Watercooled Powersupplies

[jonnyGURU]

Re: Watercooled Powersupplies

Yes. You are correct.

[jonnyGURU]

Re: Watercooled Powersupplies

EDIT: Here's the actual Engrish answer I got. It just didn't translate right, I guess.

the first heatsink will come out either AC or DC depends on what it touches. the second heat sink is almost grounded and has no power. when first heatsink touches the second one that will occur the voltage impact (one is too high and the other is too low on voltage) on FET and other parts and make PSU blow up or short.

[kc8adu]

Re: Watercooled Powersupplies

the heatsink on the primary side is often connected to "hot" ground(one side of the ac line depending on wiring in the outlet.should be neutral but can be hot if outlet is wire in reverse.
the rectifier heatsink on the secondary side is often connected to earth/chassis ground.
the secondary side is isolated via the transformer.if its connected to the ac line backwards there can be full ac line voltage between the sinks.

[japlytic]

Re: Watercooled Powersupplies

In an old Seventeam PSU, the large heatsink is used for the primary side transistors and the +5V rectifier, with the heatsink grounded. The metal tabs on the transistors and rectifiers were electrically isolated from the grounded heatsink; this caused no problems.

[jonnyGURU]

Re: Watercooled Powersupplies

In an old Seventeam PSU, the large heatsink is used for the primary side transistors and the +5V rectifier, with the heatsink grounded. The metal tabs on the transistors and rectifiers were electrically isolated from the grounded heatsink; this caused no problems.

Hmm... With the heatsink grounded to Earth ground?

[larrymoencurly]

Re: Watercooled Powersupplies

Distilled water is non-conductive

I remember a science teacher demonstrating conductivity by using a 120V lightbulb and dipping two leads into a beaker of tap water -- bulb glowed brightly. Then he dipped them into a beaker of distilled -- bulb glowed orange.

[larrymoencurly]

Re: Watercooled Powersupplies

I've never seen any FETs mounted on the secondary side heatsink - usually these are just rectifiers. (double-diodes, as it were)

The cheapo 300W King Star PSU on the right uses one for the +3.3V rail. I had the 250W version and learned of this when a friend of mine plugged a DIMM in backwards. The DIMM merely got warm, but the FET shorted out.

[tiresias]

Re: Watercooled Powersupplies

The cheapo 300W King Star PSU on the right uses one for the +3.3V rail. I had the 250W version and learned of this when a friend of mine plugged a DIMM in backwards. The DIMM merely got warm, but the FET shorted out.

Right! So (excuse my lack of electronics professionalism here ), how does that actually work? Is the FET used to modulate (PWM-style) an already rectified DC output to a lower voltage? Or in some other way?

[tiresias]

Re: Watercooled Powersupplies

I remember a science teacher demonstrating conductivity by using a 120V lightbulb and dipping two leads into a beaker of tap water -- bulb glowed brightly. Then he dipped them into a beaker of distilled -- bulb glowed orange.

Absolutely - there is a huge difference in conductivity between some ion-containing solution and distilled water. That's why most of my messy hydrogen-producing electrolysis demonstrations ( https://www.badcaps.net/forum/showthread.php?t=1680 ) tend to involve a very large amount of salt dumped into the erm... container. (Actually an IKEA-bought plastic container with graphite rods siliconed through holes in the bottom!)

Obviously this way you get mostly chlorine instead of oxygen, which has the side effect of 'stinking the place up' a little if done indoors.

[Spacedye69]

Re: Watercooled Powersupplies

The only non conductive water is deionized. Running through copper blocks would soon cause it to be contaminated, and thus conductive, rather quickly.

[tiresias]

Re: Watercooled Powersupplies

The only non conductive water is deionized. Running through copper blocks would soon cause it to be contaminated, and thus conductive, rather quickly.

For that matter, even perfectly pure water is a little conductive, due a small degree of self-dissociation into hydronium and hydroxide ions.

And, as you say, with copper tubing/contact blocks (not to mention exposure to the open air, etc), it wouldn't stay pure for long.

[Galvanized]

Re: Watercooled Powersupplies

For that matter, even perfectly pure water is a little conductive, due a small degree of self-dissociation into hydronium and hydroxide ions.

And, as you say, with copper tubing/contact blocks (not to mention exposure to the open air, etc), it wouldn't stay pure for long.

Given all that's been stated so far about water's conductivity, how do the linked water cooled PSUs manage with one inlet & one outlet?

[Per Hansson]

Re: Watercooled Powersupplies

Given all that's been stated so far about water's conductivity, how do the linked water cooled PSUs manage with one inlet & one outlet?

They do it the unsafe way; one big block and the primary side FET's only insulated with a non-conducting thermal interface material against the waterblock...

[tiresias]

Re: Watercooled Powersupplies

Per and Galvanized - It can indeed be that they use a single block - does anyone have any pictures of the inside of one of these "commercial" watercooled PSUs? I've never seen or had one myself.

Still, I guess if constructed properly, even a one-block setup could be made to be 'safer' by ensuring that the block itself be grounded, both with respect to the 0Vdc (as it were) and the wall socket. .

If I see how few wall sockets around here actually have an earth though...

On the other hand, a two-block setup would essentially maintain the same design found in "conventional" PSUs - the water's conductivity may not be zero, but even that of a salt solution is far, far less than that of a copper slab.

I say we try it!

PS. I'm already eyeing the old Fortrons here as potential "victims".

[Per Hansson]

Re: Watercooled Powersupplies

tiresias; the commercial blocks I have found when googling have all been a single block, or rather two blocks but connected together with a welded copper tube so...

[Galvanized]

Re: Watercooled Powersupplies

The technicalities of water cooling PSUs does interest me. The idea of doing it myself does not, atm. The point kc8 made about the correct polarity of mains voltage should not be taken lightly, I do believe.

I spent some time on Google and came up with one site that maybe of interest. http://zfz.com/projects.asp?request=liquidenermax550w I think leaving a low flow 80mm@25cfm would have been wise but his pursuit was NO fans. The Enermax Noise Tacker seems to be the unit of choice from what I've seen via Google.

A much more basic/crud presentation= http://www.digital-explosion.co.uk/i...p?articleID=65 Not that I could do better first try.

This fella does conversions on request. The pics of the Noise Takers would indicate one heat sink block. http://www.edvextrem.de/

Enermax is known for a rather open/orderly pcb lay out. A crowded PSU may have a rough time of it keeping it's cool with just water cooling. Now I must see the innards of an Etasis 300W passive unit

[Howard]

Re: Watercooled Powersupplies

http://www.silentpcreview.com/article80-page1.html

[jonnyGURU]

Re: Watercooled Powersupplies

Missed this one: http://www.performance-pcs.com/catal...oducts_id=4570

[MD Willington]

Re: Watercooled Powersupplies

I remember a science teacher demonstrating conductivity by using a 120V lightbulb and dipping two leads into a beaker of tap water -- bulb glowed brightly. Then he dipped them into a beaker of distilled -- bulb glowed orange.

Also called 10Mohm water when it is pure H2O...

[larrymoencurly]

Re: Watercooled Powersupplies

How are industrial water cooled PSUs built?