Have you split the orange from the valve, and the "water-on" feed from the cylinder stat, (at wherever they join before feeding the boiler CFH), to confirm 100% it is coming from the Valve orange, and not the Cylinder stat water-on, (whatever colour it is?)..
Remember..
"Water On / Heat Off" is the default position where the 3-port valve doesn't need to move.
(held in the home position via the return spring, no trigger voltages required, So the boiler CFH bypasses the valve).
"Water On & Heating On" has to stall the 3-port valve in the mid-position.
(only one voltage trigger from the heating thermostat, out to CFH via the valve orange).
"Heating On / Water Off" has to drive to motor to the furthest position.
(two voltage triggers from heating ON and water OFF, out to CFH via the valve orange).
How long is the cable run run from cylinder to valve/control centre box? Could this be a source of phantom voltages?
Whenever I encounter "strange" happenings with heating operation, I normally start splitting off bits of the control wiring, to ensure I'm not chasing my own tail up a wrong path.
I also sometimes use a bog-standard pendant bayonet lamp holder plus flex, with a GLS lamp wago'd across key points to verify the expected voltages are where they should be... As it can save a load of faff sticking meter probes in various screw terminal connectors trying to work out what's happening.
I presume this is the same physical boiler for the whole 8 years?