I had wondered whether the extra delay in voltage getting to the VCC pin due to C47 being 470uf in some kits (as opposed to 220uf per the schematic) was causing initialisation issues for the controller in those kits with the 470uf cap. And when the second switch on happens there is still some charge left in C47 so VCC gets to voltage quicker and the initialisation of the chip occurs properly.
I had the switch on issue and I had the 470uf cap. But I changed it to 220uf and still have the same issue. So there went that idea.
I've reflowed the solder on L5 and L6- no change. Will try 47uH for L5 and or L6 when I get some.