Viewing page 2 of 3 pages.
Previous 1 2 3 Next
PSU FAN CONTROLLERS
Before moving on to what the unit does when powered up, I want to tell you
something about temperature controlled fans in other PSUs. Even though Silent
PC Review has yet to publish an "official" review on a thermistor
fan controlled PSU, I have described their behavior in the forums and in the
articles on the P4 system and PSU experiments. They usually start out quiet,
but the fan(s) quickly speed(s) up to become quite noisy -- even without a load,
outside a case, without any external heat to raise PSU temperature. The thermistor
that controls the fan speed always seems to be positioned near hot components
within the PSU that cause the thermistor to heat up and speed the fan up. This
is the behavior I have seen with two different models of Enermax 350W PSUs,
a SH 300W PSU, and a Zalman 300W PSU. Judging from comments
by a members in our forums, the new Antec True Power PSU appears to suffer the
same affliction.
The Seasonic SS-300FS is different. This PSU starts out quiet upon turn on
-- and stays that way indefinitely without a load, outside a case. It
is a very pleasant and unique distinction among thermistor fan controlled PSUs
I have seen. I later measured the startup fan voltage to be 4.34 VDC
The PSU was started outside a case without connection to a motherboard. Running
a jumper wire between the green wire contact and any black wire contact in the
main ATX connector to the motherboard tricks the PSU into starting up. There
is no load, which is not ideal for many PSUs, but this is a convenient way to
listen to its minimum noise without fussing with motherboards. Later, I tried
the PSU outside a case while connected to a motherboard, with the same results.
S2FC FAN CONTROL
The 7-page documentation about the S2FC fan control circuit
explained its quiet behavior. It also explained more clearly for me the
reason for the increasing fan speed behavior of other thermally controlled fan
PSUs. Much of the following is my interpretation and rewording of the printed
material Seasonic sent; the graphs come from Seasonic's web site.
Seasonic reasons that an ideal fan control in a PSU should achieve sufficient
air flow for cooling with minimal fan noise. Thermal control of PSU fans is
not a new idea, but the concept of tailoring the response of the fan controller
to closely follow the real cooling needs of the PSU appears unique.
At low power levels, almost any power supply can be run with minimal airflow
from any 80mm fan because little heat is generated. That minimal airflow is
more than enough for adequate cooling as the power load is increased, until
a certain critical point is reached when the heat generated demands higher airflow
to cool the PSU components adequately. At this point, the fan speed must be
increased for effective cooling. But until this critical point is reached,
any increase in fan speed is not needed for adequate cooling. It only
results in greater noise and wear on the fan.
Seasonic says its S2FC fan control circuit ensures that the fan remains at
minimal speed until extra airflow is really needed for cooling, thereby ensuring
the lowest noise over the widest range of operating conditions. Once the critical
power load point is exceeded, the curve becomes exponential, apparently to match
thermal characteristics of the PSU and to keep fan RPM (and noise) lower until
needed.
The above graph (courtesy of Seasonic's
web site) shows the behavior of 3 thermal fan controller circuits under
load for a 250W PSU. The vertical scale shows the fan voltage, the horizontal
scale shows system load in watts, which presumably has a direct correlative
temperature.
The red curve represents a linear thermal fan controller. As system
load and heat increases, fan speed increases from the 5V minimum, which is the
usual safe consistent start voltage for 12 VDC fans, linearly to 12V at full
load. This circuit does not take into account any of the cooling effects of
heatsinks, and natural convection at low power loads. In this example for the
250W model, 75W is the load at which the fan must provide additional cooling;
the linear fan control already has increased the fan voltage to ~8.5V, with
accompanying increase in noise. It is my belief that all the other thermally
controlled fan PSUs I've had my hands on follow this simple linear curve. This
is why, running idle in a cool room with minimal load, they all get louder for
no good reason.
The green curve represents a thermal fan controller with exponential
characteristics. As system load and heat increases, fan speed increases from
minimum to 12V at full load following an exponential curve. Throughout most
of the operating range, an exponential thermal fan controller will keep the
fan spinning slower and quieter than the linear thermal fan controller. However,
because this circuit still ignores the fact that no additional cooling airflow
is required well past 75W, the fan is allowed to speed up much sooner than necessary,
thereby causing unnecessary wear and tear as well as unnecessary additional
noise.
The blue curve represents Seasonic's S2FC thermal fan controller. It
starts as a straight horizontal line turn curves up exponentially. As system
load and heat increases, fan speed remains at the minimum (startup) 4.35V level.
The fan is not sped up at all until the thermistor reaches a temperature where
additional cooling is needed. For the SS300, this point is reached at ~100W
(at 25° C). When the fan does begin to speed up, it follows an exponential
curve. This means that the increase to maximum fan speed occurs faster the closer
the PSU gets to maximum rated output. The area under the curve is much smaller
than for the other curves; the difference in area represents the theoretical
reduction in noise and wear and tear provided by the Seasonic.
The characteristics of Seasonic's S2FC thermal fan controller
should make for a PSU that is quieter throughout the operating range yet remains
well cooled. A particularly salient point is the 100W speedup point of the 300W
unit.
NOTE: Although Seasonic refers to power levels, the truth is that there is no direct power monitoring within the fan controller. The thermistor is the only monitoring device in the fan
circuit. However, the circuit's calibration is such that the fan speed up point
corresponds to the temperature reached by the thermistor when the PSU is delivering
about 100W in an ambient temperature of 25° C. This explains the next
graph provided by Seasonic, which shows how the speedup point shifts depending
on temperature. So keeping the case cooler, whether with cooling fans or a cool
room, should have the effect of keeping the PSU fan low to a higher power point.
Those who have read my article on a Super
Simple Power Meter may recall that the highest measurements I obtained on
total power dissipation of various systems was 120W maximum peak for a fully
loaded AMD XP1700+ system. The long term maximum for that same computer was
110W. This result was obtained with the CPU at 100% utilization. Two points
relevant to these results:
- My simple power meter may not be highly accurate for complex AC loads.
One apparently knowledgeable reader said it is likely to read too high, perhaps
as much 20% too high. Until tested against a better power meter, the accuracy of my meter is in doubt. I have to give some credence to the idea that my measurements are inaccurately high, which means even less power is drawn.
- Most applications that most people work with keep the PC at close to idle, with
occasional instances of 100% CPU usage, again not for prolonged periods. In
other words, the load on the PSU in a desktop PC is typically dynamic, with low averages
accompanied by short burst of high power activity.
All of this suggest that the fan of the SS-300FS will not often speed up past
its startup voltage in a typical PC used in a typical way. It ought to be pretty quiet most of the time.
| Help support this site, buy from one of our affiliate retailers! |
|
