The whining seems to be due to a particular load which I placed on one of the molex peripheral connectors: a new 500 GB Seagate IDE drive. The drive itself is a bit loader than quiet.
I moved the drive over to the other power 'rail' today and the whine all but disappeared.
It may be total load on the rail (which has only one other device on it), or uneven DC power consumption by the Seagate's motor causing harmonics with the PSU.
The carbide idea is to make it more accessible for repair/replacement, and to replace the potentially spillable, gradually degrading coconut oil. Recent visual inspection shows it is still a nice clear golden color.
The carbide powder has one more advantage: because it isn't solid across more than the grain size, it doesn't transmit vibrations well at all, unlike liquids like water/oil (think sonar). Yes, unfortunately coconut oil transmits very well every frequency of sound from the PSU's components. After all, just like water it is an incompressible fluid, which means excellent vibration transfer with minimal energy loss.
Temperature is a big disadvantage of non-convective fill. In the convective fill system (eg. oil), a plume of hot oil rises from the center (PSU heatsinks), spreads across the top and falls down the metal sides, exchanging heat with the metal. Finally, it travels from the periphery of the floor toward the center, beginning to rise as it encounters the warm zone of the PSU componentry.
This ensures that the components surrounding the heat sink are kept well-cooled by receiving the influx of cool oil from the outer floor. Capacitors are in particular a concern, since their lifetime is very closely correlated with operating temperature.
In the non-convective system (solid fill), heat radiates spherically from the source(s). Since the capacitors and other components are very close, they will be exposed to the majority of the thermal energy flux at the high density such closeness implies; this means high temperatures.
My calculations predict far higher non-heatsink component temperatures in the solid-fill system in comparison to convection flow.
Thinking about the differences, I recall once again that the vast majority of high-power transformers use liquid cooling. They use it because even though a solid fill could last longer and be cost competitive, it just cannot move energy from source to sink as efficiently as naturally or forced convection can.
For example, vertical or near-vertical openings (hollow cores, other vacant areas) in heat sources will act as convection pipes in a liquid system, moving heat out of hot interior spaces efficiently. In a dry system, the only heat that can escape from such vacancies is proportional to the angular degree of its opening due to 3-D spherical heat spreading.
Of all substances, water has enormous heat capacity and would be the ideal cooling material. It has only two problems: first, gradual ionization due to dissolution of immersed materials, and second evaporation.
The DC which is present in parts of the circuit is particularly troublesome; DC will cause electrolysis at even the smallest ion levels.
Some thoughts I had on a fix for ionization: why not just saturate the deionized water with sugar? By filling it up to saturation, nothing more would dissolve, including conductive salts. The sugar would thicken the water thus making convection slower, but it would still beat the pants off oil.
Solid-State PSU Cooling: Beyond the Pail!
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee, Devonavar
See fractional crystallization. I believe that a salt would dissolve forming an anti-solute causing precipitation of the sugar. With the varying temperature of the power supply there would certainly be an opportunity for a more soluble substance to dissolve faster than precipitated sugar as the temperature rises after a cooling period where the excess sugar precipitated. Salts would be a more aggressive solute than sugar.By filling it up to saturation, nothing more would dissolve, including conductive salts
I like the powder idea. The container doesn't have to be large, just enough to avoid shorting to its walls. I saw some large (6"x8") heatsinks on ebay that were cheap. Some power supplies have a rather shallow layer of components with the remaining height taken up by the fan. If the supply had this removed and was surrounded by heatsink on all sides then any component would be close to heatsink in some direction.
Water does compress to some degree, which is part of the reason we bleed our car brake pipes from time to time. Brake fluid compresses much less than water...but is hydroscopic (i think is the word) which means it attracts water, so over time the fluid becomes contaminated with water molecules and its performance drops.Isochroma wrote:After all, just like water it is an incompressible fluid, which means excellent vibration transfer with minimal energy loss.
However, I do take your point that water transmits vibrations well
No, actually, the reason why water in the brake fluid necessitates a flush is because it boils and turns into a gas.Myth! wrote:Water does compress to some degree, which is part of the reason we bleed our car brake pipes from time to time. Brake fluid compresses much less than water...but is hydroscopic (i think is the word) which means it attracts water, so over time the fluid becomes contaminated with water molecules and its performance drops.Isochroma wrote:After all, just like water it is an incompressible fluid, which means excellent vibration transfer with minimal energy loss.
However, I do take your point that water transmits vibrations well
Hygroscopic actually.Myth! wrote:Water does compress to some degree, which is part of the reason we bleed our car brake pipes from time to time. Brake fluid compresses much less than water...but is hydroscopic (i think is the word) which means it attracts water, so over time the fluid becomes contaminated with water molecules and its performance drops.Isochroma wrote:After all, just like water it is an incompressible fluid, which means excellent vibration transfer with minimal energy loss.
However, I do take your point that water transmits vibrations well
About water:
Water is for all intents and purposes incompressible. It can be compressed, but it is so slight that it is negligible, which is why 99.9% of equations will treat it as such.
I thought bleeding the brakes usually was done to get air out of your brake lines after some major brake work in which the system didn't remain closed?
Code: Select all
one part in 5×10^7 decrease in volume for each atmosphere increase in pressureI thought bleeding the brakes usually was done to get air out of your brake lines after some major brake work in which the system didn't remain closed?
The coconut-oil PSU is still running (after moving to my new home last summer). It's been on 24/7/365 since being immersed, without a single problem.
Only issue is in summer it's hard to get it to re-solidify even if unplugged for a day.
It still looks exactly like the pictures I posted here before, except that there's just a tiny bit of barely-visible dust on the surface, due to the tape not perfectly covering the top of the pot (especially over the wires).
Now I've got to move again at the end of the month, so will have to unplug it and put it in the basement for a coupla days to get it solid again. Won't fit in the fridge unfortunately.
The only thing I'd change if it was built anew, would be to put a damn quick connector(s) on it so the PSU can be detached from the huge mass of 12AWG wires going to the computer.
Coconut oil is truly the best PSU-cooling solution and is zero-maintenance. It's also much safer to have the PSU immersed because if it fails it won't catch fire, unlike a regular air-cooled one.
Finally, with no fan or other moving parts, and the transistor-area hot-spot almost eliminated due to better thermal mass, the unit should outlast its normal life expectancy by many times. I'm shooting for a lifetime of continuous service.
Until we meet again, here's an interesting PDF:
COCONUT OIL AS AN ALTERNATIVE TO TRANSFORMER OIL
http://www.elect.mrt.ac.lk/Coconut_oil_eru_2001.pdf
Only issue is in summer it's hard to get it to re-solidify even if unplugged for a day.
It still looks exactly like the pictures I posted here before, except that there's just a tiny bit of barely-visible dust on the surface, due to the tape not perfectly covering the top of the pot (especially over the wires).
Now I've got to move again at the end of the month, so will have to unplug it and put it in the basement for a coupla days to get it solid again. Won't fit in the fridge unfortunately.
The only thing I'd change if it was built anew, would be to put a damn quick connector(s) on it so the PSU can be detached from the huge mass of 12AWG wires going to the computer.
Coconut oil is truly the best PSU-cooling solution and is zero-maintenance. It's also much safer to have the PSU immersed because if it fails it won't catch fire, unlike a regular air-cooled one.
Finally, with no fan or other moving parts, and the transistor-area hot-spot almost eliminated due to better thermal mass, the unit should outlast its normal life expectancy by many times. I'm shooting for a lifetime of continuous service.
Until we meet again, here's an interesting PDF:
COCONUT OIL AS AN ALTERNATIVE TO TRANSFORMER OIL
http://www.elect.mrt.ac.lk/Coconut_oil_eru_2001.pdf
There's no reason to do so. Wax is flammable and a hydrocarbon product, while coconut oil is renewable and non-toxic.
Adding any solid such as metal oxides inhibits the natural convection that makes liquid oil such an excellent coolant.
Remember the transformers on power poles - just oil, nothing more.
Sometimes the easiest, cheapest, most natural solutions are indeed the optimal ones, such as this design.
Adding any solid such as metal oxides inhibits the natural convection that makes liquid oil such an excellent coolant.
Remember the transformers on power poles - just oil, nothing more.
Sometimes the easiest, cheapest, most natural solutions are indeed the optimal ones, such as this design.