This is my version of the AndreasL alternative dual z axis for smartrapcore (http://www.thingiverse.com/thing:896556) but remixed for the aluminium 2020 frame version.
I have reorientated the axis so that it sits front to back not left to right to prevent it interfering with the corexy gantry at the sides. So all parts labelled R are Front and parts labelled L are Back. Please ensure your software endstop settings and define_manual_x_home and define_manual_y_home settings do not allow the head to move along the x direction and hit the top front bracket for this z axis.
The parts I include here are the parts which needed to be modified for the alu version and should be fixed to the frame, plus a modified slider to make belt tensioning easier. Many of the non-printed parts you need are available from your old SCALU z axis.
Original SCALU parts:
2x 8mm smooth rods around 320mm
1x m8 x 20mm bolt
1x 608zz bearing
1x nema 17 motor
4 x m4 t-nuts + 4x m4 bolts
4x 608zz bearing
3 x m8 x20mm bolts
1 x m8 x 20mm axle/grub screw/shaft (see below)
4 x m4 t-nuts + 4x m4 bolts
4x m3 x 20mm + 4x m3 nylon lock nut
1.85m gt2 belt (I can highly recommend using steel reinforced)
Optional m3 x 20 bolt + 3mm bearing (or a larger bearing and a printed bushing)
The m8 axle (i.e. an m8 bolt with the head cut off, 20mm long. It goes in front of the motor to hold one of the bearings - it will help if you cut a slot in one end with a Dremel to screw it in with a flat bladed screwdriver).
If possible use a 0.9deg stepper instead for improved resolution, I am using a 42mm motor.
You might need another bearing if you put the additional pulley next to the motor to improve belt grip around the drive pulley but I am not sure its always necessary.
I have remixed one of the sliders with two belt clips. This means no need to join the belt with another part. (I am calling it the front one as its easier to tension the belt and attach it at the front than the back).
The remaining stl's which you might need you should obtain from the original design here: http://www.thingiverse.com/thing:896556
I have used the original plywood piece fixed to the sliders. A laser cut aluminium piece as shown in the original would be great but I wonder if it will be any lighter or stiffer in reality. I assembled everything making sure the smooth rods are aligned with the frame, The smooth rod should be 200mm from the frame on each side measured to the middle of the rod. I then strung the belts, placed the plywood on top of the sliders and marked where to drill the plywood, by pushing a little paint through the fixing holes with a fine point (these should be drilled out to 3mm). Then I attached it with m3 bolts and lock nuts.
Ensure that the top brackets which hold the top bearings, are extremely level when installed, otherwise the belts "walk" across the pulleys and do not remain centralised. They do not come off the pulleys as there is insufficient space for them to do so, but belt wear would be considerably more if friction with the brackets is continually occurring.
The springs at the bottom of the smooth rods I added to my original z axis and I used them instead of landing gear. I find this dual z axis descends on its own very gently when the motor is powered off and therefore I feel they are not needed. Without the springs you should be able to get around 170mm of z height - a big improvement!
Please be aware the original by AndreasL might change and therefore some geometry might be different. Please message me if you find an incompatibility and I will see about making changes here too. I might change things once I've had chance to extensively test it.
UPDATED 04/02/17- To improve the belt path for the front descending belt I have changed the z-slides. They now have one belt clip in a slightly different position. This means a straight belt path for the front descending belt, which should make the z axis more accurate when the nozzle is near to the bed. I cut away a small amount of my z-carriage to allow this belt access, but yours might not be in the way. I also wanted to stop the belts rubbing on the top brackets around the 608zz bearings, so I filed away the small flanges that centre the bearing in the top bracket and printed some m8 large washers to guide the belt.
UPDATED: Top brackets now attached which have been lightly reinforced, use them in exactly the same way as before. Much gratitude goes to AndreasL for the original design, this has performed admirably since I modified it and installed it.
UPDATED May 2016 - Have swapped to 16tooth drive pulley which has a default 100 steps/mm for 1.8 deg stepper or 200 steps/mm for 0.9 deg stepper. Slightly improved resolution but also a little more gearing to help the motor lift the axis.
Also when attaching the belts I found it very helpful to find two objects the same size, for example around 100mm long. Place them under the bed plywood, and rest the bed on them, then attach innermost front belt, then the back belt then the front one. This avoids the plywood tilting under unequal belt tension, and makes adjusting the build plate level easier once assembled.
UPDATE May 2016 - So someone suggested a counterweight, and since I've upgraded my machine to 300x200 the z carriage now weights in at 700g. So I designed a bearing holder to be mounted over the front lower bracket. Add an m8 bolt, two large washers and a 608zz bearing, then a short run of belt cable tied to the front-most belt and laid over the washer. You can then add weight to it just less than your z carriage/bed total assembly. I am using a 500g weight for a 700g carriage meaning the carriage is now effectively 200g. This now descends nicely without a harsh bump at the bottom and means I can lower the current on my z stepper motor.
October 2016 - Decided to do away with the Z probe and replace it with a Z endstop. Came up with a add-on to the lower front Z axis bracket. This is designed for a mechanical endstop module board. Superglue or acetone weld the endstop bracket to the lower z axis front bracket attach the endstop. Bend the endstop lever upwards slightly.
Have something under the bed like landing gear or springs or similarr so that the bed stop just short under elastic tension of the endstop, that way when homing the firmware finds the endstop trigger point, and does not press it fully home as this will be less reproducible.
Another method to improve z endstop reproducibility when used in this way is to have home z (g28 z) before home all in your start gcode. That way if the bed is resting fully on the endstop ie past its trigger point the axis will lift and then rehome at the trigger point before homing x and y.
More details of how I configured the firmware etc.. can be found here
Public files on Tinkercad if you want to copy then edit them: