Fuzion Nozzle Kit Thermal Performance Link

[Martinm210]

This is not mine, but I thought it was interesting and a good example of another way to do thermal testing that includes a few more variables.

I've been having PM discussions with him for a while now about doing pressure drop testing similar to what I've been doing. In the end his restriction measurements appear to be higher than my own, but relative comparison between different blocks looks to be right on with what I got.:up:

He used a very accurate thermocouple (.01C resolution) on the base of the processor and the other in the water, so it's a very accurate measurement of delta between the two taking out any error in the processor based location. I'm not sure exactly how this would correlate to core temperatures, but it is at least indicating that the one spot measured continued to improve even down to the 3.5mm nozzle.

To me this would indicate that for single core processors the smallest nozzle may be your best selection, and as you spread out that heat signature that that will probably change.

I particularly liked the way he plotted the temperature deltas over flow rate, it's very interesting.

The other very very interesting thing about this testing is that it shows that flow rate becomes more and more critical as the restriction goes up. Look at how steep the curves are for the 4.5mm and 3.5mm nozzle as well as the supreme. This would explain the cases where folks are seeing optimal temperatures with some really strong pumps even though heat dump from those pumps is pretty high.

Glad to see this level of testing, very very nicely done:clap:

http://www.google.com/translate?u=http%3A%2F%2Fwww.nexthardware.com%2Ffo cus%2Fscheda%2F80_701.htm&langpair=it%7Cen&hl=it&ie=UTF8

[DavidNJ]

Nice find. I would imagine it really doesn't make that much difference if it is two die or one. The small nozzle gives the highest velocities. The difference in nozzle location versus heat distribution isn't that much different between the different number of dies.

However, their pump never got to very high flow rates. One issue with just putting in the nozzle, shown well in your spreadsheet, is the reduction in flow rate my adding a nozzle. In there test the 3.6mm nozzle never went past .8 gal/min.

For example, using your chart, a XSPC Res Top DDC 3.2, PA120.3, and different nozzles we see that 6.3 to 5.5 reduces area 24% but adjusted for a 10% reduction in flow increases velocity only 15%, 5.5 to 4.4 area is down 36%, flow is down 25% and velocity up 15%. From 4.4 to 3.6 area is down 33%, flow is down 24%, and velocity only up 12%.

In their graph, temps would have gone from 15.42 to 15.9 to 16.35 to 16.5. If the flow was constant the decline could have been 15.01 to 15.89 to 16.48 to 16.71. The first was a range observed in most testing of 1.1 degree; the second with constant flow was 1.7 degree.

Still, how critical is 1.7 degree to any realistic overclocking?

[Martinm210]

Nice find. I would imagine it really doesn't make that much difference if it is two die or one. The small nozzle gives the highest velocities. The difference in nozzle location versus heat distribution isn't that much different between the different number of dies.

However, their pump never got to very high flow rates. One issue with just putting in the nozzle, shown well in your spreadsheet, is the reduction in flow rate my adding a nozzle. In there test the 3.6mm nozzle never went past .8 gal/min.

For example, using your chart, a XSPC Res Top DDC 3.2, PA120.3, and different nozzles we see that 6.3 to 5.5 reduces area 24% but adjusted for a 10% reduction in flow increases velocity only 15%, 5.5 to 4.4 area is down 36%, flow is down 25% and velocity up 15%. From 4.4 to 3.6 area is down 33%, flow is down 24%, and velocity only up 12%.

In their graph, temps would have gone from 15.42 to 15.9 to 16.35 to 16.5. If the flow was constant the decline could have been 15.01 to 15.89 to 16.48 to 16.71. The first was a range observed in most testing of 1.1 degree; the second with constant flow was 1.7 degree.

Still, how critical is 1.7 degree to any realistic overclocking?

That's true, and no much difference, actually I didn't overclock a whole bunch more going to water, I'm just at much more comfortable temps.

Those are nice heat signature charts.:up: Where did you get them?, I need to study up on that before I do any die simulator stuff. That's a ways off, but it's been in the back of my mind for a while.

[saaya]

wow, very nice! :toast:

[vkbms]

Still, how critical is 1.7 degree to any realistic overclocking?

+1....

Plus consider that with less flow the radiator should perform a little worst...

PS: I'm the reviewer.... :rolleyes:

[DavidNJ]

Increasing the water flow only has a minor impact on radiator efficiency once a threshold is past. Fans have a much greater (10x) effect.

However, if the delta in the unreported water temp to core was 10C, the 1.7C represents 17%. That could be significant in a way. Although Liquid Metal Pro may have a greater impact.

Martin, how about making some Fuzion mods instead of new block? I'm thinking a stepped base plate, different nozzles for both the inlet and scavenge could add a bit.

P.S.
Are they going to let the Seattle girl go in Perugia?

[vkbms]

Increasing the water flow only has a minor impact on radiator efficiency once a threshold is past. Fans have a much greater (10x) effect.

However, if the delta in the unreported water temp to core was 10C, the 1.7C represents 17%. That could be significant in a way. Although Liquid Metal Pro may have a greater impact.

P.S.
Are they going to let the Seattle girl go in Perugia?


Oh yes....the fans are obiuvsly more important....but we are talking about the same condition of work ....only changing the nozzle....

PS: The girl seems to be guilty.....but you know....we're in Italy....these processes will take years ....

[DavidNJ]

The girl seems to be innocent...there is no motive or evidence. The press used a discount store camera of her buying underwear after the police locked up her house with all her clothes to prove her guilt.

There is only prosecutorial abuse here.

Did changing the nozzle change the water temps? Did the size of the radiator and its effectiveness change the effectiveness of the nozzle change? I would tend to think so. a lower core temp with a lower water temp would leave the radiator less effective. If the water was at 20C above ambient that would probably mean less than if it was at 3C above ambient, as some systems are.

[nikhsub1]

I'm thinking a stepped base plate...
http://www.xtremesystems.org/forums/showthread.php?t=160566

[RealRedRaider]

Great info, thanks.:up:

[Martinm210]

Martin, how about making some Fuzion mods instead of new block? I'm thinking a stepped base plate, different nozzles for both the inlet and scavenge could add a bit.

P.S.
Are they going to let the Seattle girl go in Perugia?

I'm shying away from any CPU block mods for a little while, Im more interesting in producing some data for radiators. There are plenty of great folks on the board already doing that:up:

I like focusing on areas there is little information on. Flow rate and pressure drop wasn't understood very well at all when I got started in this. Now I feel like radiators are calling to me...:D