3
\$\begingroup\$

It is usually advised to keep the temperatures low for every component of a PC. Not only there is a fan (or similar system) for the CPU or GPU, but there are optional fans for RAM, or hard disks.

While I understand the need of a CPU/GPU fan, since those units can probably melt or burn if not kept at normal temperatures, I'm unsure about the reason to, for example:

  • decrease the temperature of a hard disk from 104°F (40°C) to 86°F (30°C),
  • decrease the temperature of RAM from 113°F (45°C) to 95°F (35°C).

I've read in several places that lower temperatures increase the lifespan of hardware. But what are the reasons behind? How a 15-20°F decrease influences the hardware when we talk about the temperatures which are not as high as would be the temperature of a quad-core CPU used at 95% with no fan? What parts of the hardware are affected? Well, in short, what's happening?

\$\endgroup\$
  • \$\begingroup\$ "The figure shows that failures do not increase when the average temperature increases. In fact, there is a clear trend showing that lower temperatures are associated with higher failure rates. Only at very high temperatures is there a slight reversal of this trend." labs.google.com/papers/disk_failures.pdf \$\endgroup\$ – endolith Jul 9 '11 at 16:55
3
\$\begingroup\$

Melting and burning are just extreme extensions of the process. Generally, changes happen faster with increasing temperature and material properties generally alter. As devices are usually designed to perform best "as supplied" changes are usually bad. For example, a lubricant will be designed to work best at the temperatures expected to be encountered. Increase the temperature above the expected range or above the range at which it was feasible to optimise performance and eg lower viscosity and increased oxidation rate will decrease lifetime. (Lowering temperature markedly may also adversely affect things like viscosity but generally for different reasons.)

  • Chemical reaction rates increase with temperature, including oxidation from atmospheric Oxygen, impurity reactions etc.
  • Plastics "soften". Materials' mechanical properties generally degrade as -they approach melting point or glass transition temperatures or ...
  • Diffusion rates increase. Both at eg metal-plastic sealing boundaries and in materials that rely on differences at boundaries.
  • Components that rely on internal liquid content dry out. Notable examples are electrolytic capacitors and batteries.
  • Lubricant properties change (eg motor bearings).
  • Defects in materials are liable to form faster and propagate more rapidly and further. (This is a major factor in eg long term LED lifetimes).

In some cases, attempts to address temperature issues can be counterproductive. eg how long will a fan last for a given bearing technology and cost etc.

"Collateral damage" can occur eg aluminum electrolytic capacitors are sensitive to heat degradation, largely through electrolyte dry out. Placing them near a hot device may cause early capacitor failure even though this could not be predicted from analysis of the circuit.

Some products are commonly used without regard to their time & temperature susceptibility. "Hot melt" adhesives are often used for component fastening but will lose their grip on many surfaces in weeks to months.

Temperature effects can be regenerative (positive feedback). eg Iron powder cored toroids can provide superior performance for power conversion inductors in the 10's of kHz to low MHz range. The cores are formed with an organic binder which is temperature affected, giving it a design lifetime. As the binder ages core losses increase leading to increased temperature rise leading to more rapid degradation leading to .... An iron powder core handling significant power can go from nearly 100% OK to a smoking ruin in a very short period relative to total design life as it hits the end of its core design life. Similarly, electrolytic capacitors handling high ripple currents will have increasing ESR as electrolyte dryout occurs, leading to increased losses and increased temperatures leading to ... .

\$\endgroup\$
  • 1
    \$\begingroup\$ Another important factor is the silicon semiconductor material usually stops being a semiconductor around 150C. This tends to make transistors behave rather undersirably. \$\endgroup\$ – Olin Lathrop Jul 9 '11 at 11:56
  • \$\begingroup\$ @Olin Lathrop: Interestingly, bipolar junction transistors and MOSfets have opposite behavior characteristics as they get hot (but not so hot as to be destroyed). BJT's tend to conduct electricity more easily, while MOSfets tend to exhibit increasing resistance. MOSfets tend to fare better in inherently-voltage-limited circuits, since the increasing resistance reduces current and therefore power; BJT's fare better in inherently-current-limited applications. \$\endgroup\$ – supercat Jul 10 '11 at 18:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.