View Full Version : Psu Watercooling
alright.. i want to wc my psu.. but i find that taking off the hs and attaching (wut i think are voltage regulators) would be too much work..
could i just take a wb and attch it to the hs currently in there..
remove the fans..
would this run stable?
and i was reading http://www.zerofanzone.co.uk/content.asp?request=liquidenermax550w&page=1 .. and i dont really get wut it means by live...
safemode
10-11-02, 09:36 PM
the heatsink has current running through it. Most likely due to being in contact with a wire or connection that is not a ground. You'd have to ask the manufacturers why they do this if it's for a particular purpose or due to cheapness.
JFettig
10-11-02, 10:29 PM
dude, get a decent setup first, i tried speeding off to this and it didnt work, you need a good psu and good tools, takes a ton of work and all to make something for a psu, and if you have leakage your screwed, you gotta make it perfect, i advize you not to do it at this time,
Since87
10-11-02, 10:35 PM
The whole supply was designed with the assumption that there would be airflow. There are likely parts that aren't attached to heatsinks yet dissipate a significant amount of power. Taking away the airflow is going to allow a lot of stuff to get hotter than the manufacturer designed for.
A rule of thumb used for semiconductors is that each 10 degree C rise in temperature halves the life expectancy of the part.
Watercooling just the hottest components of your power supply, while taking airflow away from the rest, is likely to reduce the lifespan of the supply a lot.
I wouldn't recommend Watercooling a psu to anyone. Sure there are a few special individuals on these forums that took on watercooling a psu, but what for. There are too amny risks involved in doing so. It can cause fires and even your own death. This is 110 AC that were playin with in a psu. AC means the current alternates. DC is a steady current.
AC can reach 300 volts + at some points. Thats enough to kill you if your sitting next to your computer and a leak occurs in your psu.
Did I mention eletrical fires cause by sort circuits. Don't play with eletricity like that usnless your willing to pay heavy consequences.
PS: Im not flaming you or anything Liss, just don't want to see a fellow OCer loose there system or get hurt. ;)
thx for the advice..
i'm not gonna go throgh with it anymore anyways..
:p
safemode
10-12-02, 04:09 PM
if a leak accurs in the powersupply. You are going to short your circuit while inside the psu. The psu case is grounded. You should short the circuit breaker that the computer is on before the water gets all over the place enough to damage things nearby or hurt the person sitting next to it.
We're not talking about a tv here. You may start a fire in the psu, but unless your computer is backed up against a curtain the chances of an electrical fire getting out of control are very slim. The biggest risk due to a water spill on your psu nearly the same as a water spill anywhere else in the computer. You may damage parts and you may cause a fire inside the case. It's just more likely that this i will happen when you short the psu. Also running the computer fanless is not viable most of the time. So keeping a 80mm fan in there running at 5volt or even less should be sufficient for all the other items in the psu since the pot's heat is going to the water block.
I'm not really sure what people think is going to happen when you dump water on the psu. You think bolts of lightning are going to shoot out? Perhaps a raging fire that goes everywhere? No. Your house circuit breaker should fire and if that doesn't first, the circuit breaker you plug your computer into will. The residual current should quickly drain out and any fire will stop soon after if not from the water leaking still then because it's all inside a metal case where the only fuel is the rubber and pcb resin which soon runs out. You'll get a lot of smoke and possibly dead hardware.
As long as you dont have your face right up next to it while testing the setup, you should be fine. Always plug your computer into something with a circuit breaker.
Originally posted by safemode
if a leak accurs in the powersupply. You are going to short your circuit while inside the psu. The psu case is grounded. You should short the circuit breaker that the computer is on before the water gets all over the place enough to damage things nearby or hurt the person sitting next to it.
We're not talking about a tv here. You may start a fire in the psu, but unless your computer is backed up against a curtain the chances of an electrical fire getting out of control are very slim. The biggest risk due to a water spill on your psu nearly the same as a water spill anywhere else in the computer. You may damage parts and you may cause a fire inside the case. It's just more likely that this i will happen when you short the psu. Also running the computer fanless is not viable most of the time. So keeping a 80mm fan in there running at 5volt or even less should be sufficient for all the other items in the psu since the pot's heat is going to the water block.
I'm not really sure what people think is going to happen when you dump water on the psu. You think bolts of lightning are going to shoot out? Perhaps a raging fire that goes everywhere? No. Your house circuit breaker should fire and if that doesn't first, the circuit breaker you plug your computer into will. The residual current should quickly drain out and any fire will stop soon after if not from the water leaking still then because it's all inside a metal case where the only fuel is the rubber and pcb resin which soon runs out. You'll get a lot of smoke and possibly dead hardware.
As long as you dont have your face right up next to it while testing the setup, you should be fine. Always plug your computer into something with a circuit breaker.
I know i guy over internet and he has have a waterleek thats did come in the psu and as ju said(and i as already know) its only shortcircut and killed the psu for him but all other parts where fine.
SemiCycle
10-12-02, 06:03 PM
I'm just going to add my vote to the masses. Bladerunner is one of the fewer people who could pull this offer without signing his computers death certificate. HIGHLY ***NOT*** RECOMMENDED for the average water cooler like most of us.
BTW, there is a reason his site is called zerofanzone. He set out on a project to create a totally fanless computer and succeded. Even the most die hard water cooling guru wouldn't worry about doing this. In a way, it is s pointless as watercolling the memory in the computer. I really enjoy his site, but I know I don't have the abilty, knowledge or guts to do many of the things he has.
Ohh BTW i plan on watercool my psu as well.
I would have the ability to do it, I'm studying electronics and computers(first year. so i'm no expert) So i know a bit about electronics and i can use some better tools for it in school. I will take precaution and messure all things so i know its safe and the waterblock will i make so it cant leak. All solderd and do some test to it first.
What i think of watercooling the psu that if you got the ability to do so then do it and if you dosen't have it you better don't do it becuse some parts have live voltage(the sinks for exampel) and you have to isolate the mosfets so they aren't leading any electricity to the block and water.
SemiCycle
10-12-02, 07:54 PM
Alls I can say is good luck.
athlonnerd
10-12-02, 08:07 PM
yuo can always run mineral oil through your psu, adn cool that. that would solve all problems mentioned here. since, is oil bad for any of these components, assubing its cold.
drunkmonkey
10-12-02, 09:06 PM
Well, you cant conduct heat w/o conducting electricity in a normaly constructed watercooling setup. Also, voltage regulators' heatsinks conduct voltage because the output voltage goes into the heatsink and one of the pins. You can't exactly absob the heat from the regulator unless you conducted electricity from it; thats why the heatsink has power to it.
safemode
10-12-02, 09:12 PM
so you get a very low powered tec and a cold plate to connect to the pots. water cool the tec. boom. done.
Iron Hawk
10-12-02, 09:55 PM
Arctic Alumina is not electricaly conductive, yet it conducts heat very well, whynot use a layer of AA Epoxy?
safemode
10-12-02, 10:09 PM
you could epoxy the plate to the wb too...that would work.
Originally posted by drunkmonkey
Well, you cant conduct heat w/o conducting electricity in a normaly constructed watercooling setup. Also, voltage regulators' heatsinks conduct voltage because the output voltage goes into the heatsink and one of the pins. You can't exactly absob the heat from the regulator unless you conducted electricity from it; thats why the heatsink has power to it.
Actually the mosfets(or transistors) are what i know thats generate most of the heat and they are not in electric contact with the heatsinks. They are isolated from them tho so why have voltage in them? There is a little pad or what i should cal them it's a sort of rubber like thing.
You can se them here
http://www.zerofanzone.co.uk/pictures/jpg/psuwiressmall.jpg
its the grey/white thing u see betwen the block and mosfets.
athlonnerd
10-13-02, 07:03 PM
what on earth are yall talkking about?
there is no current going through the heatsinks, or atleast, not designed to be. some are however electrically connected to the heatsinks. this is becuase, it has no detrement, but eliminates the thermal resitsance created by the rubber pads. there really in this case is no correlation between current and heat. the reason why they are electriacally tied is that most semiconductors are monted to a piece of metal, for higher thermal conductivity. the metal being live is only a side effect, and in noraml psu desing is not a problem. if it is, the rubber pads are used. all yuo need to do is use the pads, adn youll be fine
Since87
10-13-02, 08:15 PM
Actually I've seen a power supply that used a copper heatsink as a conductor. Much better to route high currents through a sheet of copper that's doing double duty as a heatsink, than to route the high current through traces on a PC board. As to whether any PC supplies does this, I don't know.
In my opinion, mounting the power components to a waterblock as shown in that picture is stupid. If the supply is a quality design, then many of the connections to the power components are very short when mounted on the PCB. Doing a mod like that is going to decrease the efficiency of the supply substantially at best. It may make the supply downright unstable.
Ct. Strangelove
10-13-02, 11:44 PM
I humblely disagree. The length increase is only 2 inches or so. If ou lengthend your ram or CPU 2 inches it will reek havoc. In a PSU the mosfet circuits are not sensitive to critical timeing or interfence. If the length was over a foot (or more), then noticable effects would be noticed. The only problem with lengthening the mosfets is if the solder joints are not 'hot' (techy term) then resistence will accour, thus reducing overall rail stablity and effeciency.
If you work is of quality then its a go. If you can solder worth a @#$ then it's a no go. I think bladerunner has done an exellent job with his PSU. I am attemping to modify my PSU in a similar fashion. But that project won't go underway for a few weeks.
A interesting pheonomon accured to me. When I was testing my Antec 550 PSU powering a 226w tec, i was shocked. I was playing my guitar to pass time, i touched the PSU to test warmth then i got shocked! It was not static, it was a long stream of electrons flowing up my fingers. When my left hand rested on the strings of the guitar i was shocked, when i released my left hand the flow stoped. I was grounding myself through the guitar and amp. It was interesting. On the a different note, the HS in the Antec psu is live.
Since87
10-14-02, 12:38 AM
Originally posted by Ct. Strangelove
I humblely disagree. The length increase is only 2 inches or so. If ou lengthend your ram or CPU 2 inches it will reek havoc. In a PSU the mosfet circuits are not sensitive to critical timeing or interfence. If the length was over a foot (or more), then noticable effects would be noticed. The only problem with lengthening the mosfets is if the solder joints are not 'hot' (techy term) then resistence will accour, thus reducing overall rail stablity and effeciency.
Having designed several switching supplies I can say with certainty that the lead length matters substantially. The main issue is the increased inductance of the connection between the gate of the MOSFET and its driver. This inductance will causing ringing in the voltage at the gate, which will cause the MOSFET to turn off and on multiple times when it is switched. Many switching designs rely on one MOSFET turning off at precisely the same time that another is turning on. (Or at least as close to the same instant as is feasible.) Unless the MOSFET's switch on and off cleanly, undesirable shoot through currents will occur, reducing efficiency.
Correct me if I'm wrong, but aren't many heatsinks for mosfets grounded to keep them from
A)Being a broadcast antena and
B)To prevent them from being too dangerous if a component should move for whatever reason and make it hot with current?
I'd rather have it grounded (and being a better path to ground than me) if it were to get some positive juice from a capacitor meltdown.
And another thing, those pads are non-conductive TIM's made for shunting off the heat, but isolating the power from the 'sink. Mosfets get hotter'n'hell yet often conduct mucho power. You wouldn't want a pad that will melt/ooze out from in between them like you would want to happen with an HSF.;)
BladeRunner
10-16-02, 04:33 PM
Some interesting points here, but as the builder of the PSU mod above all I can add is it does work fine and has been for nearly a year now in a fanless PC. I'd known and mentioned in the build article about the possible increased inductance lengthening the wires can cause, but like said in this topic the amount is small and doesn't appear to affect anything.
I think what is being missed here is the fact we are dealing with a, "low quality" Power supply type, and I don't just mean Enermax PSU's, but all of the PC PSU designs I've seen. I also have a PC Power & Cooling 600 watt unit, supposed to be a top make, but in truth I think these types of switching PSU's are all poor, in general design / layout & and implementation.
athlonnerd
10-16-02, 04:47 PM
so make tiny wb's to fix individually to each mosfet, so you dont need to move the mosftes
drunkmonkey
10-16-02, 06:33 PM
Originally posted by Penti
Actually the mosfets(or transistors) are what i know thats generate most of the heat and they are not in electric contact with the heatsinks. They are isolated from them tho so why have voltage in them? There is a little pad or what i should cal them it's a sort of rubber like thing.
You can se them here
http://www.zerofanzone.co.uk/pictures/jpg/psuwiressmall.jpg
its the grey/white thing u see betwen the block and mosfets.
I though they where just voltage regulators. lol
Iron Hawk
10-16-02, 07:26 PM
Yep, thats what they do. The mosfets are transistors that regulate the voltage. Mosfet is the technical name, voltage regulator is the more common name, keep in mind though that there are other types of voltage regulators.
Since87
10-16-02, 11:45 PM
Originally posted by BladeRunner
Some interesting points here, but as the builder of the PSU mod above all I can add is it does work fine and has been for nearly a year now in a fanless PC. I'd known and mentioned in the build article about the possible increased inductance lengthening the wires can cause, but like said in this topic the amount is small and doesn't appear to affect anything.
I think what is being missed here is the fact we are dealing with a, "low quality" Power supply type, and I don't just mean Enermax PSU's, but all of the PC PSU designs I've seen. I also have a PC Power & Cooling 600 watt unit, supposed to be a top make, but in truth I think these types of switching PSU's are all poor, in general design / layout & and implementation.
I don't doubt that your supply works okay.
I'd almost guarantee that it has lost between 1% and 5% of its efficiency. Looking around today I saw that Enermax and Antec supplies are rated for 68% to 70% efficiency. That's low by todays standards for quality switching supplies so, I'd have to agree that all PC PSU's are probably of fairly poor quality. I didn't know how pathetic they are. I'd been guessing 90% efficiency. I guess losing a few percent on efficiency is no big deal when you are starting out at 70% anyway.
I suspect its lifespan may be decreased by lack of airflow over parts that aren't watercooled. I don't believe it's reasonable to assume, that all the parts that need airflow to stay within their spec'd temperature ranges, are mounted to heatsinks.
I did skim through your article. I got the impression that you did this mod as an experiment and were willing to accept the supply blowing up as a possible result. You realize the possible consequences and go for it. Great, more power to ya.
However, there are a lot of people on this forum who barely understand what resistance is, let alone inductance, and will consider taking on something like this not knowing that doing the same sort of thing with their supply may result in destroying their supply, motherboard, CPU, etc. There are a lot of potential problems that aren't addressed in your article, and the things I've mentioned just scratch the surface.
Since87
10-16-02, 11:59 PM
Originally posted by Iron Hawk
Yep, thats what they do. The mosfets are transistors that regulate the voltage. Mosfet is the technical name, voltage regulator is the more common name, keep in mind though that there are other types of voltage regulators.
Not really.
The MOSFET's are just a part of the voltage regulator circuit. They are actually the "switches" in a switching regulator circuit. An IC in the circuit controls the signal to the MOSFET to maintain the output voltage at the correct level.
Saying a MOSFET is a voltage regulator is like saying a CPU is a computer. There's a whole lot more than the MOSFET that goes into making a switching regulator.
BladeRunner
10-17-02, 01:55 AM
Since87
Your above Summation is spot on and I still consider it work in progress / development. In heavy use (gaming) there was a coil that was getting quite hot, (over 60°c), but a mod has solved this. I also thinned out the wiring for the devices replacing with single thicker wires for the four pin plugs rather than lots of thinner ones, to help airflow. I'm pretty sure if it were run at a high load all the time in a high ambient it would die. I doubt it would cope running a high watt TEC for instance, but like said in the build it's not being at all stressed in my system. I will be water-cooling the larger PC Power & Cooling PSU and can probably do it well with the mosfets kept in their default position as there is a bit more space with this unit. It's supposed to have 1% regulation but I haven't noticed it's "reported" voltages being any better than the water-cooled Enermax... so ??
http://www.zerofanzone.co.uk/pictures/jpg/600watt1.jpg (http://www.zerofanzone.co.uk/pictures/jpg/600watt.jpg)
I've also cut out the "lid" on the water-cooled Enermax to help natural heat rise airflow as the PSU is mounted the other way up in my tower to help the heat rise off the PCB rather than being trapped by it when mounted in the default way.
using a infra red temp sensor, no component or area in the PSU appears to get any hotter than 42°C (in a 26°c ambient).
Again it's not finished yet and like you said this is not something for the inexperienced to try as I was will to accept PSU and PC death in trying this build. My site just outlines how I did it, nowhere do I suggest that anyone does it themselves.
here are a few current pics showing the cooling bracket for the coil and the PSU "lid" with cut-outs that will have a mesh over them when I find the right type of mesh.:)
http://www.zerofanzone.co.uk/pictures/jpg/psu_coil.jpg
http://www.zerofanzone.co.uk/pictures/jpg/psu_coil1.jpg
more about the mod details HERE (http://www.sharkyforums.com/showthread.php?s=&threadid=154834)
Lid cut-outs below done with a Dremal using a cut-off wheel.
http://www.zerofanzone.co.uk/pictures/jpg/psucase.jpg
pauldriver
10-17-02, 01:59 AM
Originally posted by Liss
and i was reading http://www.zerofanzone.co.uk/content.asp?request=liquidenermax550w&page=1 .. and i dont really get wut it means by live...
HT= High Tension = High Voltage, Live= Powered On, carrying current or volatge or both.
Pdiddley's British to American Dictionary.
safemode
10-17-02, 07:25 AM
how much was the infrared temp sensor?
That's a sweet psu mod by the way. I'd rather not see the plastic windows since the psu is now no longer em shielded but it's your computer.
pauldriver
10-17-02, 07:52 AM
Originally posted by safemode
if a leak accurs in the powersupply. You are going to short your circuit while inside the psu. The psu case is grounded. You should short the circuit breaker that the computer is on before the water gets all over the place enough to damage things nearby or hurt the person sitting next to it.
We're not talking about a tv here. You may start a fire in the psu, but unless your computer is backed up against a curtain the chances of an electrical fire getting out of control are very slim. The biggest risk due to a water spill on your psu nearly the same as a water spill anywhere else in the computer. You may damage parts and you may cause a fire inside the case. It's just more likely that this i will happen when you short the psu. Also running the computer fanless is not viable most of the time. So keeping a 80mm fan in there running at 5volt or even less should be sufficient for all the other items in the psu since the pot's heat is going to the water block.
I'm not really sure what people think is going to happen when you dump water on the psu. You think bolts of lightning are going to shoot out? Perhaps a raging fire that goes everywhere? No. Your house circuit breaker should fire and if that doesn't first, the circuit breaker you plug your computer into will. The residual current should quickly drain out and any fire will stop soon after if not from the water leaking still then because it's all inside a metal case where the only fuel is the rubber and pcb resin which soon runs out. You'll get a lot of smoke and possibly dead hardware.
As long as you dont have your face right up next to it while testing the setup, you should be fine. Always plug your computer into something with a circuit breaker.
ohh yeah, I smoked a 400 watt PSU real real nicely with a coolant leak.
NEVER TRUST A PLASTIC (or even a cheap metal) HOSECLAMP.
Fortunatly, the PSU went BOOM!!!!! FLAMES SHOT OUT (yes they did), and the lights went out and all was good. No real fire, lots of scorching and an exploded cap or four.
I was SOOOOO lucky, not one thing other then the PSU was blown!
pauldriver
10-17-02, 07:55 AM
Originally posted by Since87
Having designed several switching supplies I can say with certainty that the lead length matters substantially. The main issue is the increased inductance of the connection between the gate of the MOSFET and its driver. This inductance will causing ringing in the voltage at the gate, which will cause the MOSFET to turn off and on multiple times when it is switched. Many switching designs rely on one MOSFET turning off at precisely the same time that another is turning on. (Or at least as close to the same instant as is feasible.) Unless the MOSFET's switch on and off cleanly, undesirable shoot through currents will occur, reducing efficiency.
He's right.
BladeRunner
10-17-02, 05:41 PM
Originally posted by safemode
how much was the infrared temp sensor?
That's a sweet psu mod by the way. I'd rather not see the plastic windows since the psu is now no longer em shielded but it's your computer.
We use the hand held temp sensor at work, (motorsport). They are surprizingly accurate, and have a positioning laser. All you do is Highlight with a red dot just like a sniper rifle and it reads the temp. I think they are around £60 in the UK
http://www.zerofanzone.co.uk/pictures/jpg/lazer_temp.jpg
http://www.zerofanzone.co.uk/pictures/jpg/mosfet_temp.jpg (http://www.zerofanzone.co.uk/pictures/jpg/mosfet_temp1.jpg)
The Psu doesn't have any "windows", they are open cutouts for airflow, not for perspex, and like said will have a metal mesh when I can find the right type, so will still act as a good em sheild, not that it appears to matter.
CrashGawker
10-17-02, 10:39 PM
well, I just want to start by saying, BladeRunner, I aspire to the greatness that is your ZeroFanZone. Period. I dont care what dangers I face, or the flamage I will have to endure, but I too wish to someday see a complete lackage of fans. Only problem for me is, I can't build the same type of setup as you, so I'll still have to have a rad with fans. BUT I would like to try and reduce the number of fans as far as I possibly can, and this means later attempting your WC PSU mod. So, if you could oblige a fan by continuing your updates on the project on your website, along with any new insights or progress, I would really appreciate it. Also, BTW, great job, that stuff looks unbelieveable!!!
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.