Thanks for the input everybody, seems most of you favour holding to the original plan. I am keeping my options open for the moment, so all the work using the new test frame will be very carefully undertaken on the assumption that it could become the basis of the final build. (More importantly I will be attempting to avoid doing anything that would prevent me from using it that way, in the event I/you subsequently decide I should go with Plan B/C). So if I’m doing stuff that seems a bit obsessive and unnecessary for a proof-of-concept prototype, that’s probably why.
In reply to sinnedone, and for the purposes of informing you all, the immediate plan of action is as follows:
The next step will be to finish off all the custom (mostly metal) mountings that the test frame is still missing - primarily the main brackets that will support the motherboard tray. The metal needs to be soft enough for me to bend easily and accurately at home, so I’ve cut some copper sections from some spare 15mm pipe that we had lying around and flattened it out into strips to see if they might be suitable. However it looks likley they might not be rigid enough, in which case I may have to buy some bits of aluminium or thin steel instead. Each strip will need four right-angle folds, plus a set of drill holes for mounting them to the case and for mounting the motherboard tray to them. However because tapping the tiny threads of normal computer-case size screws will be awkward to do, the brackets and motherboard tray may well end up being fixed together using some familiar looking miniature bolts instead…
The other mountings that need to be sorted out are for the lower radiators. Two options here. The stands I got for the radiator testing rig are a little bit too tall, but by a stroke of luck there is enough scope (just) to carefully cut them down to size with a hacksaw. These would be screwed into the baseplate, I’d commission the backlighting spacers in black acrylic as planned, and get a set of longer screws required to mount the fans. Alternatively however I could get the spacers made in the same laminate material used in the test frame (like the front fan mount), extended in length beyond the ends of the radiators allowing it to screw in to the uprights. These could then be cut down to the correct size for the 900D if and when it was subsequently required. I don’t know which of these options would be faster, but the latter approach would definitely make assembly and disassembly of the test rig much easier.
Once these are done, then at long last it’s time for assembly of the cooling loop in the test frame. Phase 3.1 was going to be simple point-to-point connections in soft tubing to test the concept. However I have decided instead to go straight ahead and build the loop in full, with the fittings and rigid tubing - with the exception of those acrylic sections which require bends (like the pump connections) which will initially be in soft tubing. An extra piece of timber work will be made up to represent the motherboard and GPUs, simulating the mounting positions for the EK waterblocks and the Raystorm, thus allowing me to test the cooling loop without risk to the expensive electronics. This stage will also include setting up any components which aren’t part of the core system, like the Aquaero and Neopixel strips. Phase 3.2 will see the soft tubes replaced with rigid acrylic tubes as appropriate, and will also be the absolute latest that I can forestall a decision on implementing the motherboard heat sink design. Finally Phase 3.3 will involve the disassembly of my current Phase 2 rig and the transfer of the Ironbeast system hardware into the Phase 3 (requiring only draining and minimal disassembly of the cooling loop via the D-plugs) bringing the prototype to a fully operational state, hopefully with minimal downtime. It will then undergo leak and thermal performance tests before becoming my main rig until the 900D modifications are ready, or whilst I make up some case panels for it.