Why not use unsupported rails?

[Jonathan]

Very pretty but also a very non accurate picture has been painted here...

You seem to be under the impression that I'm trying to say that unsupported rails are weak. I'm not, I'm saying they are many times weaker than their supported equivalent. If 30mm unsupported rails give you enough accuracy, then I'm saying you could achieve the same result on a much smaller diameter with support.

If I'm honest, I've seen so much abysmal design in commercial machines that I simply ignore what 'expensive' machines do. Their objective is to make money not make good machines, and those different objectives result in different designs.

Yes, there is deflection in the bed, I'm not implying this accounts for all the deflection. My only point here is that the same diameter of rail is tens of times weaker when unsupported, so all other factors being equal, if given a choice between unsupported and supported, you should go for supported.

An FEA simulation is never 100% accurate, but it's certainly a big improvement on estimation and common sense. I've simulated bed designs too, here's park of one I'm working on: