Project: Measure twice cut once! (56k no!)

[matt100]

OK, got a bit cocky after the (relative) success of my first wcool (in sig) so I thought I'd try something a bit more challenging, I also fancied dual core and seeing as I had some ddr2 sat around (yes the decision really was that arbitrary) and partly the price seeing as its a plaything I went for the intel 830 dual 3ghz.

Thought I'd document this, started being a laugh but its got more and more complicated as time has gone by..

Final specs are:

P4 830 dual 3ghz
1gb crucial ballistix ddr2 5400
200gb sata hdd
xfx 6800 ultra pci-e
asus p5ld2 vm mobo
aspire xqpack case

Swiftech 6000 cpu block
swiftech 350 pump
DD maze 4 gpu block
black ice extreme 120.1
5.25 bayres

I chose the aspire because of the all acrylic sides, also because it's small (perhaps too small by the looks of things??) I also think it looks pretty funky and again, it wasn't expensive, the mobo was pretty much a choice of 1 but I have asus in my main pc and it seems pretty solid, the 6800 ultra was the first card that came up in the mm that fitted the bill, I have a 2405fpw it'll have to drive so my options for gfx are a bit limited to high end stuff.. bummer eh!

It begins..

cheers ocuk for making me daydream all day at work that aprticular day after the "there are boxes here, what have you bought now" text from my long suffering wife.

 

[matt100]

Then shortly afterwards from the US of A..

The case is pretty good, and certainly well made..

and it comes apart easily too, one of the reasons I really like it (and this is possibly going to be the only thing that makes it possible) is because it has a removable mobo tray, the split level design can be a bit of a pain considering there's going to be loads of tubing in here soon

Another reason I like it is the 120mm fan exit (useful for the 120.1 mounts), you can see here the psu too, my main source of concern for it not working at all but 20A on the 12v line and 420w overall may well justabout do it..

 

[matt100]

Down to business, and it turns out the rad has to be mounted outside (first compromise, I wanted it all inside I should have realised a 120.1 wouldn't fit into a 120 sized hole) so to not have the tubes having to do too much acrobatics I wanted to point the barbs into the case, guess what?? yep, first requirement for the £10 dremel clone I got down at the local b&q!

and then, the reason for the name of the project?? yep, offered the rad up and it the barbs are fine, but you can't get tubing on them

Other problems are the fact the already small matx psu gets in the way of the gpu outlets, I think I'll get away with this but only just..

 

[matt100]

but despite all the odds, the components are in.. I just need to figure out a flow order and then somehow make it so that you can get the mobo in and out without having to drain the system... hmm.. I don't think I'm going to manage that with the bayres but I really want to be abke to look down on the pc and see water flowing about, for that reason I might go over the top with the tubing firstly for ease of removal of components should anything pop and secondly because I got an acrylic sided case to look at water and tubes!

obviously it's going to have to be filled and leak tested in the pc so it's a good job I have a spare psu so at the least I can do it with everything switched off...

More updates soon! comments welcome..

 

[matt100]

But bless him, Mat100 stayed up until at least 1:12 am just to bring us some nice info about his new system (not for his own personal pleasure).

Looking forward to you getting up/home from work Matt as it's a good looking project

lol.. I was up at 6:30 (kids).. and obviously I'm doing this for your benefit!

Question is though (and one that has been puzzling me) when it's all plumbed in, will the tray still slide out?

Even if you have the clearance, you will need at least a foot of spare tube in a coil to allow for mobo removing.

simple answer is, I can get the tray to go in and out easily, the problem is the connection to the bayres, the case is basically in 2 halfs, or decks if you like.. and once the tubing goes to the upper deck thats pretty much game over for coming out without draining.

I could do with some draining practice anyway

Truth is once the hoses are connected to the gfx card the psu will be in the way of it coming out anyway.. I could probably remove the psu, then the gfx card and lay it on the mobo and slide it out but messing like that isn't really in the plan..

It does sort of make it all one shot though..

 

[matt100]

there is a pentium m (thats what you have isn't it?) micro atx by aopen linky but apparently the heatsink they supply is a funny fitting so you can't buy waterblocks for it schoolboy error imho. I think Pm makes perfect sense in this case..

Still don't know if a 120.1 is going to cope with the heat of a 6800U and 830D.

 

[matt100]

OK, did a bit more work tonight...

Its all ready pretty much for fill and test tomorrow

I was bored at work this morning and whilst daydreaming I realised there's a semi conveniently placed hole in the mobo tray almost directly under the cpu, so I got to thinking, enlarge the hole so you can see the whole underneath of the cpu, then cut a hole in the bottom of the case in the same place, then I can release the cpu block with the mobo in, the gpu will release anyway and presto, the mobo tray is out leaving the watercooling intact.

ta da..

it also means I can loop the tubing anywhere I like because I don't have to worry about getting it back out in a hurry..

So now its all in and the psu is connected up etc, its quite a squeeze!

the area between the psu and res is where the pipes are going to cross so I figured that is where the psu cables can bunch up seeing as you won't see them..

 

[matt100]

And its funny what you learn from cocking something up, I flew headlong into my last pc and ended up with a mess of cables and tubes totally the wrong length so this time I've tidied as I've installed.. Its difficult to do it as well as you can in a full tower but I think its ok..

incedentally, can you see the bend in the mobo underneath the cpu? anyone got any suggestions for what this might be? I've tried loosening the supports for the cpu waterblock with no effect, could it be the placement of the mobo mounting points? I seem to remember having this problem with my current machine with an asus star ice before (and because) I went for water (dreadful HS)..

and finally I had the sense to measure out and cut the tubing now, and I've even marked them for where they go! This is pretty radical for me.. my levels of organisation leave much to be desired..

so, tomorrow with any luck it'll get connected up and filled, then its overnight leak testing by which time the kvm should have turned up and we're off.. straight to 4ghz with any luck.

 

[matt100]

Nice diagram neon!

I did say I've tried loosening it, I'll try loosening it a lot more to see if it makes any difference.

I'm guessing the cpu isn't bending so it might not be such an issue.. Like I say it did it with my old star ice too but I assumed that was something to do with the stack cool on my p5ad2e mobo.

Experiments tonight!

can u get a sticky pad standoff under where the board is bending ?

you could even drill the backplate and make a mount out of a small nut and bolt attached to a hollow piece of plastic, obviously leaving a step inside it so the bolt will be able to hold it down.

what do you mean a sticky pad standoff? If you mean under the mobo there's nothing there (see earlier post) explain more please!

 

[matt100]

oh I get you, you mean like a prop? I could try that but it's not sagging, its bending under pressure I think as per the diagram above..

I'll play tonight and see what I can do..

As an aside, do we think I could get this thing into the bios with no water etc? Just using the waterblock as a big passive heatsink for literally 30 seconds just to make sure it boots before I make the working space even more difficult?

EDIT: would you not think the waterblock should just screw straight down on the cpu? its obviously going in at an angle if its causing the mobo to bend which is a bit of a pain..

Also, the cpu won't be bending so perhaps its still making a decent connection with the block?

 

[matt100]

I still find it hard to believe its sagging, I could understand it if was something like the leverage created with a star ice on a stood up mobo (eg in an atx case) but this is horizontal and whilst the waterblock is heavy it has a very low centre of gravity and no leverage..

I still think the tightening of the rods that hold it in place as created a pinching effect and bending the mobo??

 

[matt100]

the plot thickens, I sussed the bending mobo, if you look at the photo you can see the waterblock flat on 3 capacitors (is that what they are?) that was what was causing it to bend because they are higher than the cpu and they were being pushed down

took the waterblock off and it was only touching the cpu by a tiny bit (shown by tiny smudge of as5!

reorientated it so the step out that is supposed to go over the cpu bracket lever (according to swiftech) is now over these bits and it looks fine..

Onwards tonight!

 

[matt100]

lol.. ok ok!

I totally flaked out last night, fast asleep at 10pm! I must be working too hard.. right, now, serious time...

Just about to disconnect my current pc to bios test the new one.. its like going to the dark side of the moon, no comms for a bit!

With any luck I'll be running round the kitchen like some sort of demented chemist with a litre of bright green deionised water in an hour or so.. fingers crossed! See you on the other side...

EDIT: bugger me that was quick, straight into the bios no messing, got to do something aout the fan because it can't spin up against the fan guard built into the case.. other than that very good..

Waterblock got very hot, checked the cpu temp in bios.. 50-52-55-58-60 then I turned it off! lol, I guess seeing as the block got hot there must be a good connection?

 

[matt100]

good news: its all in, pump working and looking good

bad news: bad leak at the rad and I really don't want to be dreaming up solutions at 1am!

anyway, new barb tomorrow along with wet matt

EDITics..

as you can see still major bend on the mobo, I'm waiting to test the temps live to see if this makes any difference, I've stuck a small torch behind the cpu and it appears to be making full contact and as I said before the brief boot into bios made the waterblock very hot.

 

[matt100]

and here it is bleeding, with me oblivious to the rad leak (because it leaked by the res where you can see the tissue, the res in my current machine leaked in exactly the same place so I was feeling all smug about noticing and dealing with it only to find my kitchen top swamped with hydrx and deionised water from the rad

never occured to me how hard it would be to bleed considering the layout of the tubing/pump and res.. it was a nightmare! I filled the res and turned the pump on and.... nothing! there was air in the tubes stopping the water getting to the pump and it went on from there!

ANYONE WITH ANY HELPFUL TIPS ON STRESS FREE BARB REPLACEMENT PLEASE POST ASAP!

 

[matt100]

yes I've got ptfe on the threads. The leak is actualy from the thread on the barb into the radiator... I investigated a bit more last night and the barb won't tighten, it gets so far and then goes slack again..

I left it overnight and the leak is minescule but I'm going to change the barb tonight and try again.

 

[matt100]

I've got a bunch at home so I'll find something that'll fit.Got to be done tonight because the thing is currently stricken on the kitchen worktop so my missus ain't too happy

 

[matt100]

nearly there!

Pulled the barb off and stuck another on, I was tempted with a metal one I have here but I don't know what metal it is and I've gone to lengths not to mix metals so plastic it was.

Good god the water was flying everywhere! There wasn't enough tube to get the 2 ends to the highest parts of the loop so it was just flowing out of the rad... very traumatic but in the end the new one went on and all appears to be well again (touch wood) it's had the pump on bleeding for about an hour and no sign of any leaks..

The mess!

But like I said, it's nearly there! Just realised I need to cut away some of the shell that dremel is earning its keep here.. also need to cut away a bit more of the backplate because the hoseclip is in the way...

Then I get the case on the only thing left is installing some lighting.. I fancy getting some 4" uv cathodes installed actually on the shell (in the bends where it's metal so you can't see them) just got to figure some way of wiring them up without having wires trailing everywhere when the case is shut.

Lookee though, I'm getting excited!

 

[matt100]

Bone dry all night yeah!!!

Got to get to the shed to cut it, might not be feasible tonight.. especially if its raining (it's 150ft long, a total mess, on a hill with no lights=deathtrap) but I will! soon!

 

[matt100]

argh ffs!

right, all wired up, turned it on, POSTed fine, went into bios.. temps holding at 32-35 very good..

The I went into the kitchen (to get a beer anticipating a prolonged wxp64 install) came back not 30 seconds later to find it rebooting.. permanently.

Sounds like the power is cutting out, the fan dims and then it all comes back on again, sometimes gets through the POST beep, sometimes doesn't..

Any ideas???? Not experienced this one before, so far I've..

Disconnected the reset switch, connected/disconected a fan into the cpu header, touched the backside (oo-er) of both the gpu and cpu to make sure they were overheating with no contact to the wblock and both are icy cold..

Any ideas???