I had success with this design:
I used the Lulzbot fan (Shop LulzBot 3D Printers, Parts, & Plastic).
Unfortunately I didn’t think to measure how long the wires I ran were before I closed up the patient, so I can’t tell you the length you’ll need. Just grab a couple spools of 26AWG wire. A pre-joined red/black pair would be ideal. Personally I spliced together extra wires cut from an install of a replacement Y-axis stepper motor.
I ran the wires through the X-axis harness without much trouble (it was nicely cut all along its length), cutting and replacing zip ties as I went. Got to the RAMBo enclosure and opened it up (being careful of the enclosure fan).
Unscrewed the three bolts on the bottom of the RAMBo box, to loosen the clamp for the wire harnesses, which made it relatively easy to get a pair of foreceps through the X-axis harness entrance, grab the wires, and pull them in.
Cut and stripped the wires. On the board, I used the plugs already referenced by Orias here:
This diagram (File:Rambo-conn-all.jpg - RepRap) obsucres that red goes on the right-side pin. You can confirm with this one (File:Rambo-conn-aux.jpg - RepRap) or by looking at the board itself.
On the other end, I cut the existing plug off of the fan power wires, spliced it to the wires I just ran, soldering and patching the insulation with heat-shrink. Tried to be careful there. It’s going to be moving around, and I want it to last.
Once everything was connected, I was able to test in Printrun with:
M106 S255
to turn it on, and
M106 S0
to turn it off.
I attached the fan mount using two 16mm long m3 bolts. Had to drill out the holes to get them properly sized. Chicken and egg - I had to print the fan mount without a fan mount, it came out pretty messy.
The fan mount’s author mentions using heat set nuts to attach the fan to the mount; I found 6-3/4 wood screws worked fine too. The fan’s labeled side faced into the duct. Or another way to be sure you dont get it backwards; you want to be driving the screws/bolts into the side where there’s a circular indentation around the bolt hole. Or, just remember which way the air blew when you had it turned on. The duct’s shape is such that, when running, much of the air will rebound out against the fan; this might make you wonder if you got it backwards at first, as it did me.
The result looks like this:
You can see here that this fan/mount combo does not intersect either the bed clamps or the Z-axis threaded bolt at all. My compliments to the designer.
I discovered all my existing slic3r configs already had the fan enabled, so there was nothing extra to do there.
Results pretty much speak for themselves. For this STL (taken from here: Calibration · slic3r/Slic3r Wiki · GitHub):
_40x10.STL (85.1 KB)
Before:
After:
That’s an immediate, dramatic improvement in PLA output quality right out of the gate. I haven’t even started tuning settings on it yet.