[C] Date: Sun, 21 Mar 1999 16:39:15 -0800
From: miker
Subject:  Battery charge circuits

... But here's how I think the thing works. There's a chip that manages
the battery. It swithes a FET in series with the battery negative.
The load is placed directly across the battery & fet. The AC adapter goes
thru an isolating diode and is also placed directly across the battery and
fet.
There'e a green light on the front panel that shows that the battery is
charging. (ignore the flashing modes for now) The fet is normally turned
on. The only time the fet turns off and disconnects the battery is when AC
is applied AND the green light is off.

When you plug in the AC power, it is connected directly across the battery
thru the turned-on FET. (That's why it's very important to use ONLY the
supplied AC adapter. One without the correct current limit will cook your
battery and maybe destroy the computer.) 

The battery charges until the proper voltage conditions apply OR the
temperature increases to the point that the thermistor decides to turn off
the charging. At that point, the green light turns off and the FET turns
off disconnecting the battery. According to the schematics, there is no
trickle charge current. It should be safe to leave the AC adapter plugged
in continuously. The downside is that if you do it for a long time, the
battery will discharge itself and not be ready when you need it. You can
reset the charge circuit by removing the battery for a few seconds with AC
disconnected. I've not discovered all the details of resetting the charge.

Another point is that the signal that controls the green light is not the
same wire that controls the FET. I blew mine up thru a battery mishap.
After I replaced the fet, the green light would go off indicating
full charge,
but the signal to turn off the fet was broken. The battery continued to
charge and get very hot. Luckily, it's possible to synthesize the required
fet control signal from existing signals from the chip. 

FnF8 should tell you whether the machine thinks the adapter is plugged in.
Under normal conditions, 'bout the only difference is that the disk will
refuse to spin down with AC applied. There are additional power control
functions supplied by the operating system. Those seem to conflict with
the hardware power settings. I've only been able to get WIN95 to put the
machine to sleep by turning off the hardware power functions. 

So, If you can plug in the AC and the green light comes on then goes off
later, the battery is being charged and the computer understands that AC
is connected. If the battery continues to get VERY hot, you may have a bad
charging circuit. With the machine turned on, check for voltage at the
center conductor of the power socket. If there is, you may have a shorted
series diode. That would affect whether the computer thinks AC is present. 

Bad batteries can cause all sorts of strange symptoms. That would be my
first choice of something to replace.  miker

[C] From: Christian.Rausch@Physik.TU-Muenchen.DE
Date: Tue, 15 Oct 1996 18:33:34 +0000
Subject: 3 C functions

Hi folks,
thanks for your recent answers to my upgrade questions.
Yesterday I disassembled Compaqs charge.com (SP0993) and figured
out how Compaq checks for an Aero, for AC-powering and how the charge
level can be read (Hopefully this was not already discussed here recently,
I am new to this list, so do not blame me, please!)
I also wrote 3 C functions for these tasks. So, here are the 
functions (I hope you are interested. I also wrote a commented 
assembler source file of charge.com and 2  C-programs that
deal with charge.com and the following 3 functions. In case you are 
interested, just drop me a short message. I will send you the files 
then (17 kB, zipped)):

int isaero()
{
 union REGS inregs, outregs;
 inregs.x.ax=0xe800;
 int86(0x15,&inregs,&outregs);
 if( (outregs.x.cflag) || ((outregs.x.bx & -4) != 0x20c) )
   return 0;
 return 1;
}

int isacpowered()
{
 if( inp(0x1c65) & 0x40 )
   return 0;
 return 1;
}

int readcharge()
{
 int charge;
 _disable;               /* 8086 CLI instruction */
 outp( 0x2065, 0x84 );
 charge = inp( 0x2465 );
 _enable;                /* 8086 STI instruction */
 return charge;
}

cu Christian


Dr. Christian Rausch
Fakultaet fuer Physik E21
Technische Universitaet Muenchen
James-Franck-Str.
D-85747 Garching, Germany
Tel. +49 89 289 12185
Fax  +49 89 289 13776
email: crausch@physik.tu-muenchen.de