Log In   |   Sign up

New User Registration

Article / Abstract Submission
Register here
Register
Press Release Submission
Register here
Register
coolingZONE Supplier
Register here
Register

Existing User


            Forgot your password
December 2005
library  >  Application Notes  >  Ake Malhammar

Uncertainties in Thermal Design


discussions about uncertainties in thermal design are predominantly focused on the temperature calculation methods. this article takes a broader view and reveals that there are a handful of other problems which are at least as important.




heat dissipation

 

heat dissipation is rarely specified as a fix value. for some circuits it is given as two values, typical and maximum. for other circuits it is given as a function of a conglomerate of parameters: frequency, voltage, duty cycles, manufacturing variations an additional complication is that software dependencies often are difficult to predict. in essence, unless a realistically measured value is available, one can not be sure.


engineers have a preference for conservative calculations. although this tendency is both sound and understandable it also generates overshoot problems. heat dissipation predictions that are 20% to 30% higher than their actual values are fairly typical. the discrepancies can occasionally even be as high as a factor 2.


it is nevertheless fair to state that there also are many good predictions, particularly for mature technology applications. when working with latest up-to-the-minute gear, there are, however, good reasons to maintain a sceptical attitude.



 

temperature - reliability

 

the ultimate purpose of thermal design is not to push temperatures below a certain critical limit, it is to fulfil the reliability requirements. a temperature that not can be interpreted into some type of reliability measure is therefore useless.

 

one problem in this context is the unfortunate cultural gap between thermal designers and reliability experts. the latter group is very liberal with their temperature definitions. they would not go as far as using the temperature on the moon for their predictions but once they are on the pcb, or even in its vicinity, anything seems to go.

 

it is here also relevant to raise the issue of how component reliability relates to temperature. the answer is, in lots of different ways. some components have a threshold temperature above which their failure intensity increases dramatically. other components follow the ahrrenius function, while others are sensitive to temperature variations.

 

given this large blend of properties, it is evident that it is difficult to predict reliability. an additional dilemma is that it is almost impossible to measure failure intensity, or lifetime, on new systems.

 

one type of reliability can nevertheless be measured, safe function at worst case conditions. to predict it is another matter. all thermal designers have struggled with the impossible task to interpret the manufacturers maximum ambient air temperature criteria into something that makes sense on a pcb. things are however improving. some manufacturers have lately begun to specify a maximum operating case temperature. this temperature is contrary to the air temperature both measurable and predictable and, in addition, it makes sense.

 

the final test whether a design fulfils the safe function requirement or not, is made in a climatic chamber. the author’s experience is that if the temperature limits given by reliability expertise are respected, this test is passed with a large margin, well above 10 c for most cases. a margin of this size is of coarse cotton for the nerves but it also represents a potential cooling capacity increase of the order 20% - 50%.



 

 

component models

 

all thermal designers know that the traditional junction-to-ambient model has done its time and that there is an urgent need for a better model. how such a model should be defined has been a glowing issue for the last ten years. the fundamental problem is that the two desired qualities, simplicity and accuracy, are contradictory. one always has to make compromises. the newly created junction-to-board jedec standard is for example a compromise that favours simplicity.

 

lack of data is the main uncertainty problem for component models. this situation will hopefully improve as more and more manufactures accept the new jedec standard. there are nevertheless a couple of other problems that need to be addressed. the following discussion is general and concerns all models that have a defined interface with the pcb.

 

the main heat spreaders in multi-layer pcbs are the inner ground and voltage layers. the heat that is pushed out on the pcb must therefore not only overcome the thermal resistances inside the component but also the pad-to-ground layer resistance. the latter can in many cases be of the same order as the former.

 

what complicates things is that the copper pattern below a component can be shaped in so many different ways. another complication is that the distance between the surface and the first ground layer varies. it is therefore essential to know if the pad-to-ground layer resistance is included in the model definition or not. if it is, model data is only correct if the arrangement below the component is the same as for the test case.

 

another uncertainly problem has to do with the way the model treats the case-to-air heat. a component model has to purposes:

 

1. predict the chip temperature when the board and the air temperatures are known.

 

2. help to predict the board temperature.

 

the latter is typical for pcb temperature calculations. if the case-to-air heat is incorrect so will be the board temperature. there is a wide spread belief that if a model is correct for purpose 1 it also is correct for purpose 2. this is not the case. star models, (all thermal resistances originate from a common point), have for example an inherited case-to-air heat problem. this deficiency is not significant if the packaging density is low. for bad cases it can however result in errors of the order 10% - 20%.



 

inlet disturbances

 

disturbances in the inlets to the air gap between two pcbs can have a significant impact on the cooling. heat transfer coefficient increases up to 60% have been observed but 30% is probably a more typical value.

 

the only certain way to map this phenomenon is to make measurements. detailed cfd analyses, in which all obstacles in the air inlet, (pcb guides, emc grills,.), are conscientiously modelled, might have some success. large-scale cfd models will not do the job.


the cooling gains caused by the inlet disturbances do not come free. they are associated with a considerable pressure drop. if someone decides to reduce the pressure drop in a future redesign, it has to be made with caution. a lower pressure drop will result in a higher air velocity, which is the positive impact but it will also decrease the gains caused by the inlet disturbances, which is the negative impact. on a general level it is very difficult to predict what the all over result will be. it all depends on the circumstances.


in conclusion. it is tempting to fully use the cooling gains caused by inlet disturbances but there are risks involved.



 

air flow

 

the difficulties to manage airflow in electronic equipment are well known. fluid dynamics software tools are helpful but they do not take care of everything.

 

air leaks are a frequent problem. modern electronics is often designed as a collection of modules that can be “snapped” together as lego pieces. this will inevitably create non-perfect joints and therefore also air leaks. it is difficult to estimate the magnitudes of those leaks. they can sometimes be modest even though the visible air gap is quite wide, some times it is the other way around. a minimum leak assumption of 10% of the total flow is recommendable.

 

the air leak problem is particularly accentuated for airflow entering high performance heat sinks. these heat sinks have a high pressure drop and the air subsequently takes all possible chances to bypass. air leaks of up to 30% are probably not exceptional.

 

dust, particularly clothe fibres, is another problem. if there are persons in the vicinity of a fan-cooled equipment, fibres will inevitably be sucked in and jam the flow paths. the problem is particularly severe if there are emc grills in the flow paths.

 

if the air intake is close to the floor, it will in addition operate as a vacuum cleaner. airflow decreases of the order 25% can be a fact within a few months. the only difference between using an air filter or not, is that with a filter one knows were the dust can be found.

 

re-circulation of warm exhaust air into the inlet air is sometimes a problem for individual units. in rooms lined up with equipment, it can however be devastating, particularly if there are units that blow out their exhaust air horizontally.



 

production testing

 

the only correct way to make heat tests on pcbs is to place them in their proper enclosure and heat up the environment. this is often not possible in a production line. a variety of other methods are therefor used.

 

a much-practiced method is to place the pcbs in a wind tunnel-like enclosure with hot air circulation. since the same apparatus often is used for all kinds of pcbs, regardless if they are intended for natural convection or not, it is evident that the method is coarse. if one has made a design that has a 5 c margin for the worst temperature case, chances are that problems will appear at testing, even though the design is next to perfect.



 

specifications

 

specifications should ideally be crystal clear and easily understood. this is rarely the case. there are always interpretation problems. the thermal uncertainties are predominantly found on the combination level. which worst cases should be combined with which, and which should not? instructions are frequently lacking.

 

the altitude issue is a typical example. at 3000 m altitude the air has lost about 20% of its cooling capacity. if it is required that an equipment must be fully functional at that altitude, a strict interpretation would imply that it also must be combined with the worst temperature requirement.

 

the consequence is of course that additional margins must be designed into the cooling system. some specifications have spotted and cleared this uncertainty problem, others have not.

 

there are numerous other combination possibilities. most of them do not make much sense but some do.



 

calculation

 

temperature prediction methods have as a result of assiduous efforts evolved considerably in the last ten years. the calculation error has gradually decreased. when the best methods that currently are available are applied, the error is typically on the 10% level.

 

the temperature calculation error is presently not the dominant uncertainty in thermal design!


about ake malhammer



 

 

 

 

 

 

ake obtained his master of science degree in 1970 at kth, (royal school of technology), stockholm. he then continued his studies and financed them with various heat transfer-engineering activities such as deep freezing of hamburgers, nuclear power plant cooling and teaching. his ph.d. degree was awarded in 1986 with a thesis about frost growth on finned surfaces. since that year and until december 2000 he was employed at ericsson as a heat transfer expert. currently he is establishing himself as an independent consultant.

 

having one foot in the university world and the other in the industry, ake has dedicated himself to applying heat transfer theory to the requirements of the electronic industry. he has developed and considerably contributed to several front-end design methods, he holds several patents and he is regularly lecturing thermal design for electronics.



to read ake's web site for more thermal information and software tools he has developed, please visit http://akemalhammar.fr/ - see more at: https://www.coolingzone.com/library.php?read=534#sthash.y3rcxrow.dpuf

to read ake's website for more thermal information and software tools he has developed, visit http://akemalhammar.fr/.

Choose category and click GO to search for thermal solutions

 
 

Subscribe to Qpedia

a subscription to qpedia monthly thermal magazine from the media partner advanced thermal solutions, inc. (ats)  will give you the most comprehensive and up-to-date source of information about the thermal management of electronics

subscribe

Submit Article

if you have a technical article, and would like it to be published on coolingzone
please send your article in word format to [email protected] or upload it here

Subscribe to coolingZONE

Submit Press Release

if you have a press release and would like it to be published on coolingzone please upload your pr  here

Member Login

Supplier's Directory

Search coolingZONE's Supplier Directory
GO
become a coolingzone supplier

list your company in the coolingzone supplier directory

suppliers log in

Media Partner, Qpedia

qpedia_158_120






Heat Transfer Calculators