Is this a dumb idea of mine for greater cooling?

Okay, I just wanted to run an idea past others, in case I'm being
dumb... which is entirely possible.
So one of my Pi 3's needs to be silent due to location, but it's current
copper heat sink for the CPU just ain't up to the task of keeping it
from heating up too much.
But the current heat sink is pretty tightly bonded to the CPU due to
some unusually aggressive thermal adhesive tape. I know I could start
slopping alcohol about, but that would require disassembly of the case
to avoid cosmetic damage to other elements, and for reason's I won't
bore you with, that's not a task I wish to undertake, if I can avoid it.
The one ease of access consideration this case does have, is a nice big
hole for air flow, right above the CPU. And that leads me to my possibly
dumb idea: can I aid heat dissipation by simply stick another, better
heat sink right on top of the existing one with it's flat topped short
fins? Sure it'll only contact about 40% of the existing heat sink via
the flat tops of it's stubby little fins, but won't it be a decent
increase in heat dissipation, nonetheless, especially if I use a tall
second heat sink that pokes out the top of the cooling hole?
Or is there some daft flaw in my thinking?
And if this would be sensible, how to you reckon I ought to affix the
second heat sink? I could just use more thermal tape, but I don't want
dust to sick and clog around the gaps between the fins of the existing
heat sink and the sticky bottom of the top heat sink. At least I assume
that'd happen. Goodness knows dust gets everywhere else in my home...
heh.
TIA :-)
--
If you're not part of the solution, you're part of the precipitate.
Reply to
Jamie Kahn Genet
Loading thread data ...
A heat sink is a thermal resistor. It improves heat removal because it provides better contact with the air, lowering the that thermal resistance. More airflow will greatly improve the thermal resistance of the heat sink to air interface and so the total thermal resistance.
Adding a heat sink on top of the existing heat sink will add more thermal resistance to the heat sink path, possibly significantly because of the smaller interface area of the heat sink to heat sink contact. The question is whether the increased resistance of the second heat sink will be offset by better airflow over the protruding heat sink. This is possible if you have the case oriented so the convection airflow is optimized or there is a natural breeze from some other source.
I doubt you will need to use alcohol to remove the existing heat sink, but I expect case disassembly is required in any event, so if that is a problem your idea might be the best. I'm not sure what is the best way to connect the two heat sinks. Possibly a thin piece of thermal pad. Thinner is better as all materials are thermal resistors and thinner is a lower value of resistance.
--

Rick C
Reply to
rickman
What's "too much"? How hot does it get? What environment is it running in that requires a heatsink at all?
A heatsink in a case may not work well, depending on the overall thermal design. The case will have thermal resistance itself, and beyond a certain point, better coupling the heat source to the air inside the case will not improve things.
You can use a case *as* a heatsink with the right design - aluminium case with a thermal path (copper, aluminium, or a heat pipe) carrying the heat to the case. This would normally be complete overkill for a Pi though.
--
Andrew Gabriel 
[email address is not usable -- followup in the newsgroup]
Reply to
Andrew Gabriel
Case temps of 100C are acceptable for some silicon chips
A sealed case is almost bound to result in semiconductor failure. I did a design once in a 19" rack case fully enclosed.
It took 4 hours for the temperature inside to get above 100C, but it did.
if you must have a sealed case you must provide a low thermal resistance path to the case itself and use that as a heatsink.
Exactly.
--
If I had all the money I've spent on drink... 
..I'd spend it on drink. 
 Click to see the full signature
Reply to
The Natural Philosopher
OK, so thats a good outflow hole for hot air, but how big are the hole(s) for cold air inflow and where are they located on the case? For optimum flow they should be as low as possible and IIRC, should be bigger than the hot air outlet. Note that, if there are several inflow holes, their effective area will be less than that of a single hole with the same area due to flow turbulence round the edges of the holes disturbing smooth flow through them.
IOW, if the inflow hole(s) are smaller than the outflow hole or badly placed, making more inflow holes in the case or enlarging the existing holes may solve the problem.
--
martin@   | Martin Gregorie 
gregorie. | Essex, UK 
 Click to see the full signature
Reply to
Martin Gregorie
electronics in the 60's if possible, to increase lifespan.
Have you guys used the Pi 3? It isn't like prior versions that generally didn't need additional cooling. It's Broadcom BCM2837 SOC gets hot very easily and the chip speed throttles to cope.
--
If you're not part of the solution, you're part of the precipitate.
Reply to
Jamie Kahn Genet
Yeah, the vents seem adequate. Though I can't say it looks optimum to me. I've yet to see a case that really tries to optimise airflow over the entire board, top and underside.
all my systems during their downtime, and I hate to see wasted processing power due to excess heat.
--
If you're not part of the solution, you're part of the precipitate.
Reply to
Jamie Kahn Genet
I didn't mean that - just that inflow needs to be close to the top of the PCB as possible, assuming its horizontal, to get some (minimal) chimney effect past the heat sink. A case with inlets just under the top won't work well without a fan.
You can almost guarantee that a case designed for use with a fan isn't going to work well without one.
--
martin@   | Martin Gregorie 
gregorie. | Essex, UK 
 Click to see the full signature
Reply to
Martin Gregorie
[snip]
I have an RPi2B running BOINC 24/7 and the cpu temperature stays around
underclock your RPi 3 to match the speed of the 2B (900MHz).
Ray
Reply to
RRansil
The Pi2 is very different to a Pi3. The average no temperatures for mine
are:-
Pi 1B 256MB 50.7C No heatsink Pi 1B 512B 47.0C Heatsink Pi 2B 44.3C No heatsink Pi 3B 55.1C Heatsink
As you can see the Pi2 runs the coolest of all with no heatsink, and the
Pi3 runs hottest even with a heatsink.
---druck
Reply to
druck
How about fitting 6 inches of chimney directly above the existing heatsink and making sure that there is a good airflow over the board? That way you can use the chimney effect to accelerate the airflow to give better cooling.
Reply to
mick
It completely defeats the point of using something as small as the Raspberry Pi. If heat is that much a problem, use something else with proper passive or active cooling.
---druck
Reply to
druck
It also won't work. I tried measuring air flow in a 4 inch tube suspended over a 100 watt lightbulb. I could detect no airflow and the air at top of the chimney was the same as the room temperature. I'm sure the 2 watts of the rPi won't move much air either.
--

Rick C
Reply to
rickman
Pretend it's a ZX80 and put a carton of cold milk from the fridge on top of it.
Owain
Reply to
spuorgelgoog

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.