(Puts down crayons and cries...)
TL;DR - some advice on working bubbles through a reservoir-less (radiator-as-reservoir) SFF loop would be appreciated.
OK, so I'm in the process of switching my build to a custom water loop. This is my first venture into custom loops and I'm doing it in a case smaller than my toaster. Yay, for the wayward fool! I'm still in the "intransigent bubble" phase. There are probably more topical "water cooling specific" places to vent and ask for advice, but the water cooling loops for SFF are pretty situation specific, so I thought I'd start here and maybe my trevails will help others working on their Ghost builds. I did do quite a bit of watching videos and researching water cooling parts so I came into this well informed, but I think I see now why they never show more than 2 seconds of "watch water slowly drip into my tubes and blocks" in the build videos before they jump to thermal graphs.
I'm using a NexXxus ST30 radiator on top, an EK Vector GPU block, and an Apogee base with a DDC-310 pump on it. I'm using Mayhem's Ultraclear tubing because it's my first loop ever and I wanted to be able to visualize how the water is filling in the spaces. I'm glad I did that because I had a lot of bubbles in the tubing while filling it that collected in places I wouldn't have thought and were hard to move along by hand during the filling phase (multiple bubbles all want to go opposite directions) and the clear tubing helped a lot with that. Once primed, the DDC-310 moves the water very quickly through the system judging from the bubbles racing through the loop.
My loop is set up radiator -> GPU block -> Apogee -> radiator with the GPU block connected in the "normal, but not critical" direction (i.e. flows in over the chip, out through the fins, and then around the block). Everything fits with the case sides sliding easily over all parts and tubing with about 1mm to spare and nothing has leaked so far (knocks on wood). I'll detail my build later with pics if I don't throw it down the hill outside my house, but a hint is that the Koolance low-profile elbow fittings rule for this application.
What helped a lot getting out most of the air was to hook up a short (1 foot) pseudo-reservoir tube with a barb fitting on one of the radiator top fill holes and fill it with 4 inches or so of fluid. I chose the port over the return tube because I thought the returning air would immediately look for the high point and have momentum into the reservoir tube. It turns out the air bubbles slightly preferred collecting near the outlet tube and would loiter in that tube just outside the outlet. So choosing the other port over the outlet tube might have been slightly easier. It wasn't that bad because if I turned off the pump and rotated the case just a little, that bubble would travel through the radiator back to the inlet and into the reservoir tube, so the workaround was pretty easy. In terms of catching air as it shot back into the radiator - that was a bust as, if anything, the water level in the tube rose when the pump was running (I had a mark on the tube to gauge whether it was gaining or losing fluid).
At this point nearly all air is vacated, but for some tiny bubbles on the walls of the tubes that I haven't gotten to and one "big" bubble that amounts to about a 1-1.5 inch of 16/10 tubing when I manage to collect it in a tube. The reservoir was no longer helpful for that last remaining bubble because I can't get it to collect in the radiator where I can easily displace it with fluid. Instead it just loves to collect in the chamber just beyond the fins in the GPU block. The loop direction may be fighting me here, but the easiest route to the radiator would have been back through the intake and up the line to the radiator. Unfortunately, the air doesn't like to flow back through the fins (probably capillary effect in the thin fins) so I can't get it to travel that way. Instead, to move it forward to the GPU block outlet, I have to do about 20 seconds of weightlifting gymnastics to rotate the case all the way around through a loop-de-loop. At that point I can isolate it in the GPU outlet tube, but unfortunately from there it has to travel to the pump. The pump has enough power to suck that bubble through the tube and then it basically atomizes the bubble and I see tiny bits of air rushing through the return tubes to the radiator and then they go all the way through and back to the GPU block where they collect again just past the fins.
This last bubble doesn't collect anywhere near the radiator unfortunately so I don't think a tube reservoir in either radiator port will help (or even if I put it in the one port at the far end of the "out and back" radiator where I currently have a temp sensor.
Am I screwed for having reversed my loop (GPU before CPU)? Is there a better way to fashion a "tube reservoir" that is more effective at collecting air? Any advice on "Olympic level air bubble gymnastics"?