MadHacker,

I already did a check on the aluminum heat sinks and found +4.65 volts on one ... that one had a single IC that didn't have an insulation pad. I simply pulled one and a plastic screw washer from the power supply that died. (Waste not, want not... :D )

On the other heat sink there was an IC without a pad too... but it was completely incased in plastic (i.e. - no metal heat sink tab) so it's insulated already.

I will check the copper "fingers" again before they are attached to the water block.

Well I had an Antec a while back was a 400 watt I think and It was very messy inside. Someone was watercooling a truepower 550 I think recently and it still didn't look the easiest to w/c to me. Other than that I can't comment as I haven't had any others.

A lot depends on what you are trying to do. if you are water-cooling the PSU and intending to have no fans in it or near it then it can be more complex than it appears. One major thing I've learned in my fanless cooling ventures is things that don't get hot or even warm in the slightest of airflow can overheat surprisingly when the forced airflow is taken away.

As PSU's all differ in design and layout, so there's no one definate overall solution or answer, it's something that you will have to discover and cure as you go. My solution, while I accept it may not be the best way to do it, also partly cools the transformer, and I later added cooling for a hot coil and a large ceramic resistor. It still works fine though and has done fanless for over two years now, but water-cooling any PSU shouldn't be taken lightly imo.....

My PSU is quite simular in layout as MMZ_TimeLord's PSU...
since i don't have a mill... i may not make a block the same way...
copper tubing connected to a copper plate sems to work quite well...
at least well enough...
not like i'm overclocking my PSU...

have to keep my eyes open for the hot spots then...
mabe a 40mm fan undervolted to 5 volts will be enough for air flow...
i wonder if you could hear that.

Just drill alot of vent holes in to the PSU casing to get some air flow as hot air tends to rise.
Look at the picture of the watercooled PSU over at http://www.zerofanzone.co.uk
http://www.zerofanzone.co.uk/picture...su/psu_032.jpg

Great job by the way, TimeLord!

EDIT: using the [img.. thingy wasnt a good idea with such a big pic. :rolleyes:

Which components in a psu get hot? How did you test them for heat without shocking your self? take the fans out and see whats hot after 10 mins?

hmm I was just thinking. What about having a sealed copper case and submerge the psu pcb in it using a non conducting liquid and that liquid be cooled by pipes attached to the copper casing.

Basically as long as the main heat producing parts are effectively cooled, (the power fets), other parts aren't likely to seriously overheat quickly. I took the view the transformer could benefit from some cooling help so my table mount helps a little here. The coil I found getting hot later on and there was a few solutions to this, (In my PSU update article on my site). The hot ceramic resistor I found later using a infra red laser temp sensor.

There are also other variables to like what is your coolant temp? / flow rate?, and as previously mentioned each PSU design is different, what may not get hot in PSU design A with no airflow might get red hot in PSU design B with no airflow. Heat rise will work a bit, but it's the wrong way to look at it. In a similar way to heatpipes being a poor overall solution as they both require a significant heat factor to be going on somewhere for anything useful to start happening. I approached it with cooling at source in mind, (fets on a block over the coolant channel), mainly to avoid an in PSU heat up. with the other hot parts I found and cooled as I went it now runs no more than warm under gaming load with no forced airflow anywhere near it. Of course It could die tomorrow due to something I've over looked, no way of knowing for sure...........

Unfortunately this area of modding is still very much "suck it and see" after effectively cooling the obvious areas.

"MMZ>TimeLord skillfully dispatches the thread hijackers and returns to the scheduled flight plan... :D "

I got the water block for the power supply re-drilled and the fingers drilled and tapped in the proper places now. I also used the torch to heat the block and fill the "bad" holes with solder. Not perfect... but better than ugly holes in the wrong place.

Top-Back 3/4 view:

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0185.jpg

Top-Front 3/4 view:

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0186.jpg

Front:

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0187.jpg

One item of note... when I checked the "fingers" after soldering the ICs back in place, there was about 23 volts AC between the two, according to my meter.

Just to be sure, I put a 12V light bulb on it and tried to get it to light up... Nothing... not even a glow. I scratched my head. I grabbed a small LED from my parts bin... I put that across the two "fingers"... NOTHING... not even a glow. Again, I scratched my head.... :shrug: :confused: :mad: :cry: :(

I called a friend of mine who has a degree in electrical engineering... and asked him why it was showing 23V AC on my meter... but wouldn't even run a light bulb or LED.

Here was his answer... "Induced voltage, no usable amperage" ... The light bulb in my head went on at that point. :eek: :rolleyes: :dome: :p

The transformers and ICs in the power supply are mostly dealing with AC voltage and in the 115V range. So the power supply in it's normal course of operation is INDUCING a voltage on both of the fingers. Copper being the better conductor shows a higher voltage than the aluminum heat sinks (they showed only about 7V AC).

So if you are going to water cool a power supply, the heat sinks are indeed isolated and you see AC voltage on them... it's INDUCED.

Anyhoo... now just to chop up the old outer casing to permit the water block and pipes to protrude and I will have a finished water cooled 550W power supply. ( I will keep the fan and just run it at 5 volts so as not to over heat the coil and transformers. :) )

That's it for today... Cheers!

In ur PSU. You could leave the orriginal fan in and volt mod it so its inaudiable. This should giv enough air flow to any other componets in the PSU cool, as if they were cooled normally, specially as the air flow they would have got wouldn't have been particaually huge, and the chances are, the air would have been fairly warm when it got to them anys ways. Juat an idea.

Pears0,

I believe I covered this already... but here goes.

I will probably end up running the PSU fan at reduced voltage (+5V) with this mod to further reduce noise and still keep air moving.

Tomorrow I should have time to cut up the cover so I can finish the PSU up and move back to my video card solution.

Sorry, must have missed that.

Pears0,

No problem.. *HUGS* :D

JFetting,

In regards to your question in Post # 96 The pattern that I used is 4.5 inches X 4.5 inches. The same as the piece I'm covering today ... the Back block for the All-in-Wonder Radeon 9700 Pro.

Here I've started milling the passages into the curve. I mounted this on my rotary table from the get-go to make it easier. :D

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0188.jpg

Here I have the passages completed ...

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0189.jpg

Tomorrow I will drill the holes for the small cap screws and begin on the back sides of both blocks.

Cheers! :cool:

How deep are you cutting each pass?

An undervolted 80mm would be similar noise and better airflow. 40mm fans really suck.

pippin88,

Not really doing "passes" in the curved area, on the straight spots yes I did (and I took about 1/64" to 1/32" on each pass), but I found that because my mill functions like a drill press I decided to try something new (for me at least).

I drilled one hole at the beginning of the passage on the right (has to pass through to the front side) and then from there moved about 1/32 of an inch into the path and milled straight down, cutting just a little from the side of the starter hole with the endmill.

I found I could remove material MUCH faster this way without the lateral forces and risk of snapping my endmill.

In the end I think it took about the same time, but again less risk to the endmill and easier motion on me as I was not cranking on the table so much. (Just slow ¼ turns to advance the table and pull down on the handle to cut, then back up, rinse and repeat :D )

CNC language that would be called Plunge Roughing :D

Looks like you didnt cut out the whole outline?

dima_y,

Plunge Roughing... yeah ... that's the ticket. :D

I just thought it would be easier on my arm and the endmill. :shrug:

As for the whole outline, yeah, I thought maybe that one bulge was too much... so I left it. (Did the same on the front one too... )

Plunge roughing would be kinda nice, Ill be doing that with small end mills soon I think, or at least Ill give it a try.

MMZ, if you take less than 1/2 the diameter of the end mill in depth, you will be fine, so if you rusing a 1/4" end mill you should be able to take .1" without any problems. Get ahold of some carbide and youll really have fun doing that:)


Jon

Today didn't get as far along as I'd hoped... (social responsibilities ... read:wife :p )

I did get the block halves lined up, drilled and tapped for the four main cap screws... Front half...

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0190.jpg

Back half...

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0191.jpg

I should be ready to mill the back sides of each block where they sandwich the card tomorrow...

The rest of today I spent with my dremel :evilaugh: :D

I cut out the rough opening for the power supply water block to slide in...

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0192.jpg

Here is the rough fit... still need to widen the back area where the copper pipe/tube is at so it will slide in to the mounting points.

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0193.jpg

That's all for today... tomorrow we may be removing a large quantity of copper... so bring your shovel. :D :p :dome:

nice work!

Impressive stuff MMZ - keep going! Quick question; How heavy is all this lot? I have a lot of water-cooling gear in my machine (similar to how yours will be once complete) and mine weighs a ton! My blocks look like they're probably quite a lot lighter than yours so are you worried about the weight of your machine / the strain on the AGP slot etc. ?!

gr8 work , very impressive

G33k,

I'm not worried at all about the weight as PCBs can take quite a bit of weight and the video card is supported by the AGP slot, it's locking mechanisim AND the card's back plate screw.

I plan on lightening up the card too... probably by drilling holes through it in the solid areas and then soldering on the top plates last so it looks solid. Still kind of debating that, I'll see how heavy it is after I mill down the back side to accomodate the card.

Or you could cut off as much solid areas as you can, then send them to meeeee :D
does the psu still fit in your case with the tube out the side?
Amazed at how well youve made these, make me one :D

Fair enough - sounds like you've got it under control. I just thought I'd raised the issue in case you'd not thought about it :)