Swiftech MCW30 Problem

[meths]

I'd originally posted in General Hardware but this has become a bit of a cooling problem instead.

I installed a MCW30 on the NB of my DFI Infinity 975x and the computer wouldn't boot. Turned out that the hoses were tipping the block onto the small chips or capacitors (whatever the are) surrounding the NB chip. I shortened a couple of the hoses to pull the block as upright as possible and now the computer boots fine.

Trouble is that now I can't get enough downward pressure on the NB to cool it effectively and it's idling at around 50c. If I push down on the block the temps drop to about 36c.

The mobo only has two diagonally opposite loops to attach one spring clip for the block which just doesn't do the job well enough.

I'd thought about drilling a couple of holes through the mobo so I could use the hard mount kit but the tracks of the PCB are in the way of where the screws would need to go.

Has anybody managed to successfully mount one of these blocks on a similar mobo or does anyone have any idea on how I can improve the contact with the NB?

 

[Bubo]

Photo of the situation will help massively, but I would hazard a guess that the solution will be to do a proper bolt mod with a bar going across the block, rather than a spring, so you can apply a greater clampinf force. Also drilling holes in the mobo is a bad idea, even if it appears there are no tracks on the visible sides of the mobo. As I understand it mobos are made of multiple layers of tracks, so there could be some which are not visible below an apparent "free" area.

 

[meths]

Photo of the situation will help massively, but I would hazard a guess that the solution will be to do a proper bolt mod with a bar going across the block, rather than a spring, so you can apply a greater clampinf force. Also drilling holes in the mobo is a bad idea, even if it appears there are no tracks on the visible sides of the mobo. As I understand it mobos are made of multiple layers of tracks, so there could be some which are not visible below an apparent "free" area.

Point taken about the drilling.

I'll try and post a pic tonight.

Edit: Here's a pic of the mobo showing the stock NB heatsink. Don't think this fastening method allows a bolt mod from what I've read.

 

[bifday2k]

Have you also tried this, "NB/SB Core Protective Foam Pad" as it should help protect the stuff around the chip, and allow abit more pressure without going to far.

Btw sorry i can't link directly to it as its from a competitors site, but its the first link that pops up in google.

 

[meths]

Have you also tried this, "NB/SB Core Protective Foam Pad" as it should help protect the stuff around the chip, and allow abit more pressure without going to far.

Btw sorry i can't link directly to it as its from a competitors site, but its the first link that pops up in google.

Thanks, but that'll lift the block even higher. My problem now is that I've got nothing to press it down onto the chip harder.

I think I need a stronger wire clip than the one that's supplied with the MCW30.

 

[bifday2k]

ah good point, mmm think i read somewhere that someone had this kind of problem and they sorted it by using epoxy stuff to stick it on firmly, kinda long term option though, hopefully someone will be able to help you.

 

[Bubo]

Ah, from what you describe it sounds like you are not using the mounting plate on the MCW30 at all, just threading that tiny spring through the middle of the block? That is a seriously naff board design to be frank. You could probably make a frame to clamp to the mobo, using those two loops to provide the opposing force, then have a couple of drilled and tapped holes in the frame so you can use the mounting plate on the block properly, but its not going to be easy. Another thing you could maybe do is thread some wire through the loops and the holes on the mounting plate and twist until it a tightened down, this will be very tricky to get even pressure both sides however. Or you could see if you can generate more force with the spring as is by putting spacers between it and the block so that the spring has to deform more.

 

[bifday2k]

Just a thought from what Bubo said, his added wire option, could you possible push down (obviously not to hard) on the water block so that the pressure is there, whilst you tighten the wire, as that way you would be giving it even pressure, and all u have to do then is twist and twist untill it was tight.

 

[meths]

Okay here's what I'm dealing with.


The crappy wire clip from both angles...

The overall loop...

TBH I don't think there's much I can do with it and at the moment with the side off the case my northbridge is sitting at 46C according to Everest.

I'm thinking I'll have to just remove it from the loop as it's only working as well as my Thermalright heatsink did so it's not worth the extra heat it's leaving in my processor.

 

[Bubo]

Your one and only option given that setup is to try and make that spring generate more force by forcing it to deform more than it is at the moment. The simplest way looking at your pic would be to put two spacers either side of the block resting on top of the caphead allen screws, so that the spacer lies perpendicular to the bit of the spring which is going through the middle of the block, the spring would go over the spacers. You will have to play around with different thickness spacers to test how much extra deflection you think you can get away with without breaking the spring. If you can do this then it would also help to resist the turning over moment caused by the tubes, which is you main problem currently, as the spring is acting as a point contact as is. It is debatable whether this situation can be rescued however as it simply is not a mount which is suitable for any water block. I would start saving up for you next cpu/mobo upgrade if you want to make full use of that block.

edit:

And another thing, it looks like the NB on that board does not have a heat spreader and therefore the block is sitting on the relatively small area NB core? Its no wonder the block was pivoting on it when you attatched the tubes. If this is the case then I believe this is a total non starter whatever you do, you would need to generate a heck of allot of donwards force to overcome the turning moment. Given the mounting mechanism this doesn't look like it is possible, even with the spacer idea.

 

[meths]

You're right there's no heat spreader on the block but I've all but eliminated the turning movement by cutting off parts of the hoses that were causing the problem. I think at this point all I need is something to make it shove down.

I've given up hope of finding a solution involving the clip as I did a test on the other one you get in the box and it snapped without putting that much force on it.

I'm just gonna use it as it is just now and start pushing my overclock back up but I'll probably replace it with a heatsink when I can be bothered.

 

[MikeTimbers]

If the block is sitting too high and the chip it's cooling is small, you could add a cold-plate (basically a slab of copper). I did this once with a GoldenOrb where I found an early 1p piece from when they were still made of copper and lapped it smooth (technically a criminal offence I believe!), TIM on both sides and hey presto! more pressure, better cooling.

 

[meths]

If the block is sitting too high and the chip it's cooling is small, you could add a cold-plate (basically a slab of copper). I did this once with a GoldenOrb where I found an early 1p piece from when they were still made of copper and lapped it smooth (technically a criminal offence I believe!), TIM on both sides and hey presto! more pressure, better cooling.

That's a great idea.

A small 1 or 2 millimeter thick copper sheet would probably do the trick.

Never thought of trying to fill the gap between the block and the chip.