Any electricians here? I know my computer is unsafe, but don't know how unsafe

[JonJ678]

In short I've done something very stupid and would like to know if there will be particularly terrible consequences. Please avoid spamming this thread with any variants of "you're so stupid" or "you're so cool".

I've put a waterblock on a piece of copper which is unfortunately at mains ac. Between the copper and the waterblock are three layers of pvc electrical tape as I hoped these would isolate the two. Not so, I believe the magnetic field associated with the ac is generating a voltage within the copper base of the waterblock. This voltage is applied to the water in the block, leaving the barbs live.

A length of tubing is connected to each barb, and grounded at the far end. As such I'm fairly confident that any water beyond this short loop is safe (multimeter reports zero, doesn't hurt when I touch them). However my multimeter is no longer reporting sensible values at the psu barbs, set to 600V ac range it's fluctuating between 0 and about 2000. No fuses are blowing while it's operating under these circumstances, so I believe the current running to ground to be small.

How do I go about 1/estimating the risks involved here and 2/isolating the rest of the loop from the psu? Some photos attached.

 

[JonJ678]

Yeah. It all went swimmingly until it turned out the big copper heatsink in a convenient place was live.

Block's on the piece of copper shown here

 

[JonJ678]

Kapton tape is probably a winner. Any idea if it'll actually isolate it? I'm pretty certain a single layer of pvc will block 240V, hence the induction theory

 

[JonJ678]

There's a thought actually. CPU heatsinks are bolted through the board, so probably are earthed. I can run some wires from the barbs on my cpu block to the mounting mechanism as another line to earth.

I couldn't find any data on thermal coefficients of pvc tape but I can't see them being good. I think I'll have to rebuild with kapton tape in the near future.

The block is an xspc chipset one, design is a piece of copper with a lot of parallel slots milled and some acetal to force water through the slots. It looks like it'll be pretty good. It's the largest one I can fit without cutting up the psu itself, which I'm not yet brave enough to attempt. I have no estimate for heat dissipation, I'm hoping for any value greater than zero. I haven't worked out how to measure it, so I'm currently working on "without block, air coming out the psu is hot. With block?" which is woefully imprecise.

Well, it turned on, and nothing has caught fire yet. I'm clinging to induced voltage => earth means no current going to cpu like a drowning man though, not feeling very confident here.

 

[JonJ678]

If you can suggest a mechanism by which this leads to harming other people then I'm all ears and will dutifully remove this change. As it stands though the only person likely to suffer electrocution is me, and the only hardware at risk is my own. I appreciate the warning though, and think I will purchase a CO2 fire extinguisher in the near future.

The air coming out of the psu is still hot, so my current suspicion is that the process of electrically isolating the block has indeed thermally isolated it. I'll report back when I get some of the new type of tape.

 

[JonJ678]

@Justintime no pets sadly, and no kids, otherwise I'd be much more cautious. It's in a sealed, earthed metal box so there's limits to how dangerous it can be to passers by. There's definitely potential for this to go wrong, but I can't see it hurting anyone. Well, beyond the electric shocks I've already sustained.

@jaybee Well **** you too

 

[JonJ678]

Koolance make some (expensive with suspiciously short warranty) watercooled psu's, and shadowscotland has done this in the past. A couple of people have done this before me, though it sucks badly that the heatsink turned out to be live in this case. I already have an excellent psu, which makes me reluctant to buy a new one. If this one goes terribly wrong, then I'll be in the market for a new psu and will either get a passive one or a watercooled one. Until then, I want to see what happens when I try this

Cheers AMG. I've always learned best from my own mistakes. Seems in the spirit of an overclocking forum to share them, this is all a learning process after all.

Fortunately most "not so good ideas" don't involve severe electrocution, explosions and fire

Only the absolute best ideas involve these

I'm not even close to 100% sure on this idea. A large part of me wants to connect the heatsink up to the waterblock directly, just tim inbetween, and see how that works out. This may be the death of my nas. However the process of experimentation is how we become sure of things, the whole point to life is experiment and observation, preferably recorded. This is an unusual idea, and stupid on many levels, but I am learning from it.

@.walls If upon my accidental death you can get to it before my girlfriend beats it to death with a hammer you're welcome to it. You'll have to move fast though

@Rjkoneill I'd given up on the psu warranty at roughly the same time as I took a piller drill to it. Ek told me I'd void the warranty by flycutting their block, but I believe you're also in possession of a machined flat block so I daresay you understand where I'm coming from.

I do hope this doesn't kill the rest of the system, as I suspect my best shot at a very immoral rma would be to wait for an electrical storm in my area then plead ignorance as to why it isn't working anymore, and I can't see this working out now I've started this thread.

I am of course pleased by your compliment, thank you. Mental is a useful aspiration for a trainee engineer.


Ontopic after all that,

Does anyone know how to estimate the current available on the wet side of this system? Assuming 240V at 50Hz acting on an infinite plane, with another infinite plane perhaps 0.5mm away, what is the induced emf in the second plate and what current can be drawn from this? I'm going to have to track down a physicist I think.

System seems to be fine btw, it's passed stability testing and I'm now fighting with f@h.

 

[JonJ678]

I can barely cope with photography, video is far beyond my skills. Probably for the best considering that I now flinch whenever my hand brushes against the casing of the computer or any of the tubes.

I'm not sure oil is any better than air to be honest, though I'd be interested in arguments in favour of it. The main issue would be oil leeching out along the cables, otherwise it's reasonable a reasonable idea if combined with something like this, without a waterblock on the heatsink the oil temperature would just rise uncontrollably under load I think.


That's an excellent post wonko and most appreciated.

I hadn't considered the dissipation limits of the chipset block, and that's a pretty glaring omission on my part. It's very unlikely to be good for 200W. Under full load in my computer it's probably drawing around 500W from the wall (overclocked i7, two overclocked 8800gt cards, folding) and wasting 100W as heat, also rather more than the average chipset. I've contacted xpsc to ask for details on their block, though it's quite possible they're going to ignore me on the basis that they can't possibly condone doing something like this. At present the fan is running in addition to the water block, for the best really given I don't think the block is doing anything much.

Even if I get this working nicely I don't think I'll remove the fan completely, as under full load I don't care that the computer is loud (untrue, but I don't have space for many more radiators). However I would like a switch that turns the psu fan off when idling, and at idle (surely to be sub 150W, even with 8800s), passive cooling of the psu components with a block on the heatsink will probably be fine. I'm relying on pc p&c overengineering this to some extent, since it's good for 64A continuous at 50 degrees ambient and I'm placing a far smaller load on it this is probably a good assumption.

I'm happy to see another cautious vote for high voltage, low current as a result of inductance. I probably should have learned some details about ac before starting this.

Don't be silly Mr Fire Investigator, you can't have a fire in a watercooled computer, as it's full of water and water doesn't burn. Feigned ignorance probably isn't a legal defense though. Regardless, after leaving it folding overnight I'm now pretty confident in it.

I should have a 50mm square piece of thermally conducting (-60 to +180 degrees), electrically insulating (up to 1kV at 50Hz), 0.2mm thick silicon tim arriving in the post which will make this project far more successful. Fingers crossed.

note: I "invested" in an 800W ocz elitexstream from here perhaps a year ago, before starting water cooling, on the basis that it was brilliantly reviewed and I thought it would last just about forever. It died very gracefully with a common fault a few months after this, and ocz swapped it with my 860W pc p&c, a far better psu. This is how I ended up with a ridiculously overspecified, air cooled power supply.