Yeah that's what I thought. The system had exposure to the air in a capped pvc pipe used as the reservoir but no additives were used. I might have had some contamination, but most likely it was the dissimilar metals. The cooling loop also ran very hot due to low flow volume, so that didn't help I'm sure. I had always used nylon fittings before and never had a problem with buildup.
Magura said:
I think most of you guys arguing with Magura totally missed his point here. He wasn't arguing that water cooling is a bad idea. He was just trying to point out that it water as a heat transfer mechanism isn't nearly as efficient as copper or aluminum.
And he's completely right as well:
Thermal Conductivity Table
The thermal conductivity of water is terrible compared to aluminum. It takes it a LONG time to absorb and dissipate heat. You have to compensate for this by dissipating the heat over a much larger area either by having a large reservoir or a large radiator and constantly pumping cool water through your hot devices.
The main advantage of water cooling is the ability to move the heat from one area where it might be difficult to vent the heat (a cylinder head, the middle of your motherboard, etc) to somewhere where it would be easier to dissipate (your radiator, the back of your computer case, etc). There's no argument about that.
However, mathematically, you will need more of either radiation area OR forced air cooling than you would with just a straight forward heatsink for using water cooling.
--
Danny
Danny,
You haven't thought it through. There are several holes in your argument. (Amongst other things, you might want to actually take a look at the chart you linked to.)
I doubt that anyone would argue that a metal--any metal--is a better conductor of heat than air. The only thing air has going for it is that it's free and readily available. Note that it's listed right down at the bottom of the chart you linked to, next to wood. Spend a moment pondering the concept of a wood-cooled circuit. Not a pretty sight.
So the question is: Is a metal/air or metal/water interface more efficient? The answer is water, by a country mile. Water removes heat from the heatsink--which can be much, much smaller, physically, than an air-cooled heatsink--better than even forced air.
Except for the power supply components, my Aleph 2s would easily fit into a 1U rack chassis. I dare anyone to attempt that with any variation of air cooling; it would take a tornado's worth of forced air to cool the thing. Hardly something you want in your listening room. My water-cooled system is efficient enough that I use TO-220 output devices (cheaper and smaller) and they run about 105 degrees F, if I recall correctly. I can keep my finger on them indefinitely and be comfortable. And that's in spite of TO-220s having much smaller back plates than TO-247s or TO-264s.
As for the heat exchanger--the one I use is faaaaar larger than I actually need. I could cut it in half, if need be, and still run much cooler than with air-cooled devices. The total bulk of one side of the heat exchanger is about 15 inches square by about 2 inches thick. The volume is roughly the size of one monoblock chassis of a single Aleph 2 pair. I don't need a fan on it, so that part is passively air-cooled. So what gives? Why isn't it huge? The heat exchanger is more efficient at moving heat from the water to the air. What everyone seems to miss is that the heatsinks used in amplifiers aren't nearly as good at transferring heat as you think they are. Compromises are made in order to have flat planes for the devices to mount to, more flat planes to bolt to the rest of the chassis, and still more flat planes so you can stack herds of the heatsinks together to arrive at any arbitrary amount of heat transfer you might need. They also need to be physically robust and look nifty. The heat exchanger from a heat pump (in my case) has no such limitations. It's designed to move heat to air. Period.
About the only thing you can say thermodynamically is that the water pump generates some heat, but it's a trivial amount and is taken care of by the pump itself; it doesn't add to the heat within the cooling system.
Perhaps this is just one of those things you have to see to believe.
Grey
You haven't thought it through. There are several holes in your argument. (Amongst other things, you might want to actually take a look at the chart you linked to.)
I doubt that anyone would argue that a metal--any metal--is a better conductor of heat than air. The only thing air has going for it is that it's free and readily available. Note that it's listed right down at the bottom of the chart you linked to, next to wood. Spend a moment pondering the concept of a wood-cooled circuit. Not a pretty sight.
So the question is: Is a metal/air or metal/water interface more efficient? The answer is water, by a country mile. Water removes heat from the heatsink--which can be much, much smaller, physically, than an air-cooled heatsink--better than even forced air.
Except for the power supply components, my Aleph 2s would easily fit into a 1U rack chassis. I dare anyone to attempt that with any variation of air cooling; it would take a tornado's worth of forced air to cool the thing. Hardly something you want in your listening room. My water-cooled system is efficient enough that I use TO-220 output devices (cheaper and smaller) and they run about 105 degrees F, if I recall correctly. I can keep my finger on them indefinitely and be comfortable. And that's in spite of TO-220s having much smaller back plates than TO-247s or TO-264s.
As for the heat exchanger--the one I use is faaaaar larger than I actually need. I could cut it in half, if need be, and still run much cooler than with air-cooled devices. The total bulk of one side of the heat exchanger is about 15 inches square by about 2 inches thick. The volume is roughly the size of one monoblock chassis of a single Aleph 2 pair. I don't need a fan on it, so that part is passively air-cooled. So what gives? Why isn't it huge? The heat exchanger is more efficient at moving heat from the water to the air. What everyone seems to miss is that the heatsinks used in amplifiers aren't nearly as good at transferring heat as you think they are. Compromises are made in order to have flat planes for the devices to mount to, more flat planes to bolt to the rest of the chassis, and still more flat planes so you can stack herds of the heatsinks together to arrive at any arbitrary amount of heat transfer you might need. They also need to be physically robust and look nifty. The heat exchanger from a heat pump (in my case) has no such limitations. It's designed to move heat to air. Period.
About the only thing you can say thermodynamically is that the water pump generates some heat, but it's a trivial amount and is taken care of by the pump itself; it doesn't add to the heat within the cooling system.
Perhaps this is just one of those things you have to see to believe.
Grey
Another item to keep an eye out for on fleabay are pass banks out of ion lasers. They have lots of room for TO3 devices (most of the time they still have the devices mounted) and are usually crazy cheap.
Agreed. Can you imagine cooling big welders or chillers or large frame lasers with only heatsinks??!!
GRollins said:
Agreed. Can you imagine cooling big welders or chillers or large frame lasers with only heatsinks??!!
You could go one up on water cooling and use a refrigeration cycle...
To Danny: the poor heat conductivity of water doesn't really factor into the heat transfer equations. If you break down the heat transfer mechanisms, it looks like this:
1. Device => Water at edge of pipe - Mechanism: Conduction through some heat transfer compound, into the metal that the pipes will run through, then into the water right at edge of the pipe.
2. Water in pipe near device => Water in pipe at heat exchanger/radiator - Mechanism: Convection. The heat's just part of the water that is being pumped around. If you don't understand, its so simple to me that I can't actually explain it. The higher the velocity of water, the quicker the heat is transferred.
3. Water in pipe => Air at edge of heat exchanger (HX)/radiator - Mechanism: Conduction. The heat is transferred to the air through the metal of the HX/radiator. Having the water traveling through a pipe gives you 270-360 degrees to conduct heat away (the heat doesn't go 'down' very well) from the pipe, rather than the 180 that you get with a flat face of a normal heat sink.
4. Air at edge of HX/radiator => Bulk Air - Mechanism: Convection. Lower density, hot air rises into the bulk air, drawing in the cooler bulk air.
For the heat sink arrangement, the mechanisms are as follows:
1. Device => Heat sink - Mechanism: Conduction. Again through a heat transfer compound, into the heat sink.
2. Heat sink => air at edge of heat sink - Mechanism: Conduction. As above
3. Air at edge of heat sink => Bulk Air - Mechanism: Lets say convection.
The big gain of the water system are that
A. The temperature gradient between the device and the water in the pipe can be kept constant at T(device) => T(bulk air/ambient), depending on how well the heat exchanger works. This is beneficial because you're only conducting through a small distance of metal, basically pipe wall plus the contribution of the mounting block (which should be fairly minimal).
For the heatsink, you have to conduct through a comparatively large distance of heat sink before reaching air.
B. Moving the heat exchanging area away from the side of the enclosure means that you get more than just heat transfer in 180 degrees, and you can design a much more efficient system: eg thinner fins closer to the bulk of the heat and better air flow (more convection.
Essentially you're right, its just relocating the heat, however it does allow much better heat transfer by cutting down the slow heat transfer sections (conduction).
To Danny: the poor heat conductivity of water doesn't really factor into the heat transfer equations. If you break down the heat transfer mechanisms, it looks like this:
1. Device => Water at edge of pipe - Mechanism: Conduction through some heat transfer compound, into the metal that the pipes will run through, then into the water right at edge of the pipe.
2. Water in pipe near device => Water in pipe at heat exchanger/radiator - Mechanism: Convection. The heat's just part of the water that is being pumped around. If you don't understand, its so simple to me that I can't actually explain it. The higher the velocity of water, the quicker the heat is transferred.
3. Water in pipe => Air at edge of heat exchanger (HX)/radiator - Mechanism: Conduction. The heat is transferred to the air through the metal of the HX/radiator. Having the water traveling through a pipe gives you 270-360 degrees to conduct heat away (the heat doesn't go 'down' very well) from the pipe, rather than the 180 that you get with a flat face of a normal heat sink.
4. Air at edge of HX/radiator => Bulk Air - Mechanism: Convection. Lower density, hot air rises into the bulk air, drawing in the cooler bulk air.
For the heat sink arrangement, the mechanisms are as follows:
1. Device => Heat sink - Mechanism: Conduction. Again through a heat transfer compound, into the heat sink.
2. Heat sink => air at edge of heat sink - Mechanism: Conduction. As above
3. Air at edge of heat sink => Bulk Air - Mechanism: Lets say convection.
The big gain of the water system are that
A. The temperature gradient between the device and the water in the pipe can be kept constant at T(device) => T(bulk air/ambient), depending on how well the heat exchanger works. This is beneficial because you're only conducting through a small distance of metal, basically pipe wall plus the contribution of the mounting block (which should be fairly minimal).
For the heatsink, you have to conduct through a comparatively large distance of heat sink before reaching air.
B. Moving the heat exchanging area away from the side of the enclosure means that you get more than just heat transfer in 180 degrees, and you can design a much more efficient system: eg thinner fins closer to the bulk of the heat and better air flow (more convection.
Essentially you're right, its just relocating the heat, however it does allow much better heat transfer by cutting down the slow heat transfer sections (conduction).
Three forms of heat transfer (heat flux/unit area).
Conduction q=k(T2-T1) Metal to Metal, Liquid to Liquid. Liquid to Metal and vice versa
Convection q=h(T2-T1) Gas to metal, Gas to liquid and vice versa
Radiation q=e@(T2-T1) radiating object to anything and vice versa
@ is Stefan-Boltzmann constant - can't find sigma
Do not bother with Radiation (emmisivity) It is negligible. If you can't feel the heat on your face (like being next to a campfire) its not appreciable and can be omitted. Now if the heatsink were glowing hot, you would feel the radiation and it would matter.
As someone said earlier, painting a heatsink to gain emissivity is counterproductive, as it would be one more layer to conduct through - unless of course the coating has higher conductivity than the sink.
An Aluminum Sink coated with real copper?
Conduction q=k(T2-T1) Metal to Metal, Liquid to Liquid. Liquid to Metal and vice versa
Convection q=h(T2-T1) Gas to metal, Gas to liquid and vice versa
Radiation q=e@(T2-T1) radiating object to anything and vice versa
@ is Stefan-Boltzmann constant - can't find sigma
Do not bother with Radiation (emmisivity) It is negligible. If you can't feel the heat on your face (like being next to a campfire) its not appreciable and can be omitted. Now if the heatsink were glowing hot, you would feel the radiation and it would matter.
As someone said earlier, painting a heatsink to gain emissivity is counterproductive, as it would be one more layer to conduct through - unless of course the coating has higher conductivity than the sink.
An Aluminum Sink coated with real copper?
Correction:
As mentioned, three forms of heat transfer (heat flux/unit area).
Conduction q=k(T2-T1) Solid to Solid
Convection q=h(T2-T1) Fluid to Fluid, Fluid to Solid, Solid to Fluid
Radiation q=e@(T2-T1) radiating object to anything and vice versa
@ is Stefan-Boltzmann constant - can't find sigma
Do not bother with Radiation (emmisivity) It is negligible. If you can't feel the heat on your face (like being next to a campfire) its not appreciable and can be omitted. Now if the heatsink were glowing hot, you would feel the radiation and it would matter.
As someone said earlier, painting a heatsink to gain emissivity is counterproductive, as it would be one more layer to conduct through - unless of course the coating has higher conductivity than the sink.
An Aluminum Sink coated with real copper?
As mentioned, three forms of heat transfer (heat flux/unit area).
Conduction q=k(T2-T1) Solid to Solid
Convection q=h(T2-T1) Fluid to Fluid, Fluid to Solid, Solid to Fluid
Radiation q=e@(T2-T1) radiating object to anything and vice versa
@ is Stefan-Boltzmann constant - can't find sigma
Do not bother with Radiation (emmisivity) It is negligible. If you can't feel the heat on your face (like being next to a campfire) its not appreciable and can be omitted. Now if the heatsink were glowing hot, you would feel the radiation and it would matter.
As someone said earlier, painting a heatsink to gain emissivity is counterproductive, as it would be one more layer to conduct through - unless of course the coating has higher conductivity than the sink.
An Aluminum Sink coated with real copper?
Actually, even if the coating had a higher heat conductivity than the heatsink itself, its still not going to be beneficial. You said it yourself: its another layer to conduct through.
The radiation equation actually should have dT^4. Its the ^4 that dictates how useful radiation will be: while its small, the contribution will be minimal, however as it increases, the contribution increases very rapidly.
Josh
The radiation equation actually should have dT^4. Its the ^4 that dictates how useful radiation will be: while its small, the contribution will be minimal, however as it increases, the contribution increases very rapidly.
Josh
I stand corrected on both items.
I was preoccupied trying to find the Sigma symbol that I totally forgot about the power of the fourth on the differential temp...at heatsink temps we are dealing with 25C - 80C, I still think Radiation can be neglected....but you are absolutely correct - as it gets hotter, it plays a more important part of the overall heat transfer
I was preoccupied trying to find the Sigma symbol that I totally forgot about the power of the fourth on the differential temp...at heatsink temps we are dealing with 25C - 80C, I still think Radiation can be neglected....but you are absolutely correct - as it gets hotter, it plays a more important part of the overall heat transfer
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.