Thermal Management : How to go fanless in a small space?

[daslolo]

I am doing cooling for a pair of FPV goggles, the initial design uses a blower which vibrates and doesn't cool enough.
Without fan, I tested with thermal camera which shows that temperature ramps up from 18C ambient to over 40C inside within 30 minutes then coasts at 41C.
The outer shell gets toasty at above 30C after 2 hours.
The main heatsource is the RDC100 from Raeon which cooks up to 61C when the top case is open, probably in the 70C when it's closed. The RDC is facing forward (away from the face).
Another heat source is sandwitched between the top PCB and the LCOS dispkays, it's covered with a aluminium plate, probably serves as a low profile heatsink.
I am planning on testing Panasonic SSM and PGS made heat pipes to move all that heat sandwitched in tight corners. I need to find a smart way to dissipating this heat PASSIVELY and would like to hear your advise.

Power consumption is 4.9V 1.6A. Des that mean I need to dissipate 8W of heat? How do I find out all the thermal parameters?

If you have telegram, there is more details as I go : https://t.me/EV200D

https://youtu.be/xnckMM-tqUw

[SilverSolder]

Can the outer casing be made of aluminum? - it is a big surface, able to dissipate lots of heat, could have small fins etc. as part of some outrageous design...

[OM222O]

you have to somehow get rid of the generated heat. if you think the fan was useless (not cooling enough) have fun trying to passively cool it (inside a closed enclosure at that!). There is simply no way around the laws of physics so your best bet is a better fan ... or water cooling if you're crazy enough (I've done a few water cooling mods on ridiculous things just for the hell of it  )

[Marco]

Just bodge a big black anodized heatsink on the front and connect the heat generating components to it with heatpipe or PGS graphite sheet.

[daslolo]

Just bodge a big black anodized heatsink on the front and connect the heat generating components to it with heatpipe or PGS graphite sheet.

Tomorrow I get the PGS and I'll test moving the heat outside.
I can cut a hole in the plastic case or even simply a slit and glue the heatsink on the outside of the case, the PGS serving as a sort of ribbon heat transfer.
What's the best heatsink material and fin shape? Or can PGS even be used to make one, like a PGS beard perhaps?

lets see the insides, maybe there isnt space for any mods 

see photos there is loads of room for PGS or maybe even SSM.

Can the outer casing be made of aluminum? - it is a big surface, able to dissipate lots of heat, could have small fins etc. as part of some outrageous design...

It's a great idea, working aluminium is beyond my ability.

[Marco]

Never seen someone make a heatsink from PGS. You could experiment, a long curled up ribbon of PGS almost certainly wouldn't work ... but maybe a flat backbone sheet with a curled ribbon on top glued to the backbone at the valleys might. Even if it works, it's fragile, so it would need a plastic cage around it.

Thermal resistances add up fast with only 70 um thickness, even at 1000 W/mK. I'd double up on PGS and make two connections from heat source to heat sink.

[ebastler]

To give you an idea of what you would need in terms of passive cooling, browse some catalogs of off-the-shelf heatsinks, and look at their shape and thermal resistance. If, say, you want to limit the surface temperature of your gadget's enclosure to 32°C, for environmental temperatures up to 24°C, you need a heatsink with 8°C/8W max. thermal resistance. (When it is convection-cooled, i.e. without forced airflow.)

You will find that such heat sinks are available, but require a pretty large fin surface area. My gut feeling is that you don't want to accommodate that large area on your device, and need to stick with fan-supported cooling.

Any trickery within the device (heat pipes, clever arrangement of components etc.) will only help to transport the internally generated heat to the heatsink efficiently and evenly. You would need to get these things right, but they cannot work around the eventual bottleneck of the outside heatsink -- which will have to get rid of the heat via radiation and convection no matter what. So have a look at the external heatsink needs and what they would mean for the form factor of your device first, to decide whether passive cooling is a realistic option for you.

EDIT: I just realize that the estimate I suggested above will be too conservative. The thermal resistance specified for heatsinks refers to the temperature on their hot side, of course. And you can get away with more than the 8°C temperature raise there. So you don't have to shoot for 1°C/W, but can get away with a higher resistance -- maybe twice that? (Guesstimate.) But I still think that pure passive cooling would imply a larger-than-desirable heatsink.

[TERRA Operative]

I would look at fitting the biggest heat sink I could, with a small fan (any air movement is much better than no air movement) and take to the enclosure with a drill to sprinkle some strategically placed holes around the place for increased airflow.
Design it so the cool air comes roughly in the bottom and out the top (In the sides, out the front, whatevs) to both aid convection and also blow the air away from the users face and see how it goes then.

It looks like you could fit a fair sized heat sink in there (get a bigger one than will fit and file it down) with room for a small blower fan to get the static pressure up to help airflow instead of an axial fan.
I reckon a heat sink that takes up 2/3 of the space in the front, with the fan taking up the remaining 1/3 would do the trick.

[daslolo]

I finished PGS-ing the hell outta this thing. It was more arts and craft than anything else. You can probably tell what was the first patchwork I did 
PGS A-M adhesive is insulating but it's so easy to break in sharp corner that I ended up with some nice fireworks on the bench. Fortunately, and surprisingly, the goggles held on and everything is working fine. So I replaced the 40 and 70 with some of that thick SSM and removed PGS when functions were lost. One display, 70C, 3A draw no DVR were a telltale sign.
SSM is heavy so I tried limit its use.
Overall it's balancing the heat of most of the internal organs quite well but now it's time to dissipate all this heat with heatsinks. Or maybe abandon this idea and dremeling in a 20x20 silent fan (cannot find 20x20 blowers!)

Never seen someone make a heatsink from PGS. You could experiment, a long curled up ribbon of PGS almost certainly wouldn't work ... but maybe a flat backbone sheet with a curled ribbon on top glued to the backbone at the valleys might. Even if it works, it's fragile, so it would need a plastic cage around it.

Thermal resistances add up fast with only 70 um thickness, even at 1000 W/mK. I'd double up on PGS and make two connections from heat source to heat sink.

PGS70 is quite sturdy, the guys at Panasonic told me that best way to bridge two PGS is pressure. I'd have to make or find a clip
You're saying that thermal resistance goes up the thinner the PGS is?
Why is the thermal conductivity higher on 40 than on 70?
How does thermal resistance work?
What do you mean by double up?

You will find that such heat sinks are available, but require a pretty large fin surface area. My gut feeling is that you don't want to accommodate that large area on your device, and need to stick with fan-supported cooling.

I wonder if there is a way to turn the plastic case into a heatsink by covering it with a material that's not too heavy
PGS is sensitive to scratching otherwise I'd use it.

[ebastler]

You will find that such heat sinks are available, but require a pretty large fin surface area. My gut feeling is that you don't want to accommodate that large area on your device, and need to stick with fan-supported cooling.

I wonder if there is a way to turn the plastic case into a heatsink by covering it with a material that's not too heavy
PGS is sensitive to scratching otherwise I'd use it.

No, I am afraid that will not work:

• First, the heat needs to be transported to the surface. A plastic case will have bad heat conductivity, and you will not change that in any way if you treat its outer surface.
• Heat transfer away from the surface, whether by convection or radiation, is proportional to the surface area. You would still need large fins, with at least comparable area to the aluminium heatsinks.
• Radiation of energy works best for a perfectly black surface. A suitable surface layer (PGS, paint...) can probably be slightly better than black-anodized aluminuim, but not a whole lot. I dont see how you could even come close to offsetting the limited conductivity of the plastic.
• Convection cooling does not depend much at all on the surface finish, to my knowledge. Surface area and thermal conductivity (to transport heat to the surface) rule here. So the only way your "plastic heatsink" could compensate for its bad conductivity vs. aluminium is by an even larger surface area.

[ebastler]

@3roomlab, I think I understand your calculations of heat conductivity. But how did you estimate the radiation??

now then, say we have the 3cm radii PGS radiating to air. this becomes a tricky estimate. a thickness of approx 2mm seems to give plausible numbers.
C/W = thickness/(w/mk x area) = 0.002/(0.024*0.00283) = 29.4 C/W
add the disc radii resistance + 2.498 = 31.9C/W

Are you saying that you just modelled conductivity (through a layer of what material?), and postulated some arbitrary layer thickness to make the numbers "fit"? If so, fit to what?

And what about convective transport, which happens in additon to radiation?

I am truly curious about good (i.e. quantitative and simple) ways to estimate radiation and convection. If you have experience with such methods, could you  please eleborate?

[daslolo]

No, I am afraid that will not work:

• First, the heat needs to be transported to the surface. A plastic case will have bad heat conductivity, and you will not change that in any way if you treat its outer surface.
• Heat transfer away from the surface, whether by convection or radiation, is proportional to the surface area. You would still need large fins, with at least comparable area to the aluminium heatsinks.
• Radiation of energy works best for a perfectly black surface. A suitable surface layer (PGS, paint...) can probably be slightly better than black-anodized aluminuim, but not a whole lot. I dont see how you could even come close to offsetting the limited conductivity of the plastic.
• Convection cooling does not depend much at all on the surface finish, to my knowledge. Surface area and thermal conductivity (to transport heat to the surface) rule here. So the only way your "plastic heatsink" could compensate for its bad conductivity vs. aluminium is by an even larger surface area.

transport is via a heatpipe made of a slice of PGS. It seems to work so far.
surface area would be the surface of the goggles, like the idea someone floated earlier about making the case aluminium.
it's not a plastic heatsink but a heatsink layer taped on the plastic body

this is my approximations or rant

the PGS mechanism, from the spec we have 1000W/mK. assume PGS70 is stuck on a heat source 1cm x 1cm
on this end we can approximately calculate the heater to PGS thermal resistance
C/W = thickness/(w/mk x area) = 70e-6/(1000*0.01^2) = 0.0007 C/W (this did not include the glue, which they said requires some pressure)
...
but you can now try and guess things like sun heating, and it no longer dissipates heat properly.

Power draw is 8W. What heat W is produced you think?
Could you run your math with:
1- a layer of PGS70 glued to the surface of the goggles?
2- 14x14x10mm pin copper heatsink pointed 45 degrees down mounted at the bottom where the blower was https://www.amazon.com/dp/B002BWXW6E/ref=psdc_2998409011_t1_B077VM8FJW
3- a 30x30 fan blowing air out of the case
4- a 20x20x6 fan blowing air out of the case
5- arctic silver adhesive instead of the Panasonic stuff
6- use of SSM mazipan looking thing vs PGS A-M adhesive PGS

Thanks

[coppercone2]

Yea no one is gonna like this suggestion: I recommend thin walled heat exchanger with liquid or forced air (in plumbing) cooling. for the electronics and the face contact area. Possibly even two seperate loops so you can keep your face comfortable. And forehead band.

If its possible. Check the design weight when loaded with fluid to make sure its good. See if you can cool the face a bit too.

Complicated and expensive but its on your face and its gonna be fucking annoying. I can imagine now you wanna fly outdoors on a nice summer day and its 30C+ and your standing in the sun on a air field. Fuck that. I would literary go to stand by a damn picnic table with a big ass power pack that powers a peltier pump that connects to the goggles with coolant. And it might keep it lighter.

Why inefficient peltiers? so it can tune the loop to a temperature that you find comfortable on your face. Dual heat exchanger (one is peltier-air/coolant other is goggles/face/transistor-coolant. It would require a powerpack like a fanny pack at the very least and possibly a small backpack but IMO worth it. Gonna pull that thing off at 6-7PM then the fucking gnats are gonna land on your sweaty face and its gonna attract mosquitos and all that nasty shit.

Fucking hate head-lamps too for that reason, when I do out-doors stuff I like being by water (lake) in the evening and any kind of heat, sweat, light... swarms your ass. Unless its something tactical.. go for luxury here. Man I got itchy just thinking about wearing some black plastic goggles in the sun At least a welder helmet is loose on the eyes! 

[Marco]

You're saying that thermal resistance goes up the thinner the PGS is?

Yes, the thermal conductivity goes up slower than the area of the cross section goes down, so the thermal resistance indeed goes up.

Why is the thermal conductivity higher on 40 than on 70?

Dunno, structure is more regular I guess. But thermal resistance is Length/(Area * Conductivity), 1000*70 is larger than 1350*40 so thermal resistance for the 70 um sheet is smaller.

If you do the math that also shows why you really don't just want to make a long sheet into a wavy heatsink ... it has relatively low thermal resistance, but not that low. It's nothing like a real heatpipe.

[ebastler]

Power draw is 8W. What heat W is produced you think?

I'm afraid you are way out of your depth here. 

That whole goggle design looks complex enough, not like a one-man effort. Aren't there others on the team who can get involved with the heat management questions? I don't think this can be designed between just you and a couple of forum members who have seen a few photos.

[IDEngineer]

Your photos look like FatSharks, complete with the RF module and dual SMA's. My son and I have thought a lot about actively cooling the interior of that enclosure. There's an existing fan for keeping the optics defogged but you're correct, the interior gets quite warm. On really hot days my son is aggressive about keeping them unplugged when he's not actively using them (most people just plug them in and leave it, since they don't have a power switch).

My suggestion is an externally mounted fan (just like the defogger on top) on one end with vent slots on the other end, to draw fresh air across the entire interior length of the volume. You could add some baffles if you want to favor cooling one area over another. But beyond this, I don't think there's enough volume internally to embed a fan inside the existing-sized enclosure.

I suppose you could reinvent the entire enclosure as some sort of futuristic heat sink with fins everywhere, but 1) the additional weight will anger customers, and 2) you still have the challenge of coupling the internal heat sources to the heat sink. Remember, your goal isn't to chill the AIR inside, it's to chill the COMPONENTS. If you don't thermally couple the latter to the former, the dead air inside will act as an insulator and you won't get much benefit for all the extra weight on your face.

Here's a wild concept: Do the fluid cooling that others have recommended, but make it a pocket or belt module to keep the additional weight out of the goggles themselves. Then combine the fluid coupling with the battery connection so users still only have a single connector to mess with. Many professional-level pilots (including my son) are already doing something like this, they ditch the headband battery for a standard LiPo pack in their pocket and use a "rabbit in the python" SMPS cable with an XT60 on one end to drop the battery voltage down to what the goggles expect. The result is a cable running from their pocket to the goggles.

EDIT: My son has discussed internal heat dissipation with the head honchos at FatShark, so they are definitely aware that their customers are sensitive to it. One of my son's goggles lost one eye's LCD due to what we believe was excessive heat, and the leadership of FatShark agreed it's a challenge. So you're in good company.

[3roomlab]

I am truly curious about good (i.e. quantitative and simple) ways to estimate radiation and convection. If you have experience with such methods, could you  please eleborate?

edit
I have updated the convection/radiation transfer method below

[Marco]

Here's a wild concept: Do the fluid cooling that others have recommended, but make it a pocket or belt module to keep the additional weight out of the goggles themselves. Then combine the fluid coupling with the battery connection so users still only have a single connector to mess with. Many professional-level pilots (including my son)

A pretty much dead silent slim 80mm PC fan will also move more air than anything you can put on the helmet. Presumably you'd just attach the heatsink and pump fixed with the headset and then connect the battery to the heatsink.

[3roomlab]

(edit)
after several messy LTspice approaches to estimating the heat dissipation
here is the nearly scientific way
convective dissipation added to radiative

[excitedbox]

Trying to think outside the box here. Could you stuff it full of a copper sponge type material (I had a copper foam type air filter once that I think would be perfect for this) and add vents at the top and bottom? The theory is that the heat rises out the top vents causing air current to move through the copper mesh which acts as a heat dissipation surface.

Experiment with different sized vents and amount of copper mesh to optimize the results. A very slow moving small fan should be possible with minor vibration though. If you have use the sponge idea you have a huge surface area and could blow a little bit of air from the side across the front of the PCB and out the other side.

https://image.ec21.com/image/changshalyrun/oimg_GC04161679_CA04161891/Copper-Foam-As-Filter-Element.jpg

[ebastler]

Trying to think outside the box here. Could you stuff it full of a copper sponge type material and add vents at the top and bottom?

That's rather thinking "inside the box", isn't it? 

But I am afraid it will not work. The surface area of your copper sponge will not be comparable to the required approx. 2 K/W heatsink. And at the same time the sponge will inhibit airflow enough to stop most of the "chimney effect" you hope for.

[excitedbox]

That is what I was worried about. Maybe a brilliant pad type copper mesh would let more air through. As others have said there is probably no way around a fan though. Keep in mind a heavy heat sink hanging off the front will act as a lever pulling on your head and strain your neck.

[Marco]

That's why it's interesting to make a heatsink which uses PGS for fins, it could in theory be very light.

[3roomlab]

edit post #19 updated, the nearly condensed method in LTspice
the other earlier simulations are full of errors
which boils down the entire goggle mod to, having a fan or not. a fan is a must have
by adjusting the LTspice sim which includes convection, the range of surface to air resistance is from 25 to 38C/W.cm2.
by quadrupling the key internal surface areas, the internal disspation resistance goes down to 6.6 to 9.3 C/W.cm2.
but still you need a fan

with surface emissitivity over 0.8, the heat radiation and convective loss is nearly similar. they curve up at nearly the same rates. at 20C diff, the loss rate thermal resistance can go down below 5C/W

[daslolo]

the first plot is the rate heat is moving to the last 2mm
the 2nd plot is the air heat up in 1s. the PGS is still trying to play catch up, but the copper has already more or less stabilized the entire sink temperature.
PGS70 is $0.32 per cm2 vs cu 1mm is around $0.05 per cm2. I think based on simulation, the thinnest Cu to beat PGS70 by a good margin is around 0.4mm. at this, the PGS is really for a very niche use.

0.4mm cu base with how long fins?
What's the weight of both solutions?