time and again, my friends in the system design world want to predict the junction temperature of their components in their particular system. so, they grab the data sheet supplied by the part manufacturer and look for that magic parameter theta-ja. once they find it, all they have to do is to multiply it by the (often estimated) power of the part, add the result to the ambient, and presto! the junction temperature is found.
the simplicity of this method belies the fact that theta-ja was never really intended to be used in this manner. jedec, the semiconductor engineering standardization body of the electronic industries alliance, states clearly:
"the intent of theta-ja measurements is solely for a thermal performance comparison of one package to another in a standardized environment. this methodology is not meant to and will not predict the performance of a package in an application-specific environment."
the implication could not be clearer - beware the designer who uses theta-ja for predicting the absolute value of the junction temperature in their particular design!
jedec knows what it is talking about. to illustrate this with greater clarity, let us take a look at some hard numbers. let us say you are a designer who has been using a part from supplier a. the part is packaged in a common package style - a 208 pin pqfp (plastic quad flat pack). with a 1 watt heat dissipation. the supplier has reported a theta-ja of 35.3 c/w using a sophisticated cfd software tool and testing.
you, the designer, work for a demanding employer. in fact, your paymaster is a fast growing networking company. time is critical; your new product needs to be designed within a week. getting it to market could make a world of a difference for the viability of the company, not to mention your considerable stock options. you desperately want that 3000 sq. ft. house in palo alto that you have always dreamed of.
you snap yourself out of your reverie and are tempted to use the theta-ja value to predict the performance of the part in your design. however, you vaguely recall reading an article on coolingzone.com that warned you against such a thing. you also realize that your application board is significantly smaller than the jedec still air test board you had once seen. in addition, your application board has another chip dissipating half a watt on it. reluctantly, you dig out the test sample of the part, place it on your application board, and run the test.
your test yields a junction temperature rise of 47.6 c (yielding an actual junction temperature of 77.6 c). curious, you compare that against the value predicted by using theta-ja. you note that if you had been lazy and decided to use theta-ja as your performance predictor, you would have been off by about 35%.
however, your task isn't quite over yet. supplier b is selling a higher-performance version of the same part. the difference is that supplier b is packaging their part in a new technology - the pbga, and is also claiming a significant reduction in the junction temperature as a result. in fact, supplier b reports the theta-ja of their part to be 21.5 c/w.
you would like to predict the junction temperature this new part would reach in your operating environment to sign off your design. however, samples of the new part are not quite available yet, so testing is out of the question. your manager of course, is banging the door down, demanding an answer yesterday. how do you make her happy?
using theta-ja as a comparative metric is the way out. knowing the theta-ja values of the two parts, you could assume that this ratio is reasonably fixed, to within the first order. a quick calculation yields this ratio to be 1.64. extrapolating from this ratio, the pbga junction temperature rise can be calculated as 47.6/1.64, which gives 29.0 c.
thus in a few minutes you whip out your calculator and calmly tell your manager that you expect the part from supplier b to reach a junction temperature of 29.0 c + 30 c (ambient), i.e., 59 c.
let us see how close you were to the right answer. a cfd simulation for the pbga in your design environment would have yielded a junction temperature rise of 33.2 c - an error (compared to the predicted value of 29 c) of about 12 %. not bad, considering the first cut nature of your prediction!
this is the way theta-ja's are supposed to work. and they do indeed, rather well.
about sarang shidore:
sarang shidore obtained engineering degrees from iit madras (india), texas a & m university (college station), and university of texas (austin). he worked at flomerics inc. in various roles in engineering and product management with a special focus on package-level thermal modeling and analysis, a field in which he has authored several papers and articles.
in addition, he worked for mentor graphics as product marketing manager and for several years as a consultant for various organizations. he is currently a visiting scholar at the lbj school of public affairs at the university of texas focused on energy and climate policy and future strategies.