I knock off a tiny ceramic capacitor on the Asus V9520 Video card.
The capacitor is connected to pin 12 ocset and connected parallel to a
1500 ohm resistor.

The PWM is a APW7045
http://www.datasheetcatalog.com/data.../APW7045.shtml

I found the datasheet and it said the value is 220pf. I tried to
solder a big cermaic cap with marking 221 on it but it doesn't work.

The datasheet said ocset can shutdown the PWM. I measured the output
at Ugate and lgate, and there is 0V.

I look at it again and the layout around the PWM is similar to my
graphics.The Ugate and Lgate resistors(pin 14 and 16) are even the
same.

The only difference is the resistor connected to the Pin12 Ocset. On
the video card, the resistor is 1500 ohm, but it is 1000ohm on the
datasheet.

Since the 220pf cap does not work, should i try cap with higher value?
What value should i tried?

Thanks!

marslee@hotmail.com wrote:
> I knock off a tiny ceramic capacitor on the Asus V9520 Video card.
> The capacitor is connected to pin 12 ocset and connected parallel to a
> 1500 ohm resistor.
>
> The PWM is a APW7045
> http://www.datasheetcatalog.com/data.../APW7045.shtml
>
> I found the datasheet and it said the value is 220pf. I tried to
> solder a big cermaic cap with marking 221 on it but it doesn't work.
>
> The datasheet said ocset can shutdown the PWM. I measured the output
> at Ugate and lgate, and there is 0V.
>
> I look at it again and the layout around the PWM is similar to my
> graphics.The Ugate and Lgate resistors(pin 14 and 16) are even the
> same.
>
> The only difference is the resistor connected to the Pin12 Ocset. On
> the video card, the resistor is 1500 ohm, but it is 1000ohm on the
> datasheet.
>
> Since the 220pf cap does not work, should i try cap with higher value?
> What value should i tried?
>
> Thanks!

This chip operates in a similar manner. I have this chip on an
old Slot1 440BX motherboard. You can see that OCSET plus a resistor,
creates a voltage reference with respect to the high side.
The PHASE input is connected to the comparator, at a particular
point in time (presumably when the high side MOSFET is turned on),
and then the voltage across Rds of the high side MOSFET can be
compared to the reference.

http://www.intersil.com/data/fn/fn4567.pdf

What that means is, a bypass cap should be placed across the resistor,
only as a means of filtering out noise in the circuit. If there is
noise on the reference, it could falsely trigger OC. And then you'd
notice gate drive missing from both MOSFETs.

The APW7045 datasheet says the regulator uses "soft start", which means
in theory, it will deliver gate drive for short intervals under
overload conditions. This "hiccup" mode is the best, because if
you as the experimenter, manage to solve the overcurrent issue,
the circuit should immediately resume normal operation. The
"hiccups" are the device waiting for the overload to clear.

Some other regulators, latch the first OC condition they see, and
won't recover until you power cycle the PC. Regulators with a
hiccup mode are easier to work on, because they won't stay
latched off, but will continue to try to work if the
conditions are right.

The Intersil datasheet uses 1000pF for the filter. So going a
little higher probably won't hurt. The cap in question, should
not affect circuit stability. The cap should be kept small,
because the reference current source is only 200uA or so
and the cap will be charged by that current (so sticking
1 farad across it is not recommended).

Figure 4 on page 7 of FN4567.pdf, shows the regulator working
in hiccup mode, and it is retrying about every 80 milliseconds
or so. So figure 4 represents the overcurrent response - gate
drive is tried for a very short interval, and then removed
again, until it is time for another attempt at operation.
Your multimeter may not be able to detect the gate drive
pulse that results. You'd need a scope.

Paul