0
\$\begingroup\$

How is device operating temperature range determined? I supposed that for a specified temperature range all the built in components have to support it. I am in doubt because I saw some commercial product with specified the lowest temperature of -30°C, but implements some IC that has 0 - 70°C operating temperature range (see EDIT). And that is crucial part of the device. How is that possible? Is that certification violation? Is it reliable to use in range that is not covered by all the components?

EDIT: I must add that maybe I made bad assumption. Sometimes, device manufacturers have option to order specific IC variants from the IC manufacturers (like variants with industrial, mil/space grades), that are not initially available. In my particular case, the IC I am mentioned is LM380N that has specified operating temperature range 0 - 70°C in the corresponding datasheet. This IC (audio power amplifier) I found in service manual schematic (not by inspecting hardware) for the device I am referring. However, not sure if some other than commercial variant is available and implemented in the hardware. Anyway, I appreciate each answer.

\$\endgroup\$
1
\$\begingroup\$

The operating temperature range specification is generally determined by testing a component. You generate a list of critical characteristics for the component and then you measure those characteristics while the component is heated or cooled to the specified temperature. If the characteristics are within their specified limits then the component passes the test and you can be confident in using those temperature extremes in a published specification.

However, the manufacturer can make an economic and marketing decision on what temperature range they want to use in their specification. If your target market is video game controllers, why go to the considerable expense of testing your product at -40C, even if you think it will probably work just fine at that temperature?

Now, if you design a system using components with different specified temperature ranges then you can at least be confident that the system will work within the most conservative limits of any component. But it is possible that some of the components will actually work just fine, at least well enough for your system, at temperatures beyond what the component manufacturer specified. If you test your system at the desired temperature range and it meets all of your specifications then you can reasonable claim that it will work over that temperature range.

I should say that like all component testing you must test not just one unit but a reasonably sized sample and perform the proper statistical analysis.

\$\endgroup\$
0
\$\begingroup\$

Schlumberger (oil well logging company) produces components for high down-hole temperatures while measuring the well. I don't know their methods.

I do know the chip industry has at last 3 grades of temperature ranges:

(1) military, usually -55 to +125 (so equipment can sit out in the sun)

(2) industrial, usually -25 to +85

(3) commercial, usually 0C to +55

What sets these ranges for a particular IC? the marketing department initially, then the designer saying "Can't reach that Integral Nonlinearity without more area for the binary weighted capacitors in the ADC.", then all personnel get involved and the Test people get to bin the product at higher cost (if you want the best INL, you'll pay up for the non-recurring-engineering to develop the automatic test equipment needed to measure the INL).

I recall a computer company (Convex? Computer) used liquid nitrogen to chill their CMOS transistors and achieve faster clock rates, some decades ago.

I think the upper end is set by leakages in the devices, or where the standard methods for predicting deltaIout/deltaVin (transconductance) are not valid. Processes I've designed with had models that went at least to +125C; I never asked the Device Modeling/Measuring people how hot they set their hotchucks.

If I saw any sloppy waveforms at +125C, or any param trends that looked like changing from linear to exponential, that part of the circuit was carefully rethought.

\$\endgroup\$

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.