Here I ended up with a slight variation on the published solution which incidentally showed a difference in current on both sides of the circuit. It meant that in the AND state there was still a residual current on the ammeter. I played with belt tensions etc. and found in the end that it was simply due to the different gear ratios on the junctions. The LHS was on lowest gear (just like a bicycle) and the RHS on the highest gear. Placing a resistor on the RHS compensated for this, slowing the RHS down and allowing the 2 sides to equilibrate quickly in the AND state and stop any residual current.
