Tempratures climbing after addition of 6800 NV-68

[tobyed2]

Finally got my Danger Den NV-68 to stick on my BFG 6800 Ultra (I thought I was gonna go deaf !).

Rerouted a couple of pipes to make the layout cleaner (I thought ...). Ran system overnight to clear bubbles, now running clear.

I was running at 218x12.5 @ 1.9vCore 45C load, 32C idle. This worked fine for my old card (a Radeon 8500, yeah bless ...). But putting the 6800 in with the stock fan I had to clock back to 208 just to boot , hence the waterblock.

Turning on the system this morning (no didnt get much sleep), went straight into BIOS health. Temp was 32C idle, this climbed (praticaly every second) to 45C before the system froze. Turned PC off. Checked mounting screws on waterblocks. Turned PC on, temperature climbed again. Suspecting to much heat set CPU settings back to 12x200@1.6v, again watched temp and held at 34C. Booted into XP but quickly BSODed.

I'm gonna increase vCore a bit to see if I can get into XP but if thats the case there must be something wrong with my setup.

I've attached a pic of the setup after the 6800 reroute.

The order is pump - rad - CPU - NB/GPU - RES

Any ideas
:shrug:

[Etacovda]

Reseat the cpu wb?

[MaxxxRacer]

sounds like something is wrong other than cooling. if you say it got to 45c at load before and didnt crash it shouldnt crash at 45c at idle either. see what i mean.

you said that even when the 6800 was not in the cooling loop u had to back down to 208 on ur fsb. It seems you have some sort of compatiblity issue with your hardware. the 6800 doesnt seeem to like the higher fsb on your mobo for some reason.

About your temps being higher. I would reseat the cpu as etacovda said. sounds like u got issues there. but another culprit could be the small rad and high restriction system you got. But you system should not crash at those temps thought. I was running my xp2600 at 1.95v and 2.8ghz and it was doing fine. just as long as it was below 120f. granted it didnt ever get to 120f unless i turned my rad fans down to nothing... but ...

[brucoman]

look for a weak power supply first
seems you are losing water flow as well, reroute all flow in series, not parallel

[killernoodle]

Looks like he has a truepower, I dont see how big though.

[tobyed2]

Sorry, I should have a sig ...

Ah, well. Its an Antec 550w PSU.

System boots at 12x200 1.75vCore idle 44C.

I will modify to a parallel system, I guess
RBX (has to be first ...)/NV68/NB/RES/RAD/PUMP ?

I'll use the Y spitter to join the two pipes out of the RBX into 1 before hitting the NV68. At the same time I'll reseat the CPU (again !).

The system "did" work great with the Maze4 GPU. I guess the NV68 must be more restrictive or is dumping more heat into the system.

Maxxxracer - Just in laymans terms, whats making the system restrictive ? Is it the RBX block ?

Theres a pic in the Gallerys section of a system with the same blocks (except the NV68) and it uses the same route I have at the mo (NB/GPU in parallel) but the only addition is its a 2 pump 2 rad system (page 7 I think ...).

[tobyed2]

Ah, there we go, NOW we have a sig

[redleader]

That new GPU block is going to be 10x as restrictive as that tiny chipset block. You're probably not pushing any flow through the GPU and its overheating and crashing windows.

[ferdb]

redleader is right, the 6800 puts out an enormous amount of heat, probably more than your CPU does if you are using a mobile athlon. I wouldn't put anything in parallel with that GPU block. You need the full water flow going through that GPU block.

[MaxxxRacer]

Well hte TDX is incredibly restrictive in its own right (but i love mine) , and the nv68 block has alot of channeling in that thing, so its going to add some resistance. the 90 degree turns it makes arnt the best either.

Here are my suggestions redleader. First off get rid of the go forsaken NB block. there is NO need for it. if u want good cooling on it get a zalman fanless cooler. if there is some case flow then it will stay nice and frosty. Seconldy, have the 2 exhust lines from the rbx come toghether before they hit the nv68block. Third: set up ur sytem like this
Pump > CPU > NV68 > RAD > PUMP
Having the rad after the pump (before it hits the cpu) is not doing you any favors. its just killing the pressure that the rbx is getting. and if you think its keeping the water a bit cooler before it gets to the cpu.. well its not measurable... the differneces are in the 0.XC.

As far as power consumption. a barton at 2.4ghz consumes around 90-110 watts of power at load. for reference a xp3200 consumes 76 watts and a xp3000consumes 64watts.

[Blackeagle]

Yep, dump the chipset block. That 6800 block needs all the flow it can get.

[Althornin]

Originally Posted by MaxxxRacer

Having the rad after the pump (before it hits the cpu) is not doing you any favors. its just killing the pressure that the rbx is getting.

um, no.
It doesn't work that way.

[MaxxxRacer]

actually it does... its physics....

think of it this way.. is there more pressure when there is 10 feet of tubing or 1 feet of tubing. 1 foot of tubing of course.. that means that the closer the head is to the pump the more pressure it gets. its quite simple. By the time the water gets to my res (my system is pump cpu gpu rad res) there is less pressure (force) than if it is in the cpu which is only 9 inches of tubing from the impeller of the pump.

Another way to think aobut it. what happens after the water goes through the cpu block. In my case the tdx. It slows down like a mother F. thats becasue of the backpressure that is caused in the tdx. well the RAD will do the same thing. just not as bad as its not as restrictive as my tdx with the insert im using.

I hope that clears things up Althornin.

And dont try to pull the "closed loop" stuff on me. I know that things arnt exactly the same in a close loop, but physics doenst get thrown out the window in a closed loop. One way to test htat is to put ur cpu block before the pump and have the pump suckwater from a big jug of water. and then do the same test with teh cpu block after the pump. with the cpu block after the pump you will see higher flow rates.

[Althornin]

Originally Posted by MaxxxRacer

actually it does... its physics....

think of it this way.. is there more pressure when there is 10 feet of tubing or 1 feet of tubing. 1 foot of tubing of course.. that means that the closer the head is to the pump the more pressure it gets. its quite simple. By the time the water gets to my res (my system is pump cpu gpu rad res) there is less pressure (force) than if it is in the cpu which is only 9 inches of tubing from the impeller of the pump.

Another way to think aobut it. what happens after the water goes through the cpu block. In my case the tdx. It slows down like a mother F. thats becasue of the backpressure that is caused in the tdx. well the RAD will do the same thing. just not as bad as its not as restrictive as my tdx with the insert im using.

I hope that clears things up Althornin.

And dont try to pull the "closed loop" stuff on me. I know that things arnt exactly the same in a close loop, but physics doenst get thrown out the window in a closed loop. One way to test htat is to put ur cpu block before the pump and have the pump suckwater from a big jug of water. and then do the same test with teh cpu block after the pump. with the cpu block after the pump you will see higher flow rates.

lol.
Sorry, but you are totally wrong.

[MaxxxRacer]

if u think so.. prove it....

[redleader]

Originally Posted by MaxxxRacer

if u think so.. prove it....

No you're completely wrong. Pressure drops are cumulative, and depend only on the componets and flow rate.

Since the flow rate is constant in a series loop, and the construction of a block is fixed, the pressure drop doesn't depend on order.

think of it this way.. is there more pressure when there is 10 feet of tubing or 1 feet of tubing. 1 foot of tubing of course.. that means that the closer the head is to the pump the more pressure it gets. its quite simple. By the time the water gets to my res (my system is pump cpu gpu rad res) there is less pressure (force) than if it is in the cpu which is only 9 inches of tubing from the impeller of the pump.

You're confuseing pressure drop with pressure generated.

Another way to think aobut it. what happens after the water goes through the cpu block. In my case the tdx. It slows down like a mother F. thats becasue of the backpressure that is caused in the tdx. well the RAD will do the same thing. just not as bad as its not as restrictive as my tdx with the insert im using.

No it doesn't slow down. Flowrate entering and exiting a componet is equal, or the part is leaking.

I hope that clears things up Althornin.

And dont try to pull the "closed loop" stuff on me. I know that things arnt exactly the same in a close loop, but physics doenst get thrown out the window in a closed loop. One way to test htat is to put ur cpu block before the pump and have the pump suckwater from a big jug of water. and then do the same test with teh cpu block after the pump. with the cpu block after the pump you will see higher flow rates.

Ugh if you're going to be pretentious and incorrect, at least use proper grammer when you do it.

[MaxxxRacer]

I know that pressure drop is cumulative. Hence the pressure going into the first component is higher than the next component. Makes sense doesnt it. If this isnt true then please explain because as far as I know that is the way physics operates.

I know that flow rate is constant going in and out of the block, and in the entire loop itself.

The pressure of the water going into the pump is much much less than the water being pushed out of the pump. For this reason the pressure in the loop is "renewed" (spelling) as it exits the pump.

At a certain point of a closed loop there is less backpressure and more negative pressure (vacuum pressure) pushing the water through the loop. At what point that is, is entirely dependent on the lenght of the loop and the pump as well as components in loop.. But still there is this point. For myself I dont reach this point as the head on my pump is 14ft and the water in the loop doesnt travel 14 (atleast i dont think it does) feet till it reaches the intake on the pump. Anyway the point of this is that in a closed loop like our water cooling loops the pressure does drop....

[tobyed2]

Well, heres the weekends efforts ...

Changing to blocks in series PUMP->RAD->CPU->GPU->NB->RES. I've read the tail end of this thread and I will see what difference removing the NB has, as now the system is working I can see what I can overclock to and compare.

First off changes resulted in 90cm less tubing, wow ... Also, reseated the CPU.

Went into BIOS PC Health, temps started at 19C (geez thats the lowest I've ever seen it), though short lived this soon climbs and settles at 30C. Under load (ran 3dmark03 and Aquamark several times), load is 44C.

So far gone through 12x settings up to my memory max of 217 (anything higher and my memory cr@ps out). This I couldnt even do with the stock cooler on the 6800 !! Starting on 12.5x, currently at 210mhz FSB = 2613 @ 1.9v temps still 30C idle 44C load. Amazing ....

Something I really dont understand. Overclokced the 6800 first of all, the clock and mem maxes at 416mhz/1.27ghz temp under load 62C, 3DMark03/3DMark01 both consistently get higher scores. Aquamark seems to pluck a figure randomly between 58301 and 59370. When I then started overclocking the CPU changing to 12.5x210 (from 12x217) the Aquamark score dropped 1,000 points (12.5x210 faster CPU setitng) !!

I've attached a pic, excuse the technical solution to stopping the card from warping (the weight of the NV68+tubing from the NB warped the card, surprised the card still works !!!).

Lastly, something I noticed while retubing. The old tubes have a cloudy inner residue, sticking a cloth up the end removes some of it but just wondering what causes it ? Is it normal ??? (In the attached pic the RAD to CPU pipe is the oldest and the NB to RES is the newest (which is completely clear). I have had the setup a year, should I change the water ?

[nikhsub1]

Much better plumbing tobyed2. NO surprise about the aquamark, higher FSB will usually result in better benchies, even if the clock speed of the CPU is a bit lower

[MaxxxRacer]

The tubing is much better. But still try it without the NB cooler. It will put less heat into the system and make ur tubing simpler. I know it sounds liek a waste to get rid of the cooler, but you will thank us later.

That 6800 seeems way way to hot. 62C under load with WC. that is nuts. if those numbers are right its possible that got mounted incorrectly as well. Becasue that is the stock temps as far as i know.

the white residue is normal if you dont use any water addative. If you put in some zerex racing cooling (u can get it at auto parts stores) you wont get that any more. I highly suggest you get some. If you odnt you will end up getting corrosion in your system.

[JamesAvery22]

Originally Posted by MaxxxRacer

...
the white residue is normal if you dont use any water addative. If you put in some zerex racing cooling (u can get it at auto parts stores) you wont get that any more. I highly suggest you get some. If you odnt you will end up getting corrosion in your system.

From his first pic either he was already using hydrx or just green highlighter

[brucoman]

62c on the 6800? my 6800 GT at 410/1.1 with the same waterblock runs at 49c during 3Dmark03... perhaps need more flow still, try without the NB block, better yet get a second pump in series

[redleader]

Originally Posted by MaxxxRacer

I know that pressure drop is cumulative. Hence the pressure going into the first component is higher than the next component. Makes sense doesnt it. If this isnt true then please explain because as far as I know that is the way physics operates.

I know that flow rate is constant going in and out of the block, and in the entire loop itself.

The pressure of the water going into the pump is much much less than the water being pushed out of the pump. For this reason the pressure in the loop is "renewed" (spelling) as it exits the pump.

At a certain point of a closed loop there is less backpressure and more negative pressure (vacuum pressure) pushing the water through the loop. At what point that is, is entirely dependent on the lenght of the loop and the pump as well as components in loop.. But still there is this point. For myself I dont reach this point as the head on my pump is 14ft and the water in the loop doesnt travel 14 (atleast i dont think it does) feet till it reaches the intake on the pump. Anyway the point of this is that in a closed loop like our water cooling loops the pressure does drop....

First, the head of a pump is a measure of the pressure across it, not the distance it can pump. If you're getting confused by that, just convert the head to PSI or Torr or Bars or some other unit that you're comfortable with.

Second, all that matters is the difference in pressure. there is no absolute pressure in the sense you seem to think. The way our systems work is that we have a difference in pressure that causes flow. So we're only interested in the difference, not absolute pressure. You could run the entire system under 50 feet of head (relative to sealevel) with just a tiny Eheim 1048 and its not going to change anything (well water compresses a little, but thats not important). Do you understand why this is?

Finally once you realize that absolute pressure is meaningless so long as it doesn't induce a change in the fluid, its obvious why the location of the componets doesn't matter. You can get a difference in pressure from negative feet of head just as easily as positive ones . . .

[MaxxxRacer]

Head is the height that the water can be pumped to vertically. pump with 26feet of head can pump up to the exit point of hte tubing 26 feet above the height that hte pump is at... obvioulsy it can pump more than 26 feet if the tubing is at the same height relative to the pump. it could pump much much furter.. but stil there is a point when the resistance of the water in the tube will cuase the pump to stop flowing even if the tubing is horizontal. this is just because the water would be so heavy that the pump could not start moving it.

by absolute pressure do you mean the pressure at a given point???

not that this maters but water isnt compressable. if you take water that is 10 thousand feet below sea level (not that there are any places that deep i think) and bring it to sea level it wont get any "bigger". But this doesnt apply to water cooling.

[Voodoo Rufus]

Let me see if I can clarify or not.

For a given flow rate, you will have a pressure drop from the inlet to the outlet of the block. The pressure drop is relative to where you place the measuring points. It doesn't matter where in the loop you place the block, you will get the same pressure drop across those two points. You get a pressure difference across the inlet and outlet of a pump. Without it you get no flow.

Or like an individual cell in a battery. It doesn't matter where you place the single cell in the pack, if you put a voltmeter across the terminals it will measure the same voltage.