Liquid-Filled Heat Pipes, Horizontal vs. Vertical
- - -
i am writing to ask for your help making sure that i am using the heat-pipe technology correctly.
in my work as an engineer, when we used the term "heat pipe" it was always for any construction that acted as a conduit for heat energy, from the heat source, to a surface closer to the "cold-plate".
a lot of military avionics, electronics for airplanes, is cooled by conduction - the chip is bonded to a circuit board, the circuit board has 64 copper vias, below the circuit board is an aluminum housing which is clamped to an aluminum frame which has cold fluid pumped through it.
so then, a lot of the doo-dads and gizmachies (those were professional terms used by us engineers ... paid for by your tax dollars !!) that we designed to route heat from a specific chip to the cold plate, we called "heat pipes".
since we were using CAD tools to do the design & analysis, our hardware was HP UNIX workstations and Dell PC's, mostly dual Xeon & ATI-Fire GL machines.
now, with PC's, like this
http://xbitlabs.com/articles/coolers/display/noctua-coolers_3.html
Noctua heat sink, i believe that the heat pipes in it are filled with some liquid. for example, ammonia, since it boils at a lower temperature. this part is my imagination - the bubbles are lighter than the surrounding liquid so they "go up". when they get to a cooler part of the envelope to which they are restricted, they experience another phase change back to liquid.
in each of those phase changes, from liquid to bubble and then from bubble to liquid, there is a latent heat of evaporation & condensation that is very effective in "moving heat".
designed effectively, it has a low thermal resistance. this is the beginning of the "Holy Grail" when you're working as a design engineer for a defense contractor - we would sit in presentations and calculate the junction temperature. if we could show that we could show that the junction temperature was "within spec", we got brownie points, which were redeemable when pay raises were being handed out.
OK, so now that Wusy at Tom's Hardware has pointed out that the Zalman 9500 is way out of spec, doesn't meet spec, and now that i've found this
http://www.frostytech.com/articleview.cfm?articleid=1923&page=4
i'm looking at these Noctua heatsinks, especially the NH-12.
.14 degrees C per watt in free air, that is its thermal resistance. don't know what fan that was measured with.
OK but there's a big assumption in all of this - the ORIENTATION of the heat sink.
i haven't taken apart the heatsinks like the Zalman and the Thermal right and the Noctua. from the way they're pinched off, and from a lot of the posts about them, i get the conclusion that they are liquid filled and that the movement of the liquid inside the heatpipe genuinely helps heat transfer, whether or not the liquid turns into fizzy little bubbles in the process.
is the main part of this heat sink supposed to be vertical ?
this would mean that, when they are placed in a conventional tower workstation, the end of the heat pipe away from the CPU, is horizontal. so, how does the cooled-off fluid get returned to the CPU area of the heat pipe for another dose of heat ? if the heat pipe is horizontal, and it's a skinny heat pipe, it's not like pouring cold vodka into hot chocolate, in which case it mixes real quick.
so i'm wondering, since most of us computer folks are using tower cases of one kind or another, doesn't that leave these big beautiful heat sinks operating in a sub-optimal orientation, with horizontal heat pipes ?
have any of you experimented with your systems, like for example measuring CPU temperatures with the case mounted horizontally and vertically, to see if there's a difference ?
thanks for some good technical feedback. vodka-hot chocolate recipes are appreciated too.
- - -
i am writing to ask for your help making sure that i am using the heat-pipe technology correctly.
in my work as an engineer, when we used the term "heat pipe" it was always for any construction that acted as a conduit for heat energy, from the heat source, to a surface closer to the "cold-plate".
a lot of military avionics, electronics for airplanes, is cooled by conduction - the chip is bonded to a circuit board, the circuit board has 64 copper vias, below the circuit board is an aluminum housing which is clamped to an aluminum frame which has cold fluid pumped through it.
so then, a lot of the doo-dads and gizmachies (those were professional terms used by us engineers ... paid for by your tax dollars !!) that we designed to route heat from a specific chip to the cold plate, we called "heat pipes".
since we were using CAD tools to do the design & analysis, our hardware was HP UNIX workstations and Dell PC's, mostly dual Xeon & ATI-Fire GL machines.
now, with PC's, like this
http://xbitlabs.com/articles/coolers/display/noctua-coolers_3.html
Noctua heat sink, i believe that the heat pipes in it are filled with some liquid. for example, ammonia, since it boils at a lower temperature. this part is my imagination - the bubbles are lighter than the surrounding liquid so they "go up". when they get to a cooler part of the envelope to which they are restricted, they experience another phase change back to liquid.
in each of those phase changes, from liquid to bubble and then from bubble to liquid, there is a latent heat of evaporation & condensation that is very effective in "moving heat".
designed effectively, it has a low thermal resistance. this is the beginning of the "Holy Grail" when you're working as a design engineer for a defense contractor - we would sit in presentations and calculate the junction temperature. if we could show that we could show that the junction temperature was "within spec", we got brownie points, which were redeemable when pay raises were being handed out.
OK, so now that Wusy at Tom's Hardware has pointed out that the Zalman 9500 is way out of spec, doesn't meet spec, and now that i've found this
http://www.frostytech.com/articleview.cfm?articleid=1923&page=4
i'm looking at these Noctua heatsinks, especially the NH-12.
.14 degrees C per watt in free air, that is its thermal resistance. don't know what fan that was measured with.
OK but there's a big assumption in all of this - the ORIENTATION of the heat sink.
i haven't taken apart the heatsinks like the Zalman and the Thermal right and the Noctua. from the way they're pinched off, and from a lot of the posts about them, i get the conclusion that they are liquid filled and that the movement of the liquid inside the heatpipe genuinely helps heat transfer, whether or not the liquid turns into fizzy little bubbles in the process.
is the main part of this heat sink supposed to be vertical ?
this would mean that, when they are placed in a conventional tower workstation, the end of the heat pipe away from the CPU, is horizontal. so, how does the cooled-off fluid get returned to the CPU area of the heat pipe for another dose of heat ? if the heat pipe is horizontal, and it's a skinny heat pipe, it's not like pouring cold vodka into hot chocolate, in which case it mixes real quick.
so i'm wondering, since most of us computer folks are using tower cases of one kind or another, doesn't that leave these big beautiful heat sinks operating in a sub-optimal orientation, with horizontal heat pipes ?
have any of you experimented with your systems, like for example measuring CPU temperatures with the case mounted horizontally and vertically, to see if there's a difference ?
thanks for some good technical feedback. vodka-hot chocolate recipes are appreciated too.