I'm in the process of getting my Z axis working and had a couple of ideas to make things a bit easier.
The main thing is that everything needs to be pretty close to square/vertical/perpendicular with very little room for deviation and this is a bit of a pain.
I'm not particularly mechanically minded so I may not be able to visualize the consequences of the changes I'm proposing very well but I thought I'd throw a couple of things out there and see what sticks.
The simplest change would be to lengthen the Z axis idler and studding drive rods up through the top corner brackets and hold them in place with an arrangement of a nut and washer on each side as on the bottom brackets. This gets the internal spacing correct and you don't have to worry about adjustments to get the Z axis rods exactly vertical and parallel to the frame.
My corner brackets came from Bits and Bytes and the top ones don't have holes in the correct places although I think the stl files do.
This would require a bit of thinking and changes to instructions as you would have to do everything in a specific and slightly different order but I don't think it would be too hard and it may be that then you could do without the tie brackets and bed constraints.
At that point it may be worthwhile redesigning things further to do without half the vertical rods as there will be 2 rods running parallel to each other for most of their length on each corner where 1 would do (maybe just leave a small additional piece that protrudes through the bottom corner pieces as adjustable legs for each corner). Of course, it may be best to keep both from a strength and stability point of view
The other idea I had is impractical at the moment but may be worth looking into in the future when we can reprap our own motors. Rather than have one Z axis motor and a belt and pulley system connecting each corner, we could have a motor on each corner directly controlling it. This goes against the trend of simplifying things and the only advantage I can see for this is that you can then move each corner independently.
This is good for trying to get everything level in the calibration stage but during normal operation you'd just want to move the platform as a whole up or down. Can anyone think of any other advantages to this? If not, then I doubt it would be worth the added complexity but it was brought up in conversation yesterday and I could see it would have made my immediate problems easier and quicker to deal with.