Vapor: can thicker brass plate of yours impact bending of waterblock base unlike thinner ones from Bei or stock one?
Printable View
Vapor: can thicker brass plate of yours impact bending of waterblock base unlike thinner ones from Bei or stock one?
.002in difference in thickness (between stock and brass) wouldn't do that, IMO. .005in difference between Bei's and the brass plates, maybe. But definitely not 4-5C, that's just a massive difference that I can't explain. Contact has been great with every midplate, as well, so the bow is doing its job regardless of the slight thickness differences.
Final HS2 numbers are in, graph is updated a few posts up.
Stock: 76.27C, 1.61GPM
HS1: 75.25C, 1.51GPM
HS2: 75.11C, 1.32GPM
Going to make HS3 right now (and kill HS2)....1mm wide channel, 30-31mm tall.
Vapor: maybe try experimenting with different width of that channel near ends or near middle?
Goofed on turning HS2 into HS3....oh well. Testbed needs to get to testing other stuff now anyway, so not the end of the world :)
Final results:
http://skinneelabs.com/assets/images...360/Plates.png
All the Bei Fei plate results are obviously not conclusive...never figured out the source of the 4-5C fall off :confused:
And especially not with most of them being a loss of GPM, one up to 0.75GPM :eek:!Quote:
Originally Posted by Vapor
That was just the Dots plate, which was expectedly restrictive. I'm talking about the Jagged/Double (not shown since I didn't let them complete a full run) that were 4-5C behind Stock but should have performed comparably. Those two plates basically invalidated all the testing on the other Bei Fei plates.
Wait, so you tested 2 plates that were basically the same as the stock plate, yet they showed 4-5C worse temps?
Essentially, yeah. Double and Jagged have roughly the same impingement as the stock plate (Double has a little more, Jagged has a little less), use the full array of microchannels, but perform way out of bounds. Even a free-flow plate should perform within a degree or two of stock, so something is definitely awry that I can't attribute.
Some sort of flow bypass is the most fitting, but I don't see where that would happen :confused:
I'm thinking about bypass at fault aswell, afterall, most of heat exchange of modern waterblocks with micro-pins/micro-channels comes from turbulence, so if by less thick plates some 5% of flow bypasses those channels (by going least resistance way), here one gets worse temps, as water not going through those channels takes less heat from block. Can you try some mod to make those plates a tiny bit thicker? As in some silicone mode variation or something .. whatever that's easiest to do with current plates at hand? Have you tried to measure insides of block where the plate is placed?
Based on flowrate vs. temp testing I've done with the CPU-360, bypass figures would need to be closer to 80%, not 5%, to achieve a 4-5C drop. Flowrate figures weren't indicative of this (flowrate was lower than stock....if bypass were that high, flowrate would be much higher if microchannel restriction stayed constant).
I'll repeat, testing on this topic is done.
In talks with Koolance about bringing a plate with a narrowed jet to market :)
Bump...
TIM have you seen these yet???
Yes. Sending some new stuff to Vapor to try out... :)Quote:
Originally Posted by NaeKuh
You hear that Vapor? More testing and writing to do :rofl:.Quote:
Originally Posted by Koolance
Poor chap :rofl:
The best kind of testing to do is the testing you want to do....and I definitely want to do this :)
Given your experience, that's a great compliment. Thank you.Quote:
Originally Posted by Vapor
Tim