having looked at the design of the delta system that I had thought of, I realised that it will need two controllers; one for the drive and one for the printhead. Both of these controllers will have to be controlled.
I therefore revisited the gantry based design and suddenly was aware that if I used a single axis bed machine, I could remove the bed and instead use the stepper motors as a drive (the hotend would need a Z axis). All that would be needed is to change the size of the bed in the GCode software (Repetier, Kisslicer etc). This means that although the model would be limited in the Y and Z axes by the machine size, the X axis would be limited only by the quantity of filament carried and the supply of power. So, long, thin models!
My next stumbling block is that of layers. If I have to produce a model one layer at a time, this would mean the machine moving a long way for each layer. This in itself is not a problem apart from the battery life. The main problem is that of the temperature gradient experienced by the part causing warping. This usually occurs in the Z axis as most entry level machines have a relatively small bed size (relative to my proposal) and have compensated this with heated beds and / or high temperature deposition for the first couple of layers. These layers will then be able to accommodate any subsequent expansion / contraction through the model's Z axis. However, my concern is now that the temperature gradient will be found in the X axis rather that the Z, producing long, thin warps along the length of the model.
Therefore I am now looking at whether it is important to redesign models to allow for the material's thermal coefficient by introducing thermal breaks and also whether these can be overcome before a redesign is necessary.