Once again, the caller was confused about a section of one of our data sheets. The conversation started something like, "what is with these three temperatures? You give temperatures for specification, operation, and storage. Why so confusing and why do I care?"
First, lets tackle the easy part. You care because if these temperatures are exceeded the performance and even the reliability of the system could be compromised. Now I had his attention.
Now, the specification temperature should be straightforward. Over that range of temperature the part will perform within the minimums and maximums of the data sheet. When within the operating temperature range the part will still function but one or more parameter may be outside of the performance limits. An operational amplifier will still act like an operational amplifier. For many of the devices from Texas Instruments the typical performance curves are given over the operating temperature range. This assures the designer that there are no wild variations in the performance curves just outside the specification temperature range.
The storage temperature range is the extreme temperature limits where the devices will not degrade even after extended storage. At these extreme temperatures the part is not powered. Over the years, plastics have been improved so storage and operating temperature ranges have been widened.
"OK, but how do I know the temperature of the junction?" was the next logical question.
Here things do get a bit tricky. Lets start with the assumption that the junction is at ambient temperature. But, we know that the quiescent current could cause some self-heating. How much of that stays in the die and how much escapes to the ambient environment?
Here I needed to introduce the thermal resistance parameter (theta J-A for the junction to ambient resistance). This specification predicts the temperature rise for a given power dissipation in the die. It is different for each package style. The part in question here was a dual in the SO-8 package. The theta J-A for this package is 150 degrees C/W. Operating with plus-minus 15V supplies and at a quiescent current of 4mA per amplifier this amounts to 240mW of power being generated in the die.
Power = 30V * 4mA * 2(dual) = 240mW
Considering the theta J-A of the package this will result in a temperature rise of 36 degrees C over ambient.
Junction Temp Rise = 240mW * 150C/W = 36C
This looks like a well-defined situation, however, there are some factors that can be trouble. The thermal resistance number is approximate for any given package. The geometry of the traces on the printed circuit card is very significant. For some packages the lead frame is the major heat transfer mechanism. The question remains, "what is ambient?" This can be influenced by the topology of surrounding components, airflow and heat generating neighbors. The existence of a shield plate can change the heat transfer scheme.
With all of these variables we still have not looked at the heat caused in the device by current flow into the load. That is enough for another day.
I hope this little discussion saved this engineer from serious headaches in the future.
Thanks for readingComments invited, Please.
And be sure to copy, you know, firstname.lastname@example.org