This started out as a fitting review and morphed into more of an elbow/fitting restriction guide. As most of you know…changes in system flow translates to very little CPU temperature change. This leads to the assumption that fittings and elbows have no impact to performance at all. While that’s generally true from a practical +- a degree doesn’t matter perspective, you can split hairs if you want.
So with that, I’m splitting hairs. I wanted to see what if anything I could measure between various elbows, fittings. While it is relatively insignificant, we all use these little bits and pieces, so I thought it would be worthwhile measuring what I could.
Special thanks to the many sponsors in my past testing that contributed to the various bits of fittings I’ve collected. This includes Danger Den, Bitspower, The Feser Company, Swiftech, and Koolance. Without their generosity, this review/guide would not have been possible..
In many ways, these juicy little shiny bits are like the legos of our builds. They assist in fabrication of our many ideas and serve in that critical junction between flexible tubing and rigid blocks. We also now have many compression fittings options that clean up an otherwise less than perfect zip tie or clamp,etc.
I actually tested barbs as well, but found that the differences between 1/2″ high flow barbs were too small to even properly measure. To measure differences in barbs, you would have to string 10 fold barbs together. There is a very minor loss in compression fittings however, so I’ve tried to capture that here.
Standard pressure drop except I tested on a Danger Den MC-TDX and simply switched out the inlet barb. In the end, I subtracted out the restriction of the standard barb test, giving you just the “Added” restriction over standard barbs.
I also “Attempted” a barb retention type test, but in the end found they all passed with flying colors. I was only able to test up to about 50PSI, and using as little as a single zip tie was more than enough to exceed 50PSI. Beside that, I was getting tired of getting wet! Just make sure you use clamps or zip ties, or the compression ring, and you should have no problem exceeding the rating of the tubing or most pumps. Here is a snapshot of that test, since nothing would remotely fail, I even resorted to testing without clamps…even with that, the least amount tested was still nearly 20PSI, larger barbs would get up to 30psi without any clamping.
I’m giving you both the pressure drop measured and a translation to CPU temperature gain.
This took a while and requires many assumptions, but I feel it is important to convey flow rate in terms of CPU temperature impact. Thanks to great tests like Vapor’s CPU block testing, I was able to pick a typical CPU block and estimate the change in CPU temperature due to the flow rate impacts.
The pressure drop testing was an actual measurement, but the CPU temperature results were derived by calculation based on a variety of test results and should only be used as an approximation.
I’ve just come to the conclusion that the universal language in the overclocking community is degrees celcius..so I did my best to translate..
All of the barbs worked relatively the same, although I did notices slightly more retention by the longer barbs. Restriction differences were too minor to really measure with just one barb. My personal favorite is a tie between the DangerDen/Bitspower Fatboy and Hardware Labs kit barbs that came in my SR1 radiator package. D-tek barbs are another testing favorite of mine because they are a bit easier to remove 7/16″ ID tubing (My personal favorite). This just makes taking things apart while testing easier.
My favorite for function/performance would be the Koolance fittings. I like the flat spot in the threads for wrenching the barb on/off. I also prefer the flat style knurling as it’s less harsh and equally effective in giving finger grip. Finally, I just like the larger/deeper barbs on the Koolance compression best. BP and TFC probably get my pick for visual looks, but they measured a touch more restrictive and I’ve had trouble using them without resorting to some sort of pliers to remove them. Regardless, I rarely use compression fittings of any kind, and I’m a bit old school and like my barbs and clamps or zip ties.
I didn’t test too many, but the 45s as you might expect were roughly half the restriction of a 90 (Who would have guessed?..). Also using the combined compression/elbow fitting was very slightly less restriction than separate 45+compression pieces. My suggestion is to use elbows when you need them, but don’t let them become the “Spare tire in the woods” where they multiply like crazy. Tubing CAN bend you know.. Temperature has very minimial effect, but it does add restriction. One elbow on the low restriction DD MC-TDX adds almost 40% more resistance, so it will have impacts to flow rate…but it’s not going to add up to much in terms of temperature.
In the end, fittings are very much a personal preference thing. Elbows and fittings do cause some restriction, but it’s very much fractions of a degree and impacts flow rate much more than it does temperature. All of these fittings and options are winners in my book, just another color/shape lego block for you to build with. But…if performance is absolutely #1 priority, and you want to take on the hair-splitting performance alway first perspective… less is more and generally standard barbs with tubing bends and longer tubing runs will generally net you the very highest flow rates. Many people assume that tubing itself is restrictive and will assume that adding an elbow to shorten the tubing length will be a net benefit. Unfortunately, the it rarely is…one 90 elbow is about equal to 4′ of 1/2″ ID tubing so the few inches you saved is lost and then some. Do not add an elbow to save a few inches of tubing, it’s counterproductive in performance.
However, if you are building a piece of art which many people do….by all means add the fittings as needed for the cleanest possible look. Sure some extra fittings will add restriction, but it’s very minor. As the chart shows, it would take almost 30 each 90 degree elbows to add up to 1 degree in CPU temp rise. Half a dozen to get to aesthetic perfection is perfectly fine from an art priority perspective.