HardwareBanter - View Single Post

**Paul** · December 28th 14, 07:57 AM posted to alt.comp.hardware.pc-homebuilt

wrote:
I have built and bought dozens of PCs over the last 20 years.
I have had at least 3 PCs with this problem, they were LGA775 or LGA1156
boards, but different makes. The battery would go flat after a few months.
So eventually it would boot up and say I have lost my settings,
hit FX to load failsafe defaults and continue or something like that.
I replaced with known new good batteries and they expired too.
Yet these same boards had no other flaws at all.
So I am puzzled what sort of fault causes this.

Stick your multimeter across the 1K ohm current limiting
resistor that comes from the battery.

If the load is drawing 10 microamps, times 1K ohms, that would be
10 millivolts (if normal). Any value higher than 10 millivolts,
you'd have to figure out where the current was going, and that
would not be easy.

It could be excessive Southbridge leakage, but just as easily,
a mistake in the motherboard design, causing the Southbridge to
leak. Big chips like that, you have to be really careful about
how the logic inputs are strapped when they're not being used.

The CMOS/RTC circuit block sits in a "well", with transmission
gates for interconnect to the rest of the chip. That scheme is
intended to reduce leakage out of the well area, into other
(unpowered) circuits. And keep the current draw under 10
microamps.

*******

I learned about I/O leakage, on my first job. We had a
setup like this.

Power Power
Supply Supply
| |
Circuit --- I/O ------- Circuit
Board signals Board

We turned off one of the supplies in the lab, expecting
the circuit board to be safe to pull out of the system.
Instead, the circuit board continued to run. Sufficient
power flowed through the I/O signals, to keep the
second circuit board running, including lighting up
up the status LEDs on the faceplate. One side was powered
by 5V. And the "phantom powered" circuit on the right,
it was running at somewhere between 3.6V and 4V or so. The
leakage doesn't charge up the other circuit to 5V or anything,
but enough power is available, that it was still performing
its logic function. This holds for "lightweight" circuits,
like stuff using CD4000 series logic. Any logic family with
a huge appetite for power, the leakage isn't enough to keep
them running.

The method Intel is supposed to use, is like this. The
CMOS/RTC would be the circuit on the left. The circuit
on the right would be the rest of the Southbridge. Killing
the PSU output, causes the power to go off on the right hand
side. The transmission gates are then set to "open circuit",
so the current can't leak across. That helps the battery
holding up the stuff on the left, to not end up powering
the rest of the Southbridge on the right hand side.

Power Power
Supply Supply
| |
Circuit --- transmission---- Circuit
Board gate Board

HTH,
Paul