|
Fanless or not
|
|
|
Introduction:
Every now and then we receive the (somewhat surpised) question or statement that our system uses fans for cooling.
This is quite understandable as there is a major tendency in industrial computing to go fanless.
However, at White Bream, we have some very good reasons not to follow that same route, and to stick with good ol´
fan-based active cooling of the system. Read on to learn why, and to learn why fans don´t have to be so bad
at all!
Cooling requirements:
The size of these computer systems is among the smallest in the industry. Needless to say that this is
one of our unique features. But when considering fanless, one needs to start with a rather big case, especially
if the demand on operating temperature range is high.
To cool a 35W CoreDuo processor at an ambient of 60°C, the cooling solution needs to have a thermal resistance
less than ( Tcpu(max) - Tamb ) / Pcpu = (100°C - 60°C / 35W) = 1.1°C/W. A quick browse in an electronics
components catalog learns that this requires a passive heatsink no less than 150x150x27mm (6x6x1.1").
That´s already ~75% of the total volume of our system, and that is just the heat coming from the CPU...
There is one more reason not to rely on passive cooling, and that is the power distribution throughout
the system. Next table lists some system configurations and the estimated total (internal) power consumption,
and the power consumption that is attributed to the (passive) cooled components, typically just the Intel
chipset and processor:
| System: | Options: | System power: | CPU Power: |
| UNIGO | T2050, 1GB, GPRS, HD, QUSB | xx.x W | xx.x W = xx % |
| UNIGO/MINI | Z510, 512MB, GPRS, SD, DVI | 15.9 W | 3.5 W = only 22% |
With the UNIGO/MINI as worst case example, a passive cooling solution which provides a thermal coupling between
CPU, chipset and the system case will dissipate only 22% of the heat generated in the system. The other 78% is
cooled only by internal passive convection, which is usally far from enough. Only a rather big case can provice the
necessary material surface to accomodate this heatflow. Pity that having exceptional small cases is one of our
unique selling points...
This overview makes it pretty clear that a cooling solution that only targets the ´classic´ hot-spots,
CPU and chipset, is not sufficient anymore with modern low-power microprocessors.
Reliability & quality:
It is a very common misconception that fans in the system have to reduce reliability of the system. Of
course it does, but so does every other component that is added to the system. It is also true that fans
are very oten the least reliable parts of a computer. But the truth is that this really depends on how
much you are willing to spend on these fans...
The fans that are used in the White Bream mobile computer systems are Sunon Maglev units, one of the most
reliable fan architectures available. In addition, we have chosen to configure these fans as ´inlet
compressors´, which means they can work at ambient temperatures, instead of the much hotter temperatures
found at the outlet air vents. If we assume that the fans operate at a temperature of 30 °C, then they
have a typical lifetime of no less than 224752 hours. Yes; that is equal to a lifetime
of approximately 26 years! Hence much longer than the lifetime of the average vehicle the system
is build into, and certainly way much longer than the practical lifespan of the computer system itself
(particulary in terms of obsolecence due to new software, new communication technologies, etc).
Just like :
Despite the long life expectancy of the fans, they still can fail prematurely. An undetected fan failure will
jeopardize the systems reliability due to elevated temperatures, which ultimo results in premature complete
system failure. Therefore the system controller monitors the fan speed setpoint and effective RPM speed. When
these two deviate too much, an alarm is signalled to alert the user or operator of a possible fan failure. (Which
in most cases will be a loose wire or something stuck in the air intake, and blocking the fan.)
Pros & cons:
Of course there can be very good reasons not to use fans. The best reason is probably ingress protection (as
in IP44, IP65, IP67, etc). Our fans transport the ambient air through the system, and this air cointains some
dust as well. Over time, this dust can accumulate and cause problems inside the system, especially when combined
with high humidity.
A fanless cooled system (with sufficient IP rating) does not have this problem since all the dust remains outside.
However, this is not a warranty against problems. After all, the dust will settle on the outside of the case
equally well. After a while this dust will form a thermally isolating cover over the case, which will cause less
cooling capacity, increasing temperature, and thus early failure.
The thermal block of the fanless system is easily solved by cleaning, but this is getting less easy if that
system is hidden away in a difficult spot. And then the difference between just cleaning and a preventive overhaul
gets rather small...
References:
Please check the download button above for links to Sunon reliability data, and to a report on fan reliability
from another Sunon customer.
|
|
| |
