Watercooled Antec Truecontrol, voltage on the blocks.
 

Hi all,
I have just finished up watercooling my Antec Truecontrol 550W PSU, pictures can be seen here:
http://www.nskjaer.dk/antec/
Now for the problem, there runs current though the heatsinks, alot of current. I have touched them and felt nothing (before I knew how much current it is), I have also connected the to with a piece of wire, no problems there, but there was still a current across the heatsinks.
The psu is running fine, the rails are very precise, so the currentproblem does not affect the rails.
The original heatsinks were live, the new ones are not. I have isolated all components and cannot see where the blocks are in contact with anything that has current running through it. I have watercooled other PSU's without problems, is there something special with Antec?

At what two points did you connect the ampmeter, how many amps ( or what portion of an amp ) did you measure?

actually I only meassured voltage, the points was each heatsink so I connected those with the multimeter.
When I am measuring DC, the reading is either -70V or out of range as in over 1000V which I find very strange. When I measure against the 0 in the wall outlet, the volts is about 65V on the little block and 13V on the big block.

I do not understand what the problem is, but it runs fine even when it is watercooled, so the amps must be very low or am I missing something?

Well u could replace the koper plates by alu one as those don't guide elektricity...
Other then that I really like the Idea :drool:
Is it hard to solder those mosfets off and resoldering them?

I have the TP380 ,only I have modded the voltage rails :p

Alu does guide electricity, so that wouldn't help. The soldering part is not hard, it just takes time with a desoldering thing. The hard part is to make the blocks precise so the mosfets can be soldered back on.

Now I have measured the amps, between 0V in the outlet and the blocks there are about 0,05A. Another guy was talking about, that the currency could be generated by the coils, by induction. What du you guys say about that?

By "guide" I presume you mean conduct electricity.

To answer your suggestion, ALL metals conduct electricity. And to make things worse, there is a very strong relation between electrical and thermal conductivity. In other words, you will have a hard time finding something which condcuts the heat but NOT the electricity.

8-ball

Beat me to it

There are supposed to be mica shims that insulate the heatsinks/waterblocks from coming into contact with your mosfets and toroids. Did you replace those? I can't tell from the pics.

#8 > I reused the ones from the original heatsinks.
http://www.nskjaer.dk/antec/cwdata/antec37.html

When I measure the volts between the 2 waterblocks, it depends on the setting of my multimeter. If I measure with the setting at max 200V dc, the volts is 67V, when max is 20V, the volts is 6.7V.
It is very strange I think.
I think I will take it apart later and recheck all connections, but at first I will try and connect the 2 points where the original heatsink was attached.

anodized aluminum then?

I doubt that would be non-conductive either. But it doesn't really matter, because either there is a spot where I haven't isolated a component from the waterblock (which I doubt) or else it is because of the coils making noise planting in the blocks and then I will need to find a solution to that.

Hi,

The voltage you measure is there because the insulation between the mosfets and the heatsink act like a small capacitor. A small current can run across a capacitor if the voltage is AC. The capacitance acros the mosfet to the heatsink should be very small and only a few micro Amps should run between them. If it is more the insulation you used might not be suitable for this purpose. Mica is a good insulator that can handle a good bit of voltage too.

You were lucky to survive touching the heatsink though, if the insulation was not up to the job you could have been electrocuted. The mosfets chop the AC current from the mains to a higher frequency to convert the AC mains more efficiently to 5V .. 12V than a regular transformer could. The result is that some of these mosfets are directly connected to the mains voltage, also the metal plate connected to the mosfet (very) possibly carries full mains voltage. :eek:

The voltage of about 65V is in the middle between 110 mains, suggesting that one of the mosfet is connected to 110V and the other to 0V. The capacitance of the insulation between the two mosfets act like a voltage divider and the resulting voltage on the heatsink is 65 V.

0V ---||---heatsink (65 V)---||--110V

You better make sure these insulator pads are made for this job...

Thanks for the explanation, the voltage is 230V though, I am living in Europe. The micashims are the same as before, so they should be up to the job. But still, I think it is weird that the problem is with both heatsinks.
Can I connect the PSU to my computer without frying anything?

I guess if they came from the powersupply they should be fine. I don't think there is a danger of frying the computer, just a danger that the mains voltage might end up on one of the watercooling parts. It might not fry the computer but it could kill you. The original heatsinks are supposed to be contained within the power supply, within a metal shield. The power supply was not designed with the possibility that someone could touch the heatsink, either directly or through a watercooling setup.

Are you sure the bolts keeping the mosfets on the heatsink are insulated as well ?

You said the voltage is 65V, but the current to ground you measured as 0,05A. I assume you mean .05A which is 50 milliamps. If this is true what you could do is ground the blocks. This will protect you and may protect the rig as well, although you would have a little leakage current. IMO and thats all it is, if you cant ground these blocks you shouldnt even try this at all. Maybe you could rig some sort of monitor for it like a cheap ampmeter in the circuit all the time, or a low value resistor in series to ground that you could monitor the heat in. I say this because perhaps if your insulators break down gradually you could have no increase in voltage but still have an increase in current sourcing ability, which is the really dangerous part of it. Good luck.

there is one issue i don't know what would happen.
water conducts electricity as well.
what will happen when when the power in your powersupply leeks out through the water... will you have that much voltage going to the waterblock on your cpu?
also I imagine your heatercore will be grounded to the case. will it short through there?
Just questions that mebe someone will have answers to...

:eek: 50 mA is much too high for leakage through linked capacitors. Are you sure it is that high ? It is probably AC current so you have to set your meter to AC current. Probably the meter can't handle 20kHz AC. I would check those insulating pads and bolts again....

Did you insulate the bolts...?
I can't see any washers to keep the bolts from making contact between the diodes and the copper plates...

Ok, lots of replyes, thanks.
I know the heatsinks are supposed to stay inside the cabinet of the psu, but this is not the first psu I have watercooled, I have never before had a problem like this, I believe that I have insulated all the components, that is what I don't understand, where does the power come from?
I do not really have good options in grounding the psu, also I do not believe this to be the best solution, I would rather find the reason for the problem. I mean 50mA, I am from Europe, we use , where you americans would use a . so when I write 0,05A you should read it as 0.05A
I have run it with water, there is nothing to measure on the other blocks.
I have connected everything as it was, I have only replaced the heatsinks with waterblocks.
Tomorrow I will take everything apart and check all isolations and maybe add some shims.
So far, thanks a million (1.000.000,00 ;))

I dont think you got what i was asking for.
You need to insulate the bolts that hold the parts (power diodes if not in error) on the waterblock.

example from BladeRunner's work :

http://www.zerofanzone.co.uk/picture...su/psu_012.jpg

Seen the washers on the bolts...? Try that.
I dont know if you used any but the pictures dont show any.

Ok, I will try and do that, but they weren't before so therefore I don't think it should be nescesary know. Btw, the components where the bolt would be in contact with metal, they are insulated now as they were before.

TerraMex is right on this - looking at your picture the trouble is very clear. the back of the Mos usually have some sort of electrical connection. In your picture you can tell that the screws are making contact with the back plate.
have a look here as one example notice that the back also works like the drain:
http://www.usinternet.com/users/kyle...ic/IRF-510.pdf

#22 I would say that it depends on the mosfet. Some of them have a piece of metal as back, those do work as a drain, but some of them don't and should therefore not need a mica shim. The ones that have metal backs also have mica shims. But the strange part is, that I have connected everything just as it was, only thing is that I have made other blocks. So the mosfets should be insulated just as well now as before.
But I have bought new mica shims and nylon bolts today, so now I am really ready to insulate those suckers, that I will do tomorrow.

I read through this topic the other day and didn't post because I couldn't see that I could add any help. :shrug:

I'm still not sure I can help but to clarify some things, all my fets bar one use mica shims to prevent contact of the metal fet back plate on the block. This was not really a requirement as all the fets tested "earth" on the back plate and as my water block is also earthed it wouldn't matter. I just did it for additional safety / piece of mind. My water block is earthed via a wire to the PSU chassis, so in theory I could have left the mica sims out as they were originally stopping the live 240volt mains sink from shorting out on the earthed fet back plates. Those that worry about the mica shim not isolating the voltages should realise this is what they do and are designed to do in the original item. Also look upon the mica shim in a similar way to the plastic coating around the wires, no one questions this thin plastic coping with the voltage isolation.

The fet that has no mica shim has a plastic backing with no exposed metal. On my fets in my PSU, (and you have to appreciate all PSU's differ in design, components and layout), the washers I used were purely to spread the loading on the fet package and not for isolation. The metal area around the fet hole on the front is not connected internally or externally to the metal backing on my fets.

I'm puzzled as to what is causing your sinks to show a voltage, and can only assume it is as Prlwytkovsky said, capacitance, and the voltage wont have any amperage, (or a very small amount), I'm only guessing though as I don't really understand enough about how they work to advise :confused:

A long shot the sink is not touching the tops of the large coils??

The sinks are touching the large coils, but there is some plastic between:
http://www.nskjaer.dk/antec/cwdata/antec40.html
The white stuff is doublesided tape and you can see some of the plastic.
But what about conductive electricity? Could that be the problem?

Tomorrow I will take it apart and replace the mica shims and the screws with nylon screws, that way I should be certain that there is no connection between the mosfets and the sinks.