. And the various nodes in the circuit are then
> not isolated in the desired way, and when one side of the K45
> drops, it drags down the other side. The other side having
> the 1K ohm resistor in the path, and having "weak" impedance
> characteristic and being easy to drag down.
>
> I think you can see the disadvantages of this "suicide design"
> on the two CLR_CMOS pads, when the users don't heed the real
> reason the paragraph in the manual is worded the way it is.
>
> The instructions make it seem almost "optional" to remove
> PC power while fooling around. No, it's not optional.
>
> And another bizarre part of this 20+ year old problem, is the
> number of times that specific page in the manual was WRONG!
> This resulted in PDF files, for a one page insert with
> the corrected text telling people to remove the line cord
> before touching CLR_CMOS. Various incorrect manuals had
> people leaving the jumper on the pins, not removing the
> power and so on. To their credit, a lot of copy and paste
> today of manual content, means a correct description can
> be copied without too much trouble.
>
> And maybe sometimes this is apparent, when a manual talks
> of "pins" and you only find "pads". The manuals can at times
> be a bit too "generic", and the copy paste should have been
> the "pads" version and not the "pins" version.
>
> I hope you can find that broken three pinner. You can try
> to do front-to-back ratios with your meter, to spot the
> defective side. Since there are two diodes, you can compare
> their conduction behavior as a means to spot that there
> is damage present.
>
> And nothing in the circuit should involve high voltages.
> A 12V on the board, could be for a fan header.
>
>    Paul
>
>

There is a part close to the battery labeled S43 KO, with the KO at
right angle to the S43. With no power to the board, One of the 2
terminals on one side shows battery voltage and the other 0V. The single
terminal on the other side shows .376V drop from battery voltage. That
same voltage shows up on one end of a .8 Mohm resister very near the
CLR_CMOS pins next to the "front panel" connector. The CLR_CMOS pins
show 120Kohm across them or to ground, which one of those pins is
connected to. The voltage drop is an additional .636V between that .8
Mohm resister tab and the CLR_CMOS pin. The other end of that resister
is down 1.4V additional drop below the battery, connecting to a 140Kohm
resister to ground.

ABORT!!!

Inspecting those resisters with my old phono stylus 4" long 50X
microscope, I saw something funny. It looked like one end of one of the
resisters was sticking up higher than it should. That little 4" long
microscope made it hard to see exactly what was going on there.

I remembered that I had stashed in my basement something I got 20 or 30
years ago cheap at a yard sale - A very high quality Bausch & Lomb lab
stereo microscope. I dug it out, dusted it off, and set it up on my desk
with my ancient "high intensity" folding arm desk lamp.

Looking through the microscope at the 6 tiny resisters next to the
CLR_CMOS pins, it looked like a disaster area. That one resister I had
noticed was indeed standing up at maybe 20 degrees angle higher at on
end, with flakes of solder and what looks like ceramic flakes hanging
from it's sides. The other resisters look really rough, like they were
covered with solder shrapnel or had been manually re-soldered. One other
resister is just missing - just 2 rough solder cones. Perhaps the result
of someone getting sloppy with a screwdriver trying to clear the BIOS
settings.

Too bad! This seemed like a really good motherboard. It works great once
you get it running, as long as you don't power it off. It even
re-programmed the BIOS memory fine.

Is there any chance of finding out what value these resisters are (if
that's what they all are). I suspect, with this stereo microscope, I
might be able to replace these parts myself. It looks like only 2 of
them would need replacement.

I just realized that in cleaning up my basement yesterday, I have 2 big
boxes of over a dozen ancient motherboards and many other boards that I
can steal parts from if I can figure out what I need. I was going to
recycle them all - maybe not yet.

I wish I had a way to photograph this through the stereo microscope. It
really would be a graphic picture.

I might be able to hard wire a suitable connection to get the right
power connection , but it would be nice to be able to figure out what
parts were there initially. Putting the battery in shows 2V on the
CLR_CMOS pin now. The question will be whether the connection from that
pin voltage goes through one of the missing resisters and the connection
from that to the CMOS is still good or can be replaced.

If I had access to another board like this, that would be easy. It is
possible that my always on GA-Z270XP-SLI uses the same circuit for this
as this GA-Z390-AORUS-PRO-Wifi, but I will have to complete disconnect
many wires and move it to get that access. Blowing up the photo on the
motherboards Gigabyte webpage for the board is fuzzy, but it does look
very similar, so maybe that's my next step.

Cleaning up the trashed area around those resisters might remove a
solder bridge that is pulling down the voltage. Is there likely to be a
place nearer the south bridge to see if any voltage is getting to it.

On 4/8/2022 11:11 PM, Bob F wrote:

> Cleaning up the trashed area around those resisters might remove
> a solder bridge that is pulling down the voltage. Is there likely
> to be a place nearer the south bridge to see if any voltage is getting to it.

You would practically need to buy a second board, to
get reference materials to work with. And with some
of the companies, their designs have "revs" and without
too much trouble you could end up with something that
has a slightly different BOM.

With the components standing at an angle, maybe the
board got bumped while traveling through the solder tunnel
(double-IR reflow, meaning components on either side
of the board, components fitted with glue dots, and so on).
It's more likely though, that the previous board owner
got careless with a hot air gun.

https://forum.allaboutcircuits.com/threads/how-are-double-sided-pcbs-soldered-in-a-reflow-oven.55282/

I tried to stay away from our factory. I hope
my boss didn't notice :-) I would have shown our
factory more love... if it wasn't such a long drive
from home. Working at the factory, gives you a
splitting headache, and then... you have to drive home
again. Not good.

When Asus does boards, there are three revisions. You
don't panic if the board does not say "1.0". The third
revision is "strap free" and "production ready". The
rev number then, should reflect it is the final revision
of the first revision (as it were).

Some of the other companies then do further revisions.
In which case there could be Bill Of Material changes and so on.
There isn't much incentive to changing the CMOS circuit,
but since the damage area on your board involves more than
that, who can really say.

There was one Asus board I bought, perhaps a P4B, where
Asus added a red LED next to the AGP slot. It switched
off the board, so if you installed the "wrong kind" of
video card, the board would not start. This was only
on "PCB 1.04" or maybe "PCB 1.05". At the computer store,
I asked the staff to open the box so I could verify the
PCB revision and the presence of the red LED. Then I paid
for it and took it home. (Of course the red LED never
came on, but at least I had the protection feature there
if such a situation were to arise. There were some mis-keyed
SIS video cards that would burn the board if you installed them.)
That's an example of where the Rev number has zoomed up.
And if you had the board without the red LED, you bought
a second board with the red LED, the Bill Of Materials would
not match.

Paul

On 4/7/2022 9:05 PM, Paul wrote:
> On 4/7/2022 10:23 PM, Bob F wrote:
>
>> Could a fused diode on the (K45?) line from the power supply be
>> draining current back towards that supply from the battery, resulting
>> in that drop to 1.72 volts (On the pins, not the pads).
>
> It's possible the original owner cleared the CMOS with ATX
> power present, and damaged the item marked K45. Sometimes
> there is visible damage (enough burnt to obscure the K45 label).
>
> One poster fixed a ruined K45 with a couple 1N914 switching
> diodes. And those have more Vfb (forward bias) and that
> causes the battery to apparently look like it is worn out,
> a bit early. The K45 thing is a dual Schottky, with low Vfb
> for best performance in the battery life domain. The reverse
> leakage is supposed to be less than 1 ua or so.
>
> Diodes can fail short, so yes, the ORing function could be
> ruined. And the various nodes in the circuit are then
> not isolated in the desired way, and when one side of the K45
> drops, it drags down the other side. The other side having
> the 1K ohm resistor in the path, and having "weak" impedance
> characteristic and being easy to drag down.
>
> I think you can see the disadvantages of this "suicide design"
> on the two CLR_CMOS pads, when the users don't heed the real
> reason the paragraph in the manual is worded the way it is.
>
> The instructions make it seem almost "optional" to remove
> PC power while fooling around. No, it's not optional.
>
> And another bizarre part of this 20+ year old problem, is the
> number of times that specific page in the manual was WRONG!
> This resulted in PDF files, for a one page insert with
> the corrected text telling people to remove the line cord
> before touching CLR_CMOS. Various incorrect manuals had
> people leaving the jumper on the pins, not removing the
> power and so on. To their credit, a lot of copy and paste
> today of manual content, means a correct description can
> be copied without too much trouble.
>
> And maybe sometimes this is apparent, when a manual talks
> of "pins" and you only find "pads". The manuals can at times
> be a bit too "generic", and the copy paste should have been
> the "pads" version and not the "pins" version.
>
> I hope you can find that broken three pinner. You can try
> to do front-to-back ratios with your meter, to spot the
> defective side. Since there are two diodes, you can compare
> their conduction behavior as a means to spot that there
> is damage present.
>
> And nothing in the circuit should involve high voltages.
> A 12V on the board, could be for a fan header.
>

Thanks Paul for your ideas.

I have been doing more tracing of the battery voltage path. The current
measured at the battery socket is 90-100 uA, which seems to be about 10x
what it should be. The voltage from the battery (3V) after the 1k
resister (.09V drop) and diode is 2.63V, and I have traced that to a
resister near the CLR_BIOS jumper, and the other end of that resister is
at 1V. The CLR_BIOS pin however is at 2.1V currently (No power). I have
yet to find where that 2.63 is reduced going to the CLR pin, but maybe
that does not matter. Or maybe, it is the problem, signaling something
to drop the voltage to the southbridge by being only 2.1V. No, I tried
jumping battery voltage through a 1K resister to the CLR_CMOS pin, which
took it to 2.9V but did not change the 1.3V discussed below.

I removed the southbridge heatsink, and managed to find several pins on
both sides of the board by the corner of the southbridge with the full
2.63 V, so it appears the extra current drain must be either in or near
the southbridge. I did also find voltage on a 20K resister on the top of
the southbridge and on a diode? to GND (to prevent negative voltage?) on
the board near the edge of the S.B. That voltage seems to vary in the
1.3 to 1.6 volt range at different times I measure it. Jumping the
CLR_BIOS pins only drops that voltage to around 1.23V. So far, I have
not found any component where the voltage is dropping from 2.63 to 1.3V.

After hooking up the power supply, with the power switch on it set to
OFF, that 1.3V jumped to 1.52V, and seems to be hanging out there
several minutes even after disconnecting the supply from the board
again. Turning off the power supply for 3 seconds after shutting down
windows is sufficient to cause it to go through several resets before it
will boot to BIOS for resetting again.

I also tried jumping a 3.6V Lithium battery to the CMOS battery socket,
with no change in power off boot ability.