Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations waross on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Wanted: EE guest authors and bloggers for Electronics Cooling Magazine 1

Status
Not open for further replies.

Tunalover

Mechanical
Mar 28, 2002
1,179
US
As electronic products are becoming smaller, the circuit heat dissipations have not reduced accordingly. EEs are becoming more concerned with thermal management and MEs are becoming more concerned with heat dissipation.

Now I've been an ME for a while doing electronic packaging and my experience has shown that worst-case circuit heat dissipation estimates provided by the EE design engineer are ultra-conservative and lead to over-design and unnecessary expense for thermal management. I've seen first-article product testing show real worst-case heat dissipation can be as little as 1/3 that of the estimate provided at the start of the design! But by then it was always too late to spin the design again to remove the unneeded thermal management hardware!

Guys I'm looking for EE guest authors and/or bloggers with experience in the estimation of most-likely maximum circuit heat dissipations who would be willing to contribute in the Electronics Cooling Magazine (ECM) publication and website blogs. Analog, RF, digital, or mixed, doesn't matter. The theory and general approaches are what we want to know about. I don't think the software used is important. This subject may take more than one blog so us electronic packaging guys and ECM would be really grateful if he/she would be willing to do a series if necessary. In fact one or more feature articles are also a possibility. BTW I'm not on the ECM staff but I'm becoming a first-time contributor soon. If it's too global to think the subject can be addressed for all circuit types at once, then maybe a few guys will be needed: one for digital, one for analog, and one for RF. Hopefully multiple treatments like this can be combined to address mixed-technology circuits.

Granted, the ECM target audience is thermal guys (usually MEs) and they may not be able to follow it all, they can at least share it with their EE colleagues who will surely benefit from it. ECM is non-profit and contributions are not compensated.
Thanks,
Bruce


H. Bruce Jackson
ElectroMechanical Product Development
UMD 1984
UCF 1993
 
I think you're on what is mostly a wild goose chase. Typically, a power supply is designed concurrently with the rest of the circuit, which means that the design has no idea how the circuit will behave in actual usage. This is further made difficult by the fact that most components are not characterized for their modal behavior, because there's no monetary return for the manufacturer to do all that work. A final issue is that while the end result is overdesigned for the baseline conditions, we all know that users tend to abuse baseline conditions, so what you see as overdesign as it leaves the factory might actually be marginal in actual usage.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list
 
Thanks IRStuff, I understand. Since this forum category seems to involve only power supply design my request may be misplaced. I could not find a more general EE forum in eng-tips.com. Am I missing something? Are you aware of other forum topics in here? What about engineering forums that have active EE groups or even forums just for EEs?


H. Bruce Jackson
ElectroMechanical Product Development
UMD 1984
UCF 1993
 
I wasn't suggesting that the forum wasn't right, just that the information that you're hoping might exist probably doesn't. I think that even if it did exist to the degree that you need it, at the start of the design, there's usually insufficient detail in what the software is doing, and how that affects the potential power consumption. In a standard aerospace environment, coding doesn't start until after CDR, which is when the hardware design is finalized. This means that whatever assumptions were, or had to be, made are cast in concrete for the design and there's no time allowable for what would amount to be a re-baselining of the design against a better understanding of the actual power modes in the hardware as influenced by the user and the software.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list
 
I think as electronics become smaller, thermal management becomes increasingly important and more of a concern than a larger design. However, as electronics become more efficient, the heat generated becomes smaller and less of a concern. But, if they are more efficient, the manufacturer's can make them smaller than they could otherwise and they do.

EE's have to design for the worst case scenario's within the specifications of the product. Things like applied voltage tolerances, ambient temperature specifications, etc. all play into what the EE uses when considering thermal management. For example, if the voltage is listed as +- 10% and -10% is applied, the current needed typically goes up as result, producing more heat. If a product specs say it is good for up to 40C or 50C then that's the temperature used in considering thermal management. Couple this with a lower than desired voltage and any number of design parameters that could affect the heat dissipation requirements, it can add up quickly. Depending on the product, there are many other factors that play into heat calculations. Then there is a safety margin to consider. Perhaps, this is where opinions from EE's will vary the most. Obviously, safety applications require a higher safety margin than non-safety or non-critical operations.

I can see, now, how this could lead to a series of articles...
 
And there are so many variables all with tolerances that, when all assumed worst-case at the same time, would lead to unreasonable, ultra-conservative heat estimates. This kind of analysis needs to be done rationally with a statistical approach.


H. Bruce Jackson
ElectroMechanical Product Development
UMD 1984
UCF 1993
 
Unless you know exactly what your software and operating modes are before you design, that's just not going to happen, particularly if it requires any significant amount of engineer time, which it will.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list
 
I understand what you are getting at tunalover. However, as pointed out by IRstuff, time is never on an engineer's side. Even so, we have to design them to operate with the parameters and tolerances identified at project inception. Yes, these parameters can still be changed but the bottom line is, management won't be happy if I tell them only 10% of the products will operate in the high end of the temperature range set by you so I designed it to be okay at 10 degrees less than you want. Higher temperatures will lead to premature failure, thereby reducing its life expectancy. I think this area could be improved but requires the component manufacturer's support and a clear understanding of the product life cycle by managers in order to facilitate and justify more thorough analysis of thermal management solutions. In all but the larger electronics, the time involved doesn't seem to support such an analysis, especially given lack of test data from manufacturer's. High sales volumes may also warrant a more complete thermal analysis.
Interesting subject.
 
Of course I respect both of your opinions on the matter. But I think that statistics are sorely needed to prevent overkill.



H. Bruce Jackson
ElectroMechanical Product Development
UMD 1984
UCF 1993
 
I think the subject needs to be more clearly defined. If were talking about a heat sink that costs $0.30 more than another for 10,000 products, $300 doesn't cover a lot of engineering time. If were talking 1,000,000 units, then it could be worth the extra time with supporting data. Similarly, if were talking a $400 difference in thermal systems but were only making 100 a year, it may not be worth the effort as opposed to making 1,000 of these a year. As an EE, I would love to analyze every detail like this but management don't seem to want us too. I think each design needs to be rationalized if this is something worth considering. Even if I think it does, management may not think so. The biggest hurdle is getting them to approve the time and getting reliable data for such a statistical analysis to be performed. Determining data for the components, without data from the manufacturers, would never be approved given the great deal of time and testing involved. If this data were widely available from most manufacturer's of electrical components, it would be much less time consuming and more feasible.
Don't get me wrong, I think this should be considered and analyzed but only if the economics justify the consideration.
 
Of course the production volume and NRE must be factored into every situation. The higher profit potential and/or the stiffer the customer's requirements the higher the NRE for thermal management. This is a constant in our world. I'm talking about researching, identifying, and advancing methods that could be automated for use as any design progresses. Imagine a PSPICE add-on that could do something like this...


H. Bruce Jackson
ElectroMechanical Product Development
UMD 1984
UCF 1993
 
I think that laptop and portable electronics designers might have a chance at a go, but whatever they come up with, it would only be applicable to a specific point design.

I don't disagree that statistics would be useful, but I guess I'm having trouble getting across that it's highly unlikely that anyone or even any committee of engineers could come up with statistics that have more applicability than the point designs they came from.

To wit, my laptop is cranking its fans full bore at the moment, mainly, I'm guessing because I'm streaming Netflix and Casting it to the widescreen. How is knowledge of what's going on in my laptop going to help the EEs designing our payload for a satellite? Different processors, electronics, peripherals, and functions all lead to differing power consumptions, and whatever functions we could enumerate for a laptop might have nothing whatsoever to do with a satellite payload. And even then, satellite payloads are different and have differing functions, sensors, and electronics. Now, we've usually got one guy doing the detailed design for the electronics, and, by the way, he's also designing the power supply. The power supply is roughly 1/20th of the total effort, so should spend gobs of time trying to optimize 5% of the design effort, when optimizing the remaining 95% could buy way more for the overall project. Any, the designer never needs to worry that some added new functionality or modification of an existing functionality could inadvertently push an optimized power supply beyond its statistically valid design.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list
 
10,000 x $0.30 = $3,000, I believe, not $300.
 
Compositepro,
Are you sure you posted in the right thread? I don't see how your post has any relation to the topic here.


H. Bruce Jackson
ElectroMechanical Product Development
UMD 1984
UCF 1993
 
Even with the missed decimal point, $3k isn't a lot of engineering time. I fight with myself over what's important quite often... coming from a world where even a single breakdown can mean dire consequences, there is a LOT of decision-making into what gets scrutinized further. Had one just yesterday, in fact, on the safe operating area of a FET; my advice to the other engineer was "Design it simply now with the understanding you'll want to verify with the customer what even a single inopportune failure could mean... if it means bad things, plan on doing a full study to know just how much you need to oversize it".

But to the point of others here, the study would only apply for this specific product and circumstances, not products as a whole.

Dan - Owner
URL]
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Top