The Mini-ITX build begins.

[Lazzer408]

Well she's up and running. My TIG welding buddy is nowhere to be found so best I could do was whip up a quick bracket to attach the heatsink to the psu's 'spreader'. I'm having some thermal transfer issues though. I applied some compound between the plates but it wouldn't suprise me if one of them is warped slightly. It's not machined material it's just cold roll. I found the aluminum in my scrap bin after all. On the bright side, I have the bottom fan plugged into the psu's fan header and it's whisper quiet. I've been running Prime95 for an hour now and it hasn't spun up. The air from the heatsink is cold as is the heatsink itself and the little bracket I made is also cold although the spreader below it is warm. See pics.

 

[Chicken Patty]

WOW, great job bro. Glad it's working out. Showing some skills there buddy

 

[pantherx12]

Could you make the top bracket thing one piece of metal?

See if you can get hold of some copper.

 

[Chicken Patty]

Could you make the top bracket thing one piece of metal?

See if you can get hold of some copper.

hmm, good idea. Copper should reduce temps even further, correct?

 

[pantherx12]

Its about twice as efficient at dissipating heat

 

[Chicken Patty]

Its about twice as efficient at dissipating heat

hmm, that would be interesting. Good thinking.

 

[Lazzer408]

Copper was the next option on my list but it's expencive. I figured I'll do it out of aluminum and if it's -that- bad I can go to copper and I'll already have patterns. The bottleneck is how narrow the thermal path gets because the main caps are in the way. I really need to stop touching those.

The aluminum seems to be able to cool the mosfets but what does worry me is the coupled output inductor is HOT. Almost to hot to touch. There's little to no airflow around the other components of the supply.

 

[p_o_s_pc]

Copper was the next option on my list but it's expencive. I figured I'll do it out of aluminum and if it's -that- bad I can go to copper and I'll already have patterns. The bottleneck is how narrow the thermal path gets because the main caps are in the way. I really need to stop touching those.

The aluminum seems to be able to cool the mosfets but what does worry me is the coupled output inductor is HOT. Almost to hot to touch. There's little to no airflow around the other components of the supply.

may have to rig another fan to get air to blow across the PSU then or you could rig up water cooling for it

 

[Cuzza]

Its about twice as efficient at dissipating heat

Actually, it has about twice the thermal conductivity, which doesn't actually translate into twice the amount of heat being able to be moved. Dissipating the heat is a much more complex prospect, surface area, airflow etc must be taken into account. Also heat transfer coefficients are different, that is how well each metal releases heat into the air. And don't quote me on this but as I recall aluminium is actully better in that regard? Not sure.

Anyway, not that you were wrong, just thought I'd complicate matters.

 

[Lazzer408]

may have to rig another fan to get air to blow across the PSU then or you could rig up water cooling for it

riiiight. It would need to be external and if thats the case I might as well put an atx supply in the pump box. I think it would only be 7 wires if I used some 12ga and stuck extra caps in the case and... damn it your making me think up more ideas.

There's no more room for a fan near the psu. The best I could do in that respect is have all the other fans blowing out and drill holes in the case near the psu. If push comes to shove I'd rather just etch out a new pcb and make my own psu that custom fits the case. I really want to keep this self contained but a m-atx psu in a diy power brick would be the easiest.

Actually, it has about twice the thermal conductivity, which doesn't actually translate into twice the amount of heat being able to be moved. Dissipating the heat is a much more complex prospect, surface area, airflow etc must be taken into account. Also heat transfer coefficients are different, that is how well each metal releases heat into the air. And don't quote me on this but as I recall aluminium is actully better in that regard? Not sure.

Anyway, not that you were wrong, just thought I'd complicate matters.

My heatsink is cold as a cucumber in January. The issue would be getting the heat to the heatsink in which case copper would be better suited.

 

[Lazzer408]

This is the problem inductor. Do to the height of the main transformer the psu has to go in the case this way to clear the videocard.

Now here's another power supply I has that I hadn't looked at before. Now I'm wishing I had. With the way it's laid out the inductor is on the other end. That might make it easier to cool. Also the main caps wouldn't get in the way of the head spreader and a nice wide spreader could be used to get to the heatsink.

EDIT - The transfering channel could be increased from 1-3/8" to 2-1/4" which is just 3/8 short of being the width of the heatsink plus have the choke in a position that makes it easier to cool. Damn this is taking forever to finish.

 

[Lazzer408]

My computer in the background is louder then this thing. You can almost make out the harddrive clicking. Next time I'll take a vid in a quiet room with the tv playing low in the background or something.

http://tinypic.com/player.php?v=mx0xoh&s=4

 

[lilkiduno]

VERY nice set up Lazzer, to be honest i don't think i have the balls to take apart my PSU's...

 

[Geofrancis]

i take it lazzer you have a lot of experience with high voltage electronics?

 

[Cuzza]

My heatsink is cold as a cucumber in January. The issue would be getting the heat to the heatsink in which case copper would be better suited.

agreed!

 

[Lazzer408]

VERY nice set up Lazzer, to be honest i don't think i have the balls to take apart my PSU's...

Supplies are like anything. If you know what not to touch they can be safe to work with. There are LETHAL voltages in the psu even after you unplug them. Handle with care.

i take it lazzer you have a lot of experience with high voltage electronics?

Yes actually. Experience is -exactly- what one should have when dealing with HV. Even with the experience I have I've already had one of the psu capacitors bite me. Bad capacitor! Oh the worst was when I worked on an old tube linear amplifier for ham radio. The 400v supply had something like 2200uf worth of caps and well... You can't let go of what your holding on to until their discharged. That's DC for ya. Lucky for me I know to work one handed with HV. That one rule alone can save a life. It helps keep the electrical path out of your chest.

EDIT - Geo you reminded me to put a warning in the first post before someone starts "opening things" lol

 

[Geofrancis]

Supplies are like anything. If you know what not to touch they can be safe to work with. There are LETHAL voltages in the psu even after you unplug them. Handle with care.



Yes actually. Experience is -exactly- what one should have when dealing with HV. Even with the experience I have I've already had one of the psu capacitors bite me. Bad capacitor! Oh the worst was when I worked on an old tube linear amplifier for ham radio. The 400v supply had something like 2200uf worth of caps and well... You can't let go of what your holding on to until their discharged. That's DC for ya. Lucky for me I know to work one handed with HV. That one rule alone can save a life. It helps keep the electrical path out of your chest.

EDIT - Geo you reminded me to put a warning in the first post before someone starts "opening things" lol

i have opened a few power supplies to change fans and been zapped by capacitors a couple of times. not fun

 

[Lazzer408]

Do you guys see any disadvantage to the bottom fan blowing outwards? It could be an advantage to have all the fans blowing out. The only case vents are on the gfx side. There is another slotted opening between the side fans but I'll be blocking it so the fans don't keep recirculating the air. Doing this would force all the cool intake air to flow to the gfx card and flow over the hdd. It may help the psu as well but it's pretty tight over there.
There's even an area above the io-shield about 1/2" wide that I could put a row of holes in. That would help the cpu get a little more cool air. So far the cpu temp hasn't been an issue.

 

[Lazzer408]

Ok this is more like it. This is the 3rd power supply I'm testing. This looks like a winner. Not the location of the choke and if I reverse the psu fan it'll suck cool air right over the choke. This power supply allowed for a much wider channel to the heatsink as well. My welding buddy is still m.i.a. so I had to do with the bracket to attach the heatsink again. Just as well it's removable I guess.

I've modified this supply with schottky diodes for the 3 output rectifiers. The OEM used standard diodes. To explain what this means. The forward drop across the standard diode is about 0.7v. I'f I'm pulling 10a through it that will be 0.7v*10=7w of heat. The schottky diodes have a 0.32v forward drop which would be 0.32*10a= 3.2w. That one diode is now >50% cooler.

This power supply's mosfets and diodes are flip flopped compaired to the other psu. This put the hotter diodes closer to the heatsink. The mosfets don't put out half the heat those diodes to. That should help some. The wiring harness is also relocated. It was a slight obstruction where it was near the gpu's fan intake. gpu airflow is now improved. Every little bit helps.

Time to solder on the wires.

 

[Lazzer408]

Ok here's the first test results. The gpu is running 85c@84% fan. I have the video bios set to 100%@95c. It was 100%@108c from the factory. 85c is hotter then it was running a few days ago. I wonder if all the moving around losened the thermal compound. I'll have to check that. The psu is definitely warm. The heat just isn't making it to the heatsink (which is cold to the touch). Copper time? Heat pipes? I might make some heat pipes with freon and see how that works out.

Here's a pic and a screenshot.

EDIT - The reason for 2 furmarks was to set the afinity for each core. It doesn't quite load the cpu like prime95 would and is a more accurate representation of gaming but having the gpu maxed out.

 

[Chicken Patty]

looks good man, just gotta get the heat issue worked out

 

[pantherx12]

Wow gpu is certainly toasty!


As for heat pipes, may as-well just try copper for now, although I think even having a plate that completely covers the heat sink rather then only a little bit will help..

Sorry if my spelling or wording is wrong, not slept.

 

[Lazzer408]

Wow gpu is certainly toasty!


As for heat pipes, may as-well just try copper for now, although I think even having a plate that completely covers the heat sink rather then only a little bit will help..

Sorry if my spelling or wording is wrong, not slept.

The little plate covering the heatsink doesn't even get warm. The base on the heatsink is 1/4" thick so it should spread the heat ok once I get it there. It's a long way for heat to transfer through aluminum but it is transfering. If it wasn't the psu would have burned up by now. It's just a slow process. If it wasn't for this being switched mode, I'd just mount the mosfets right to the heatsink.

 

[Lazzer408]

What do you guys think about this solution for the AC connector? It would get a strain-relief gromet ofcorse.

 

[Cuzza]

I like the AC lead. might not be pretty, but when it's plugged in it's all the same.

I'm guessing the reason the heat isn't getting from the psu to the heatsink is that the connecting plate simply isn't making enough contact, ie it's just not big enough. Can you make one that's he same size as the two pieces it is connecting, so to cover them completely? Or maybe one long piece. Not so heavy that way. Something like this
(excuse my crappy shoop skills)

Mind you if you're going to that much trouble, perhaps it's better to just custom cut one single block. I suppose you've already thought of this but hey, i'm bored. Remember I said I lost the plot of this thread? Well I've picked it up again.