I've got an AGP 6600GT video card with very little in the way of aftermarket
cooling options. To make matters worse, this is a desktop case with less
than 1/8" clearance at the top of the card, so the Zalman or Vantec coolers
will not fit.
This card is quite noisy (I can hear it one floor away) so I need to quiet
it down. At this point I'm going to underclock it and add a resistor to the
fan.
Sometimes even a newer fan may not be lubed very well, it
can be worthwhile to lube it now rather than later with a
high viscosity lube- never a thin oil on a small video card
fan.
Any suggestions for a resistor value? Can I get away with 1/4 watt or do I
need to go with 1/2 watt? I doubt that the fan draws even close to .1 amps.
I'd expect it does, being a small fan does not decrease
current, rather in similar fans the difference in RPM is a
result of more current. Certainly some video card fans do
not draw that much but since yours is reported as being loud
onto the point of hearing it a floor away, it's likely using
more if not defective or needing lubed badly.
FWIW, the card is an eVGA brand with a really crappy "Doom 3" heatsink.
Fortunately it does have a seperate HSI chip sink.
Any suggestions would be appreciated.
It may have a thermal sensor on the card that is
deliberately running the fan faster than it otherwise would-
it is common in mid-grade or higher video cards.
It is better not to try and calculate out a near-wattage
resistor value rather than leaving some margin, particularly
if the resistor is free-floating and covered (to prevent
shorting out with anything). I usually use a 1W, common
values are 22 to 68 ohm, rarely 100 ohm but depends on the
fan (and whether it has the thermal sensor mentioned above).
If you put too large a value on the card, it could work when
the card is warm IF it had been warm already, in testing but
if the system were cool (previously turned off or if there
is a significant seasonal ambient temp change for whatever
the reason) the fan might not spin up at all. That's not to
imply it's difficult, only possible if greatest RPM
reduction is attempted.
For specific resistor values-
Fan current rating (on a label) is not average current, it
is peak. Since it is not constant, it will overshoot the
actual, average effective resistance of the fan some and the
resistor will not need be as large as the peak current
rating calculation would suggest. Even so, a little extra
margin can't hurt.
The other issue is the thermal control on-card. It would
have to be considered to determine the power available in
the fan subcircuit. If there were no thermal control we
could assume (as it's by far most predominant today) a 12V
fan circuit, 12V dropped across the fan (and with a resistor
in series, across the two). With a thermal control the
power must be measured, it is a variable too significant to
ignore if one is "trying" to reach the lowest resistor value
possible instead of fudging on the side of over-spec'ing it.
Since the prior to issues are unknown variables, for this
example only an upper limit is calculated-
Supposing a fan rated for 0.22A, and Ohm's law V = I * R
12 (volt) = .22 (amps current) * R (effective resistance of
fan)
so,
12 (volt) / .22 = 55 Ohm (effective resistance of fan).
Supposing the fan runs at desired RPM with a 47 Ohm resistor
in series, (the higher the current rating of the fan, the
lower your expected resistor value will be)
12 / (55+47) = 0.12A (current through circuit)
0.12 * 47 = 5.6 (voltage across resistor)
5.6 * 0.12 = 0.67 (watt resistor needed)
Now the above wattage is certainly higher than necessary
from a perspective of average current rather than peak (so
it wouldn't heat up as much), BUT resistors are seldom if
ever selected based on their (max) wattage rating, unless it
is only a short-lifespan device and there is no other choice
(not enough real-estate for a larger value). Given the .67
being sufficiently lower than 1.0W, a 1 watt resistor would
work in this case, but different fan and resistor values
could result in a 2W resistor being prudent... or possibly
not, since the factors of average vs peak current and of
whether there is the fan control circuit on-card have not
been resolved. 2W is the safest value to choose, 1W usually
enough. 1/2W is a borderline from a conservative
perspective for most implementations. If you really want to
nail down the wattage then measure the voltage drop across
the resistor you try in actual use but it's easier to just
grab a 1W resistor, 2W sometimes and feel how hot it's
getting. A resistor can run pretty hot ok, in general, but
for this kind of inline soldering if you can't hold it for 2
seconds I'd think about using a larger wattage value.
Again I caution you about reducing the card cooling in a
difficult environment such as wedged into a small case which
presumably also has other airflow limitations for noise
reduction reasons. Underclocking will help offset that.
