What is the accepted limit of power density?

[treez]

Hello,
What is the accepted limit of power density (in W/cubic inch) for a SELV (all voltages inside <42V) SMPS without a fan, but has potting around all the electronics, and has a metal heatsink.?

Is it about 100W/cubic inch?

Ambient temperature 85 degrees C max

(My 230VAC laptop power supply is 65W in 7 cubic inches. (10W/cubic inch))

 

[andre_luis]

Re: Power density of SMPS without fan?

I may be wrong, but presumably the main issue concerns to gradient temperature felt by each internal components, once each one exhibits different dissipated amount of energy.




+++

 

[treez]

Re: Power densityof SMPS without fan?

Yes, and there is an accepted maximum for W/cubic inch, but what it is I am not certain...does any reader know?

 

[FvM]

Re: Power densityof SMPS without fan?

Why should there be an "accepted limit"? Feasible power density depends on efficiency which isn't fixed to a specific limit.

And accepted by whom? A standard?

 

[treez]

Re: Power densityof SMPS without fan?

Yes but there are general guidelines, and "rules of thumb".
-I mean, if your boss tells you to design him an SMPS to power a fuel pump in an aircraft, and tells you that the output power is to be 700W...and then tells you that you have a volume of 5 cubic inches in which to place this SMPS, then surely you are going to tell your boss...."no , that's not possible...at least , not with standard, reasonably cost-effective methods"....am I right?

...how many cubic inches do you tell him you need?

 

[andre_luis]

Re: Power densityof SMPS without fan?

What is the accepted limit of power density (in W/cubic inch)...

Whereas the factor that affects the electronics components is the temperature, the most significant parameter to be considered in order to define a metric for evaluating the thermal condition of a design, actually is the thermal resistance of the surrounding surface of this "volume" on which is confined electronic circuits, and not properly the "power density".

Note that for the same heat amount generated inside enclosure built using different materials, yield different internal maximal temperatures.


+++

 

[treez]

Re: Power densityof SMPS without fan?

You are of course absolutely correct, but managers want to know the "real deal".

Yes, and i'm saying that for a reasonable cost of enclosure material , and potting material, "what is the accepted maximum limit for power density"?

The concept of Watts/cubic inch is indeed a real concept in power electronics. The Chief Engineer at Xyratex, who design server power supplies, asked me the question at interview about what is a reasonable watts/cubic inch for an smps.

..OK then, turning it round, if your boss asks you to do a 700W SMPS, and tells you that you have a volume of 3 cubic inches in which to fit this SMPS, what is your reply to your boss?

I say that above 20W/cubic inch, special expensive cooling methods and components start being needed....so 20W/cubic inch is a good benchmark, do you agree?

 

[FvM]

Re: Power densityof SMPS without fan?

I see that "accepted" in the original post should be translated as "feasible" or "economically producable", depending on the point of view.

Did you review state-of-the-art for compact brick-style power supplies?

 

[treez]

Re: Power densityof SMPS without fan?

Yes the state of the art is things like potting compounds, Special thermal pads for heatsinking fets, external heat sinks with multiple fins, fans, special low_rdsON fets etc etc....
But really, when one is speaking of using generally widely available components and materials, we are talking about 20W/cubic inch, would you agree.....I know that things can be made very small if dunked in liquid nitrogen, but that is an extreme.

I would ask the question, "why is my laptop power supply 7 cubic inches when its only 65W"?

so for a laptop power supply, we can say 20w/cubic inch is a worthy figure?

 

[andre_luis]

Re: Power densityof SMPS without fan?

...i'm saying that for a reasonable cost of enclosure material , and potting material, "what is the accepted maximum limit for power density"?...

ok, in fact, this value can be computed, but in my opinion is strongly associated not only to the composition of the material from which surface it is made, but also to geometry associated to the enclosure.

For small apparatus we can expect a more efficient dissipation rate of generated heat, due no much air fill internal space, and consequently give smaller thermal isolation effect.

So, it seems a little bit hard to define a standardized "accepted maximum limit" due to these dependencies, and different applications can yield different limits.



The parameter R dictates how much internal temperature rises, assuming external as a constant reference , so that the model for R value must to be obtained in order to estimate some value.



+++

 

[treez]

Re: Power densityof SMPS without fan?

I am not requesting an absolute, but rather a rough rule of thumb. If your manager asks you if you can do a 680W smps in 7 cubic inches he is going to want a simple answer.

I believe that that simple answer is as follows "sorry boss, but to achieve that power density would require extremely expensive thermal measures, allow me 20cubic inches and I can do it at a reasonable cost."....would you agree with that?....the people who own and fund engineering companies don't understand highly technical answers.

 

[Velkarn]

Re: Power densityof SMPS without fan?

Hello,
What is the accepted limit of power density (in W/cubic inch) for a SELV (all voltages inside <42V) SMPS without a fan, but has potting around all the electronics, and has a metal heatsink.?

you mean DC/DC with galvanic isolation? for example, 27V to 27V?

 

[treez]

Re: Power densityof SMPS without fan?

no theres no isolation.,,,,its 18 to 32v in, and 34v out , at 680W.
Anyway, is 300W/cubic inche achievable easily with regular cooling techniques?
If not, where do you roughly draw the line?

 

[mtwieg]

Re: Power densityof SMPS without fan?

There are no "regular" cooling techniques, or "regular" operating conditions. Each application and specification should be investigated by itself. I've had clients ask for power supplies in ip67 enclosures, where the achievable power density is of course relatively low. In such cases you just have to do the math to figure out what's achievable.

 

[FlapJack]

Re: Power densityof SMPS without fan?

treez:

Here is a chart of power densities in a document i have called "keynote 2 (3)-(15).pdf"

Also if you remember any more interview questions can you please post them.

 

[FvM]

no theres no isolation.,,,,its 18 to 32v in, and 34v out , at 680W.
Anyway, is 300W/cubic inche achievable easily with regular cooling techniques?
If not, where do you roughly draw the line?

100-300 W/inch³ is the range of available isolated brick modules from major manufacturers, without the heatsink. The module design involves a number of special techniques, in so far it's surely no easy.

 

[treez]

Thanks Flapjack, you confirm that in 2014, the Power Density is at 40W/inch cubed......anybody who tries to do 100W/inch cubed will get their fingers burned, unless they employ very very expensive cooling techniques.

FvM, thankyou, though as you confirm, that's without the heatsink, so I can't count it in my personal investigation, though that of course in no way belittles the data you kindly offer.
I remember using a Fuji power module for a 10KW three phase bridge drive...the Fuji power module was about 10cm by 5cm by 1.5cm....but the heatsink that it needed was 25cm by 13cm by 7cm...and this had two fans blowing over it....there was also a huge high voltage capacitor bank, and other necessary circuitry......the drive itself must have been a good 750 cubic inches....13W/cubic inch.

It is beginning to emerge that in 2014, we are looking at 40W/cubic inch....that is the benchmark, the figure which all marketing and management staff will be ogling and slothering over....do you agree?

 

[Velkarn]

i think 20-40 W/cubic inch. its for module with no external cooling.
for example, i had experience in development of 600W dc-dc with 18 W/cubic, 100V-100V, galvanic isolation, 95% efficiency.
for 40 W/cubic inch you should archive more than 98 % efficiency

 

[FvM]

FvM, thankyou, though as you confirm, that's without the heatsink, so I can't count it in my personal investigation, though that of course in no way belittles the data you kindly offer.
I remember using a Fuji power module for a 10KW three phase bridge drive...the Fuji power module was about 10cm by 5cm by 1.5cm....but the heatsink that it needed was 25cm by 13cm by 7cm...and this had two fans blowing over it....there was also a huge high voltage capacitor bank, and other necessary circuitry......the drive itself must have been a good 750 cubic inches....13W/cubic inch.

I was simply referring to your original post which is obviously talking about a heatsink-cooled supply module. Looking at the flip-flopping in your later posts, I'm no more sure if the question is serious, respectively seriously considered.

There's obviously a range of applications that demands for contact cooled "brick" modules with high power density, e.g. aviation. Just as obviously, a heat sink is needed for this modules, either natural air convection, forced air, or even liquid cooling.

The question title "power density without a fan" is equivocal in this regard. If it means "power/size of a convection cooled power supply", it should state it clearly. Most examples discussed in thread, including some introduced by yourself, include a fan and are off-topic anyway.

The W/cubic inch metric is questionable when talking about convection cooled units. How do you acount for the free space necessary to allow a free air convection? Or do you consider placing it in a fan-cooled instrument case?

 

[treez]

40W/cubic inch is , I take it, with a fan, since server supplies have fans, I know this, I went for the interview at xyratex, and was taken into the room of 10 thousand fan cooled power supplies....all on soak test.
Ive seen convection cooled telecom base station psu/amplifiers of total power = 1KW.....they are enormous hunks of finned aluminium......an alu block of about 70cm, by 50 cm by 25cm, with the pcb on top of it.....