I was supposed to test the CPU-360 rev1.2 on my old testbed back in 2010 when it was released. But I had issues with the eVGA motherboard I used and the testing never got underway before the board literally went up in smoke. One thing Koolance did provide me with for that test, however, were a range of injection plates with different slit widths. The stock injection plate has a 2.0mm wide slit (and 31mm long, but that’s not as important), Koolance also sent me plates with 0.7mm, 1.0mm, 1.3mm, 1.6mm, and 1.9mm wide slits to test.
Since then, I’ve lost the 1.0mm wide plate and have decided the 1.9mm wide plate is too close to stock to be worth testing, so below is a comparison of the 0.7mm, 1.3mm, 1.6mm, and 2.0mm injection plates. Each plate was tested in each orientation as well (I guessed, correctly, that some injection plates might prefer the ‘wrong’ orientation due to perfect the positioning of the injector over the cores when oriented incorrectly). In these tests, the larger the data point, the more error there is due to an incomplete set of testing (not every plate and orientation combo was worth the time to test a full 7-mount set). Additionally, if a data point has a thick border, that also means it’s an intentionally incomplete data set.
First up, the raw output:
Looks like the 0.7mm plate is a bit of a lagger in performance, it’s both restrictive and not great thermally (especially when the block is mounted the ‘correct’ way–vertically). 1.6mm and 1.3mm plates need further examination.
So the first step to looking into the plates further is to isolate their size vs. individual performance characteristics. First up, slit width vs. temperature:
Looks like 1.3mm and 1.6mm really do separate themselves, each being about a degree ahead of the stock plate. The trendlines that I added show that a plate with roughly a 1.4mm slit might be the sweet spot thermally, but that both 1.3mm and 1.6mm are pretty good and close in general. So how do they compare in terms of flowrate?
So the flow rate complicates things further. The 1.6mm slit injection plate is barely more restrictive than the stock plate but the 1.3mm slit injection plate is noticeably more restrictive. In typical fashion, it’s a flow vs. temperature tradeoff–but the differences are really small. My gut says the 1.6mm plate is the winner (against stock, it’s a big gain thermally with a tiny drop in flow rate), but thermally the 1.3mm plate is definitely the winner.
How should this be sorted out? A composite score. For this chart, I subtracted the flow rate from the best temperature each plate produced (regardless of orientation). Lower number is better. In essence, 0.1GPM is worth 0.1C, a pretty even trade-off, in my opinion.
So what does the composite score say? 1.3mm and 1.6mm slit injection plates are really close, with the 1.6mm slit plate winning slightly. The trendline shows that the sweet spot is between 1.45mm and 1.5mm, according to this metric. I forwarded this information to Koolance well before this performance preview went up, it will be interesting to see what conclusions they draw from it and if they end up bundling multiple plates with the CPU-370 or if they make a single plate something existing and future owners can get easily and cheaply get. I know this much though: if I were building a system with a CPU-370, I’d want to replace the injection plate with one that has a 1.3-1.6mm wide slit.