The pertinax kind of material used for the frame of my i3 "Pro" B appears to be somewhat brittle, that the wish arose to fortify it. Because I had two 8mm spindles left - the result of replacing the Z spindles of this printer against trapezoid spindles - I used those, with two custom Z axis top brackets, modified with two blocks each to carry those rods.
Purpose of added rods is enabling the frame to carry heavy stuff, like filament reels, box feeders, tool trays, which I wouldn't trust my pertinax frame to carry without eventual breakage.
You can use these blocks with or without 608zz (8x22x7) ball bearings. If used without, the ball bearing insert offers a wide enough diameter to allow unrestricted wobble of the Z spindles, or indicate the necessity to align, straighten or replace the rods.
The photo still shows the previous version of these brackets. Later versions join the blocks for horizontal rods, for extra near rod support.
Goal of these brackets is to strengthen the top horizontal bar, not necessarily to introduce bearings. But before the endless pro/contra bearings discussion seems to get retriggered by the fact that I use these with bearings, I've added two photos: the G clamp (http://www.thingiverse.com/thing:1673030) has been printed without bearings, and shows the wavy pattern which I mentioned and added my thoughts about their origin in my reaction to a comment. The cylinder was printed with bearings. It (and other objects printed as far) don't display this wavy pattern, more clearly visible, by light reflection, in the photo which shows the G clamp alone. I'm sorry, I don't have identical items printed with and without bearings, for better comparison. My current opinion in the top bearing discussion: Don't restrict with wobbly spindles. Restrict with perfectly straight and aligned spindles. I'm likely and willing to change my opinion if new evidence supports it. For practical considerations, I opt for an agnostic position: any design which may be affected by it, will be able to operate either way.
.scad presets are suited for 5.5mm thick frame material. Customiser parametrisation still needs some work. If you're not ready to print and replace the top brackets, here's a stand-alone version of the blocks: http://www.thingiverse.com/thing:1789180
An early version of a top filament holder, which rests on the horizontal rods, has been published as http://www.thingiverse.com/thing:1801124
Prusa i3 Pro B cheap China clone
0.4mm nozzle, 0.3mm/layer
2016oct01: added strengthening rails along X axis. Meant to become rails for guiding the display, if mounted not centered, but moved to the side, the bridge between blocks obstructs a display at that position. Instead of weakening the bracket, by inserting a cut into the bridge, I accepted that display on that position and height is not recommandable. Instead, it could be moved up a bit, resting on the bridge.
Adjusted screw holes which weren't frame centered due to having introduced the variable "slack" for widening cutouts, but not considering that variable for hole positions.
Widened diameter for bearing recess