The Sound of Silence - Everything Water cooled - Update Feb 23

looking special so far, keep going

 

You are very welcome. Just please just give me an honerary mention in your next update.

Looking forward to seeing your implementation of water cooled RAM

 

i have a concern over your parallel cooling circuits.

Basically i think they are fundamentally flawed!!!!!!

The water will always take the route of least resistance around the system and so the parallel route with the least resistance with have the majoirty of the flow through it. Looking at your tubing diagram i would say that is going to the through the CPU, which isnt a bad thing. However, this wil cause a distinct lack of flow through the rest of the system. I am especially concerned that the GPU, NB and mosfets are all on the same circuit, there is quite a lot of heat all being cooled by the circuit which looks to have the most resistance and will therefore recieve the lowest (if any) flow!!!!!! You seem to have a good grasp on electronics and circuit design for them, think of the water system in those terms with the water being the current and the length of tubing/route (sharpe turns etc) as the resistors, and the pump being the voltage. That should allow you to see where im coming from in terms of heading towards a major meltdown.

I think you may need to take a step back and consider either a second complete circuit (including pump) and have them both run serially through the componants, i guess they could have a common rad, although with the amount of cooling you are doing i would like to see a 3X120mm one in there.

Im guessing with your plans to run the system into your houses cooling system you will be able to use a pump external to the computer to get a large flow and use the internal one as a boost, however if you decide to run the unit as a stand alone then i very much doubt the single rad will cut it.

Sorry to sound so negative, but i thought i'd lend my fluid mechanics mind to the blend.

I do really like the work you have done to allow water cooling to all the componants, i just think a little more thought is needed as to how to run the flow throught them.

 

I too will have parallell loops in my watercooling system, and I plan to first connect one flow indicator to eatch loop, and then connect a flow regulator to the loop with all-/most of the flow.

btw: anyone know where I can get or how I can make two simple but identical flow indicators?

 

I just read your log in great detail and the idea od having regulators is much more sound.

I am also well impressed by your flow paths and the external big rad

This project will just take a bit more thinking in my opinion.

 

Thoughts about the waterflow and cooling ability

Yes, I have some of the same concerns, and I thank you for bringing them up.

I have thought about them, and have so far just taken a very simple approach to dealing with them, but I will deffinately make short tests first, when I turn on the whole che-bang, to make sure I do not fry any of the chips.

These are the thoughts I have had so far for the tubing:

1) By making the tubes through the pump/rad larger than the rest of the system, I hope to achieve more pressure/flow than will naturally be needed by the thinner CPU Loop.
I have placed the CPU feed tube right opposite the Pump inlet in the spliter box. Also the CPU return is right opposite the outlet/suction from the Radiator loop.

2) I have made a rough estimate of how many watts to expect from each of the different elements that needs to be cooled.

I know then entire PC is using ~110W (meassured at 220V socket)

CPU 56 Watts max.
PSU app. 30% of 110W = 36W

36+56 = 92 which leaves only 18W for the rest of the system
If I guestimate that the PSU only burns 18W, then the rest of the system uses 36W

The specs on the HDs say - Full load: 2x (12V@0.8A=~10W + 5V@0.7A=~4W) = 14W
At Idle/normal load they probably use half or less: This is 5-7W per HD/DVD
I have four units which ends up at 20-36W

If the HD/DVD used 36W that would leave nothing for the NorthBridge, VGA, RAM etc. so the HD/DVD probably only uses half which is 18W

This would leave 18W for NB, VGA, RAM etc.

When just air cooled. some Motherboards come with passive cooling for the NB, and NO cooling for the MOSFETs.
This tells me that they use very little power compared to the CPU.
The size of any vented cooling block for the NorthBridge, is somewhat like what was used on early 486's which used app. 10W at the most.​

But these are just some of the thoughts so far.
The LAB test will show if I am way off the mark.


Also I will be making quite a few temp messurements when I fire up the whole system. And I will only fire it up for a few minutes at a time, until I feel sure that I have not made a serious bluder.

During normal operations I plan to use 8-16 temp probes to monitor the system.
The last two years I have been using the 2x MAX1668 temp chips. They each have one internal and four external temp sensors.
The MAX1668 can be directly connected to the SMBus of a motherboard (Intels version of the I2C bus running at max. 100Kbps)

So I will definately be watching to see if my primitive splitter block will be enough to ensure adequate waterflow through all the cooling loops.

I have been looking at this flow indicator.

StudieData in Copenhagen has this
But I find it a bit prizy.... and they don't have it in stock right now.

I do however like that it gives both a visual AND a digital indication of the flow speed.

I have also considered this sensor

Studiedata ID

And I have been thinking of maybe making my own flow indicators.

But what I REALLY want to know about, is the temp of the different units in the system, as that is the main parameter anyway.

 

It should be very easy to make a simple flow indicator with a peace of clear tube, some fishing line and a twisted peace of plastic.. the problem will be to make them behave identically to make them usefull in comparing the flow in the different loops.

hm.. i guess you could make as many as you need and mount them in a series first.. to compare them to eachother on the same loop with the same flow. If I could only find a way to make them so that I could tune them... hmm...

 

Keep thinking, as I think you are on the right track.

Here is another thought:
One way I have thought of to have the digital signal is to use a LED and a photo trasistor. Maybe IR diode and IR sensor.
I have used this for making the speed indicator of a small robot car. It workes quite well. Anything rotating that will cut the light beam, will work.
And the LED could probably shine through the PVC tube.

For now I will however keep working on connecting all the tubes, so I can start testing the system, and only if it is not under control will I spend time/enegry adding flow indicators.

 

Good to hear you have been thinking of the things i mentioned. I will be following this to see how the system and flows turn out

 

Please...Tell me

Who is the maker and what model is the case? It looks so much like two I bought from a supplier in the US. It has no name on the box it came in and nothing on the case itself. Nope, no paperwork was in the box either.
Please and thank you...

 

Btw the abit vt7 mobo is a gr8 mobo for your scopes as it is a cool running board and the chipset even tho it has a heatsink its on there just for looks and not for any real cooling.

I was reserching that mobo as a buddy of myne has it and he is very happy with it.

good work btw

subscribed

 

Actually I am not quite sure, as it was originaly given to me. I used to think it was an AOpen case, but I have gone through the AOpen cases museum, and have not been able to find it.

The power supply has a Q-Tec marker, and as far as I remember it is the original power supply that came with the case. So it is probably a Q-Tec case. Q-Tecs web site does however only show cases being produced, and it is not on that list.

This is probably as close as I can get with a producer/name of the case.
Hope this is some/any help.

 

Thanks! Good to know. I expected good things from Abit, but it's very nice to hear from others with some live experience, and that it performs well for them.

 

Always nice to have someone giving one's ideas some serious thought. Two brains DOES think better than one.

I'll keep you all posted as to the progress. And testing!

 

flow regulation

i just had an idea as to how you can make a simple and cheap flow regulator.

Take a short price of your tubing, and put a screw into it. Adjusting the screw in/out with increase/reduce the resistance due to a turbulent flow being setup by the irregular shape of the screw int the regularly shaped tube. It sounds crude, but with your flow indicators it would allow you to get a decent flow in all loops and also enable you to fine tune things to get the balance of flows correct for the amount of heat discipatin needed.

 

thats sounded very crude indeed. A better solution might be to start every loop with a peace of copper/brass pipe, drill a hole in it and make.. um.. dont know what its called in english.. reversed screw tracks.. in the hole, and put a screw in it. I'm not sure if this will be able to keep completely sealed though. (You could actually just use the nipple to drill a hole in I guess).

 

Well, that probably would work well.
It does however have the disadvantage of breaking the watertight surface of the PVC tube.
Another way would be to just squeeze the pvc tube. It too will create resistance and cause less flow.

 

i was thinking more of a piece of that alu pipe you are using, as then you can adjust everything to work riight then seal the screws with teflon tape, or even just solder it watertight.

 

ah, I guess we were thinking the same thought then I thought you ment putting a screw in a pvc tube.. which I think would most probably cause a leak

 

Okay, that migt be an idea. I'll give it some more thought, if it turnes out that I need to limit flow in one or more of the loops.